lectures/L08_pytest_pip_cprofile.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Code development: pytest and pip\n",
    "\n",
    "*Davide Gerosa (Milano-Bicocca)*\n",
    "\n",
    "**Sources**: \n",
    " - Michael Zingale at Stony Brook University: https://sbu-python-class.github.io\n",
    " - Googling around (a lot); my own research.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Plan for today. I'd like to showcase three more things you might need in your research in computational physics.\n",
    "\n",
    "- **Part 1**. Package and share your code\n",
    "- **Part 2**. Testing done right\n",
    "- **Part 3**. Find the bottleneck"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Part 1. Package and share your code"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os, sys"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "So far, we've been writing our code all in Jupyter.  But when it comes\n",
    "time to write code that we want to reuse, we want to put it into a\n",
    "standalone `*.py` file.\n",
    "\n",
    "Then we can load it on in python (or Jupyter) and use the capabilities\n",
    "it provides or make it a standalone program that can be run from the\n",
    "command line.\n",
    "\n",
    "Jupyter is great for interactive explorations and sharing your workflow with others\n",
    "in a self-contained way.  But if there is an operation that you do over and over,\n",
    "you should put it into a separate module that you import.  That way you only need to\n",
    "maintain and debug a single instance of the function, and all your workflows can reuse it.\n",
    "\n",
    "\n",
    "## Editors\n",
    "\n",
    "Use your preferred editors. Some popular ones include:\n",
    "\n",
    "* VS Code: https://code.visualstudio.com/\n",
    "* spyder: https://www.spyder-ide.org/\n",
    "* emacs / vi\n",
    "\n",
    "Rest in peace [Atom](https://github.blog/2022-06-08-sunsetting-atom/), you were the best.\n",
    "\n",
    "\n",
    "## Standalone module\n",
    "\n",
    "Here's a very simple module (lets call it `hello.py`):\n",
    "\n",
    "```python\n",
    "def hello():\n",
    "    print(\"hello\")\n",
    "\n",
    "if __name__ == \"__main__\":\n",
    "    hello()\n",
    "```\n",
    "\n",
    "There are two ways we can use this.\n",
    "\n",
    "* Inside of python (or jupyter), we can do:\n",
    "\n",
    "  ```python\n",
    "  import hello\n",
    "  hello.hello()\n",
    "  ```\n",
    "\n",
    "* From the command line, we can do:\n",
    "\n",
    "  ```python\n",
    "  python hello.py\n",
    "  ```\n",
    "\n",
    "Additionally, on a Unix system, we can add:\n",
    "\n",
    "```python\n",
    "#!/usr/bin/env python3\n",
    "```\n",
    "\n",
    "to the top and then mark the file as executable, via:\n",
    "\n",
    "```bash\n",
    "chmod a+x hello.py\n",
    "```\n",
    "\n",
    "allowing us to execute it simply as:\n",
    "\n",
    "```bash\n",
    "./hello.py\n",
    "```\n",
    "\n",
    "Here we see how the `__name__` variable is treated by python:\n",
    "\n",
    "* If we import our module into python, then `__name__` is set to the module name\n",
    "* If we run the module from the command line, then `__name__` is set to `__main__`\n",
    "\n",
    "\n",
    "## Changing module contents\n",
    "\n",
    "If we make changes to our module file, then we need to re-import it.  This can be done as:\n",
    "\n",
    "```python\n",
    "import importlib\n",
    "example = importlib.reload(example)\n",
    "```\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Command line arguments\n",
    "\n",
    "For standalone programs, we often want to have our program take\n",
    "command line arguments that affect the runtime behavior of our\n",
    "program.  There are a variety of mechanisms to do this in python, but\n",
    "the best option is the [argparse\n",
    "module](https://docs.python.org/3/library/argparse.html).\n",
    "\n",
    "Here's an example of using `argparse` to take a variety of options:\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# %load argparse_example.py"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "\n",
    "```python\n",
    "#!/usr/bin/env python3\n",
    "\n",
    "# to get usage: use -h\n",
    "import argparse\n",
    "\n",
    "\n",
    "def setup_args():\n",
    "\n",
    "    # simple example of argparse\n",
    "\n",
    "    parser = argparse.ArgumentParser()\n",
    "    parser.add_argument(\"-a\", help=\"the -a option\", action=\"store_true\")\n",
    "    parser.add_argument(\"-b\", help=\"-b takes a number\", type=int, default=0)\n",
    "    parser.add_argument(\"-c\", help=\"-c takes a string\", type=str, default=None)\n",
    "    parser.add_argument(\"--darg\", help=\"the --darg option\", action=\"store_true\")\n",
    "    parser.add_argument(\"--earg\", help=\"--earg takes a string\", type=str, metavar=\"test\",\n",
    "                        default=\"example string\")\n",
    "\n",
    "    # extra arguments (positional)\n",
    "    parser.add_argument(\"extras\", metavar=\"extra\", type=str, nargs=\"*\",\n",
    "                        help=\"optional positional arguments\")\n",
    "\n",
    "    return parser.parse_args()\n",
    "\n",
    "\n",
    "if __name__ == \"__main__\":\n",
    "\n",
    "    args = setup_args()\n",
    "\n",
    "\n",
    "    if args.a:\n",
    "        print(\"-a set\")\n",
    "    print(f\"-b = {args.b}\")\n",
    "    print(f\"-c = {args.c}\")\n",
    "    if args.darg:\n",
    "        print(\"--dargs set\")\n",
    "    print(f\"--earg value = {args.earg}\")\n",
    "\n",
    "    print(\" \")\n",
    "    print(\"extra positional arguments: \")\n",
    "    if len(args.extras) > 0:\n",
    "        for e in args.extras:\n",
    "            print(e)\n",
    "```"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "A nice feature of `argparse` is that it automatically generates help for us.  If\n",
    "we place the above code in `argparse_example.py` then we can do:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "usage: argparse_example.py [-h] [-a] [-b B] [-c C] [--darg] [--earg test]\n",
      "                           [extra ...]\n",
      "\n",
      "positional arguments:\n",
      "  extra        optional positional arguments\n",
      "\n",
      "options:\n",
      "  -h, --help   show this help message and exit\n",
      "  -a           the -a option\n",
      "  -b B         -b takes a number\n",
      "  -c C         -c takes a string\n",
      "  --darg       the --darg option\n",
      "  --earg test  --earg takes a string\n"
     ]
    }
   ],
   "source": [
    "os.system('python argparse_example.py --help');"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Paths\n",
    "\n",
    "How does python find modules?  It has a [search order](https://docs.python.org/3/tutorial/modules.html#the-module-search-path):\n",
    "\n",
    "* current directory\n",
    "\n",
    "* `PYTHONPATH` environment variable (this follows the same format as\n",
    "  the shell `PATH` environment variable)\n",
    "\n",
    "* System-wide python installation default path (usually has a\n",
    "  `site-packages` directory)\n",
    "\n",
    "We can look at the path via ``sys.path``.  On my machine I get:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['/Users/dgerosa/Documents/reps/scientificcomputing_bicocca_2023/lectures',\n",
       " '/opt/homebrew/Cellar/python@3.11/3.11.4_1/Frameworks/Python.framework/Versions/3.11/lib/python311.zip',\n",
       " '/opt/homebrew/Cellar/python@3.11/3.11.4_1/Frameworks/Python.framework/Versions/3.11/lib/python3.11',\n",
       " '/opt/homebrew/Cellar/python@3.11/3.11.4_1/Frameworks/Python.framework/Versions/3.11/lib/python3.11/lib-dynload',\n",
       " '',\n",
       " '/Users/dgerosa/box/lib/python3.11/site-packages',\n",
       " '/Users/dgerosa/box/lib/python3.11/site-packages/modulescicompclass-0.0.2-py3.11.egg']"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sys.path"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "My PYTHONPATH is empty..."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "0"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "os.system('echo $PYTHONPATH')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "You can add things explicitly to my the `PYTHONPATH` shell variable.\n",
    "\n",
    "Using `PYTHONPATH` to quickly add a module to your search path is an easy hack,\n",
    "but if you are developing a library that will be used by others, it is better\n",
    "to make the modules installable to the system search paths.  This is where\n",
    "_packaging_ comes into play."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Packaging\n",
    "\n",
    "\n",
    "\n",
    "Let's look at the structure of creating an installable python package.\n",
    "The python packaging system is constantly evolving, and the current recommendations\n",
    "of tools is list here: https://packaging.python.org/en/latest/guides/tool-recommendations/\n",
    "\n",
    "Eventually, you'll want people to do  `pip install myawesomepackage`. One thing at the time...\n",
    "\n",
    "\n",
    "## Our example\n",
    "\n",
    "We'll work on an example that builds on the Mandelbrot set exercise\n",
    "from the matplotlib session.  Our example is hosted here:\n",
    "\n",
    "https://github.com/dgerosa/scientificcomputing_bicocca_2023/tree/main/lectures/modulescicompclass\n",
    "\n",
    "The directory structure appears as:\n",
    "\n",
    "```\n",
    "modulescicompclass\n",
    "├── modulescicompclass\n",
    "│   ├── __init__.py\n",
    "│   └── mandel.py\n",
    "└── README.md\n",
    "```\n",
    "\n",
    "This is a rather common way of structuring a project:\n",
    "\n",
    "* The top-level `modulescicompclass` directory is not part of the python\n",
    "  package, but instead is where the source control (e.g. git) begins,\n",
    "  and also hosts setup files that are used for installation\n",
    "\n",
    "* `modulescicompclass/modulescicompclass` is the actual python module that we will load. To make python recognize this as a module, we need an `__init__.py` file there --- it can be completely empty.\n",
    "   \n",
    "   \n",
    "* The actual `*.py` files that make up our module are in `modulescicompclass/modulescicompclass`\n",
    "\n",
    "Right now, this package does not appear in our python search path, so\n",
    "the only way to load it is to work in the top-level `mymodule/`\n",
    "directory (that's because the current directory is *always* in the path). Then we can do:"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## setuptools\n",
    "\n",
    "The current python package recommendations are:\n",
    "\n",
    "* Installation:\n",
    "\n",
    "  * `pip` to install packages from PyPI\n",
    "  * `conda` for disctribution cross-platform software stacks\n",
    "\n",
    "* Packaging tools:\n",
    "\n",
    "  * `setuptools` to create source distributions\n",
    "  * `build` for binary distributions\n",
    "  * `twine` to upload to PyPI\n",
    "\n",
    "Let's look at how to use [`setuptools`](https://setuptools.pypa.io/en/latest/index.html) to package our library.  See the\n",
    "packaging guidelines here:\n",
    "https://packaging.python.org/en/latest/guides/distributing-packages-using-setuptools/\n",
    "\n",
    "The main thing we need to do is create a `setup.py` that describes our\n",
    "package and its requirements:\n",
    "\n",
    "Here's a first `setup.py`:\n",
    "\n",
    "```python\n",
    "from setuptools import setup, find_packages\n",
    "\n",
    "setup(name='modulescicompclass',\n",
    "      description='test module for the SciComp class',\n",
    "      url='https://github.com/dgerosa',\n",
    "      author='Davide Gerosa',\n",
    "      author_email='davide.gerosa@unimib.it',\n",
    "      license='MIT',\n",
    "      version='0.0.1',\n",
    "      packages=find_packages(),\n",
    "      install_requires=['numpy', 'matplotlib'])\n",
    "```\n",
    "\n",
    "\n",
    "The packaging ecosystem is always evolving (too quickly if you ask me...).  There are 2 special config files that can help customize the package and contain the defaults for other tools: `setup.cfg` and\n",
    "`pyproject.toml`.  See:\n",
    "\n",
    "* [Configuring setuptools using `setup.cfg` files](https://setuptools.pypa.io/en/latest/userguide/declarative_config.html)\n",
    "\n",
    "* [Configuring setuptools using `pyproject.toml` files](https://setuptools.pypa.io/en/latest/userguide/pyproject_config.html)\n",
    "```\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Installing\n",
    "\n",
    "We can use setup in a variety of ways.  Two useful ways are:\n",
    "\n",
    "* Install:\n",
    "\n",
    "  `python setup.py install`\n",
    "\n",
    "  This will copy the source files into your install location (likely\n",
    "  `~/.local/...`) putting them into your python search path.  Then you\n",
    "  can use this package from anywhere.\n",
    "\n",
    "* Development mode (https://setuptools.pypa.io/en/latest/userguide/development_mode.html):\n",
    "\n",
    "  `python setup.py develop`\n",
    "\n",
    "  This doesn't actually install anything in your user- or site-wide\n",
    "  install location, but instead it creates a special link in that\n",
    "  install directory back to your actual project code.\n",
    "\n",
    "  This allows you to continue to develop the package without needed to\n",
    "  re-install each time you change the source.\n",
    "\n",
    "  You can uninstall via:\n",
    "\n",
    "  `pip uninstall modulescicompclass`\n",
    "\n",
    "The above put the package in the system-wide path. Should be ok for your laptop, but if you're on a cluster where you don't have admin privileges use the `--user` flag.\n",
    "\n",
    "\n",
    "## Using our module\n",
    "\n",
    "Once the module is installed, we can use it from any directory.  For example, if we do:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "/Users/dgerosa/box/lib/python3.11/site-packages/modulescicompclass-0.0.2-py3.11.egg/modulescicompclass/__init__.py\n"
     ]
    }
   ],
   "source": [
    "import modulescicompclass\n",
    "print(modulescicompclass.__file__)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAe4AAAGiCAYAAAAhjSVBAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjguMCwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy81sbWrAAAACXBIWXMAAA9hAAAPYQGoP6dpAACnZklEQVR4nOz9d5wkV33vD7/POZW6e+LOZm3WKucAYkUSRiDLQhewLxcw9yKDja8fo8eAbAPyYyzCtWVetrF8bS6Cn01GBowN+GcwWBJXAiyBMpIQWmXtanOa1KHCOef5o2p6pyft5J3ZOW+96qWd6qo61d3V9a3v53yDsNZaHA6Hw+FwLArk8T4Bh8PhcDgck8cZbofD4XA4FhHOcDscDofDsYhwhtvhcDgcjkWEM9wOh8PhcCwinOF2OBwOh2MR4Qy3w+FwOByLCGe4HQ6Hw+FYRDjD7XA4HA7HIsIZbofD4XA4FhFzarhvvPFGXvSiF9He3s7KlSt5wxvewPbt24+53z/90z9x+umnE0UR55xzDt/97nfn8jQdDofD4Zg3BgYGeO9738vGjRsplUpceuml3HvvvZPef04N95133sm73/1ufvKTn3DrrbeSpimvfe1rqVar4+5z11138da3vpXf/M3f5MEHH+QNb3gDb3jDG3j00Ufn8lQdDofD4ZgXfuu3fotbb72VL33pSzzyyCO89rWv5fLLL2fXrl2T2l/MZ5ORAwcOsHLlSu68805e8YpXjLnNm9/8ZqrVKv/2b//WXPeSl7yE888/n5tvvnm+TtXhcDgcjlmnXq/T3t7Ot7/9ba666qrm+osuuogrr7yS//W//tcxj+HN5QmOpK+vD4Bly5aNu83dd9/Ndddd17Luiiuu4Fvf+taY28dxTBzHzb+NMRw+fJienh6EEDM/aYfD4XDMK9ZaBgYGWLt2LVLOnTDcaDRIkmTGx7HWjrI3YRgShuGobbMsQ2tNFEUt60ulEj/+8Y8nPeC8oLW2V111lX3pS1864Xa+79tbbrmlZd0nP/lJu3LlyjG3v+GGGyzgFre4xS1uOcGWnTt3zpoNGkm9XrerV6pZOc+2trZR62644YZxx962bZt95StfaXft2mWzLLNf+tKXrJTSnnrqqZM693nzuN/97nfz6KOPTv6JYpJcf/31LR56X18fGzZs4JVb/j94cvTTjsPhcDgWNpmJufOZT9He3j5nYyRJwt79mmfv30hH+/S9+v4Bw+aLnmfnzp10dHQ014/lbQ/xpS99iXe+852cdNJJKKW48MILeetb38r9998/qTHnxXBfe+21/Nu//Rs//OEPWbdu3YTbrl69mn379rWs27dvH6tXrx5z+/HkCE+GeMoZbofD4ViszMd0Z0e7nJHhbh6no6PFcE/EySefzJ133km1WqW/v581a9bw5je/mS1btkxq/zmNKrfWcu211/LNb36TH/zgB2zevPmY+2zbto3bb7+9Zd2tt97Ktm3b5uo0HQ6Hw7FE0dbMeJkulUqFNWvWcOTIEb7//e/z+te/flL7zanH/e53v5tbbrmFb3/727S3t7N3714AOjs7KZVKALz97W/npJNO4sYbbwTgPe95D6985Sv5q7/6K6666iq++tWvct999/GZz3xmLk/V4XA4HEsQg8VgZ7T/VPn+97+PtZbTTjuNp556ij/8wz/k9NNP5x3veMek9p9Tj/tTn/oUfX19XHbZZaxZs6a5fO1rX2tus2PHDvbs2dP8+9JLL+WWW27hM5/5DOeddx7f+MY3+Na3vsXZZ589l6fqcDgcjiWImYX/pkpfXx/vfve7Of3003n729/Oy172Mr7//e/j+/6k9p/XPO75oL+/n87OTl699b1ujtvhcDgWIZmOuf2pm+jr65v0vPFUGbIVu7evm3Fw2trTXpjTcx3JvOZxOxwOh8OxkNDWomfgv85k3+niDLfD4XA4lizHY457prjuYA6Hw+FwLCKcx+1wOByOJYvBoheZx+0Mt8PhcDiWLE4qdzgcDofDMac4j9vhcDgcSxYXVe5wOBwOxyLCFMtM9p9vnFTucDgcDsciwnncDofD4Viy6BlGlc9k3+niDLfD4XA4liza5stM9p9vnOF2OBwOx5LFzXE7HA6Hw+GYU5zH7XA4HI4li0GgETPaf75xhtvhcDgcSxZj82Um+883Tip3OBwOh2MR4Txuh8PhcCxZ9Ayl8pnsO12c4XY4HA7HkmUxGm4nlTscDofDsYhwHrfD4XA4lizGCoydQVT5DPadLs5wOxwOh2PJ4qRyh8PhcDgcc4rzuB0Oh8OxZNFI9Ax8WD2L5zJZnOF2OBwOx5LFznCO27o5bofD4XA45g83x+1wOBwOh2NOcR63w+FwOJYs2kq0ncEct+vH7XA4HA7H/GEQmBmIz4b5t9xOKnc4HA6HYxHhPG6Hw+FwLFlccJrD4XA4HIuIoTnumSxTGk9rPvShD7F582ZKpRInn3wyH/vYx7B28pK787gdDofD4ZgnPv7xj/OpT32KL3zhC5x11lncd999vOMd76Czs5Pf+73fm9QxnOF2OBwOx5IlD06bQZORKe5711138frXv56rrroKgE2bNvGP//iP3HPPPZM+hpPKHQ6Hw7FkMUXJ0+kuQxHp/f39LUscx2OOd+mll3L77bfzxBNPAPCzn/2MH//4x1x55ZWTPmfncTscDofDMUPWr1/f8vcNN9zAhz/84VHbffCDH6S/v5/TTz8dpRRaa/70T/+Ut73tbZMea0497h/+8IdcffXVrF27FiEE3/rWtybc/o477kAIMWrZu3fvXJ6mw+FwOJYosxWctnPnTvr6+prL9ddfP+Z4X//61/nKV77CLbfcwgMPPMAXvvAF/vIv/5IvfOELkz7nOfW4q9Uq5513Hu985zv51V/91Unvt337djo6Opp/r1y5ci5Oz+FwOBxLHDNM7p7e/nk0eEdHR4vdGo8//MM/5IMf/CBvectbADjnnHN4/vnnufHGG7nmmmsmNeacGu4rr7xySrr9ECtXrqSrq2v2T8jhcDgcjmFoK9Az6PA11X1rtRpStj4oKKUwxkz6GAsyOO38889nzZo1vOY1r+E///M/J9w2juNRQQEOh8PhcCxErr76av70T/+U73znOzz33HN885vf5BOf+ARvfOMbJ32MBRWctmbNGm6++WYuvvhi4jjm7//+77nsssv46U9/yoUXXjjmPjfeeCMf+chH5vlMHQ6Hw3EiMBQdPv39p1ar/G//9m/50Ic+xO/+7u+yf/9+1q5dy//8n/+TP/mTP5n0MYSdSrmWGSCE4Jvf/CZveMMbprTfK1/5SjZs2MCXvvSlMV+P47gl7L6/v5/169fz6q3vxVPhTE7Z4XA4HMeBTMfc/tRN9PX1TWreeDr09/fT2dnJZx+4gHK7mvZxagOad1744Jye60gWlMc9Fi9+8Yv58Y9/PO7rYRgShs5AOxwOh2NpsOAN90MPPcSaNWuO92k4HA6H4wRkvqXy2WBODffg4CBPPfVU8+9nn32Whx56iGXLlrFhwwauv/56du3axRe/+EUAbrrpJjZv3sxZZ51Fo9Hg7//+7/nBD37Af/zHf8zlaTocDodjiWKYemT4yP3nmzk13Pfddx+vetWrmn9fd911AFxzzTV8/vOfZ8+ePezYsaP5epIk/P7v/z67du2iXC5z7rnnctttt7Ucw+FwOByOpcycGu7LLrtswlZln//851v+fv/738/73//+uTwlh8PhcDiazLwAy/xnVS/4OW6Hw+FwOOaK6fTUHrn/fLMgC7A4HA6Hw+EYG+dxOxwOh2PJMt/9uGcDZ7gdDofDsWRZjFK5M9wOh8PhWLLMPI/bzXE7HA6Hw+GYAOdxOxwOh2PJYqzAzKQAywz2nS7OcDscDodjyWJmKJW7PG6H41iI+X+6nTbz03jP4XAsMZzhdiwOhMCGHjb0OQ7K1JQRqUE2EtDHo5Kxw+GYLMZKzAwiw2ey73RxhtuxaDCRT9YeYNXCt9yqliFSjXCG2+FY0GgEega52DPZd7o4w+2YH4QAOYMLXAiskhhPLA65XIlimeLTuCnkdSezOxyOcXCG2zH3KIkpB1hfTfsQVghMqBaH0QaMJ9FtIUL7U9zRIhspopHOzYk5HI4WnFTucIyFlOiST1aZ4eW2SIw2gPELdWCKyMziG+sMt8MxT2hmJnfr2TuVSeMMt2PmDMng4xhW68lcNl5EhndWmMb7tQqsFOBNUp2w1gXAORxLDGe4HTNDCGwpQJd8xntotVJgPFekbzJYAbrs5w87k0AmGllLIDsez/0Ox+LHSeWOpYcUmMgja58gTWupedozQQh0pNCRmlSAmleTiEaGOC6CncOx+HFNRhwnHqKIjB5PBhcCK0VutJ2Bnl0m8XlaJcCTYIf9lK11ErrDMUnsDNt6WpcO5lho2NBDV8IJc6dNsHiivU80jCfIOkPQR71zYUE20lxCd8bb4TjhcIbbMSFWKXTJwwRujnohYj1JNnI+3Fp8a5F1F5nucBwLJ5U7FheFDG7l+Bee9SXH4bo8obCSMRULYSyY3EOeVYpiNdZXMM53K6wFY5xH7ljyuO5gjsWFyouEmHD8y8B6YlGUGF2oWAEmEKRliRn2OQpr8eoWr25gDoqk6VBCVwTj2GVhLKqaIGrx7A/ucDjmFGe4lzBWSkzozbwwimNCjCfIIoEZ9jELI5AZUJ+bMceU0IchMoNMFEIIV17VsaTRM2zrOZN9p4u7Yy8FhiTxETdy6yvnTU8SK3K52yrGzVcfD+OJZqrccNk81YBVCGORupDOp4jQILRFTFXxLnLrReS3Gm4DQmsnoTuWDE4qdyxMpMBUQnTZww6P/hZ5aU7HJJCQlSVZNPUfqVH5/gA6FGTl/N9pmyDOBMKAqlu8+tTnu1Vi8QfNlI2+lQJd9vL678OQqUENNFxXM4djAeMM91JACIwvyUru654uVgi0Dzpk+v3ARV7S1Pj5/5urDfhS5AFjU3W6rciro2ZTPx3jSxjRA8VKgaq6hznH0sEgMTOQu6e676ZNm3j++edHrf/d3/1dPvnJT07qGO5OfqLhqVFduKyU2Km2l1xCNGVwj1ZFYvg240SGj3tMWQT2SUjbIF5mMD4txlnGApWQP1h5uTc+3HALCyKbWAa3CnSQF8EZjsxsvu9UHwQEmMhDjpz31haRZm4+3HHCoa1Az0Dunuq+9957L1ofrXT46KOP8prXvIY3velNkz6GM9wnEkJgSj66LWi5kVshsNPoVLWU0KEgrUyc+mbU5L1t4wuyChgPGqc0eMu593FytJ+7+rZy3971NGKf5HCEPKTA5p68GfFrFAb8QYGKxzeWxoOkTbYaaAt+3eJX9ZRbFxlPkFV8xAh1RsYar89COg3X3uFwNFmxYkXL33/+53/OySefzCtf+cpJH8MZ7hMMqyQ6VKM8MMfEWJVL4Xb6LcNbjycpvGhLZ1eN/9Z1L2cFHr7I2FHt5rAqcXgwAKGK8UePLbNjn4+VjHrYEBbMdGuvCIHxx845t0ogXE0XxwnGbAWn9ff3t6wPw5AwDCfcN0kSvvzlL3PdddflGR6TxBnuEwHfa0aIm0BNfw52CZBLy7IlMtwK0D5TjhaHo5K4GWFgdQl0yaKD/If9eLKahP3864Hzeebhk/CqAi+AtM2CMIhMIDMBhjzCPBNFDjhkI75QYUCmE0voxstzx+UIj1um05PQjZKYyEeo1jcqUu0kdMeixs6wO5gt9l2/fn3L+htuuIEPf/jDE+77rW99i97eXn7jN35jSmM6w30CYAKPrDMs0pWWYN/rKaADSdImR8nSVjI9w60EaSUPOGsiQEeWtMNgfYsF7hrYyqPeOh742cmc/M8NvIGYPa/oZnBbnSDIaNQCspqH0AKqEqVzDzeLQIS0zH3LDPwBMW4kuRXkDwyebJ0zNxBUDV42dSNrfEHaEYyS5L1aiuo3rq2oY9GiEegZNAoZ2nfnzp10dHQ01x/L2wb4h3/4B6688krWrl07pTGd4V7sCAFK5LnCrud1C2MpD0MStvFHvzYtRHHMEcczHljfgmewVnA4qVDNQvwjkuDZ/ZjePvyLulDKEAUpWapIfdnstgYChM0D5ka8IWFGy+Njvc9RErohr97WGgN39PWJ7HkRJzFyExu7a87hAOjo6Ggx3Mfi+eef57bbbuNf/uVfpjyWM9yLCU9hQh9GRDfryAM3p93CkCQ+co44j8Ce4bFl7oFarwhYO8Y8dJYpDsdlPGlQiQBjsFlGx3MJ8T1tNEptmC4DHTqP6g7s+PK4nsH3LPIHjLQyOphNxbmEPlWMn5fNFdrksnnsZHPH4sLYmRVRmUbdJAA+97nPsXLlSq666qop7+sM9yLC+grdEaBHFk2Ro9OBljr5HK/IU6yGYQXMtEKhlSKfww4nd7wsVRysVQBQMWAtNkkI73uS9U+0Yzsq7Ht5D4cvEKAsNjLooPVuIDIBgxI1A0XaCsgigQ5GPBRoCK3JK7BN8SakQ4nxcgndG0hQqW5pMepwLHTMDOe4p7OvMYbPfe5zXHPNNXje1M2wM9yLgWLO2kqJGaN06VJnzIflIjd75Fz2jClUbKNGp2+NhwUynX9nQoO1FqxF9/dDfz+qqxOv3pNvLAujp1qNn7W5lyw0TZ3bykLenoKdHEtClxyd4x861KQN+JCEbm1+XYoxYiycB+5wtHDbbbexY8cO3vnOd05r/zm1AD/84Q+5+uqrWbt2LUIIvvWtbx1znzvuuIMLL7yQMAzZunUrn//85+fyFBc+nsKUQ0xHCVP2XYvNEVgJOpKkbZK0XR1dSnJUpPeMEEO53nnJ0tlKG8uPLUnaBOUVVbqWD1LqbOCX05YFoPKCYPkjmvadBh1CY4Ug7srjG2aCFZCFgrRNkZXl9OrXizyjQXdE+bU6tLRH2GC2AgocjtnHIGa8TJXXvva1WGs59dRTp3XOc+pxV6tVzjvvPN75znfyq7/6q8fc/tlnn+Wqq67id37nd/jKV77C7bffzm/91m+xZs0arrjiirk81QWL9SS6LUCXijQvFzHegpV55620POJzKTzj2RsHdARZafaPjeeRdMIZK/dR9hL21DrprZdaNjnYG9L1dErpru3o0zcysKGd+toMr0+h4qLT2HSRkJXyCHaVCoQxyGnMd+tQYoKgZZ3ILD64lDHHgmW+K6fNBnNquK+88kquvPLKSW9/8803s3nzZv7qr/4KgDPOOIMf//jH/PVf//XSMNxjGWUpi9KZzmAP0fydFAbUSkDOoIb4ZMeVM/e0jQ90tKHiGNuIMY0GWIPUMJiGSGFRwlAJEjIjqSU+Seblc9zYPGffk/lxfIP1ZH5tDL33adhGO+xBxJgRcrqdomw+8jvw8ocrZJ6j7oy3wzFzFtQc9913383ll1/esu6KK67gve9977j7xHFMHMfNv0dWr1k0KImJAuyIbk1WyTyFx3EUkXt3QxHixh/DYCxAPGXoPT3mqXeswh9czYqHUkp3/Bxbq7PioYQ96Uae64LKhQd5y+b7eXxwDXf85Gw6HxeoULDnUkn2mlPzVLMwd7Gtl6sAVgpkWkSHz6Cxly28b+MppLaoxvSizZvHE2BKHshSHvxWTxGJK7/mWDgcj+C0mbKgDPfevXtZtWpVy7pVq1bR399PvV6nVCqN2ufGG2/kIx/5yHyd4twhJabsk1W80UbIyeOtFPPNSUXMvmw9hyhpePEpz3LBRTvpy0p8038Zm+8poQ8eIvzho5x0T4jZvI5nTi1xTefD/JNMuP/pc1n91Z9Tv+QU0vcd5rNnfIk7aqfwqSdeQe+BNqxn0WWL9QSqATIdvzDLZLAqN9xEoBKBzPS0Oo81EYIsUhApZGrxjXGG27GgMMyw5OlxuAEt+lCn66+/nr6+vuayc+fO431KU0MIUHlAkB1K6xIjFkfewauQqvPPiVwen2uJfGjc4XL0uNtarLRYZYsobQsSpLAIYZHCUlEJa/wjnBQewfgWRP4TNI0GurcP1VclHQh4JOlgR9yTR453dZK2KbqiOiukoEtVkdI0zy+PcLe5jD8kc8/gM2l+1sX7zj/zmbQzza/j/DuT4ClQ0l3bDsc0WVAe9+rVq9m3b1/Lun379tHR0TGmtw2TK+S+YCnkcVMUUDHBon+OmlN0KNFhbgBG5iLPCUXeswmOVlybCKvypiLWAxMZVHuK72s8T+dTvFYwkIXsSJbTl5VQsQDbqmvbI32svXUV1+76n+jQkm3VPPb/W4XfXmeD3+BTRy7gydpKao2wGNOiywZj8oIw2LzmuUxmLpsbBWlZkIUKmVm8+gxlcyXQZQ/jS0RmkDUnmzuOP3aakeHD959vFpTh3rZtG9/97ndb1t16661s27btOJ3RHCMEpuSRtvvO+zgWIjfWSUU0daI5D0YbavJRzsc/1nhWgSlZbGCQUUalHBP6GXJYdFdD++yNO+lNS8iUUcFa+sgR2v/1ITq+66HPP4UD72/w3Qv+nmeyMl88+FK+u/usPGCtUfx0lcWWLNbmef5CS2Qm8AR5r++ZvP8h2RxQDYFKZkE2L3kQWVRikImGGZ6jwzFTZqs72Hwyp4Z7cHCQp556qvn3s88+y0MPPcSyZcvYsGED119/Pbt27eKLX/wiAL/zO7/D3/3d3/H+97+fd77znfzgBz/g61//Ot/5znfm8jTnn0ImPBoR7Iz2cJrFQIZ/LmL4+nkaf0iSn8KYVuRh2KKQyIdkclX8O5AZntT4Uo97TBvH2DhGNnJvtF16VESKsZLMSLSRucKubB5FbshPcijKvjmtUESi5w3IpvdZiKOfyVBBG2EtmClEm4+kqMduPYnwimBMbVzEueO44ILTRnDffffxqle9qvn3ddddB8A111zD5z//efbs2cOOHTuar2/evJnvfOc7vO997+Nv/uZvWLduHX//939/YqWCCZHL4yUPKwUmnM1KHouf3MsVpCU5qtiM8ScxzzxTivF1dFQen+mDQuRldIc1AqVZHfVzcnSAw16Fn4RM+NAmMkOtEfBUKtiRdVPXeSETT2lK5Rgdyrw5ScODrPU4xoe0DYQRyBi8hp1WqljzeB4kFYkoWVRi8WsGZlJ+VQmytgAZeohUo6oJpDNx5x2OpcOcGu7LLrssL+84DmNVRbvssst48MEH5/CsjjNSYCInj0+E8QRZSYzKmZ4vRcr4QylWs3O8UGX0hDUqXsxJ4RHWB4cIZdoSnDYm1pIlil26k71ZF0lRCs5XBr/QwQcbIVmiRtlk41F4x+BZsPEMPGSK+fsIhjx7ryEQM6hJbqVARwodKVRDIuMM4aa7HccBJ5U7xkfJvJiKEq6T1xgMj9w2Ks/Lns/fQ3P8oSIrkx1b5NHcVuSR3XgW4VmUp6mECW1BTFdYp8OrE8qMnY1lPDa4lgONNoIjArLxvUzRSBH7uvjKvm0YBH1xicjLMFaQGom1AjEJa2wleREfA8JMP2DtqGwucvWD/Hgzks2L41lfITIvl8udbO6YR6ZbtnT4/vONM9zzwZA8XvaxnsB4LhVmJMYXpGWZ9xUv0prmdfxANL3sIUM8GfIALov1LTY0+O0JQZixvK3KRT076PGrlFVMWSakVvF3P7+M6P+2UzpsOOmpfkytNv7B9x9k07918syDp1FbJYkvGeSidTtpaI/DjQqN7Ng/36EAu0Tm89yqIWYsm2sfkjaJMBYVW/yqmdHxrJfL5qLkI1ONGoidbO5wTIAz3POE9WVeXMV522NiVV5z3ByPfhQil5V1OPWSplaCDS021MhQE5USykHK6ko/p5X3ssE/1Ny2V5eJ95bZ8t0XyJ7bcUxbp3v7UHc8QKcQdF18NtvPiFgeDjKQRfQlY6dHjoXxAK/o6T0L9tAqyBQIK3LvfYaX9JBsDuDVQdaEk80d84aTyh2tKIlVClQRQe5sdgtDkcp5ENg8BJ6NM37eApSpy+MSTGCxJY0XZYRRyrJynbYgpj+J+Jc9F7Ts1sh8SrsVNOJxDjzeiVqsABllrI8OM6gjJJaqDthXayeOPTLA6vzBR8hcDs9bgI6IzFd5Wl2LtG3Jy5FOQ0I3SqBDme+vmVZP75a3WsjmkB+LTDvZ3DGnOMPtaMGGPllbkBtt5dK+RmJ8QVrJ5XGjZt7AY8rjB0WLzmZFtsnt15THw9xor1jVx+q2ATr8BidXDtCp6nzp6RfD95fRvuuoiysMbNjZi+ntm/q5hh4regZ4ffvDpFZS7fBIreKnta38qz2XQ9UydT8gI0BkAtkQqHqrgbYiVxWMR4u0LQx4NYGKp2YgrQAdQCwlwoBfN3i1GUav+5KsI0QYi4w1qr+RG2+Hw9HEGe65osjT1pHCeq4i2lhYKdCBQAfH3nZuxs8jyI9VEW3UfoJiTjv3tFe3DbCl7SDdXo2t0T561CD12GfjfQNwzyMt+063kJn1BMtKNU71Ky3rU/scd5W20Mg8jBHUI4VNJUYrVGP0cYZk8+HILI86n9Z5KdCqqO5aFH6ZCVYKdJEi6RlQ7mHXMcc4j9uRy+O+ypuGBMpFkI+g2c2riEye72u+GWEtcqM94fhDkrhqdSKtV1RHK+a015T6ODk6wAtJN198YRuH62XsE22ogYMzSXVuweuNefyx9fx68KqW9YlRDKYhlSAhTj3q09Cphz4LHeYeushmVip1trBKYEo+wstLpDrZ3DEXOMPtwAYeWUeUR48r10d7JFbl8rj2xaz0t54qxhdklbwON0WjkvHIpWWblzEdbhA9i9+eUColrOno5xWdT/Dy0vN8bM8V7P23DSz/WczyQ73YF/bO2nnLZ1/glC9sYPd3trasP3yGj3/ZQc5dsYdUK/plaepKtcxztHUAUoM/KBDJ8TeQJpCkHSHCWlQ1RQ3GTjZ3OHCGe9axnsQEEuM7eXwsrMoD0fRx6gszVA1tUtHrAqwHNmhNVBaeJQgzSkFKR9BgvX+IDV4bde3T/USK94P7py2Jj4fu7YN7HmHkx7ZMvIiDL/Vp9xqEXjatAD9bvE8A0tkrPDNTrBTYopmMTBaABOA4IcmrBs+kycj84wz3bOAprK+wUmLCMfppL0Gakrhq/TC0nweiHZdzkTQ7fbVQSOJmRGR5LqsXP0sBwjNIzyKlQRVtNRPt8VyynB65kz21jhl1z5opocoIwowEMAZ0qopI7yINbBIXppVFWtyIuWWpjy2h5xXv8kA1meadxGYSYd5ybF9iygFoi0g1Is2cbO6YFZxUvkSxviLriDC+LG72C8RlOZ4IyEqSLBoxjy3m36OzXiGPezT7eI/E+KArBqtGGIOiaQgC/CijHCV50xCZb1fNAu4b3MyetJsXDnWxoXH8pNyyl9BZqVP3fWoyzMuXa4GqSYQWkzOiMi/3Kka49ioBb1DkldLGYCjC3HhFhHnN4FdnFmE+HBNI0s482lxVU9SAhhmUXHU4hnCGe6kiRO4RuH7aR8tiityzPmYA2Dyd09C5AGPW57TKYj2Td9waC2FRyhD6w9K7hCUzMm/RKSxp7OXlOo8TnjAESkMASeqR+flEfjNHXdhjfhlWAGOk5gnNMavZDQXyCUMztmNouJl63nmnNgHWImNXedCxtHGGe7ooiQ39PPI19BbMvODxZKizV162tPj/PNcbz+X51vXG52gPb88e9bxbtrGj54eH5HHfIIXF83JvWklD5GW5NC01oVzAAVMif886yjuFicxOWjYfjlWQRSA9gdCgkgm86SJKPa3k3rdKLMyWbC4E1peYSgjGIGKNSFyZNcf0cR73EsL6iqw9RIcSpIseB3KZNZKk5aN9rOcTKwW6xKjANytoGm6jQLeNIYnLsQ23F2ZUSkkxr53vo4SlO6zRE1ZRwuKJhWy489Q17QmwoKoSmU39i7EeZOXcc1YxCC2Q48znWwFZlOfoCwPBoMGbxbn/LFJov5DkBxI33+2YEc5wLxWEKAqsCDefTas8blURiT0f1/IYhtaoYQVVxnLxZC6L401C0hYWpSy+p5HDjiWFLTztrGX9ZDp1zTVD5yOERYjCKZY2X0yRgidATHby2Q6TvIt8dpFRSO+M63Xn3dZyiX3Wr4Wh3561R7vtGZzxdiwZnOGeCkPyuK+KQDTnZVsBJhRFXnb+/zlH5PW2R6Z02WFzs1bZfH59hGdtAnvMCVcRGJRnENLg+63GeSS+0HR6dcoywQ8zrHc8uqTklFTKsqhKYjyEsBgj0JlCZxKbylw2D2we0zUJGyeMQCYWoVu/07zka/F6egzZfI4xoYL2KG9X2sicbO6YMtYK7AyeLmey73RxhnsKWKXI2gJ0Of/YnOGmMKKSpE00Jem5vo6Ham5n5dHnMuSFWwW6bPIc7HG2GRMBnq9pqzRQ0h7Tiw5lxhq/lx5vkEopxqrjU79VCkvFiwlVhrYCTxisFSSZYrAWkmbBUdl8ks8WIpUIIxk5E2A9yCqALbp5ZYzaZl4QgixSiFAhMosPTjZ3TBnXj3sp4KqhAYVxLpYhWXTOA/SGjLIUzapntoiWbj4sNLcpcrC9YXPXTRl52PsYusfbQleWFiHzdC8lDULYVjmcPDjNkxpf5EtFxpRFjK/McYmgF8KiMOSqscVYQaCyPMIcUMqSCpu/8eEKxDHsm9UWC1jZuqGF5mdq0vz30EwTG+uYIv++bOHpz+qMgigeGL2h68LJ5o4TH2e4HVOmKY8Hed56Foq5D0QbkseLAiq6SDOz0mKCIipcgglNbqyVRUUZShmUMkRBSuBpAqVpC2I8aahnPoNJgDaSOPWIEw9rBWGYFnPElrYgoeIleFLT7se0ew18oenxq5RVTJeqsck/yApVpyuqE3vdc/xBtGJ8QWepwfroMDUdciQrE1uPdi/GlAeasnkvYMzRL8kaQZbkDUnGReaqhRkRzGZ9W1STA93nIaxApgKZjCGbi/z6sFIhtMVrzGKE+fBzEmBKHshS3lq0njrZ3DEpXHCaY8mgfZHL40X0+HxcuyaAtEJrPrEEHVlspMG3RO0xpTApPOV8k0qQsLbSR7vfoMevcnK0n4qM2Zt18kx9BXUdcKDRxr5aG5lWTS9bCku732B1aYCSTNgUHWKtfwQlDLIoatqlamzxG/TIMsvCGrvnu/Z60TVsS7CfvVkXgzokFYqKF1NSKWkhg1gr0OboB6eNYMCWyCYy3MpiIzvKiZaljEpb/vDT67WRJj4yzjuDybTVo7ZDddBDgcwEwsxuhHmTQjYnUsjU4hvjDLdjUrg57hMVUUhwSowqBbmUGJLHh4phzKo8fgzjPzTekDxui/QtK8nlXz+XuNNUYUyI72vaSw3KfooUlt6kxGAWUtd+bpBVg4bxKakUX2jqvs+AH6JV62RtNQ3ZYTwilZJaxYCOUMPqfrapBjWznxVqgN3VTmQ6vxKtTC17B9v5ReMkUquQwlKWSfP11Cp2mm4G6iHGSKIgpRwmpFohVVFwZtwPfnjTbnt0ikFAHHvEANlQJkEe4Z1PIxUe9ZB6PmxKZU6VmeIER0r7DsdCY9euXXzgAx/g3//936nVamzdupXPfe5zXHzxxZPa3xnuY6HyGskm8JpFRZYiQ8VVsiiXx3Uwu/K48QQ6muBBQIwhjwc2n8duywjLCfFgSPhImcpuy+A6Qdsv9fOmtffz3f3n8NhdW6jsEjzvwYNBnjZWOzXm9ef+jFNK+1gZDHBSqbdF9jqYtHHHw6fT8ZiP0PBEMEYluCJQznqWyk5Y/ey+WWvlORnanjhC7VvL+eLy11A/o8G1F97BeaXnm6/36gr/9vTZlH7QhtBw+HzNmjNewCtqrQ9IO6npYN/XVMIEJQ27d/TQ/kiJYMDS6BHUV1nSdlt8dwKhBSo+vtHmDsdkmW+p/MiRI7z0pS/lVa96Ff/+7//OihUrePLJJ+nunvw0mzPcx0JKdMknq3hLvsyi8UVeXKW4amZTITJenmJkJ7gihwegmcBiKho8S1RJaC83ONAfsvyRjPIPHqV82VlUXptwTcfz3HH4NFb/RFO59dFif4HwPfb8+llkZys2BQcA0COeGh6ub+CeZ8/hpK88ia1Wx//+ZbGf1uhGPINPYero7c+w4vldCM9j1zvPpvNFNV4RHfW4X8j6aBwqseV7uyHNqK/cSNd5ddr9BpmRzWC2Y9EexpxU6aWkUvY8sYI1PzwCz7xA7ZVnUFsrMWVNZhXCgMjyuugqOeZhHY7jznxL5R//+MdZv349n/vc55rrNm/ePKVjOMM9HkoWhR5ks+DKUuRoitcweXwmH4U4Kp22MIm58qY8ror87CKifGhOWviGRrdPZf1akg7FgXqF2+rtPHlkBZ39GaZWGzaeIuw3PHx4LRUvZkN4iFOCvShheDpZxXON5Tw+sIqgH2y12rrvQsLo/NyEIOiz3NW3ldV+b/PlXel6/F6FrdUhSQn64ckjK2gLY7SRTc/7WJS8lDYvoSSTvBRskmHqdYTJO6vhW6xvMb7Ic8iLfvTCFh3FhmRzWXSMsxZau6XOLlKCVwQcaOOizB3jYmfocQ8Z7v7+/pb1YRgShqP7F//rv/4rV1xxBW9605u48847Oemkk/jd3/1d3vWud016TGe4x2JIHo98rBRLunmICfJWjabooz1TedyoXBIfs3jKRB+zKOTx0GKlxQajb8Q9PYMMvD7hyK9USGsZwX2r+OAP3klltyF8ZhfZ8I2tofuhXg6p1Xyvaw2NSwf59MVfpkvWuf6Zl1L/vysIj1hWPTyATRaB62gtyx/s46HPncM9nec2V8sU1j0SYwcGsVqz6icD9B1eRm+npPfChAtOeX5Sxnt5WGVz6QBlmWBLuqky6FBiKpqoPSaWAZnwEFlxnYhCNm+AinMpPS0JtC+RGXgNg4pn36BaJcgqPjJQCG2Qg4kLVHPMOevXr2/5+4YbbuDDH/7wqO2eeeYZPvWpT3HdddfxR3/0R9x777383u/9HkEQcM0110xqLGe4x0LkjUPStqUtj1uRG+u0JGavjGnR71lH0zgX32LKelRe9hBr2vt552k/5vLSQf5438v46W0X0/mN+8EaMj2yiojF/Hw7yx5XyLYKz3adhXyRoVOmHHxmGWf843PoffuxWi8ab8387BeseFSBaH0CslpjTPH+732UrgcUav1JVE9aS8/ZNUqT0LR7/Crr/cOUZYwMdbPPuvFAlTPayw2MESRaYDOB1sNkc53L5kO9vk1QVFxL80C2WUcIdKTQkUImBt81InFMgGVmP/GhXXfu3ElHR0dz/VjeNoAxhosvvpg/+7M/A+CCCy7g0Ucf5eabb3aGe8YsMXncitxTGelRD62bjtG2ilFR+OZYnvXIY8g8P9vKo+VMx9NX65nPfw6cwv6sg5/s20TUr7HpBEbJWmyWYRsx4WH4ysFtrA77CQ8obKOBzbLx912IFO9nUtsYg1WwIhhAI/l53xr2DbaPu1slSPh5eQ2B1Ij9IaLRjwX8qsHuDzkQq3zaQtniWqKoiT7GKcxT+qDDMRkMAjELldM6OjpaDPd4rFmzhjPPPLNl3RlnnME///M/T3pMZ7gdAFhPkJblKAnbKDGtlC8rISsVBVNGRGFPFIDWQiGP6ygvqGKHV0Ebgx0Hu9n5n+uo7ILosKHy8z1MxvSaJGXNj3p5sP98rIL1j1cxg9VJnuQiRSmy7oxf7nyY+2pb+MYDL2H13eMXRkmlYIe/GisFm16IYf9BrNZUHtvHpsYKsori4NkeyTk1lKeJYwmD85zU7nAsAl760peyffv2lnVPPPEEGzdunPQxnOF2AEVt71CQTVHCHheRz2NnpZkdxnpFcZWRbTjHIBkI2Hh3Rvj9B3J5fLL6l9GYhx6j+2dD9VJHFx050bBKoioZ5wV1nkkG6XhG0v7Ne/KpgWPubJspb9lzO/Ce20FQLlPpOQ99gaYcJcRqti4kh2Nume+o8ve9731ceuml/Nmf/Rn/7b/9N+655x4+85nP8JnPfGbSx3CGeziewkoJ3tLo/DUkj9si8GxqEvbY0voQU5XEW49dRCoPyeNjjaHygivSs4R+RuRl+JWUvs0Vll96Dt5ggnh+D/rIkSkMfKKb66OIRoLcGXHDvpfzwMH1RIfMzObzzTgBboVsbvy8FrzQRZR5ET+hw7xnd3P9bCPA+hIb+QhtIVs8MQuO+cFYgZjHPO4XvehFfPOb3+T666/nox/9KJs3b+amm27ibW9726SP4Qz3EEJgygG65C+ZQitWCdKKRIe50TaTlrBz71yXJpirFFM4Xst+eTtOXTZ5/2c1hjwuwQs1YZTiK82q9gGWR1WWlwbZdXUne18TUnu2g5O/4SPumoLhXkKYQ4fZ9O0e7nnoYry6ofvn+9BzYdBEHohoPBAGvJo4GmFezsuUytTi1+Ymwtx4gqwtQJY8RGpQ/bELVHMcd173utfxute9btr7O8M9hBQYX5GV1dIJSisad0xHHs+l9VkseTri2Daw4I3vgnm+phwmRF5GV1hnWVClTcW8bsXDrPb6+OuO11C/cy0zVOpPWEythrj7Z7Tfnf89V9Xe7FBMg5e3/7RFfZq8+l7+byUEXmPuIsxNIDBIlNQotUR+245JY+0Mo8qPg4DjDPcSoCmJj/i2JyOP5+0xxfQDzCZ7jso2I5HzIDQnZy54hMDbsI54ywoaFY/qWoExgjj1wCw8A2llnuYpAbR1vbsdwPzPcc8GznAvBQRkZUlaEqPWH0vONr4gq+Rz1sM5ZsGUKWI9yMomN9pFMxHHwkYoxZGXnMTBN9ZY2dWLrEfoakjW8BGxXHB1yodkc1HxkbHG67OQLrKUP4cDZ7iXBoWBHpWaNQmszKPDpzVfPZVxxLHlcccCQ0jqKyRvPO1htrU9xVf3v5h7nt6ETSRyPjutTJZCNgfA5iqUcNPdSx7ncS9GfA/rq0JKPnHcvFxyFhi/CDwba25P0Hx9vGNMx9hP+hyVbVZkM2OUMB2J8CzCM0hpKYUJnVGDUGUsC2qsDAZoUw1WeP2sVv20Bw1qSyAz4HgjtOVwUmFv2kktC4736TgcU2a+o8png3mxVJ/85CfZtGkTURRxySWXcM8994y77ec//3mEEC1LFM1RTqgQmJJP2h2RdkXo0gkUmCYgiwRxuyRul3kg2Yi3ZpQgK0PSMfaSthd52HN0lQzJ41mnxpTMxLnaAlSgaWtr0NFeY3X7AOsqvWyqHGZraT9bw32cHu7hdP8gW33L6mggT0FyzCkygV21Tp5urKQ/jhacPO5wHIuh4LSZLPPNnHvcX/va17juuuu4+eabueSSS7jpppu44oor2L59OytXrhxzn46OjpbKMmIOjamVAuMv/rztsR76rBJ5/+gJelwbNbrhx7wg8iIn1p+kPC5AKkPoZwRKU/YS2r0GJZVSljGRTIlESiggFD7egtRqTyCEQCiJsBBrj6oOSc0YreOEnbRkM7SZi0t0OCZmzg33Jz7xCd71rnfxjne8A4Cbb76Z73znO3z2s5/lgx/84Jj7CCFYvXr1pI4fxzFxfLQH8sjWaic6QxHjJmiVvK0A7Q91acol8ZHz1Hl7zHk9XayyGK8Y+xglTCGXx6VvkNJQKcV0R3VKXsrJbQc5o7SbA1k7X95xCbv3doO0BFFKEGRUn+3k5F2N+XlTSwx16sn0nb+CuFPQe5qlA9jfaKORek3vw/oWHR0triKz/CFZhwD5eplahD5atc8qmTckic3cFGNxOMYg95pnMsc9iyczSebUcCdJwv3338/111/fXCel5PLLL+fuu+8ed7/BwUE2btyIMYYLL7yQP/uzP+Oss84ac9sbb7yRj3zkI7N+7osJHQqSNjnKCA/1zrYCsgj0iKTmoV7b84lVeXEVGxQpX8cYXwaaSjkm8DQ95Sonlfto9xtsa3uKS6N9/KC2jgP3ruK0b/SCtqDy5jCidgD2Hpiz/OQlixD0n7Oc2tt6uXj1To4kJQ41KuytdlCPg2ZnERsatJ/33FZVVRjuohhLACIDf1CgdJ4GmFTyeUYVW2QmEMa53Y75YTEGp83pbfvgwYNorVm1alXL+lWrVrF3794x9znttNP47Gc/y7e//W2+/OUvY4zh0ksv5YUXXhhz++uvv56+vr7msnPnzsmdnBAgi2WRMGSEhy8ImlXPWhb/aLlQK4fKmhatMYvFenlva8Qc5U0LO8YCKHJ5XB1jXAFSGgJPE3oZZS+h4sW0qZhIpEggsYrokMA+9jTm0ccxP/sF5qHH0E88jV5i6st8kbRJLlz1Am/ouZ/T2/fl67TCGHH0e5Z2zO946FodXsp2qJtYvn56TW2mzYkS0+JYUiy4qPJt27axbdu25t+XXnopZ5xxBp/+9Kf52Mc+Nmr7MAzH7Xs6HjbwsZFXFGRQi6bFoFWiWZ60iRB5BariPRi/6Mg1LI0r93TshJHbwghkYhF6dj4MK/PSpdZrHdN6RRnTCRCeRQYaKQ1t5ZiecpVIZZzSfoBzyzupmpD/vfPVbN+1CnsoZMP2dHLNMRxzhhSWIMgwRmCNRKcSmy7cLA2rBLoSIkMfkWpEnIJ2+vxSxDKzmMrjoQ3NqeFevnw5Sin27dvXsn7fvn2TnsP2fZ8LLriAp556anZOSghs5JF2hEcrgi2Sp27jF7XFRwaTiaOLCSCtDEVsW7I2Pez1CQx3KhEmn2OcFSSYyGIiM3rcY8njvhlTHr+o8iy/VNrNo0k7T9y/gVM/34scOIg90oc2znAfbyI/w1caawWD1Yg0W7iG2/gS2+6DsXh1jZdqZ7iXKE4qH0EQBFx00UXcfvvtzXXGGG6//fYWr3oitNY88sgjrFmzZvZOTOReq5ViwRjtpvwtx1+GZPGhoLLmMrRNs9CKRQcWHRmIDKKkIdTg23zx8v7WwxerbDG+nZ2lONbIcVCFjDrsYWPkIoTFUxpPaQKlCVVGKDM6ZIN2GRCJFK8q4dmdZM/vdJL4PCMzy4G4jZ1JD/1FoXslDb7SRH5G4GmkMq3f61QQR6/nubwnWpnXbrBSYBfIfcDhmAxzLpVfd911XHPNNVx88cW8+MUv5qabbqJarTajzN/+9rdz0kknceONNwLw0Y9+lJe85CVs3bqV3t5e/uIv/oLnn3+e3/qt35rrUz2+CNCRzCPBx8F4o0uPQu6J6zA34kmXJenR4BvaltU4qbMPKSy1NKCW+qSZYqAaoWsjvnoBumRmL/dZFqleI71tZVFhLoOPRxDoxRR6sLSwlq4nqjz7nS18Ytlm0lUJW9YfYEWp2twksxJjBf1AlnqYRvHUOQmMB2lZIkJQqUU1jMsNd8wti1Arn3PD/eY3v5kDBw7wJ3/yJ+zdu5fzzz+f733ve82AtR07diDlUcf/yJEjvOtd72Lv3r10d3dz0UUXcdddd3HmmWfO9akeV6wUZJEgLR8rP2r0KuNDVs7/nyzTtK0apBImnNOzh5d2PonE8Gy8kn1JB4eTMo/qNVRHGm5lsZGd3WtwrGJt0hJFKaVg/FqTQlikS+ZduDz8BOufqSCiiD3/ZSOsh3Xl3ubLaRGEIYWllvgM9Pt5sPkkvlKraP4GvDrI1EWYO+aYGUrlxyNIal6C06699lquvfbaMV+74447Wv7+67/+a/76r/96Hs5qfrHHkA2HS94THmeM1/Oc6CIQTOZGT0lDSSV0qRraSmomYG+9nWoaYm0eAAZFDqKdektFIQvJG8AI7CS7QQllkdKgJvC4hwx3nq3mbtoLDRvH6DhGeB5B/wYODlbYHXTQHeTtVUMrSLVioB6SxH4e8FhMnYjm9TbOsYf/RubrfjjU1EZJMMepFJbjuOHaejrGxAowoSAL5YRVzCaSySHvbqSjEUVTRF6WNG0zeQU0AbVaSKYlA10R2kr2Zp1849ELaHughA6gdmrC6rVHsFZQjQPSdGpVWISArrYaayr9SGHZU+3gyGA5Twc6Bkrlc6ETEXkZnUEDT+YV0pzxXphYrel6bIDs/+3iqc5uGi8e5I/O/x4Sw7d+cR5dP4ooGaitEsTLTJ4llghkcrzPvBXjSXR7BMYiGxmykbhANceCxhnueUIHkqRNzCgc0KrcSNthUeVWgC5ZdNvRnNms7qEzyUCap8kdTNspP1LipC9tx65ewROrujj77L3ERrGn1klfY2q14JU0nNm9j1d2PY4Slv/sP4WfsZZUT+4BYCIZXAhLqDJWRIOEMkMK42TzhYq18Mh2VmwPkJ0dPLFqM2desosAg9gTsfrfngch2HflBurrckMotCo084UTxGB8ifHy8/GlQMauZdhSYjFGlS8dwy1ELoUJgVWzW3hlqALZuJGpYlhU+HQibIcKrShGH0MURVSKf+cnlC9SWCKZ4Audl5YcrCLiThBQ8WJCozis8trfAHqMC1gIiyqOW40DqrUQawVPSpMfQ2b0JiXKfkpWpAJlRo57vPGOPYQUlkBppDB4UuMLTSgzfKHZkfZwG4IHa5vwqzhJcwFgswybZaAUQa/gzurplGWCsKBXdoMq4jaKmuULtmaCiypfusw0fcEZ7rnDhh6mHGCVxARyVj9r6wnS8uiSo83Xi1rh03EyjBLo0tFCKpOuLS6g02+wxTvMQJhL5Mj8wcVEhpOjA6RWYRCUvYSG9umPIxpZ6yUxJFtLYXhw5yZ6fuoR9lsapRI/qqxBl2DwvAavO/MR2lTMobRCX1oi0Yq+pEQ1Gb/VY6g0HWGDSLV6OJFK8YVBYdgUHeSscBf9JuJj21/H4D3LCfpgzX01bLLANNcljG3ErP1xnS/Xr8gDJVcanvz9ACHApDFkIl8cDseMWTqG21dkFR/jz37quvEFWUlM2GVr2g8KEnRI0Zxh8ghh6fDqrPUEB8yRowZfSQgM64NDNIxPzQR4QjOQRVTT0UY2kJqesIonNd4Rj5V37sM8uwNE/jmqnm6e6t7CKRftZ61/hGeSFeyKu6lmIQ3tUxP+uF63r/Jjt3ujm4EMyeOrvT4uCKvs0wMceqabMz77PPrAQWyaYV3RlQWDTRPkfz7Mmns81OqV/OL9a/l/X/Z/kFhueOFq7n1i8/E+RYdjTFxw2glOUxIfIbMbNTvFIqwaLbcbb+zjWmkLeX5Ihi+uHplHfMuirGhsDTUTNoPGRZziHfT5xoGLyaxkX62dahLQSD1q1QgdK/xSyoruAdr8hM6wTrvfQGHyqOBM59Lo0Hk0GoSHBd8/cCZrSv2UVEKbilGeIY4UnjSYQj63VmDI/62NJFQZlWL78XgqXsUh3caupJvwoMJWa9h4/O0dxxGjsbHGVmsEhyTf6LuISKbsHuw83mfmcIyPy+M+sbFKkJVlLjsPwygx4/aYVpJ77cHo9XaMb8n6ef1xq2yRBpavl4EmjFJCPzeuz6QRTycrUSlgDPbgYTZ+bxlPPHZGHuWbWnwNgbF0pRaZWQ6dWSG64ggvX/EUoUxplw00EhNZrGpVLEy1ztofDXBw30ZeWCapv2yQa8++k3ZVp9weU5ExDROwO+2mT5eo6YADSTvVLKArqLOldIBOVRvzM6mZkE9vfxniJ52ERyzrH6thamNv61g42MEq6/+jzr++8EqykqC+2kJPduwdHQ7HpHCGewpYBVkoyErH3nbKx5a50Z7ssa0krwPuDUtbEXm6VSVKKPkpGsle3cnBtB1R3Dd1fz/qjgfpnuDY3f6L8K/UXFx+trmuakKsb0YF8dg0gXsfofM+Qc+GdTy+5SSWnT/ISd4Rtvr9rFFl6raXp9ID7NUd9Ooy2+UaDqUVevwqa/0jdI1juA/pNurPt3P613aS7XgB7CwXiHHMCabRQP74IZb/p0CtXMH+q0/mULeb33YsTFxU+QlC3mZwtBdtvNlvOZgXXhGFTD4LBxwWpZ0ZRb+OMAjqqyzxK87G70/wnt9PtmfstqoAQV/G9udX83/Uq5rrEq0I9/qIZJxUGWuxjZjSPsnnX7iUzqDO2lIfPX6VmgnY0+igNylT9hJWR/10eA3KKkaK0fmyvsiIREpiVf55u6IYCx+p8NavJV3Xg/UEqpYh6ilZR0jafjSqfCpYKTB+/psTGoS2c9J9tmVMT2ADD6REGAOZi6NYEiyy24sz3GNRyNZZJFsiwYeaeMwWQ/K4Dsk7e83isY0VHE7KPB2vQlvJy175KAcuaeOp/ctZ/rWNVL55AMYJ7gof382WL5/E4a6NzXXCwqadA5iDh8d/P339rPt+H9Xta6kr2O0JTJG2K7P8xvvsZsWhK3fyP066m0BoIjH6QaBL1lmrEpbJBjbSi6pn+lJFRiH7XrMO+/pDdJUaPLOnB7mnDQBd1vlUzhRtoPYhbpcIA15s8at6yseYKtqX2K4IoS2qniKrsSvG4lhwOMM9DsYvqpTNpc0Q+fy1nlr9k0lhgYb2OJS00eXXeMfKH/HS0PBPa3r48x+/lTYpsOPcj7I9e/H27B11cRwrhsM0GvDgzyk/OP420WUXcviVJVZ6A+NvIzTLVIgvkryTmcuxXfAIz6N/C3z97C+x1bd8ZvWZfGvleVQTn4HBEmm1KOs3BayCoXotwoAnBGKOXSPrSbQHWIvQBlk95i6ORY6TyhcaSmJ9lRdd8dWERrgpj3t5sJlRc/dl5O04c290rG5fU0aC8Exe0ESNvrEZK6iZkCPmML26zBjq9Lzh9cX0PbGMj/lXjbtNexCzsjRAQ/uUng2g4aLIFwNCwyFTZpnpozYyytLhWKi4qPKFhfUVWUeE9YueuxMZY1HI46W8nriZrTnnMcbRkSCLplhQZQJkoCmVE1TRvGNkAw+NZH/WzlOyxrPxCmRiscep45J89gW2/uN6kmXLxt0mloLnvbwr1MYXDqOP9M7fCTqmhbUWFQueTvKuf0fSMmbBlklzOIYznabxI/efX05ow42UmEBhgslFlFlV9LWew+9haJ7czOI4UlpCP8VXY7vSQx73IVOhmoXH1ePWvX1wXx8T1Kppwc0uLh6EhiNZhV5VIZ6oGpHD4ZgRs19GbBGRG9Hcy85KclaDw0aSd/bKl7kcZwgBRCqjJxhkuT9IWTq52TGHpCntOwyffuDl3PDo1dyzfwMlP6U9TJDKLKSeIg5HK3YWlinw4Q9/GCFEy3L66adP6Rgntsc9CXQkSCqy2Qt7ruRxE0BWPiqPz7WKKIVlWVDj5HAfkUyJRIpabDkPjkWDiWN67txJ1xPLSDojdr6mgw2XPoknDYdrJWKmWLPX4ZgvjsMc91lnncVtt93W/NvzpmaKl6zhHuq4ZVReY3y287NHjaeYvXEmmcwaypQO1SASKRpBYhWplYsuZ9GxCLCW7IVd8MIuop5lBC86nUBqSipt6QDnOrQ6HLmhXr169fT3n8VzWRRYkRdZGCrsYKaepTJpjFc0HpmlcayyucwuQZcNopQhfYMfZEgx1MtaE6qMkpeyMhhgrXeEARPx97tfxYPPr4cDIRt3poybC+ZwzBRjkSkcjssESlNrBIi6QmQij69wQWuOhcQstfXs7+9vWR2GIWE4ttL05JNPsnbtWqIoYtu2bdx4441s2LBh0kMuOcMNoANB0iaLqmVzdx9pyuOzNI71QLcZrGcRpYxKRwNfaZS0zUjysp+wqjRAu9fgtGgPp3gpj6Q+D92zlVP+cRDZ3w8HDqNdJTLHXGENXgP2D7bhKUPSH+IP5IVUXLShY6ExW93B1q9f37L+hhtu4MMf/vCo7S+55BI+//nPc9ppp7Fnzx4+8pGP8PKXv5xHH32U9vb2SY25ZAx302gW+drGm51UrHEpunbNyjiFvpirBRZ8g/QNgZcReEdLSUlhCaRudtySGDSWARMRHpKIx55BuyYdjrlCKoRS4Oc53KlWaCNBC0QGwjhP23HisnPnTjo6Opp/j+dtX3nllc1/n3vuuVxyySVs3LiRr3/96/zmb/7mpMZaOobbE+ggl8d1wNzK40FhtGdhHKtsPjeuLLpiCJfVKYUpUZDSEcT4qtVwryr1szbsJbWK/7PzVfzBvuWkfSHrfqGxqevQ5JgbhOfBOafRe1Y7aUUwuN4i4rwnrUiFm9t2LFxmKTito6OjxXBPlq6uLk499VSeeuqpSe+zJAy3FXkJ06StSPka6mE9Bxg/l8dnaxyr8vlsGxj8joStKw+yKhpACosUpiXwR2JZG/ayNdrLC0kPT9+3gZP/sR/V24s90otOk5mdjMMxDiIIOHBhB+nVvXSX6zQGKtQHImwmkKnztB0LmFma454ug4ODPP300/yP//E/Jr3PkjDcMFTSdA7l8RYp/hjjHMP9aJH1JVjfgm+RymCsIDYeocwoqQxftHrc7apBRcb4QuNVBfLZF8h6+2b01hyOYyIEWVmwsfsIa0p91NJ11PpKuUzu5rUdjiZ/8Ad/wNVXX83GjRvZvXs3N9xwA0op3vrWt076GEvGcM8lxhN5JTQ5iZQvYTEBGN+O/6AmwYQGlEWEmkp7TClIOdJX4am7N+IPCmqbU1573qNsqRxo7qawbAwOsMk7grYSE1gQS7rGjsPhcEyIsDNLU5zqvi+88AJvfetbOXToECtWrOBlL3sZP/nJT1ixYsWkj+EM9yxgPchKR1O+jtXMRAcWUzYgx/nGPUvUHlMKE8pBytq2Pjr9BrftP4ONtyWEP9/JvtefTHKOx5Zgf3M3JQybvCNs9hSaw7iqkw6Hw3EM5rkAy1e/+tUZDJZzQhtuK/JgNOQczWkXRrpZdU0CYgJPmvxc8sWONtxFURjpazxPE3ia0Mto92K6/Dwa3B9I0PsPEPZv4YVqF0+1HU3il8LQMAENe5Cn01W4KqeOuUaWy4j2NkSljC7l6zKj5qTVoVXDyg/M0EuaFEKALG4cZoY5Q46Fy3Ge454OJ7bh9iVZmyItSYwnZtV4W0Wz7vjwlK+8gYgdfywBJhijdrMEFWo8P0NK24wW96VmVdjPSeER/FKK8STCWjof72fvv6znHzqP5g5aAdYHHeadmtbck2Lr9dl70w7HcKQiftmZvPBLPmmXJuoZINGKPfUOGuns3VpylUqQVvJccJVYVMPMbQVAITCBJOuMEMYi6ymikTrj7VgQnNiG25OkZUlaya3kbD4YWZW35jRFqt7Qsa2XR4EzRl9sYFgQ20hv2xKEKW2l3E2WxeuB0qwM+tkUHCCMUpB5jqx99AlWPxnmebMjkQKMxcQxJnZut2NuEL7HwXMCPvD6f+Gy8lP8y8B5fG/vWQzEIUnizaphtR6kngALwSCoebisdajQgURoi28sqpHO/aCO+cf1415gCDG2CjKJNK28NOPo9bkcXnjZHhjPjnjd5sZ5vPnrYvyRrwtlkdIihUWI4v9AIDMUFkW+3op8d5tl2MzlZTuOL1ZCu6rTLgUKS6oVaaaws1xsZeg3LJhnZVKIXE2TIl+087hPOJzhXhzoQKAjxi+OYkE1QMUjjLKCrCRyrzqAtN2iwxHfmjfG3PVwBMhQE4StRlfKvBIagBKWrqhO2UtYHlZdS07HwsRYvCr8uP9UGsbngf4NHByokCYeOnXNbByOuWJJGu6hKPDxcq2FBpkBI+yllQIdgo7ABHklMxuMkaQ6UdSMAM/XtJXiZn3xkXjS0BXUWRZU6fZrRNJJdI6FiVe3/KIvD5B8rn8ZjWqATV0KomMR4TzuhYVt1iVvda2NoultW2nzKO/h+0nIIoEV4qisLvL9snJegtQEFusbhD/J6hIyl7qFhDBMCb2sOY8N+Zx2qDJ8pQtPe5Dl/iAN4/PTgS3ExmNgfxtrG3XnyDiOK6qrE1b0YMshcbfAWMFAFpEZeTRC1wJG5E1FXDcwx0LGRZUvLKwvSdoE6YjysU3DXRRD0dEIeVtAsqJYpyyqlOH7GmtBF/N3QloiXyPH8ZqHIwSUw4T2MMYThpKXEnkpcpgJrngJZ7XtYkuwn0Bo2mWdSGR86fClfO+2i+ncDhv3Zcid+9ETjOVwzClCkFxwMs/9SohZGVNqy6vy7a52Uo2DPOjagoglspFHgQt3wTocs8oJbbiNyueys9LYr+c1zC020q1et8oLoLSXGwRKs669l+VBlaoOeKHaxUA8dteX8RDCsqJcZV25F2+cu1iPX+XS8pNcFCiUGEr2DvgH47HyXkPlm/eBNa4dp+O4M7A+5NKX/ZxfXf4A9wxu4e6DmzkSl0hT1fQ+RCpQjWLWyHncjgXMfFdOmw1OaMNtlWjK2sOf/Iciwq20mMhAYBDSIpRFSIuUhvZyg2WlGtpKnj6ynEeTNWSZJK4GECtEOWPl8n6WlVrbZHrSNI3z833dHN7fkUuG6+Dczl20q8a453vbwNn8x4iIuTue38qa3gyMc1scxw+1vAezcTVZe8jAekFJpTSMT3qs4v/OaDsWOm6Oe2w++clP8hd/8Rfs3buX8847j7/927/lxS9+8bjb/9M//RMf+tCHeO655zjllFP4+Mc/zq/8yq9MeVzrCbKKJWs3yFig6nnDg6xiEctilGfwlMHzTG6so5jOsEGkUjZUjrAm6OPB/vW88KP1rLwvQ2iLTA1SZxw+PUT8F8urVmxvGXOZqnKSfwSN4A/2vYk1/+ERDGie/y89XLz1WU7z9495rj+qb+XPf3A1K37aGo278lBG9PguXOKX43gSn7eJZ94mOG3THs4Pa/T4VbY31nAwact7bjscjnljzg331772Na677jpuvvlmLrnkEm666SauuOIKtm/fzsqVK0dtf9ddd/HWt76VG2+8kde97nXccsstvOENb+CBBx7g7LPPntLYzd7YocFYiWoUed2hob29QSk4Gq2tpGFVeYC1pX4qXszm8ADrg0M8VVtJ2/OW6Lv3t3i9Xf7FxMJyTvRCy5hrvT62ehKDIU0VXQ8egL0H8F92Fpu8Q5wRlMc81wfimI7HFd233DMqP9sZbcfxprra5+0X/YiPrPg5D8Ux/zZwHvuTdqpZgHFetcMxr8z5o/InPvEJ3vWud/GOd7yDM888k5tvvplyucxnP/vZMbf/m7/5G375l3+ZP/zDP+SMM87gYx/7GBdeeCF/93d/N+b2cRzT39/fsswJdn56E1rj5rAdDodjvijilKe/HIdznlPDnSQJ999/P5dffvnRAaXk8ssv5+677x5zn7vvvrtle4Arrrhi3O1vvPFGOjs7m8v69evH3M7hcDgcjlE0UxhnsMwzc2q4Dx48iNaaVatWtaxftWoVe/fuHXOfvXv3Tmn766+/nr6+vuayc+fO2Tn5kbi+1g6Hw+FYACz6qPIwDAnDsdOzZGaRSZFTmohmK0ARSwYHI+oqQHkGpfLgNAE0tE+k0rzgiYkYyEIGNwgqv3xhEZxmkdrQu8UntILHGie1jLnX62SvdwSDxPc1fectJ9iyjHSZ5rmsh1CMHZxWNWvpP00T/fqL8jrpBdFhTflnO8n2jP3g4nDMB+V9KV968CXcv2kDy8Iq66JeevwqA1nUUkjI4Vh0uKjyVpYvX45Sin379rWs37dvH6tXrx5zn9WrV09p+4kQmcWrCoxfFIIoDKJXFdg4wihLWjKIsgZpqQ5G7JMdSGXY3d7JstJKUqMoX3yQA+f6pKkirQaQ+ohSzEoruO3A6S1jetIQyDycrBzF7L4iBC1Ytf4w91U386S3auRpAmCs5J2vuJP05a3pNV/bfiEnffokPGe4HceR6KHnOLVvLfX2tdz30pC2//IAL+14kkEd8pxcdrxPz+GYPs5wtxIEARdddBG33347b3jDGwAwxnD77bdz7bXXjrnPtm3buP3223nve9/bXHfrrbeybdu2KY+fe8jk3vYwZAqkeSlTK2Te4UvavEojgIQ+aTFWEHkZpyw7yMpwgGoWsqPa3VKA5XB97ChxAF8ZVq3tbRZgOZxW6M+iMbft8au8ruMhzg+8ogBLzoGknYe7zsMTQzVanXfjmH/0wUNw8BCeELSvewmDWUBFxviuLJrDMe/MuVR+3XXXcc0113DxxRfz4he/mJtuuolqtco73vEOAN7+9rdz0kknceONNwLwnve8h1e+8pX81V/9FVdddRVf/epXue+++/jMZz4z5bGltnmXrxFKulV5ARYsyFRgG7I1NFBAIynR6I1AWfZEnfhBhrWCLFVYKxDC4k2h5Kk2knrmo4QhUtmokqd1HXCXdwp79T58NB2yQSQyApmx/yJJV9tLqOzLKN3/HPrAgSl/Fg7HbNH2QsJdd53Jf646mVI5YVXHAJ1Rg3riEwsLCKxv0ZFAGIvMiuJHLm3MsQBxldPG4M1vfjMHDhzgT/7kT9i7dy/nn38+3/ve95oBaDt27EDKox7mpZdeyi233MIf//Ef80d/9EeccsopfOtb35pyDjeASA1+Nb+RHF2Zl0DNKoAVyMQiM9lyTxEavLpAFd3BrAxy71yBHNZkpFHRiLG6g41BTYYckm0IaSmXYzqiuGVuUEnDzmoXgTqDSKWsLx2h26/RpmIuf82DxL/k8YOfncFpB1eBM9yO44W1hA88xam7VmDKIbt/qROuHGBdpZeBOKQqSnmgbWDQvgUD3qBCOcfcsVBxUvnYXHvtteNK43fccceodW9605t405veNONxhS0C1EZUMJGBaH7Youhg1OJwa/Dq4FVbvxHj56H/Osrny20gx3YixnwEE/mQEmJliD3d2tZTK+qpD0DkZQRSk1pFt1/jpR1P0qMGuXfVenQUzn3yvcMxAbq3D3r7EH5AeP5FSGGpqARPmqPuiwKwoEXege+4ZLs6HCcmiz6qfDqILDfMdhwLKEy+zej1FhXnZVNNJrBKorMRNyQvb/fZ0m1sOEWHscF6iBhm4IWwhH5G4GkyI+lNSiRGkVnFhuBQcSN0OBYWWUlwWsd+Xtz+NL1piUMDFdJUYVLp+nI7FgfO414cqGS0Fz4SMYYCLjR4tbxUjg7zft0ybd1Gh6CVmNhwNxQ6ab2pCWURbRB4Gm0FRxoleilhrKBWnlo3ModjXpCCrAKv6Hycy0q7OZh18Fz/MqpxQLUWkmbOcDsWPm6OewEy/Etpytp2+j2Chwy6zIrodClaji18iie4CaTBMZ7wLGCMyOs+W4EtTjoxHhqBRuRBcUMSv+chggDUaFdcCIG1Fpsk2Die3ht1OCaBMFAzIdWiVG/oZSRaIcZ7cJ3uOEOHm+FNdspYm49nbL44HAuAE9pwi1Tj1S3YvOGIDseXx6eMARXnBtx45M1MZP5AoGqS8bodDgXu4NF6B7KCJPHoNwKlTFM2T7Rif9KBLzRxw2/ePMRZp7BvWxdxd+sDgvVABxaVCFbelxL94GFnvB1zgk0zeh5N+NN//TU+0pVRWlZnw7IjLCvVqMUBySxOa4sMvDhvz6sSOy/ypEwMqpHlkfGJi647YZlp2dLjkC1xghtugz+QoRJJWpYYX86a4ZaZRRStuHUkSFVuuGUKYuS8d8uOkFUkRunWeB0Duu6hhYf0NbLNEnia1Cj2xR2kVpHWfWQWY4Wg7/ROuv/rLl676hdHDy0M6/3DbAn282Symo8G/42T747QznA75gKjiX70GKf+rB3bXmHnG1cTXHWQFeEgh2oVqpRmZRhhc2PtVw0yszOfk5wM1qJijdfXAGPyB2ZXQ+HExM1xLyyEzZ/QMWPPWc/4+Obo/5uV2YSYMIDWkqfI5Nr42BsaIckyRaIUnvSo6gAvy5/4szafoGcZcafgtLYjnBHtau6nhGGTd4TNniISuzAhrsa6Y04xtRqmVkMOVFD1vLqhJ3VL4OVsIfT0p7imhbW50dbz0xnQcXxwc9xLlKEodZNQ5HhPELFuQSUCYdX4CotUNGJFQ0X0Bpq+ekQpSJGBZuerI/xLTqO6JSWQGc8kR3uaKyzaSuAQz6U9owLnHA6Hw7H4cYZ7FpCZRRaeQFYSpN4Ec+lF0ReRHmNeZDCfJDeBx2DdYzAwhG0xmy7dwbKwRkmlhDJjZ+Noneihgi5KGHamPUVjFectOBwOx7g4qXzhIopIciEAMYtBakMMi3ptSvPjxTwMiw4/FkLnRt4KgdESTxpKKkUKS2w8MnE0Ck5iqZmAqgmJjU9WttiNa/E6O7BH+tD9/TN8kw7HOFiLV7fs7O2imgbU4gAhLVZOPHXkcBx3Zpqp4Az33CAsyNQSVA1WCnQAWSRm33iTB6f51fzBwATMOJK9GaUeC1Ib8pRazu6og9DP6AwbebWqobGFJbUSbSWpVWy6+AWeXLcc09vBulvXUP7OQ9g0mYV36XC0YpOE5Q/2cyTp5kilm3gzeBuqEAqyRgRVZ70djrH48z//c66//nre8573cNNNN01qnyVhuKFo8ZnlxVNoU2RzVNOkWWJVQEr+kDAThBYok5eMFFoRUyL2Q7xSRtKuCLyj0TpSWCQWXxg6vDr/3w2385JTD/CTxgqu3/UbVP7Dc4bbMSfYLIMHH6f7YYVc1oV901bSrXlZ334/xIrjE8TjcByT4yiV33vvvXz605/m3HPPndJ+SybkuBk5aGl2LJLZ3ESbw9A4RyNhZzROkWcobJFqlkp0Kkkyjzj1yHTescxYQWIUde0zqEMMklBIulSNeJnBnrYZdcoWVFfnrL1Nh6OJ0fmDYZF+GHh5LQKUxXpgPVvULXc4FhB2FpZpMDg4yNve9jb+n//n/6G7u3tK+y4Zwz0cFVuCQUPYb/Aads48AZmAPwjBQP7vmY4jMlCDEjWgYMCnOhDRP1ii2gjQJv8qa2nArlonO6rLeDJexfbUI7WKs1/0LE+8L+SJ31lJ4+KtxWS/wzEHCEkWwYpKlTXt/fjtMVm7IatYrH+8T87hmBv6+/tblvgY9TPe/e53c9VVV3H55ZdPeawlI5UPISwwTDY3SkHEnATQDJfNrRSYYGbjHJXNQVtJ5nlo35ACJkywVtDIPBqZR6A0e+NOdofdRCLlXSfdiV4r+c6R87j/sfPpERKsqwblmAOkwPjQE1UJVUallNBbCrCpRCaq+BG6B0fHwmC28rjXr1/fsv6GG27gwx/+8Jj7fPWrX+WBBx7g3nvvndaYS85ww7Da5YDUFpnmgWpDPbfnZEyTB64NjTHtgLXhBdeH/W8sYuPTryMSqYhESiA0vjDunumYfYTAW7sGvbqbuDMi6bQkRmEQ6GGpFcehOqTDMS/s3LmTjo6O5t9hOHYg1c6dO3nPe97DrbfeShRF0xprSRru4ajYEhqDlZCWJDqag5uLBRmDr3ODnZXAhHN7EzNWcDgp82y8kkimrPGP0OMNzt2AjiWNDEMOv2IDe1+bErUldJUG6EtKpFqRJEv+NuNYAnR0dLQY7vG4//772b9/PxdeeGFzndaaH/7wh/zd3/0dcRyjxmgeNZwl/YsSFkRqkanFStB+3oVrLpCZhaxIE/NFXo50DrFAQ3scSNopqYROVaMHZ7gdc4Tv079J8jsX/5Ct4T7+b98ZPHToJOqph9HyuOS6OhyTYp6jyl/96lfzyCOPtKx7xzvewemnn84HPvCBYxptONENtzHIRCNsXgjCeGLCoCyZWVScy+ZGzY1sLmzREjQpgsUnqrI2SYwRxKlPpg2e0viqNYRdCktZxvTIKhUvnpP89cmiujoxJ68j7RxfIrKewKg8ir60sx/z1HOuw9kiwCroVDW6VJXQ1dt1LBLmu1Z5e3s7Z599dsu6SqVCT0/PqPXjcUIbbpFqvP4GCIEuB9iKh/XGMdwWvLpFpRqjBGl5jmXzwvvOyqCnN83RxKSS2mCIkJYwSpFRghpWmEVhWOkNsNVvsDk8gAkEQorjUg3VbjqJp97SwbIzD475uhSWSpCwujRIohWPf/8UNv9DL9neffN8po6pIIRAh5ZTwr2c4vfR7deaJXgdjgXPIrtUT2jDjTaIorOP9CSi7I37/eQ50rmcLZVFh/MgmyswoWDGsd1aYLXACsi80Ucb8ri7ZYkuVTuuHnfaFdF5+iE+etq/jrvNeq+Xrb7HgEl40Zb3QDTH8wqOWcEq6JE1lkmPsnSFfhyOyXLHHXdMafsT23DPAJlaVKO1zrIVYLxZlNBtkZvdIE9Nm8VjCyBSGT3BINpKvrD/ZXwirvDEgRUs262xZvxHTG/1KuIzTiLpHHZ5WCjvqiEeewZTq409ZhgiTt9CbUM7xhPF+xG5ypBZpLb0bvFYVT7A/qydQGjKMiYY0auxYRWHdcyAFZAJ1wd5EWCzjPbn4Lcf++90R3We3rsC9uQRmKZsIJi6vCP00SJJKs3b9M41Ist7fgtrkYlLl1wSuCYjJwimkM1H/HCNJ0jaJHq2jOvQOHFu5LIys3ZsgK6gzsnhPp6OV3Hnj85h/e0Z6/oT/GefJzPj35Ti00/i6f8uOWPLC811qVE8/8P1bDnQg3l+bMMtuzrZeUU3PZfvpiess6rUT49fJTYeuxpd9CcRK72EtaU+nmisodOrsTE4OMpw95oSDevRa8qIhoIJHjIcCwPTiFn9H7tIftGD9to5uZ4h633otpB9Ly7Tf/rUDbdKwa8WhlST97GfY1Rq8PpiyAzC9eJeErh+3CcIw2XzUa/N8u84L4lqp9QxbEKKSXkpLIHM6FANJJbyHkF0xyOYRiN/W2MF6RUeTdLlcdrmXfze+tubL1VNyB+sfTM2GKf0lRCIIKC2xvDBjT9ivX+ILd4g67w2Bk2Dp1LBXt1Bry6zvbGGQ2kFX2jMGLp9aj1S69GrK/nnLYugQud5L1yMJntuB/K5HUdXCYG3fDn+6VunFSwijEUledbHfCEyi0gyyJy37Vi4OMM9BYQGlYwukWpUXilqRl3AjEUmAo8R9zjBmMcWBmRDYpXAeha8XO/RWhYlUAUKw2rVxy5/AFt806qjg/SCkxlYn88byzQvQiMMeZ9wbTmy1cM3igdqm4hkSptqkFqFyOQo4yn8AM49hb5T22l0S7y1gxzO2tBWcCDr4FEZ07Bt7Eq76cvK1EzAvriDahZgrKRdNaiOkxtXMyHRhgF2/tf1hEfW0f2LGvKBx12E+QJHhCHm4jM4fHqJrCyorbEgLRhXfcWxAHFS+YmN0Ba/OrryWBblEvrMDHcum9tG68FNAGkF7IguYyIFT+d9unVkMRUNCkyiaGSSNMi/2i1+g16zH+0DUiKWL+O5KyMuevl2MiPZX2unmvjEqU+9GmAShVeq0cg87jxwCp1hnbWlPhQGWRfNYL8hZKXErld0subq5zm93E+712BP0skOu4x9cTv9SQmDIDMSbWTz39YK+tM8nL7Dq4/5mUhhedsp99F5Ro1dcTff+ueXsenxMtoZ7gWNbG/j+deU+bU3/IhIpvz77jPZvXsZJM5wOxYeTio/wWlK6COQvjjmlz8ZpVBoECMf3+TYErowAgwIYTFmaAALBqzJ86ABQiEpy7j5sGFDn2xFyltW3kPD+DwVr2J/0s5AFrFjsJtqkj8haCPpi3PD2uk3kMIgjMAqifCKy0ZIRBTR6LFcvfphVnt9PJOsYFfcTTULOdho40ijhB3nzftKU9UBcoz5ByksCsMF5ee5vLyPfdrw1ZXbEOUSYrCK1RommKd3HAekQiiFKJWIV2je1vVTlLA8OrCW3Sw73mfncJwwLBnDLVKNqmVIJbC+RAdy1jpkydTi1S1mgpoTxi/k9KkOaUDFuVG3MvfAJ+PZWyvoz0rsziy7026a8V/aQCrZmfSQWsXuuIvDSZlG5jc7jA0nMYpDcRkpLFl3xoFXrCY8fyVpSZKVISsJ0vUNttdWs0t1cyit0JeWSLQi1t64Rhsg1YojcZm6bp03L6mUispz0fdmnTwYD1K1Ad2bj/Dsb2wi7NvEigeqyHt+nveBdhx3hB+gLzmTAxeUySogUsN/uet3EcKSpWpeAsscjmnhpPKFi4gzvMzkxVjaQ4wfzFpxFZFZgkGNHe9BQEBakSSeGCWzHwupLaKaH0OHgnSyDUos9KURz2TLeD5ZjkoAY8BaZF3ydGMFmVHsqHbTH0cY29oMYohG5rFftwPQtXKA3ssjLLBqWT8X9ewmkBl7Gx38om91fgwjMUVv8LGON5xYKw7WK6MKdXRFdYJShodkR9zD/rQDieVdJ/8nq0/v42e1DXzjy5ex7qHAGe4FgohCdr+8zNveejtlFfPJ71zJyZ/IsFKw99IK/We4SmqOBYoz3AsYa/NIUSHyKG5j82jlWUBYYCyZe2hoUUSjT2IuZZStG7bPWMewiKOR7lYUej55UJsVNExAw/hYBaKtAoEPFqpZSGYlsc7bgE50QkPOuu9pujurKGk4vWs/L2l/ikBofmhOY+dAF+kkc9mGDLW1gmzEGxbCkhmJsfkDQF37xMYjlBmbgoO8qjTIatXHP7Zd5nqKLwCE5yE8D1kpk3RaLm9/lEho/k5cidp1EJTCr7Y1L+wFW0zNZSw4FhFLx3AfZ2RqCarjS+VWgg4EdpxsK8iLtXh1sMOLUgkQNp97Nh7Ykka1pQRBRrufB3Et9weont1g96+fhg7BdCQ8dmQV2kiqcUCaTi15XAh4jFX0pRFSWHYPdnJksIyZRNSwUoYoSEfVUx/CWkE98znQaMOTmrKX0OHFhNJ51gsOIRBnn8qh8zuJuwRiY5XH4zUA2NUN9r1uMxiorRTIet5oRGgWXG9PmRpkrBHaIpPMGfElhgtOc4yJsKAaJperx8F4gqRdYvyJm6CIkbVPBAgtAIkOLGkElXJMWxTT7jVQwrDa6+NN5zzA9o2rqGYBu3s72LOvC6zI71HTuJHuqwbsO9BZnLzATjLVRwYapcy4hhtyeT7RFQSwvFSlzXPlMxciQil6z+yg/vo+Ni87TE9Y5ef1dWRGsnXtAXZe0UWSeJjDIX6fzIMpF6BNlJnBG4jz/G1jneFeajip3DEeQ3L6uK8XcvqEBV5EEaQ2wkYKnXvjUgJGNOeaY+PRX3QwKcuEteU+Didl9ogObDpiotyKKV2AdjKT9YJRd2prBKZIDRt3N2GxwuYR83NUL94xfYQfICulPHq8Q7CyfZAN5SNAPgWTWomvNB3lBjXPZ6AvyI32JB7uxPCb6HzdEA2grauStlRxhtsxXYSxeA2LmMC4Gy/PGWeEsi1T8GpgY4HxFIOijcHAcHfis7PajRSWwSTMPdlM0WiMocdrgUgEIpslQynBhAb8EYZbCxoNf0J53vc1pSDFUy7da0Fy3qm88MoOGj2WdFVKN/BCrav5cmYl+6tt9FcjstRDpJOvCig0zd+BzCzClbt1OEbhDPdCwYKqm7zhyDjoKJfSR8aAydQisyFPXCAzlRvwage/OFJurRY6nmdtQdUls9XUySqKqm6i1evWAl33xhcfBJgoI/RdFPKCRAj6Tm1jw9XPcuXKR9leW80jR9ZyoF5pbqKNpL8aEVcDSCUqnfzDoMzArxlkYhddq0XH4sTNcS8WbF7iE2NzY7cAopPFZORBW8jpI62eyIPbBPmNT6Z5pLlpSLRSR9PHJrjChC6KyMxWWUppEXqo5/ewYwrGlNCb2CLaXCukoJkPHkiPqgkZMIdp2HayskVsPAlvoIbp7cMMDMzOeTuOifYFq6IBNgUH2JN05euKNEBrBdoIjJZ5idPpyJB29nsCjMWQNy+MnZfOY44FyiKUyue0M/Phw4d529veRkdHB11dXfzmb/4mg4ODE+5z2WWXIYRoWX7nd35n9k7KWkQjw++P8XtjvIZeNMEoMs1LroYDtmXxGraZKiYT8AfBH4DokKC0xyPa4+EfUagBhRocZ6lJxGwGbhuQDTFqTFlToCd+ODCppFoL6RuMOFCtsKvWyY5qN/dXN3FbbR1PJKvZetEOHv9AG4+/bx2NS08DOYtt1RzTopF69A1GDAyW8qIrC/hnJVODN5gSHIlR1SSvceBwLBLm1ON+29vexp49e7j11ltJ05R3vOMd/PZv/za33HLLhPu9613v4qMf/Wjz73K5PKvnJZIUkWZ5HrcsIUK10DJUxkRoi1cbcTcUQEWShUVXsEI2H45RgrRdkJXFxLrOLH4IwghUYkfVpzYBWF80m56MeRqZQGceWsAA+dc06OWNSGomoE3FvHfDbVx4ymG+W93MX+/4r6z5gcK6EqjHDWMFSeKR1v1jPpgtBIS2qMEYERc/lkXy8O6YfZxUPoxf/OIXfO973+Pee+/l4osvBuBv//Zv+ZVf+RX+8i//krVr1467b7lcZvXq1XN1ajk2r+u9mHo9j3WBWIpOYWOkn1pZSOgml61zg370pmrzLLK5e2gZ68DW5tH1mczfkGRi2dxIkiz3pmtZwEDRmKRRJLxHIiVeZhFnbEEWlfGsEohajN130Enoc0BQNTx0YC2hOp/DSf5QHSiNlLaIoSgk8qKe/vDroFlESNP0yIWlmSomik5184Yz2I5FKJXPmeG+++676erqahptgMsvvxwpJT/96U954xvfOO6+X/nKV/jyl7/M6tWrufrqq/nQhz40rtcdxzHxsG5R/f39s/cmFgkqsQSDrV3LrIAskugIsOA1GO2Je6AjJiz6MtsIDaomsTFY32IiU7QkHRudKKo2oi4N2uRzqCUv5W65lV5d5kDWTs+L9/H4umUgLH6U4QcZ8XOr2PLPHYi7fjZ/b24pYC0djxzCfGU5P+nqof8Uy8bzdrO60k899aiLEBCIWCJjgTCiOQUjDKgGqIRm1DgUkeR1i0rzaHIXSe5wTMycGe69e/eycuXK1sE8j2XLlrF3795x9/v1X/91Nm7cyNq1a3n44Yf5wAc+wPbt2/mXf/mXMbe/8cYb+chHPjKr576YEBbILN7IrmUyj+rWoSjmvu2oiPGhxifMq+EWqOLGrLVgnFbcTXLZXKGloiotnjJUVYAnl1PXPiWV8j82/JTVJ/fRLuuc4vexSoV8YNM27v3pxbTNw3taaujtT9H+5LN0Bj7q1y+A82BlNMgev6OZwSBSgaoXaWBD5U6Lhjkjp3uGUsBUI3e1F2KRFscJzFLwuD/4wQ/y8Y9/fMJtfvGLX0z7hH77t3+7+e9zzjmHNWvW8OpXv5qnn36ak08+edT2119/Pdddd13z7/7+ftavXz/p8YSxyNRgZS6x2lmqXz6fjCmhF7KjzESuXI7VnMSC1GAnyLyyktmX04du5FhEJrCiODE1wWSTBaMlcZp3HKtlAdUsxFhJzQ9JrCK1HqmF1Goy44LV5hSjsTqPDQlVRkXF+HKM0mhTuHCcwXYcD4YSXWay/3wzZcP9+7//+/zGb/zGhNts2bKF1atXs3///pb1WZZx+PDhKc1fX3LJJQA89dRTYxruMAwJw2O4beNhLbKe4mcm907LAVnlBMmQs7kXI7XBSkjLEh223keltng1gYzHP4wJICsxqujLbCAy8KoSK8AEFlOaQDa3uWw+aCKktGgjSI2i5OVPHXnQWp4EP2D62dtob0qxjrnD+HBSuY+To/1sD1cdn7uYw7HEmLKVWrFiBStWrDjmdtu2baO3t5f777+fiy66CIAf/OAHGGOaxngyPPTQQwCsWbNmqqc6OdIMkWYIJTGBV7iqi//uIyyI1CJTi1V5AxMdjnhfQxL6BMfJEIhobtQgoUVLTropTby9zQQ2UxgJdT+gTxliz6PsJUhhqKmALpUXcx9IorwLnGNOsZ5gWVBltd9HpFzRHMciZClI5ZPljDPO4Jd/+Zd517vexc0330yaplx77bW85S1vaUaU79q1i1e/+tV88Ytf5MUvfjFPP/00t9xyC7/yK79CT08PDz/8MO973/t4xStewbnnnjtXp3riY4vCLAmjAtjGlNCHIUxeUpURkb55ZbTZO0VRzItaI/NI84lkc8fCwBpKBw3ffvIcftq1icO1EkJaCAxWLcCpCmubZVRlokdd046liUsHG8FXvvIVrr32Wl796lcjpeTXfu3X+N//+383X0/TlO3bt1Or5V5SEATcdttt3HTTTVSrVdavX8+v/dqv8cd//MdzeZonPha8ukGOKD1pvKMS+njI1OIPitapSpHL57o0e3PfIgNvsJDNI4sp6TmR5x2zh9Wa7p/spnRgBWn7SrKzPNQ5g3hRSr2hQCysL1BmFm8gQcYZaJvXcnA4nMfdyrJlyyYstrJp0ybssDzK9evXc+edd87lKS1JhsvmwzG+IIuOsa8BkYwu+mI8kaeazZbhHpLNhcUqAZFgQZfecoC1ZM/twHtuB0EU0ei6gP7zLFGQUpcL77sTxiLjDFF3bWIdi5sTJBJrFjAWmWq8hsQKgfEE1pvTirDHHwsqzef0rcw98Ilk8+EIzcQNSURxvGk4XUKDSARohVV2TNk8SxX1xCfTkl6/hCcMu3UnP9y3lWoc0P9sFycfnKBjyxJHlsuYs09mcGMZr25p+/k+smefn/VxRFFsJc/PPlpnf6hoUD4VM4ddwKxtHl8mOm/f6XCMZJFdFs5wD2EtspYgkryxddYekp3ghlvovPa5V8+976RNooNJ7Gjzoi8yE+N63EZBVmZUJ7NjH1vk85DVvGuKjiymrFvHMZDFimoWIZXFWkGsPXYd6qTyozaWP1pn1eAA8rk9E7VAX9LInmU8/cY2Lr/8QX52aC1HPr+G9ud2zH4lMZsXXfHqFBXThhp7gF+zqNgUBn1u7pwys3iDCbKR5tXZtLsiHK24Oe7FTqYRmQZPIcxkLNjiZkhCz/+YWkem3Iua4Ir1xLRLVw6XzY3P2E/DWmC1QCtLnHrUU590MKDzmRT5w4ew1k7daA9lEyyBMpg2CjAbGty45v/yzfaN/M3y/0qHUthsgnnfiT4fOc5DbuFxy9S2fo82N6oqnuPP2oJIDaLhIt4dJw7OcDuAohRpYkdFm1k1NQm9eTxjkakY3V9c5Lm/k5bks7x8JhKsN7ZsPkTQnrDvkjYqJ72E0mFDx/27yZ7feexBpEKecypHzunEeILux6uIB7dj4wkS3Bc5Qht01eNnSYleXWZgs6Hy+ovGfdiyKm/niYC2XTH+z55B9/XjbVhHvHkFcUVRXZNXuavHwaJoNOJwAC44zbF4EZnFHzT4I+63WUmStIlpGO6i/nSjVU43PqQVsJOS5AUysQidy+ZZ2WJLI2TzYazr6eWcqx/j5OgAX3z2ElRjNeEkDLcMfPa+vJsL//vDrAr7+da/vIxNT1bQJ7DhRmu8Xo/v9Z2LRnLqhTvYe0r7uJuXg5Q15QECpbnvrtM4dc9yxGCV6pmr2fkahenI8w1FosgaHjKWi27e0LE0mW+p/FOf+hSf+tSneO655wA466yz+JM/+ROuvPLKSR/DGe6JOEGKsUwGYXPjPRwrCjnc5jWnp5r6JXRe1rR15dQkdGHy7a3MG1DY5omN/rW0+zG/1PEYl5d6eXrNCn7aeTGR52GNhbFafgoBQiKCgMYyuGblj1mvBvnqym2IKEJ4HlYvnn7tCIEYI3+65f0X7xmRN/84kLRTUglb2w+wtf3AuIfu8aucFu2hLGPuXb0RG+UF7tM2iVgVs6JrkP5qRDwYQppPdYx3Q5vpjdLhWMysW7eOP//zP+eUU07BWssXvvAFXv/61/Pggw9y1llnTeoYznCPhbXIRooPWCnQoTzxI8zHQBTzkH7dYlSeAmaCGeZum6I71BgFXSaS0IUFmQpsXYG0WL+QzYc9GOwa6OSPHnkj788kcTUgPENQfc+Lqew2LPvxC2Q7Xxh2QIE681QOX9BNo1uiz6hirGTAevRsPsLON28iPLKRnkcG4KHHJ577XSDIc07jwCXdJB1HvyCZQs/PY4K7f5E/hJx7Kn2nVIg7BenKlENxmUAdu2RwZhXtqkFZxphYIbTFkkeG67piIIhIagGiphBZ3mBE1QVSFwV87NFiPjKzyGPFSMwEWwS9ZQaZGUTqAtIcEzDPUvnVV1/d8vef/umf8qlPfYqf/OQnznDPCG2QtQRZT7G+gq7ohI8wHw+ZWoJMY4UgK0tiT8yoMIrUFlFllNytI0E6URW3QjaXWd4IJqsYbKn1F3PoUBsrbg/pue8QR86rwNv384Gt3+ejv7iKeM9KVIvhlhw+vxvvv+/n1SueY1N0iMQqDugKv7nlP3l67Uq2D6xi59e3sPqxYOEbbiE4dGE3Z77z5/zq8geaq3el3fzNt1/H1kfbEEnK7pe0I19ziLYwoQ04WJ9c/7TBoh9ARcWIhgKTP3mp2CAHPRoSRFXhDUpEBv6gwB/MYx2GHtKEAb9u8WoGYe2cVS4T2qJqKaqaFP3fXYk0x/jMllQ+sqX0ZPpoaK35p3/6J6rVKtu2bZv0mM5wj0fxYxdC5LmfS0g2H07TMxZ5LqywovmEOS3Pe7wfiZngteY2It9OWYQRuYJt8x7dxgpsKokOa+wzOwg3dbKs0s9V5T6+3rOfXe2n4EdFtRkpEZ5H3Cm5uGcXv9T5GAAN64OFtf4R1vpH6PTqPN2xBVGpII0ZP3J66LrQGpOkY8vyc4VUyMAH3yfuEryy6wmuLh+9gezI9vKX3VciShF4HkkHnN+zn3a/wc5qN4frZcwkvkhf+QxmAdrm34H1FSIMsbLIAMhEHkiY5cqIzIp+2yNnSoxFaDvnUrnQBlxlNMc8MrIr5Q033MCHP/zhMbd95JFH2LZtG41Gg7a2Nr75zW9y5plnTnosZ7iPhTGoeoowNpfNI7UkZXMoyp9WbTPCeMay+fBjZ3mu70QFW7SfdyvDgowFoLCeJVZ+bsQlHDrbo9F9AYPrBX4a8ZWBNdSygL0vUVTWXZBXfAvzcaqnJnhS81yygkEd0ZeV0MNarhyM26htStn9llMQJh/bjOhdbiXoID9eZZdl1Q/2op96dnY+lEmgTtvC/kuX01guqJ3ZoE+X+VHj6M/6kF5L1FNn7xUnITQ01mp6kxKDWcjBWoWB2jFK5xXUE59qEuBJg40Me1++jOD8burLRd4+tqrwqgK/Ko5dnMfhWEjMklS+c+dOOjo6mqsn8rZPO+00HnroIfr6+vjGN77BNddcw5133jlp4+0M97EYks1rCTbwsF4JvQQ/NVF0EhuSzdOKJPHHL8AyVfKiK+O/bqWAClgfMIVsnuY91DN8Yi0R0tI4t0ZyvsH3NYNJwFd2XYLEsnnbDqSwdAV1NpUPUVYJqVU0jM8z9RXsizvYV28nHVYxRgjL1q178U/VBFKzrtzL6rCv5bzaVINTwr30yCrvf/K/0niuB38eDXd1azfijYd4y4aH0VZyOKtw28DRebLYeGxefojnXmPRWtIepvTFEdpI+gZLJDV/ck9fwlIVJRAWWc6oviyhCqT9Af4RD28wN9r+AEc9aheA5lgMzJLh7ujoaDHcExEEAVu3bgXgoosu4t577+Vv/uZv+PSnPz2p/ZegCZoGQ3Nk2uZzc0sUYYGiMEo+d1nc8MUseN7HkMmFLcYzeVA0zbFtnjOcCawHvq8pR3kaV6YVvfUSlSBhbaWPdr/BimCQzeEB2lWd3Wk3zzWWU81CBtOQwSRAm9zjlsKipKE7rLG6NEBJJmyKDrHWP4ISBllM0HapGqf7VXpkiXVtvez2l8/wg5gaOhSsae/n3NIO9mZdPNdYTs0EGCswVmIQBFLTWamTFQ8lSeahjcBoOXG+9bBpkeH5AaKUEZUSPGk40vDAgCxkcqHtqMDDISMuiumQOcMercrmcEyWhVA5zRhDPIX0U2e4HdNCpZZg0GIlZOHsyubjIROKSP9cnjYBeZR6Q2C0wipoJJKGF4KyqFCjlKEW+wzEAb4yRF7GQ8E6PGGoZz6DaW6s64lPnOQ/hyDIiPwMYQUDaYSxEk9qetMyT3orCWVGt1ejXTXoVVUCNAOqxuG43NJffD6QmeVwvczzyQoGdERqFcYKqllIXxqRGI8DtQq9g2WMOfoFWSPQ2TGmfLRAJAIxwrjbWDJQ90CA6lP4AwKV5N/PyJuYKLIIVGKL+uRzF0nuNTQiKSLJYxdJ7liYXH/99Vx55ZVs2LCBgYEBbrnlFu644w6+//3vT/oYznA7poywIOO8aYOVAiskZhZl8zEp6qOrBIzKxxqSzVVskcmQB64AhfEh69DowJAIqImjc7nDYwybAsqQdynz2ueRn2GtYDAJqKX5xPZechms5KesKfexIhikw6sQCE3V9tMXR5Sy+XX3ZGrpbYQ81+gBaM7RD2QhB+ttNDKPvmqJxmBwVKUY4lg21ICqyTydq4UiFxzwqoKgf5hBHnlMC17D4ld1U5Kci8A0YUHWM9RAA+aqYYnjxGSe08H279/P29/+dvbs2UNnZyfnnnsu3//+93nNa14z6WM4wz1V9NFORlYuvSjzIY7OYRY5uUMyqJxDz3uorLqxzQjz3AiLo88MhWwvlEUUqWPNF4a9Pq71MAJrBJmWKGkRwmJHbJtqRWYUqc2XqgmJTIkkU5SPg80wRqKRGCvIrEJbQaI9Yq1IMoXW4uiDiRGTvtEIk3+uYoTBb373dqjD1+jo8RaK1K85iSS3tlk8KL8u7OIpmONYEAg7synQqe77D//wD9MeawhnuKeA0BpVTVGxxviSrOwt2QjzJhZUYggGivzqaO5lc2FBxWMXcdFh7okLnXuLNmk9EeNbbGjGz0W3ecvQgcESQhqiKKUcjN+gIjYe+9JO+nSZaj2k+zi1jRySx48kJRLjcahe5shAGaMlWaoKQysQiSwi8o+N0Pm89aj1WfH5F8VVjtucciGPy3qWx1w0Mme0HUsCZ7ingjbIWh5AICIfE6olGWE+nKZsHlusElglMcEcKxEWVJzL5sMxnsB4gJ8bHWUsjDBSOhJkwcTamE0kWSpBWpSyGD9DjuMuplaxP21HYUhjDzHPUvlw6trncKNCI/Por+WVzKwe5mFbEIlA1cTkvd8xnsCEBq82dp72fCNjfVQed0bbMR3mWSqfDZa42ZkGQzcHa3N5ThmQwsnmkH8mOvfIkOSy9FwKEmPMp0o9vP7JUXm8KZEbi9ACy4gTk7bYbthcrRVoncvmQliUtEhhC0laEhsPJSye0CAk1ooxTmp+GSqmYu1QgRqKaQXRjOwWxXubCsLSrHQmNce82QkzbJntj8TaZvS60E4ed8yMhRBVPlWc4Z4mIjN4AzGqLjGBcrI5gAGvYZA6f5BJS3mxk/lCGJvXxx6hbBsPdATWK4zOoGx9oBAWExYS+nB7ZiFLPPqNRA6TzbUVHGpUGExDIpXSE9aoeAu0k9iQPN4Q+Zz1NIuJiQxUI//8RJaXrh1322Iqw2uYIqp8du9sXkOjahmYInrcGW3HEsMZ7umSaUSm8/t8JUJE3vFWDY87Q0VaZJLL5trPZfO5ThNrjm9yCZ0RNlRHAl20ERWZQI0wOlbkUep6ZKtRm8vmOpFoZfE8A0GKNpJqElATPpHn0+bHVObubc0MW7znRtGVbZpfhtDgNfLvdzJjqsTiV83RgMLZ+nFYi0gNshpD5lK+HLOAk8qXJsIYZGqacuxS9ryHy+ZD3lmLqRB5ENl8GXMoulJpIC2ke9n6MCHI5XOGoq+HI22+WIHWkjj1EMLiKYMEPGno8uus9AfwggzGaKs5X2RWkmhFnHp5jnaaF1gRTWn72B/6kCQ+qoiK4ZhNQXLpekgit0ePNwsM1TjPH0SM87Ids4aTypcoItV4/Q2slJiS52RzyPN36waZjojq9gRpWeQ52POEyCxeVWBlXrQlKzEqqlym4A2olqcMK8CUTC6hW0va8BjIFFIaSqWEUpBS8RIubHueC6Md3L78NLJoJSMd9/milgX0VUsksYepeqjBvMXmRL2xR2HymvEja43nKX/Hlsf9uinad85u4JpMDN5gkqdjptrlajuWNM5wzwbDZXMZIUpONh+SrVXc+knoSKBDNapZx3ycC0CGQIQjmpnYfO5XjZj/tcpifYENyWXzVKJTMJ5Eh/nGgcrYEuzn3CBiTbmfF7xV8/OmxiDWXm606x6ioVD10VXPjkX+WYFqTP0KlpnFq5s5qR4n07xngJPHHbOOk8odQltkYhDGYpRwnvcIhM5v8CrJ5Wo7UQ/u/3979x4dVXn+C/z77r3nknuIhAQqYAAFj1K5KDSxR6BQCXqKUUuLdSlY6oUiS4rVgm3hh7+6Ui5HXV6q9rSSuk7V1nNET2mrC9FAtUgF4acipASj4ZYgwVwnM7P3fp/zx54ZMkkmzCV7MkOez1p7tdnZe/aT7TDvvM9+3+e14/rSKhoSKhbTV9qerHihi7ByayQJfp+GTkWi1e/GCX0I6rWjyFB1fDXOgcKZU+A44wHqjkO2tfVL3GpuLuQlo+C7IHw1r68uccDl0NFmuOEztLg+RETw75Q4W0wnBQgZmLlBgV42YzbgVDmD8BvQWqy1m81MB4xsx6CeKtadMK1BS1qngOkMpM2T2HArOsHRbl3TdAFGJiKWahVk1UEXenhenTSCDidMQ8FxEtiZMx5NZjYKnB0Y+p1jaPpWJjr/awjGvGQCB/un4aYxF+Lfi7IwZdKRsP15UkWn4cCxjny0eV2g7mVNoyGtEePBojbCrnriMQqmx4VuAiadXeyHsf7EPW4WTJtDCCiaYs1F4oY75Gza2uryGhnJvTdCAiIwMpoUa6R1xC8OEVLoUgLSp0AqKryqE8c8+XAGFh754cj3cIHajhW+78HMiW6t62joQ9yYcPlR/J+xb4Xtr+5U8NSJb+FYWz78Rnyj/gRZz/jjSY/bSZgEpVMH9DjnsDF2nuKG2y5EEIaE6pMgw5oeJR2cNu9KSAqsGiUgVWueddJHm+tWmphU6ztWVOcRIHQBKCoMBTjVkQ0FhDPOTEgIDNE6kOHS8eWUIcgZNq3L9QgZx9pANXWgGJbwA6xe8JnOTPxb74BJAh7SoJOK/d6L0OTNQofPCb/PAfKqEKY4+zigG8VAj2fQwdHg8bAefQQfQSQ+IE1IsmZoEKDoPEebJcdApLsTwQ23jYRPh2ZIQBUws5yQmiPsWelgp+gEZ5sEqYDhUqBnWWtqJ+36foLDDKTNM6yedzRpe2ECWqcA+QSkV6AR+TiTkQWny8DRnCHIcfowPLcV5oJ2yC55+E7dgaPVxRh96isYDY2xxeoz8GVTDra2TUSb6cYx7xB0GE40dubgxJk86D4N5NGgtqmhhrTHEpuBkd9qZ7ffUaACWYwEWUt2OjukVcksOO0sAYouobX5IfyGFROnx5ndKMHKewPw5ZIbbjuZEsK0lrASLg2CktujTHWh+b46QApBT/LNCV1fwFqWNFqBFHrwDNmpQpeAaShoUjLRqTtwUd4ZfKvgEEY5mkKnNZuZeOjw9wB37OXkhCRIr4bPvRegzXCjvn0IOvxOdPic0L0ayK9C6Na62MKI8LcEBqGp/v6bqqWYBNXXfyPJhSSr0fZGXtiFscGOG+4kEbq0yjSqAlJTOG3ejZAEzUuQhrDS1o4kfskhK92regO9bi2GtLkEhM8aHi+lgFdzQkoFjY4c1LiL8ZUzCy5FR47ihZcccBZ5cLL8a8g4MwI5h9uATw6DdH+vr63m5kK/YizaL3TBU6TAkdOO075seE0Npoz+/aMYgUI4hLhLnob9zSag6sG0duLfAILpcWEQFN2Mq/fPWLx4VDnrHREUrx+K37BKgea4ITXBafMuFD/BaQTS5m4FflVEXnrTpus7DAEo1kjzaKu7hdLmXgHpFNDhhOFXcdJQ8IEcjUyHH/muTnzN3QyXYmBOyb/RfHsGTnuzcOKvo3Hh51kwv+q94UZxIermuzGl9N+QJNDmd+NERx4kCehRNtxWGVrA0YFARbTEe9uqDjjbJRQ9uAZ2Yq8nDLLS414DQkpOj7Pk4lHlLCLT+kASUuGqT70Ipq3JBBQHQZDo8Q/C1jW+u6TNgytpRZom1j2oUNpcAUxDgBQFhqahzeuC37S+feRoPmRpPozOOI05+Y04ZeTifxaMArTI/wTJ7QAVe7Go+D0c1wvwVtOlqG8dEn5MNGVMA1O8+mN+tqCzPe3+6G0DgfS4bkL4OT3OWDS44U42SVC8BhyBpUClk9Pm3VkVuKjbCl6B9baTkEJX/bCKswRm8yVa5U2XKpp8megwnFAh4RYGzhhZVjlY6qM1FQKa08TX1BbopMGpWA+STanAq2swTQWGrvY6dzuUHpdWjzvRdJ71bLzrTIDEXlDIQKEiQ0Ix5ICuY84Gt+Dys4mcn2zccCdbMG3u00GaAiPXzQ13F8EVxpzdP8gF4M9WAo8YbAwguMKZDpAqoGclPk3NozvgMzRoioRfqjChoE13Q/GhzxGppCnIdPsxzkEAmpChWj1Sv6Gi0+OC6VcCBdV7nqvogKMdoYU5Eh7tbVijxxW/Vcks4fS4SdDa/dY8beLR42wAcaqcRSXwISUQfOZI/Ly7i66rSwWREvhm22UgiV0979A3aLLmmCOQQY/qmTcJEAkQAZIEJAkoAAwSMEnALzUYUoVPahF7wcLhhHBoMF0aABMeMuElNxQhoSkSiiCro959NbMuS2hac7Mp4dHewRiDr6f0U1U1q5yp5OIqjMXBtq7eI488grKyMmRmZiI/Pz+qc4gIa9aswfDhw5GRkYE5c+bg8OHDdoU48IigdBpwtOnQ2nWr8ATrXXCNZw9B81g9YrtHcwYHdmkea+6zco5HsMIEFK+A0qFCtjvg8bjQ5nGjw+cMjQJ3qzqKXS0Y7m6xUvDdvrCpQ4agff5kfLFyEr6Yl4mWo3mYvu0+LN6zGG26G+UjPsXEYSfgCCxyAlNAeBUoHSq0DgVau4DmQaA3n+Dfb1oj7R0dBM3bD1O+iKB6TTha/VA7dIDT4ywFBEeVJ7Ilm20Nt9/vx4IFC7B06dKoz9mwYQOeeOIJPPvss9i9ezeysrIwd+5ceL1eu8IcWKaE4vFB+8oDrdULxc8fZH1RfRKuVhOuVmnNRbYbAZqX4GgjONoDi5P0QZiA6hHQ2gXUdgVGmwPedid8XgckAYog5Gg+jHKexkXu0zBdBIjwf4JiSB6OXyvx+OL/hfnXv4+cWhX/7T+/xNA/Z6JVd2PZkA/xPy74L2Rl+ALXFFADDbajTcDRDjharal1iT57U0wrPe5qMeEIFFlJhDAJWocOrbkTapsXgnvbLBUEC7AksiWZbanydevWAQCqqqqiOp6I8Pjjj+MXv/gFbrjhBgDACy+8gKKiIrz22mtYuHChXaEOLCJrjWFFOZs2747T6Na32tCtCTRKMnBrRHRp7LgEv1FHUz2HxNnUsmn9DNNKl1Mgbd5pOnDayEWLmRE2OE24XFBcLph5WXDk+jHJ1YzD7tNWOdamr+BoK0Srz41mKeGRrrOjyclqYJXgutv9sLpX8F4HH1n0S3EVCv43k7w0J0spPI87AXV1dWhoaMCcOXNC+/Ly8jB9+nTs2rUrYsPt8/ng61L3ubW11fZYbSElFI8OZ7dBOlJTYLpVXh60K7IWKnESAqt8CZjOgQ7q3Eyp4P3aEuw5OQFah0DRhzrI0wklMxP+0ktx+nIX/HlATvZp/O+WiTjYMRxtYyQaF14G0y3Q+UEx5hz8qbVOuFMCKkEYAqpHQPUF6q4nOtrbBDQfQTGs0f0JF2whguY1IfyB0eM+brQZS1TKNNwNDQ0AgKKiorD9RUVFod/1prKyMtS7T2uBtDm84b06xeUAOdwwU+a/VAogQPVKqD5r5LdPVSAdIuXLyRqmgoxDboz+v43Al2dAnZ2QXi/UoRfg1BQXRpTXI8fpRbvuwv878XUYUkHumGZ4R6noPJ6NUX+XyNzzOYxLvoaj386E90I/hAFondZzaAD9sMgHoHUStE7ZP69HgNJpQG3zcv0ClprScFR5TN24VatWQQjR53bo0CG7Yu3V6tWr0dLSEtqOHj2a1Ov3q+C0mK6blIHiGQPzLCUVhQaFBEeZm7DS5ibC0sX9ncISlPhrq34ALW0wv/oKMjh2QyiQGpDj9CJT88MkBR1+pzWFTJXIdOkgh/UeIK8Xit+E0AEYCoQhQvch3g+Qs6PQz85pDW2xvCZZMYZtwfdu8P3L72GWYtJxcFpM/bj7778fixcv7vOYMWPGxBVIcXExAKCxsRHDhw8P7W9sbMSkSZMinudyueByxb5oQ7oQhoTa7ofiU0AOFUYGp827CtY4V7plYKUKGG4BSrB4ytnrBFbVMqxyqKYL/fbaMAw4W4CDp4rgchjw+h0wDKXHM/XmsU7kZFwKf5YC1Qe4T2hQ/ece7X5OgUF4wQIr8VZEU30SqtcI61kLSRBegxtsxvpRTA13YWEhCgsLbQmkpKQExcXF2L59e6ihbm1txe7du2MamX7eMUwogefeMtMF4VIHIjOTuoJp824TD6RLQDrUhKuehV3HZzVuUhPWIiT99toSznZC85dZ8DhlxJ5zx0iC7wINwrCKqzjaqdugvfgIaTXcmifytc+JCIrfhNrq7VlMhRttlsp4Wc+z6uvrcebMGdTX18M0Tezfvx8AMG7cOGRnZwMAJkyYgMrKStx4440QQmDFihX41a9+hYsvvhglJSX45S9/iREjRqCiosKuMNND4I0hpIRiSlD3AUOB8qmDUcQ0FQV6j92WuCQBq5xpvLcr0FAqJiBNBKro9P16AoCmWo2ZrgAiMEtAycmBkpsDys2C4Q4cbAbqpHcrYypM6xm+1AA1kF2IZ/R49yI21mv3vj8qRFahF7KyQ5wOZ+mGR5V3sWbNGvzhD38I/Tx58mQAwDvvvIOZM2cCAGpqatDS0hI65sEHH0RHRwfuuusuNDc345vf/CbeeOMNuN1uMEDoJtRWP1Q1/EPddGswM7VB23j3RjEoUDjk7L8qCqy7rWcktvKYkAS1U0DRrZS86e47ba45TAzN7ICmSNS6CmGtz+6CfuXF+HKSG0YG4BtCgEJW4+lVoPjDvw0IiR6PA2KOmwDVB6uYStcPG0KglGnsr6n6JFSPHlpHmwegMWY/2xruqqqqc87hpm7fzIUQePjhh/Hwww/bFVZ6M0wo3efACgGIDJhuFeCGO8Sq+tWzS2pkKjBc1prfcb+2tNLm8AHSaS180lfaXNNMFLg8cCkGalwEKAqEpqHlIhe832hHptsP3eOC6XEAhoDiF9A8vfy3THTYPAWqz3UkkBLvRtEl1HYfz81m6SvJo8orKyvx6quv4tChQ8jIyEBZWRnWr1+P8ePHR/0aPMop3REBhoRiEBRdJjyP93zS6whQaVVAU/TwTZhxpry6vmaXTZiA0IU18lsQCpwdKHK1Qs+X0C8aBpSMhJElYJoKfLoGQ1etPLoprProQGDJUAFFD2zB145itHevf6eB0HrcMY+MJau2uKJ32fwSghcHYWku2aPKd+zYgWXLluH999/Htm3boOs6rr32WnR0dET9Gjw7+Dyg+A04mgmkCpgZDhhZGrjaWu9Uv4SzHeFLhsIagW5kxL7ymDAJmkeAug2OU3TrYbp0AKIYKMupxRjnKRy6oggf+cdC68iA6SbI0y544bIKqRjW823FgNVom4DmDdQd73rNwPraEWMia9qZ1inDnoMLQtwjxhXdWs1LdKunL0yTV/ZiLAZvvPFG2M9VVVUYNmwY9u7di2uuuSaq1+CG+3ygGxC6ASEESFUgMrWUL0YyUIRpNWjd7w+p6tkBYrG8XjBt3vM30DQB00VQBGGCswGXOTVUDNuHtklunPFk4ExjLpyNjrPlUbu/AgUWOemMvbENrmnevb54vANpFFNC8eoQ3kTnnjGWYoJ1BhI5Hz2rdkY7VTk4zqugoCDqS3LDfZ4RppXC7NpzJCFA2uAded6brg0YCavnrOoCFOFRLQmrlx7ts/FgqhpCoKUlE1tap+AjVyP+0TIep9qy4fU5AEMJPF8TYSO7u77GuUaOhwqndBtsZu2LY8AZBZbuJOv80HX0xBcZYSwl9dMz7pEjR4btXrt2Lf7jP/6jz1OllFixYgWuvvpqXH755VFfkhvu8wkRlE4dDkOGpcpJFTCznTAy+D93JKqPoJgmKMIjBlIBPVOBGeVUMkUnONqFlZKvceOPX/53SGegQQys8a35ROh5teqzSpd2FVwDu8/rGICjQ/ZYuUwxCIhnulhgBS/F2+0FDV4chJ2fBBKcDhb436NHjyI3Nze0P5re9rJly/DJJ5/g3Xffjema/El+vjFMiG4fsEJTIV0akDFAMaU4aw4yAQYQ6au31AQMFyHah+BCAiKw9GiGF8j4UgBCQM8S0LO79dwDz7VVH8X8zV+Y1kjx3tP1sRMSULwGhMd37oMZYyG5ublhDfe53Hvvvdi6dSt27tyJCy+8MKZrccM9GJA14lzrNMJ7lAKQDoVT6FEQRFB1gGJsz0Ipds1qk4UZKFEaHHkeSJFrHoLqjf2bv6rHv+ym0kv6W9Gt+viMDRpJrpxGRFi+fDm2bNmC6upqlJSUxHxJbrgHA0lQPH4IX/jAInKoMHJc1hxw1jcJaB5pLRISI8OtwJ9lFX1R/QQlUB1N6yQ4PBRKiStxPEO2Gv44zpME1WNA9fjDP3hkYKQ4Y4NEsiunLVu2DC+++CJef/115OTkhFa/zMvLQ0ZGdGlRbrgHA6LAyPNu+yUgMnnAUTTC0+nRIwEoGkGQCPW4hWk11g4PwdlqxlW6NGGSrPWxvTqXKGUsiZ555hkACFUQDdq8efM5F/EK4oZ7EBNSQvEZfb4JSBMwnQrPC0+AYlilV2WXUrUiOHrbJsKQUHUZcYCakAShm9xoM5bkymndK4bGgxvuwcy0ylUqnshzc2WmA1J1gjRuuOMhAnXAnb0UKQlWMbOD6pPQWr1AhPS7IOJn2YwhOG0y/n+IiZwbL264BzMiaxQ6Ij/TFA4Vov9KWw9K1nzsJN5BIghTWj1qnsLF2HmHG27WJ2GaUDsNa7WqXg8QkA4F0sll7wdCKCXefXS4z+A0OGPRkIir5kHY+UnGDTfrk/AZ0LoVdOmKhICZ64J0OPg5+ABQDILW4rN6111JyTXEGYsCp8rZ+SeQTo9EqIo10IkSX3WSxU6Y1upw0GMc7s4YS1vccLPESILiNaCJyCtrkSJgulWQxun0cyKC6pNQ/NE9m1b8JgQPMmMsfkkeVd4fuOFmiSGC6PRD80fu8ZFDBVQ3DG64z0kQoHp0qO1Rlmgj4pQ4Y4lIcuW0/sANN0vcOdLpUBRr8FSib/B0e4Yex99rFWc5x/1kjPWbZFdO6w/ccDPbWSPTdSiGjLj6VjSkS02b8qyKLqF6zZjX+RWmhOgje8EYY9xwM/uZEkq7D0hkMRMhYORlwHSlRxU3xbCK28TVCCdzzjdjgx2nyhmLgChiFa+oCGH1Rk0C0uFRuRn4e/n5M2MpzSqQlNj5ycYNN0sbileHM02mnQld8ipbjDFbcMPN0gMRhFeH8KXR81+uXMZY6uNUOWM248aQMdaf0nAedzo8LWSMMcZYAPe4GWOMDVpcq5wxxhhLJ2n4jJtT5Ywxxlga4R43Y4yxwYuQ2JraXPKUMcYYSx5+xs0YY4ylE0KCz7j7LZKo8TNuxhhjLI1wj5sxxtjglYajyrnhZowxNnhJAImsfzAAi4xwqpwxxhhLI7Y13I888gjKysqQmZmJ/Pz8qM5ZvHgxhBBhW3l5uV0hMsYYG+SCo8oT2ZLNtlS53+/HggULUFpait///vdRn1deXo7NmzeHfna5XHaExxhjjPEz7q7WrVsHAKiqqorpPJfLheLiYhsiYowxxtJfyj3jrq6uxrBhwzB+/HgsXboUTU1NfR7v8/nQ2toatjHGGGNRCfa4E9mSLKUa7vLycrzwwgvYvn071q9fjx07dmDevHkwTTPiOZWVlcjLywttI0eOTGLEjDHG0tr53nCvWrWqx+Cx7tuhQ4fiDmbhwoWYP38+Jk6ciIqKCmzduhUffPABqqurI56zevVqtLS0hLajR4/GfX3GGGMs1cX0jPv+++/H4sWL+zxmzJgxicTT47WGDh2K2tpazJ49u9djXC4XD2BjjDEWnwGYx71z505s3LgRe/fuxcmTJ7FlyxZUVFREfX5MDXdhYSEKCwtjjTFux44dQ1NTE4YPH560azLGGBs8BmKRkY6ODlxxxRX44Q9/iJtuuinm820bVV5fX48zZ86gvr4epmli//79AIBx48YhOzsbADBhwgRUVlbixhtvRHt7O9atW4ebb74ZxcXFOHLkCB588EGMGzcOc+fOtStMxhhjg1k/TQfrPjC6r2zwvHnzMG/evLgvadvgtDVr1mDy5MlYu3Yt2tvbMXnyZEyePBl79uwJHVNTU4OWlhYAgKqq+OijjzB//nxccsklWLJkCaZOnYp//OMfnApnjDGW0kaOHBk2ULqystK2a9nW466qqjrnHG7q8i0nIyMDb775pl3hMMYYYz1JAkQCPW5pnXv06FHk5uaGdtvZ4eRFRhhjjA1e/ZQqz83NDWu47ZRS87gZY4wx1jfucTPGGBvEEi2ich7VKmeMMcZS3gAsMtLe3o7a2trQz3V1ddi/fz8KCgowatSoc57PDTdjjDGWRHv27MGsWbNCP69cuRIAsGjRoqgW5uKGmzHG2OAlCQmlu2Xs586cOTNsVlWsuOFmjDE2eJG0tkTOTzIeVc4YY4ylEe5xM8YYG7wGYHBaorjhZowxNngNwDPuRHHDzRhjbPBKwx43P+NmjDHG0gj3uBljjA1ehAR73P0WSdS44WaMMTZ4caqcMcYYY3biHjdjjLHBS0oACRRRkckvwMINN2OMscGLU+WMMcYYsxP3uBljjA1eadjj5oabMcbY4JWGldM4Vc4YY4ylEe5xM8YYG7SIJCiBpTkTOTde3HAzxhgbvIgSS3fzM27GGGMsiSjBZ9w8HYwxxhhjfeEeN2OMscFLSkAk8Jyan3EzxhhjScSpcsYYY4zZiXvcjDHGBi2SEpRAqpyngzHGGGPJxKlyxhhjjNmJe9yMMcYGL0mASK8eNzfcjDHGBi8iAIlMB+NUOWOMMcb6wD1uxhhjgxZJAiWQKidOlTPGGGNJRBKJpcqTPx3MtlT5559/jiVLlqCkpAQZGRkYO3Ys1q5dC7/f3+d5Xq8Xy5YtwwUXXIDs7GzcfPPNaGxstCtMxhhjgxhJSniLx9NPP42LLroIbrcb06dPx7/+9a+oz7Wt4T506BCklHjuuedw4MABPPbYY3j22Wfx0EMP9XneT37yE/zlL3/BK6+8gh07duDEiRO46aab7AqTMcYYS6o//elPWLlyJdauXYsPP/wQV1xxBebOnYtTp05Fdb6gJCboN27ciGeeeQafffZZr79vaWlBYWEhXnzxRXz3u98FYH0BuPTSS7Fr1y584xvf6HGOz+eDz+cLe41Ro0Zhxpil0BSXPX8IY4wx2xjShx2fPYPm5mbk5eXZco3W1lbk5eXhm7gOGhxxv44BHe/ibzh69Chyc3ND+10uF1yu3tug6dOn46qrrsJTTz0FAJBSYuTIkVi+fDlWrVp17otSEv385z+nqVOnRvz99u3bCQB99dVXYftHjRpFjz76aK/nrF27Nlj2hjfeeOONt/NoO3LkSH82QWE6OzupuLi4X+LMzs7usW/t2rW9Xtfn85GqqrRly5aw/bfffjvNnz8/qtiTNjittrYWTz75JDZt2hTxmIaGBjidTuTn54ftLyoqQkNDQ6/nrF69GitXrgz93NzcjNGjR6O+vt62b2p2aW1txciRI3t8c0t1HHdycdzJl66xp2vcwcxpQUGBbddwu92oq6s757iraBARhBBh+yL1tk+fPg3TNFFUVBS2v6ioCIcOHYrqejE33KtWrcL69ev7PObgwYOYMGFC6Ofjx4+jvLwcCxYswJ133hnrJfsUKR2Rl5eXVm/UrnJzc9Mydo47uTju5EvX2NM1bkWxt9SI2+2G2+229Rp2iLnhvv/++7F48eI+jxkzZkzo/584cQKzZs1CWVkZfvvb3/Z5XnFxMfx+P5qbm8N63Y2NjSguLo41VMYYYyylDB06FKqq9pgtFUs7F3PDXVhYiMLCwqiOPX78OGbNmoWpU6di8+bN5/z2NHXqVDgcDmzfvh0333wzAKCmpgb19fUoLS2NNVTGGGMspTidTkydOhXbt29HRUUFAGtw2vbt23HvvfdG9Rq2PeM+fvw4Zs6cidGjR2PTpk348ssvQ78Lfqs4fvw4Zs+ejRdeeAHTpk1DXl4elixZgpUrV6KgoAC5ublYvnw5SktLex1R3huXy4W1a9dGfL6QytI1do47uTju5EvX2Dnu1LRy5UosWrQIV155JaZNm4bHH38cHR0duOOOO6I637bpYFVVVRGDCF7y888/R0lJCd555x3MnDkTgFWA5f7778dLL70En8+HuXPn4je/+Q2nyhljjJ03nnrqKWzcuBENDQ2YNGkSnnjiCUyfPj2qc5M6j5sxxhhjieHVwRhjjLE0wg03Y4wxlka44WaMMcbSCDfcjDHGWBpJ+4Y7nZcPfeSRR1BWVobMzMweZV4jWbx4MYQQYVt5ebm9gXYTT9xEhDVr1mD48OHIyMjAnDlzcPjwYXsD7cWZM2dw6623Ijc3F/n5+ViyZAna29v7PGfmzJk97vk999xja5yxLvn3yiuvYMKECXC73Zg4cSL+9re/2RpfJLHEXVVV1eO+DkQVq507d+I73/kORowYASEEXnvttXOeU11djSlTpsDlcmHcuHGoqqqyPc7uYo27urq6x/0WQkQsJ22XyspKXHXVVcjJycGwYcNQUVGBmpqac56XKu/xVJD2DXc6Lx/q9/uxYMECLF26NKbzysvLcfLkydD20ksv2RRh7+KJe8OGDXjiiSfw7LPPYvfu3cjKysLcuXPh9XptjLSnW2+9FQcOHMC2bduwdetW7Ny5E3fdddc5z7vzzjvD7vmGDRtsizHWJf/++c9/4pZbbsGSJUuwb98+VFRUoKKiAp988oltMfZH3IBVirPrff3iiy+SGLGlo6MDV1xxBZ5++umojq+rq8P111+PWbNmYf/+/VixYgV+9KMf4c0337Q50nCxxh1UU1MTds+HDRtmU4S927FjB5YtW4b3338f27Ztg67ruPbaa9HR0RHxnFR5j6eMeFdWSWUbNmygkpKSiL9vbm4mh8NBr7zySmjfwYMHCQDt2rUrGSGG2bx5M+Xl5UV17KJFi+iGG26wNZ5oRRu3lJKKi4tp48aNoX3Nzc3kcrnopZdesjHCcJ9++ikBoA8++CC07+9//zsJIej48eMRz5sxYwbdd999SYjQMm3aNFq2bFnoZ9M0acSIEVRZWdnr8d/73vfo+uuvD9s3ffp0uvvuu22Ns7tY447lfZ8sAHqs2tTdgw8+SJdddlnYvu9///s0d+5cGyPrWzRxv/POOwT0XH1xoJ06dYoA0I4dOyIekyrv8VSR9j3u3rS0tPS5qszevXuh6zrmzJkT2jdhwgSMGjUKu3btSkaICamursawYcMwfvx4LF26FE1NTQMdUp/q6urQ0NAQdr/z8vIwffr0pN7vXbt2IT8/H1deeWVo35w5c6AoCnbv3t3nuX/84x8xdOhQXH755Vi9ejU8Ho8tMfr9fuzduzfsXimKgjlz5kS8V7t27Qo7HgDmzp2b1HsbT9wA0N7ejtGjR2PkyJG44YYbcODAgWSEm5BUuN+JmDRpEoYPH45vf/vbeO+99wY6HLS0tABAn5/Z6X7P+1vSlvVMFruWD00V5eXluOmmm1BSUoIjR47goYcewrx587Br1y6oqjrQ4fUqeE97W8Yumfe7oaGhR1pQ0zQUFBT0GccPfvADjB49GiNGjMBHH32En/3sZ6ipqcGrr77a7zHGs+RfQ0PDgN/beOIeP348nn/+eXz9619HS0sLNm3ahLKyMhw4cAAXXnhhMsKOS6T73drais7OTmRkZAxQZH0bPnw4nn32WVx55ZXw+Xz43e9+h5kzZ2L37t2YMmXKgMQkpcSKFStw9dVX4/LLL494XCq8x1NJyva4V61a1etAiq5b9w8EO5cPtTPuWCxcuBDz58/HxIkTUVFRga1bt+KDDz5AdXV1SsdtJ7tjv+uuuzB37lxMnDgRt956K1544QVs2bIFR44c6ce/YvApLS3F7bffjkmTJmHGjBl49dVXUVhYiOeee26gQzsvjR8/HnfffTemTp2KsrIyPP/88ygrK8Njjz02YDEtW7YMn3zyCV5++eUBiyEdpWyPO12XD4017kSNGTMGQ4cORW1tLWbPnh3369gZd/CeNjY2Yvjw4aH9jY2NmDRpUlyv2VW0sRcXF/cYKGUYBs6cORPTf/dgPeHa2lqMHTs25nj7Es+Sf8XFxQktEdgf+mOpQofDgcmTJ6O2ttaOEPtNpPudm5ubsr3tSKZNm4Z33313QK597733hgaInivDkgrv8VSSsg13ui4fGkvc/eHYsWNoamoKaxDjYWfcJSUlKC4uxvbt20MNdWtrK3bv3h3ziPreRBt7aWkpmpubsXfvXkydOhUA8Pbbb0NKGXVxfwDYv38/ACR8z3sTz5J/paWl2L59O1asWBHat23btqQuhdsfSxWapomPP/4Y1113nY2RJq60tLTHVKRk3+/+sn//flvex30hIixfvhxbtmxBdXU1SkpKznlOKrzHU8pAj45L1LFjx2jcuHE0e/ZsOnbsGJ08eTK0dT1m/PjxtHv37tC+e+65h0aNGkVvv/027dmzh0pLS6m0tDSpsX/xxRe0b98+WrduHWVnZ9O+ffto37591NbWFjpm/Pjx9OqrrxIRUVtbG/30pz+lXbt2UV1dHb311ls0ZcoUuvjii8nr9aZs3EREv/71ryk/P59ef/11+uijj+iGG26gkpIS6uzsTFrcRETl5eU0efJk2r17N7377rt08cUX0y233BL6fff3Sm1tLT388MO0Z88eqquro9dff53GjBlD11xzjW0xvvzyy+Ryuaiqqoo+/fRTuuuuuyg/P58aGhqIiOi2226jVatWhY5/7733SNM02rRpEx08eJDWrl1LDoeDPv74Y9ti7I+4161bR2+++SYdOXKE9u7dSwsXLiS3200HDhxIatxtbW2h9zAAevTRR2nfvn30xRdfEBHRqlWr6Lbbbgsd/9lnn1FmZiY98MADdPDgQXr66adJVVV64403Ujruxx57jF577TU6fPgwffzxx3TfffeRoij01ltvJTXupUuXUl5eHlVXV4d9Xns8ntAxqfoeTxVp33Bv3ryZAPS6BdXV1REAeuedd0L7Ojs76cc//jENGTKEMjMz6cYbbwxr7JNh0aJFvcbdNU4AtHnzZiIi8ng8dO2111JhYSE5HA4aPXo03XnnnaEPxlSNm8iaEvbLX/6SioqKyOVy0ezZs6mmpiapcRMRNTU10S233ELZ2dmUm5tLd9xxR9gXju7vlfr6errmmmuooKCAXC4XjRs3jh544AFqaWmxNc4nn3ySRo0aRU6nk6ZNm0bvv/9+6HczZsygRYsWhR3/5z//mS655BJyOp102WWX0V//+ldb44sklrhXrFgROraoqIiuu+46+vDDD5Mec3CaVPctGOuiRYtoxowZPc6ZNGkSOZ1OGjNmTNh7PVXjXr9+PY0dO5bcbjcVFBTQzJkz6e2330563JE+r7vew1R+j6cCXtaTMcYYSyMpO6qcMcYYYz1xw80YY4ylEW64GWOMsTTCDTdjjDGWRrjhZowxxtIIN9yMMcZYGuGGmzHGGEsj3HAzxhhjaYQbbsYYYyyNcMPNGGOMpRFuuBljjLE08v8BJsZfm52zpEsAAAAASUVORK5CYII=",
      "text/plain": [
       "<Figure size 640x480 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig = modulescicompclass.mandelbrot(128)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## pip Deployment \n",
    "\n",
    "Making your code pip-installable is now trivial. The one thing left to do is uploading your code to https://pypi.org/ which is where pip looks for information. \n",
    "\n",
    "Pypi (Python Package Index) is beautiful because it's open and free, everybody can make their code available! (but that's also free territory for [frauds](https://www.welivesecurity.com/en/eset-research/pernicious-potpourri-python-packages-pypi/), so careful with what you install).\n",
    "\n",
    "\n",
    "You need to create an account on pypi and generate an API token. Instructions [here](https://packaging.python.org/en/latest/tutorials/packaging-projects/).\n",
    "\n",
    "This involves writing a `.pypirc` file in your home directory\n",
    "\n",
    "```\n",
    "[distutils]\n",
    "index-servers =\n",
    "    pypi\n",
    "    testpypi\n",
    "\n",
    "[pypi]\n",
    "username = __token__\n",
    "password: YOUR TOKEN GOES HERE\n",
    "\n",
    "[testpypi]\n",
    "username = __token__\n",
    "password: YOUR TOKEN GOES HERE\n",
    "```\n",
    "\n",
    "\n",
    " Then create the distribution bundle with \n",
    "\n",
    "```python setup.py sdist``` to create the distribution bundle\n",
    "\n",
    "Pypi has a test server https://test.pypi.org/ where you can try things out before going to the real pip repository (don't want to mess things up...). Upload with\n",
    "\n",
    "\n",
    "```twine upload --repository testpypi dist/* #for the test server```\n",
    "\n",
    "When you're confident everything works\n",
    "\n",
    "```twine upload dist/* #for the real pip```\n",
    "\n",
    "and you're done! Everybody will be able to pip-install and use your code. \n",
    "\n",
    "Here we are: https://test.pypi.org/project/modulescicompclass/\n",
    "\n",
    "Anyone can now install this code!\n",
    "\n",
    "\n",
    "```pip install --index-url https://test.pypi.org/simple modulescicompclass```\n",
    "\n",
    "If you upload it to the real pip than this becomes\n",
    "\n",
    "```pip install modulescicompclass```\n",
    "\n",
    "\n",
    "\n",
    "For a complete example from my research have a look at https://github.com/dgerosa/precession which is indeed pip installable. \n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Important** \n",
    "- Modules names on pip are unique. When you upload a package for the first time, that name belongs to you and you only. For this reason, when trying to run these instructions you won't be able to upload a package called `modulescicompclass` because that name is mine now. Just change the name.\n",
    "- Versioning is also unique. You *cannot* replace a package on pip, only release a newer code bumping the version up.  \n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "\n",
    "<div class=\"alert alert-block alert-warning\">\n",
    "\n",
    "Let me say here that producing good, well documented, and reusable code is *crucial* for being influential in modern research. No point understanding a beautiful piece of physics if nobody else can use it!  \n",
    "    \n",
    "</div>\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Tools to Make Your Life Easier\n",
    "\n",
    "## Version control\n",
    "\n",
    "Put your project under git version control. Just do it. \n",
    "\n",
    "\n",
    "## Code checkers\n",
    "\n",
    "There are a number of tools that help check code for formatting and\n",
    "syntax errors that are quite useful for developers.  Many projects\n",
    "automatically enforce these tools on changes submitted to github.\n",
    "\n",
    "Many editors have plugins that can automatically run these tools\n",
    "as your write your code.\n",
    "\n",
    "* [flake8](https://flake8.pycqa.org/en/latest/)\n",
    "\n",
    "  `flake8` is a checker for [PEP 8](https://peps.python.org/pep-0008/)\n",
    "  style conformance.  You can turn off checks that you don't like\n",
    "  via a [`.flake8`\n",
    "  file](https://flake8.pycqa.org/en/latest/user/configuration.html#configuration-locations).\n",
    "\n",
    "* [pylint](https://pypi.org/project/pylint/)\n",
    "\n",
    "  `pylint` is a static code analyzer.  It can find errors and also suggest improvements\n",
    "  to your code.  You can [generate a configuration file](https://pylint.readthedocs.io/en/latest/user_guide/configuration/index.html)\n",
    "  to customize its behavior (or add a section to `pyproject.toml`).\n",
    "\n",
    "* [black](https://pypi.org/project/black/)\n",
    "\n",
    "  `black` is an _uncompromising code formatted_.  It will automatically rewrite your code\n",
    "  based on PEP-8 style.\n",
    "\n",
    "* [pyupgrade](https://github.com/asottile/pyupgrade)\n",
    "\n",
    "  `pyupgrade` will upgrade source to a later python standard, making\n",
    "  use of new features where available.  For instance, you can run as:\n",
    "\n",
    "  ```\n",
    "  pyupgrade --py39-plus file.py\n",
    "  ```\n",
    "\n",
    "  to update to python 3.9 support.\n",
    "\n",
    "* [isort](https://pycqa.github.io/isort/)\n",
    "\n",
    "  `isort` simply sorts the module imports at the top of your modules,\n",
    "  grouping the standard python ones together followed by\n",
    "  package-specific ones.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Part 2. Unit testing with pytest"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Testing is an integral part of the software development process.  We want to catch\n",
    "mistakes early, before the go on to affect our results.\n",
    "\n",
    "## Types of testing\n",
    "\n",
    "There are a lot of different types of software testing that exist.\n",
    "Most commonly, for scientific codes, we hear about:\n",
    "\n",
    "* Unit testing : Tests that a single function does what it was designed to do\n",
    "\n",
    "* Integration testing : Tests whether the individual pieces work together as intended.\n",
    "  Sometimes done one piece at a time (iteratively)\n",
    "\n",
    "* Regression testing : Checks whether changes have changed answers\n",
    "\n",
    "* Verification & Validation (from the science perspective)\n",
    "\n",
    "  * Verification: are we solving the equations correctly?\n",
    "\n",
    "  * Validation: are we solving the correct equations?\n",
    "\n",
    "## Automating testing\n",
    "\n",
    "The best testing is automated.  Github provides a *continuous integration* service that can\n",
    "be run on pull requests.  You write a short definition (a Github workflow) that tells Github\n",
    "how to run your tests and then any time there is a change, the tests are run.\n",
    "\n",
    "## Unit testing\n",
    "\n",
    "* When to write tests?\n",
    "\n",
    "  * Some people advocate writing a unit test for a specification\n",
    "    before you write the functions they will test\n",
    "\n",
    "    * This is called Test-driven development (TDD):\n",
    "      https://en.wikipedia.org/wiki/Test-driven_development\n",
    "\n",
    "  * This helps you understand the interface, return values,\n",
    "    side-effects, etc. of what you intend to write\n",
    "\n",
    "* Often we already have code, so we can start by writing tests to\n",
    "  cover some core functionality\n",
    "\n",
    "  * Add new tests when you encounter a bug, precisely to ensure that\n",
    "    this bug doesn't arise again\n",
    "\n",
    "* Tests should be short\n",
    "\n",
    "  * You want to be able to run them frequently\n",
    "\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# pytest\n",
    "\n",
    "`pytest` is a unit testing framework for python code.\n",
    "\n",
    "Basic elements:\n",
    "\n",
    "* Discoverability: it will find the tests\n",
    "\n",
    "* Automation\n",
    "\n",
    "* Fixtures (setup and teardown)\n",
    "\n",
    "## Installing\n",
    "\n",
    "You can install `pytest` for a single user as:\n",
    "\n",
    "```\n",
    "pip install pytest\n",
    "```\n",
    "\n",
    "This should put `pytest` in your search path, likely in `~/.local/bin`.\n",
    "\n",
    "If you want to generate coverage reports, you should also install `pytest-cov`:\n",
    "\n",
    "```\n",
    "pip install pytest-cov\n",
    "```\n",
    "\n",
    "## Test discovery\n",
    "\n",
    "Adhering to these naming conventions will ensure that your tests are automatically found:\n",
    "\n",
    "* File names should start or end with “test”:\n",
    "\n",
    "  * `test_example.py`\n",
    "  * `example_test.py`\n",
    "\n",
    "* For tests in a class, the class name should begin with `Test`\n",
    "\n",
    "  * e.g., `TestExample`\n",
    "  * There should be no `__init__()`\n",
    "\n",
    "* Test method / function names should start with `test_`\n",
    "\n",
    "  * e.g., `test_example()`\n",
    "\n",
    "## Assertions\n",
    "\n",
    "Tests use assertions (via python’s `assert` statement) to check behavior at runtime\n",
    "\n",
    "* https://docs.python.org/3/reference/simple_stmts.html#assert \n",
    "\n",
    "* Basic usage: `assert expression`\n",
    "\n",
    "  * Raises `AssertionError` if expression is not true\n",
    "\n",
    "  * e.g., `assert 1 == 0` will fail with an exception\n",
    "\n",
    "## Simple pytest example\n",
    "\n",
    "Create a file named `multiply_test.py` with the following content:\n",
    "\n",
    "```python\n",
    "def multiply(a, b):\n",
    "    return a*b\n",
    "\n",
    "def test_multiply():\n",
    "    assert multiply(4, 6) == 24\n",
    "\n",
    "def test_multiply2():\n",
    "    assert multiply(5, 6) == 2\n",
    "```\n",
    "\n",
    "then we can run the tests as:\n",
    "\n",
    "```\n",
    "pytest -v\n",
    "```\n",
    "\n",
    "and we get the output:\n",
    "\n",
    "```\n",
    "============================= test session starts ==============================\n",
    "platform linux -- Python 3.11.3, pytest-7.2.2, pluggy-1.0.0 -- /usr/bin/python3\n",
    "cachedir: .pytest_cache\n",
    "rootdir: /home/zingale/temp/pytest\n",
    "plugins: anyio-3.6.2\n",
    "collected 2 items                                                              \n",
    "\n",
    "test_simple.py::test_multiply PASSED                                     [ 50%]\n",
    "test_simple.py::test_multiply2 FAILED                                    [100%]\n",
    "\n",
    "=================================== FAILURES ===================================\n",
    "________________________________ test_multiply2 ________________________________\n",
    "\n",
    "    def test_multiply2():\n",
    ">       assert multiply(5, 6) == 2\n",
    "E       assert 30 == 2\n",
    "E        +  where 30 = multiply(5, 6)\n",
    "\n",
    "test_simple.py:8: AssertionError\n",
    "=========================== short test summary info ============================\n",
    "FAILED test_simple.py::test_multiply2 - assert 30 == 2\n",
    "========================= 1 failed, 1 passed in 0.04s ==========================\n",
    "```\n",
    "\n",
    "this is telling us that one of our tests has failed.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# More pytest\n",
    "\n",
    "Unit tests sometimes require some setup to be done before the test is run.  Fixtures\n",
    "provide this capability.\n",
    "\n",
    "pytest provides `setup` and `teardown` functions/methods for tests --\n",
    "see https://docs.pytest.org/en/6.2.x/fixture.html for more details\n",
    "\n",
    "Note:  By default, pytest will capture stdout and only show it on failures.  If you want\n",
    "to always show stdout, add the `-s` flag.\n",
    "\n",
    "\n",
    "## Example class\n",
    "\n",
    "It is common to use a class to organize a set of related unit tests.  This is\n",
    "not a full-fledged class -- it simply helps to organize data.  In particular,\n",
    "there is no constructor, `__init__()`.  See https://stackoverflow.com/questions/21430900/py-test-skips-test-class-if-constructor-is-defined\n",
    "\n",
    "We'll look at an example with a NumPy array\n",
    "\n",
    "* We always want the array to exist for our tests, so we'll use\n",
    "  fixtures (in particular `setup_method()`) to create the array\n",
    "\n",
    "* Using a class means that we can access the array created in setup from our class.\n",
    "\n",
    "* We'll use numpy's own assertion functions: https://numpy.org/doc/stable/reference/routines.testing.html\n",
    "\n",
    "\n",
    "Here's an example:\n",
    "\n",
    "```python\n",
    "# a test class is useful to hold data that we might want setup\n",
    "# for every test.\n",
    "\n",
    "import numpy as np\n",
    "from numpy.testing import assert_array_equal\n",
    "\n",
    "class TestClassExample:\n",
    "\n",
    "    @classmethod\n",
    "    def setup_class(cls):\n",
    "        \"\"\" this is run once for each class, before any tests \"\"\"\n",
    "        pass\n",
    "\n",
    "    @classmethod\n",
    "    def teardown_class(cls):\n",
    "        \"\"\" this is run once for each class, after all tests \"\"\"\n",
    "        pass\n",
    "\n",
    "    def setup_method(self):\n",
    "        \"\"\" this is run before each of the test methods \"\"\"\n",
    "        self.a = np.arange(24).reshape(6, 4)\n",
    "\n",
    "    def teardown_method(self):\n",
    "        \"\"\" this is run after each of the test methods \"\"\"\n",
    "        pass\n",
    "\n",
    "    def test_max(self):\n",
    "        assert self.a.max() == 23\n",
    "\n",
    "    def test_flat(self):\n",
    "        assert_array_equal(self.a.flat, np.arange(24))\n",
    "```\n",
    "\n",
    "\n",
    "Here we see the [`@classmethod` decorator](https://docs.python.org/3/library/functions.html#classmethod).\n",
    "This means that the function receives the class itself as the first argument rather then an instance,\n",
    "e.g., `self`.\n",
    "\n",
    "\n",
    "Put this into a file called `test_class.py` and then we can run as:\n",
    "\n",
    "```\n",
    "pytest -v\n",
    "```\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Modern workflow\n",
    "\n",
    "Modern code tested is collaborative and automated. Say you found a bug in numpy and want to fix it. You can only pull request to their git repo only if your patch passes a whole suite of tests.\n",
    "\n",
    "That's for big things, but I argue having some basic unit tests is important for every code, even your little PhD project (and well... if you end up having a pull request merged into numpy, that should go straight into your CV!!!)\n",
    "\n",
    "\n",
    "## Other types of tests\n",
    "\n",
    "Unit tests are only one form of testing - they test a function in\n",
    "isolation of others.  Sometimes we need to test everything working together.\n",
    "For scientific codes, regression testing is often used.  The basic workflow\n",
    "is:\n",
    "\n",
    "* Start with the project working in a way you are happy with\n",
    "\n",
    "* Store the output of one (or more) runs as a _benchmark_.\n",
    "\n",
    "* Each time you make changes, run the code and compare the new output\n",
    "  to the stored benchmark.\n",
    "\n",
    "  * If there are no differences, then your changes are likely good\n",
    "    (but there is always the case of some feature not being tested).\n",
    "\n",
    "  * If there are differences, then either you introduced a bug, in which\n",
    "    case you should fix it, or you fixed a bug, in which case you should\n",
    "    update the benchmarks.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Part 3: profiling\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Profiling is the art of figuring out where you're wasting time. If your code is too slow, it's very very rare that the entire code is slow. Often you have one or a few bottlenecks, and rewriting those little pieces will make your life beautiful again. You could then try to make a smarter use on numpy, jit them with Numba, or make them run in parallel, as appropriate."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The standard python profiling tool is called [cProfile](https://docs.python.org/3/library/profile.html).\n",
    "\n",
    "This is a random example:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "from scipy.optimize import minimize\n",
    "\n",
    "\n",
    "def myfunc():\n",
    "    # c.f. https://docs.scipy.org/doc/scipy/reference/generated/scipy.optimize.minimize.html\n",
    "    func = lambda x: (x[0] - 1) ** 2 + (x[1] - 2.5) ** 2\n",
    "    x0 = (2, 0)\n",
    "\n",
    "    constraints = (\n",
    "        {\"type\": \"ineq\", \"fun\": lambda x: x[0] - 2 * x[1] + 2},\n",
    "        {\"type\": \"ineq\", \"fun\": lambda x: -x[0] - 2 * x[1] + 6},\n",
    "        {\"type\": \"ineq\", \"fun\": lambda x: -x[0] + 2 * x[1] + 2},\n",
    "    )\n",
    "\n",
    "    bounds = ((0, None), (0, None))\n",
    "\n",
    "    result = minimize(func, x0, method=\"SLSQP\", bounds=bounds, constraints=constraints)\n",
    "    print(f\"result:\\n{result}\")\n",
    "    print(f\"best fit parameters: {result.x}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "result:\n",
      " message: Optimization terminated successfully\n",
      " success: True\n",
      "  status: 0\n",
      "     fun: 0.8000000011920985\n",
      "       x: [ 1.400e+00  1.700e+00]\n",
      "     nit: 3\n",
      "     jac: [ 8.000e-01 -1.600e+00]\n",
      "    nfev: 10\n",
      "    njev: 3\n",
      "best fit parameters: [1.4 1.7]\n"
     ]
    }
   ],
   "source": [
    "myfunc()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This will tell us exactly where the time goes... \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "result:\n",
      " message: Optimization terminated successfully\n",
      " success: True\n",
      "  status: 0\n",
      "     fun: 0.8000000011920985\n",
      "       x: [ 1.400e+00  1.700e+00]\n",
      "     nit: 3\n",
      "     jac: [ 8.000e-01 -1.600e+00]\n",
      "    nfev: 10\n",
      "    njev: 3\n",
      "best fit parameters: [1.4 1.7]\n",
      "         2848 function calls (2789 primitive calls) in 0.010 seconds\n",
      "\n",
      "   Ordered by: cumulative time\n",
      "\n",
      "   ncalls  tottime  percall  cumtime  percall filename:lineno(function)\n",
      "        1    0.000    0.000    0.010    0.010 {built-in method builtins.exec}\n",
      "        1    0.000    0.000    0.010    0.010 <string>:1(<module>)\n",
      "        1    0.000    0.000    0.010    0.010 3251406122.py:4(myfunc)\n",
      "        1    0.000    0.000    0.008    0.008 _minimize.py:45(minimize)\n",
      "        1    0.000    0.000    0.008    0.008 _slsqp_py.py:214(_minimize_slsqp)\n",
      "       12    0.001    0.000    0.004    0.000 _numdiff.py:275(approx_derivative)\n",
      "  242/198    0.001    0.000    0.003    0.000 {built-in method numpy.core._multiarray_umath.implement_array_function}\n",
      "        3    0.000    0.000    0.003    0.001 _slsqp_py.py:483(_eval_con_normals)\n",
      "        3    0.000    0.000    0.003    0.001 _slsqp_py.py:492(<listcomp>)\n",
      "        9    0.000    0.000    0.003    0.000 _slsqp_py.py:293(cjac)\n",
      "        4    0.000    0.000    0.002    0.001 _differentiable_functions.py:254(_update_grad)\n",
      "        1    0.000    0.000    0.002    0.002 _optimize.py:244(_prepare_scalar_function)\n",
      "        3    0.001    0.000    0.002    0.001 _differentiable_functions.py:170(update_grad)\n",
      "        1    0.000    0.000    0.002    0.002 _differentiable_functions.py:86(__init__)\n",
      "       12    0.001    0.000    0.001    0.000 _numdiff.py:11(_adjust_scheme_to_bounds)\n",
      "        1    0.000    0.000    0.001    0.001 _optimize.py:181(__repr__)\n",
      "     10/1    0.000    0.000    0.001    0.001 _optimize.py:106(_dict_formatter)\n",
      "        1    0.000    0.000    0.001    0.001 _optimize.py:116(<listcomp>)\n",
      "       12    0.000    0.000    0.001    0.000 _numdiff.py:566(_dense_difference)\n",
      "        7    0.000    0.000    0.001    0.000 _optimize.py:344(eval)\n",
      "       20    0.000    0.000    0.001    0.000 <__array_function__ internals>:177(concatenate)\n",
      "        3    0.000    0.000    0.001    0.000 arrayprint.py:1571(_array_str_implementation)\n",
      "        3    0.000    0.000    0.001    0.000 arrayprint.py:561(array2string)\n",
      "        3    0.000    0.000    0.001    0.000 arrayprint.py:506(wrapper)\n",
      "       71    0.000    0.000    0.001    0.000 <__array_function__ internals>:177(atleast_1d)\n",
      "        3    0.000    0.000    0.001    0.000 arrayprint.py:523(_array2string)\n",
      "        3    0.000    0.000    0.001    0.000 _differentiable_functions.py:270(grad)\n",
      "        4    0.000    0.000    0.001    0.000 _slsqp_py.py:464(_eval_constraint)\n",
      "       18    0.000    0.000    0.001    0.000 {built-in method builtins.next}\n",
      "       33    0.000    0.000    0.001    0.000 _numdiff.py:455(fun_wrapped)\n",
      "       18    0.000    0.000    0.001    0.000 arrayprint.py:334(printoptions)\n",
      "       18    0.000    0.000    0.000    0.000 arrayprint.py:116(set_printoptions)\n",
      "        3    0.000    0.000    0.000    0.000 <__array_function__ internals>:177(append)\n",
      "       69    0.000    0.000    0.000    0.000 {method 'reduce' of 'numpy.ufunc' objects}\n",
      "        3    0.000    0.000    0.000    0.000 function_base.py:5444(append)\n",
      "        9    0.000    0.000    0.000    0.000 contextlib.py:132(__enter__)\n",
      "       21    0.000    0.000    0.000    0.000 arrayprint.py:64(_make_options_dict)\n",
      "       71    0.000    0.000    0.000    0.000 shape_base.py:23(atleast_1d)\n",
      "       10    0.000    0.000    0.000    0.000 _differentiable_functions.py:132(fun_wrapped)\n",
      "        6    0.000    0.000    0.000    0.000 <__array_function__ internals>:177(vstack)\n",
      "        4    0.000    0.000    0.000    0.000 _differentiable_functions.py:264(fun)\n",
      "       27    0.000    0.000    0.000    0.000 fromnumeric.py:69(_wrapreduction)\n",
      "       16    0.000    0.000    0.000    0.000 _optimize.py:351(_check_clip_x)\n",
      "        3    0.000    0.000    0.000    0.000 arrayprint.py:780(_formatArray)\n",
      "        2    0.000    0.000    0.000    0.000 {built-in method builtins.print}\n",
      "      9/3    0.000    0.000    0.000    0.000 arrayprint.py:789(recurser)\n",
      "        6    0.000    0.000    0.000    0.000 shape_base.py:223(vstack)\n",
      "        3    0.000    0.000    0.000    0.000 arrayprint.py:452(_get_format_function)\n",
      "        4    0.000    0.000    0.000    0.000 iostream.py:608(write)\n",
      "       34    0.000    0.000    0.000    0.000 {method 'any' of 'numpy.ndarray' objects}\n",
      "        2    0.000    0.000    0.000    0.000 _minimize.py:961(standardize_bounds)\n",
      "       12    0.000    0.000    0.000    0.000 <__array_function__ internals>:177(any)\n",
      "       34    0.000    0.000    0.000    0.000 _methods.py:55(_any)\n",
      "        4    0.000    0.000    0.000    0.000 iostream.py:530(_schedule_flush)\n",
      "        8    0.000    0.000    0.000    0.000 _differentiable_functions.py:249(_update_fun)\n",
      "       13    0.000    0.000    0.000    0.000 <__array_function__ internals>:177(all)\n",
      "       12    0.000    0.000    0.000    0.000 <__array_function__ internals>:177(ones_like)\n",
      "        1    0.000    0.000    0.000    0.000 iostream.py:241(schedule)\n",
      "        1    0.000    0.000    0.000    0.000 arrayprint.py:411(<lambda>)\n",
      "        4    0.000    0.000    0.000    0.000 _differentiable_functions.py:154(update_fun)\n",
      "        1    0.000    0.000    0.000    0.000 arrayprint.py:907(__init__)\n",
      "        9    0.000    0.000    0.000    0.000 contextlib.py:141(__exit__)\n",
      "        1    0.000    0.000    0.000    0.000 arrayprint.py:934(fillFormat)\n",
      "       12    0.000    0.000    0.000    0.000 fromnumeric.py:2333(any)\n",
      "       12    0.000    0.000    0.000    0.000 numeric.py:219(ones_like)\n",
      "        4    0.000    0.000    0.000    0.000 _optimize.py:93(_float_formatter_10)\n",
      "       12    0.000    0.000    0.000    0.000 <__array_function__ internals>:177(diag)\n",
      "        1    0.000    0.000    0.000    0.000 _constraints.py:249(__init__)\n",
      "        4    0.000    0.000    0.000    0.000 _slsqp_py.py:473(<listcomp>)\n",
      "       13    0.000    0.000    0.000    0.000 fromnumeric.py:2432(all)\n",
      "        1    0.000    0.000    0.000    0.000 socket.py:621(send)\n",
      "        7    0.000    0.000    0.000    0.000 <__array_function__ internals>:177(array_equal)\n",
      "       10    0.000    0.000    0.000    0.000 3251406122.py:6(<lambda>)\n",
      "       11    0.000    0.000    0.000    0.000 <__array_function__ internals>:177(copy)\n",
      "        1    0.000    0.000    0.000    0.000 _constraints.py:241(_input_validation)\n",
      "       12    0.000    0.000    0.000    0.000 {method 'copy' of 'numpy.ndarray' objects}\n",
      "        1    0.000    0.000    0.000    0.000 <__array_function__ internals>:177(broadcast_arrays)\n",
      "       12    0.000    0.000    0.000    0.000 twodim_base.py:240(diag)\n",
      "        1    0.000    0.000    0.000    0.000 stride_tricks.py:480(broadcast_arrays)\n",
      "       12    0.000    0.000    0.000    0.000 <__array_function__ internals>:177(where)\n",
      "       15    0.000    0.000    0.000    0.000 <__array_function__ internals>:177(ravel)\n",
      "        7    0.000    0.000    0.000    0.000 numeric.py:2407(array_equal)\n",
      "        5    0.000    0.000    0.000    0.000 stride_tricks.py:340(_broadcast_to)\n",
      "       46    0.000    0.000    0.000    0.000 {built-in method numpy.array}\n",
      "       17    0.000    0.000    0.000    0.000 {method 'astype' of 'numpy.ndarray' objects}\n",
      "       21    0.000    0.000    0.000    0.000 {built-in method builtins.locals}\n",
      "        1    0.000    0.000    0.000    0.000 stride_tricks.py:546(<listcomp>)\n",
      "       11    0.000    0.000    0.000    0.000 function_base.py:871(copy)\n",
      "       12    0.000    0.000    0.000    0.000 <__array_function__ internals>:177(empty_like)\n",
      "        4    0.000    0.000    0.000    0.000 <__array_function__ internals>:177(isposinf)\n",
      "        7    0.000    0.000    0.000    0.000 <__array_function__ internals>:177(atleast_2d)\n",
      "        8    0.000    0.000    0.000    0.000 ufunclike.py:48(func)\n",
      "        9    0.000    0.000    0.000    0.000 contextlib.py:287(helper)\n",
      "       15    0.000    0.000    0.000    0.000 fromnumeric.py:1781(ravel)\n",
      "        1    0.000    0.000    0.000    0.000 <__array_function__ internals>:177(clip)\n",
      "      117    0.000    0.000    0.000    0.000 {built-in method numpy.asanyarray}\n",
      "       85    0.000    0.000    0.000    0.000 {built-in method builtins.isinstance}\n",
      "       12    0.000    0.000    0.000    0.000 <__array_function__ internals>:177(copyto)\n",
      "        1    0.000    0.000    0.000    0.000 fromnumeric.py:2111(clip)\n",
      "        7    0.000    0.000    0.000    0.000 {method 'all' of 'numpy.ndarray' objects}\n",
      "       12    0.000    0.000    0.000    0.000 _numdiff.py:193(_prepare_bounds)\n",
      "        1    0.000    0.000    0.000    0.000 fromnumeric.py:51(_wrapfunc)\n",
      "       52    0.000    0.000    0.000    0.000 {built-in method numpy.asarray}\n",
      "       21    0.000    0.000    0.000    0.000 arrayprint.py:72(<dictcomp>)\n",
      "        4    0.000    0.000    0.000    0.000 _ufunc_config.py:33(seterr)\n",
      "        2    0.000    0.000    0.000    0.000 _ufunc_config.py:430(__enter__)\n",
      "        1    0.000    0.000    0.000    0.000 _constraints.py:389(new_bounds_to_old)\n",
      "        1    0.000    0.000    0.000    0.000 _optimize.py:200(item_sorter)\n",
      "        7    0.000    0.000    0.000    0.000 _methods.py:61(_all)\n",
      "        1    0.000    0.000    0.000    0.000 {method 'clip' of 'numpy.ndarray' objects}\n",
      "       53    0.000    0.000    0.000    0.000 {method 'reshape' of 'numpy.ndarray' objects}\n",
      "        1    0.000    0.000    0.000    0.000 {built-in method builtins.sorted}\n",
      "        7    0.000    0.000    0.000    0.000 shape_base.py:81(atleast_2d)\n",
      "        1    0.000    0.000    0.000    0.000 _methods.py:127(_clip)\n",
      "        9    0.000    0.000    0.000    0.000 contextlib.py:104(__init__)\n",
      "        4    0.000    0.000    0.000    0.000 <__array_function__ internals>:177(isneginf)\n",
      "       18    0.000    0.000    0.000    0.000 arrayprint.py:295(get_printoptions)\n",
      "        3    0.000    0.000    0.000    0.000 arrayprint.py:403(_get_formatdict)\n",
      "        6    0.000    0.000    0.000    0.000 arrayprint.py:753(_extendLine_pretty)\n",
      "        4    0.000    0.000    0.000    0.000 ufunclike.py:127(isposinf)\n",
      "        2    0.000    0.000    0.000    0.000 _constraints.py:407(old_bound_to_new)\n",
      "        3    0.000    0.000    0.000    0.000 _differentiable_functions.py:240(update_x)\n",
      "        1    0.000    0.000    0.000    0.000 _slsqp_py.py:329(<listcomp>)\n",
      "       27    0.000    0.000    0.000    0.000 fromnumeric.py:70(<dictcomp>)\n",
      "        1    0.000    0.000    0.000    0.000 <__array_function__ internals>:177(asfarray)\n",
      "        1    0.000    0.000    0.000    0.000 <__array_function__ internals>:177(amax)\n",
      "       12    0.000    0.000    0.000    0.000 _numdiff.py:204(<listcomp>)\n",
      "      161    0.000    0.000    0.000    0.000 {built-in method builtins.len}\n",
      "        2    0.000    0.000    0.000    0.000 <__array_function__ internals>:177(broadcast_to)\n",
      "        4    0.000    0.000    0.000    0.000 arrayprint.py:1050(format_float_scientific)\n",
      "        1    0.000    0.000    0.000    0.000 type_check.py:84(asfarray)\n",
      "        9    0.000    0.000    0.000    0.000 _optimize.py:84(_indenter)\n",
      "       24    0.000    0.000    0.000    0.000 {built-in method numpy.zeros}\n",
      "        1    0.000    0.000    0.000    0.000 fromnumeric.py:2703(amax)\n",
      "        6    0.000    0.000    0.000    0.000 shape_base.py:218(_vhstack_dispatcher)\n",
      "        3    0.000    0.000    0.000    0.000 arrayprint.py:989(<genexpr>)\n",
      "       18    0.000    0.000    0.000    0.000 <frozen abc>:117(__instancecheck__)\n",
      "       14    0.000    0.000    0.000    0.000 3251406122.py:10(<lambda>)\n",
      "        2    0.000    0.000    0.000    0.000 stride_tricks.py:367(broadcast_to)\n",
      "        9    0.000    0.000    0.000    0.000 _optimize.py:188(key)\n",
      "        1    0.000    0.000    0.000    0.000 threading.py:1185(is_alive)\n",
      "        2    0.000    0.000    0.000    0.000 _methods.py:93(_clip_dep_is_scalar_nan)\n",
      "       14    0.000    0.000    0.000    0.000 3251406122.py:11(<lambda>)\n",
      "        1    0.000    0.000    0.000    0.000 numerictypes.py:356(issubdtype)\n",
      "       10    0.000    0.000    0.000    0.000 numeric.py:1878(isscalar)\n",
      "        3    0.000    0.000    0.000    0.000 arrayprint.py:984(<genexpr>)\n",
      "       87    0.000    0.000    0.000    0.000 {method 'append' of 'list' objects}\n",
      "        4    0.000    0.000    0.000    0.000 ufunclike.py:199(isneginf)\n",
      "        6    0.000    0.000    0.000    0.000 shape_base.py:207(_arrays_for_stack_dispatcher)\n",
      "        1    0.000    0.000    0.000    0.000 {method 'max' of 'numpy.ndarray' objects}\n",
      "        1    0.000    0.000    0.000    0.000 <__array_function__ internals>:177(amin)\n",
      "       14    0.000    0.000    0.000    0.000 3251406122.py:12(<lambda>)\n",
      "        4    0.000    0.000    0.000    0.000 _ufunc_config.py:132(geterr)\n",
      "        1    0.000    0.000    0.000    0.000 _methods.py:39(_amax)\n",
      "       18    0.000    0.000    0.000    0.000 {built-in method builtins.abs}\n",
      "        2    0.000    0.000    0.000    0.000 numerictypes.py:282(issubclass_)\n",
      "        6    0.000    0.000    0.000    0.000 arrayprint.py:739(_extendLine)\n",
      "        2    0.000    0.000    0.000    0.000 _ufunc_config.py:435(__exit__)\n",
      "        4    0.000    0.000    0.000    0.000 iostream.py:503(_is_master_process)\n",
      "       71    0.000    0.000    0.000    0.000 shape_base.py:19(_atleast_1d_dispatcher)\n",
      "        6    0.000    0.000    0.000    0.000 {built-in method builtins.max}\n",
      "        8    0.000    0.000    0.000    0.000 ufunclike.py:22(func)\n",
      "       12    0.000    0.000    0.000    0.000 {built-in method numpy.empty}\n",
      "       21    0.000    0.000    0.000    0.000 {method 'update' of 'dict' objects}\n",
      "        1    0.000    0.000    0.000    0.000 fromnumeric.py:2829(amin)\n",
      "       49    0.000    0.000    0.000    0.000 {method 'items' of 'dict' objects}\n",
      "       18    0.000    0.000    0.000    0.000 {built-in method _abc._abc_instancecheck}\n",
      "        1    0.000    0.000    0.000    0.000 stride_tricks.py:416(_broadcast_shape)\n",
      "       15    0.000    0.000    0.000    0.000 {method 'ravel' of 'numpy.ndarray' objects}\n",
      "        5    0.000    0.000    0.000    0.000 function_base.py:346(iterable)\n",
      "        2    0.000    0.000    0.000    0.000 _constraints.py:420(<listcomp>)\n",
      "        2    0.000    0.000    0.000    0.000 arrayprint.py:1018(__call__)\n",
      "        4    0.000    0.000    0.000    0.000 {built-in method numpy.core._multiarray_umath.dragon4_scientific}\n",
      "        1    0.000    0.000    0.000    0.000 _minimize.py:973(standardize_constraints)\n",
      "        1    0.000    0.000    0.000    0.000 _constraints.py:401(<listcomp>)\n",
      "        1    0.000    0.000    0.000    0.000 threading.py:1118(_wait_for_tstate_lock)\n",
      "        2    0.000    0.000    0.000    0.000 <__array_function__ internals>:177(ndim)\n",
      "        4    0.000    0.000    0.000    0.000 {built-in method numpy.core._multiarray_umath.dragon4_positional}\n",
      "       21    0.000    0.000    0.000    0.000 {method 'copy' of 'dict' objects}\n",
      "        5    0.000    0.000    0.000    0.000 {built-in method builtins.any}\n",
      "        1    0.000    0.000    0.000    0.000 _methods.py:108(_clip_dep_invoke_with_casting)\n",
      "        4    0.000    0.000    0.000    0.000 {built-in method numpy.seterrobj}\n",
      "       18    0.000    0.000    0.000    0.000 {built-in method numpy.core._multiarray_umath.set_legacy_print_mode}\n",
      "       10    0.000    0.000    0.000    0.000 _optimize.py:194(omit_redundant)\n",
      "        8    0.000    0.000    0.000    0.000 _constraints.py:12(_arr_to_scalar)\n",
      "        6    0.000    0.000    0.000    0.000 {built-in method builtins.hasattr}\n",
      "        9    0.000    0.000    0.000    0.000 {method 'index' of 'list' objects}\n",
      "       20    0.000    0.000    0.000    0.000 multiarray.py:152(concatenate)\n",
      "       17    0.000    0.000    0.000    0.000 arrayprint.py:898(_none_or_positive_arg)\n",
      "       10    0.000    0.000    0.000    0.000 {built-in method builtins.getattr}\n",
      "        1    0.000    0.000    0.000    0.000 {method 'flatten' of 'numpy.ndarray' objects}\n",
      "       11    0.000    0.000    0.000    0.000 {method 'split' of 'str' objects}\n",
      "        1    0.000    0.000    0.000    0.000 _slsqp_py.py:354(<listcomp>)\n",
      "        1    0.000    0.000    0.000    0.000 iostream.py:125(_event_pipe)\n",
      "       12    0.000    0.000    0.000    0.000 fromnumeric.py:2328(_any_dispatcher)\n",
      "       10    0.000    0.000    0.000    0.000 {method 'join' of 'str' objects}\n",
      "       15    0.000    0.000    0.000    0.000 fromnumeric.py:1777(_ravel_dispatcher)\n",
      "       12    0.000    0.000    0.000    0.000 multiarray.py:345(where)\n",
      "        9    0.000    0.000    0.000    0.000 {method 'rjust' of 'str' objects}\n",
      "        1    0.000    0.000    0.000    0.000 {built-in method builtins.all}\n",
      "       15    0.000    0.000    0.000    0.000 {built-in method builtins.issubclass}\n",
      "        2    0.000    0.000    0.000    0.000 _ufunc_config.py:426(__init__)\n",
      "        8    0.000    0.000    0.000    0.000 {built-in method numpy.geterrobj}\n",
      "        1    0.000    0.000    0.000    0.000 stride_tricks.py:538(<listcomp>)\n",
      "       12    0.000    0.000    0.000    0.000 multiarray.py:84(empty_like)\n",
      "       13    0.000    0.000    0.000    0.000 fromnumeric.py:2427(_all_dispatcher)\n",
      "       13    0.000    0.000    0.000    0.000 {method 'lower' of 'str' objects}\n",
      "       12    0.000    0.000    0.000    0.000 twodim_base.py:236(_diag_dispatcher)\n",
      "       12    0.000    0.000    0.000    0.000 multiarray.py:1079(copyto)\n",
      "       11    0.000    0.000    0.000    0.000 function_base.py:867(_copy_dispatcher)\n",
      "        2    0.000    0.000    0.000    0.000 _methods.py:103(_clip_dep_is_byte_swapped)\n",
      "        6    0.000    0.000    0.000    0.000 {method 'splitlines' of 'str' objects}\n",
      "       12    0.000    0.000    0.000    0.000 numeric.py:215(_ones_like_dispatcher)\n",
      "        7    0.000    0.000    0.000    0.000 numeric.py:2403(_array_equal_dispatcher)\n",
      "        5    0.000    0.000    0.000    0.000 stride_tricks.py:25(_maybe_view_as_subclass)\n",
      "        4    0.000    0.000    0.000    0.000 {method 'write' of '_io.StringIO' objects}\n",
      "        1    0.000    0.000    0.000    0.000 _base.py:1301(isspmatrix)\n",
      "        3    0.000    0.000    0.000    0.000 _slsqp_py.py:292(cjac_factory)\n",
      "        4    0.000    0.000    0.000    0.000 stride_tricks.py:542(<genexpr>)\n",
      "        2    0.000    0.000    0.000    0.000 {built-in method builtins.sum}\n",
      "       11    0.000    0.000    0.000    0.000 {built-in method builtins.callable}\n",
      "       10    0.000    0.000    0.000    0.000 stride_tricks.py:345(<genexpr>)\n",
      "        3    0.000    0.000    0.000    0.000 arrayprint.py:993(<genexpr>)\n",
      "        2    0.000    0.000    0.000    0.000 _optimize.py:338(_clip_x_for_func)\n",
      "        9    0.000    0.000    0.000    0.000 {built-in method _operator.index}\n",
      "        5    0.000    0.000    0.000    0.000 {built-in method builtins.iter}\n",
      "        3    0.000    0.000    0.000    0.000 {method 'add' of 'set' objects}\n",
      "        8    0.000    0.000    0.000    0.000 ufunclike.py:68(_dispatcher)\n",
      "        4    0.000    0.000    0.000    0.000 arrayprint.py:427(indirect)\n",
      "        3    0.000    0.000    0.000    0.000 {built-in method builtins.id}\n",
      "        4    0.000    0.000    0.000    0.000 {method '__exit__' of '_thread.RLock' objects}\n",
      "        2    0.000    0.000    0.000    0.000 _constraints.py:422(<listcomp>)\n",
      "       10    0.000    0.000    0.000    0.000 {method 'get' of 'dict' objects}\n",
      "        1    0.000    0.000    0.000    0.000 _optimize.py:172(__getattr__)\n",
      "        3    0.000    0.000    0.000    0.000 {method 'rstrip' of 'str' objects}\n",
      "        2    0.000    0.000    0.000    0.000 arrayprint.py:431(<listcomp>)\n",
      "        3    0.000    0.000    0.000    0.000 arrayprint.py:994(<genexpr>)\n",
      "        5    0.000    0.000    0.000    0.000 {method '__exit__' of 'numpy.nditer' objects}\n",
      "        6    0.000    0.000    0.000    0.000 {method 'keys' of 'dict' objects}\n",
      "        1    0.000    0.000    0.000    0.000 _constraints.py:402(<listcomp>)\n",
      "        7    0.000    0.000    0.000    0.000 shape_base.py:77(_atleast_2d_dispatcher)\n",
      "        3    0.000    0.000    0.000    0.000 {method 'discard' of 'set' objects}\n",
      "        1    0.000    0.000    0.000    0.000 {method 'acquire' of '_thread.lock' objects}\n",
      "        4    0.000    0.000    0.000    0.000 {built-in method posix.getpid}\n",
      "        3    0.000    0.000    0.000    0.000 {built-in method _thread.get_ident}\n",
      "        1    0.000    0.000    0.000    0.000 threading.py:568(is_set)\n",
      "        3    0.000    0.000    0.000    0.000 function_base.py:5440(_append_dispatcher)\n",
      "        1    0.000    0.000    0.000    0.000 {method 'disable' of '_lsprof.Profiler' objects}\n",
      "        2    0.000    0.000    0.000    0.000 fromnumeric.py:3152(ndim)\n",
      "        2    0.000    0.000    0.000    0.000 fromnumeric.py:3148(_ndim_dispatcher)\n",
      "        1    0.000    0.000    0.000    0.000 {method 'append' of 'collections.deque' objects}\n",
      "        1    0.000    0.000    0.000    0.000 _optimize.py:216(_check_unknown_options)\n",
      "        2    0.000    0.000    0.000    0.000 stride_tricks.py:363(_broadcast_to_dispatcher)\n",
      "        2    0.000    0.000    0.000    0.000 arrayprint.py:428(<lambda>)\n",
      "        1    0.000    0.000    0.000    0.000 _optimize.py:324(hess)\n",
      "        1    0.000    0.000    0.000    0.000 _slsqp_py.py:327(<listcomp>)\n",
      "        1    0.000    0.000    0.000    0.000 fromnumeric.py:2698(_amax_dispatcher)\n",
      "        1    0.000    0.000    0.000    0.000 fromnumeric.py:2107(_clip_dispatcher)\n",
      "        1    0.000    0.000    0.000    0.000 fromnumeric.py:2824(_amin_dispatcher)\n",
      "        1    0.000    0.000    0.000    0.000 type_check.py:80(_asfarray_dispatcher)\n",
      "        1    0.000    0.000    0.000    0.000 stride_tricks.py:476(_broadcast_arrays_dispatcher)\n",
      "\n",
      "\n"
     ]
    }
   ],
   "source": [
    "import cProfile \n",
    "cProfile.run('myfunc()', sort='cumtime')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "result:\n",
      " message: Optimization terminated successfully\n",
      " success: True\n",
      "  status: 0\n",
      "     fun: 0.8000000011920985\n",
      "       x: [ 1.400e+00  1.700e+00]\n",
      "     nit: 3\n",
      "     jac: [ 8.000e-01 -1.600e+00]\n",
      "    nfev: 10\n",
      "    njev: 3\n",
      "best fit parameters: [1.4 1.7]\n"
     ]
    }
   ],
   "source": [
    "# This is the same but dumps to file\n",
    "\n",
    "with cProfile.Profile() as pr:\n",
    "    myfunc()  \n",
    "\n",
    "pr.dump_stats('example.prof')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "There's a super nice interface which is called [snakeviz](https://jiffyclub.github.io/snakeviz/). Try this:\n",
    "\n",
    "```snakeviz example.prof```\n",
    "\n",
    "(might need to `pip install snakeviz` first of course)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## A complete example from my own research: merging binary black holes\n",
    "\n",
    "Let's evolve some black-hole binaries from large separations (presumably where they form following stellar collapse) to small separation where they merge and become detectable as gravitational-wave searches.\n",
    "\n",
    "This is a python module I wrote: https://github.com/dgerosa/precession\n",
    "\n",
    "The mathematical problem is described here https://arxiv.org/abs/2304.04801\n",
    "In short, we need to solve some ordinary differential equations with a RHS that contains some elliptic integrals, and in turn some of the parameters of these elliptic integrals are solutions to some (very long!) polynomials.\n",
    "\n",
    "**Warning**: don't look at the appendices of that paper before getting a good cup of coffee. And if you do, of course those equations are machine-generated! I used Mathematica to compute the equations, and then exported the results to python."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "import precession # You need to \"pip install precession\"\n",
    "import numpy as np\n",
    "\n",
    "\n",
    "def blackholes():\n",
    "    \n",
    "    N=100\n",
    "    q=np.random.uniform(0.1,1,N)\n",
    "    chi1=np.random.uniform(0.1,1,N)\n",
    "    chi2=np.random.uniform(0.1,1,N)\n",
    "    theta1=np.arccos(np.random.uniform(-1,1,N))\n",
    "    theta2=np.arccos(np.random.uniform(-1,1,N))\n",
    "    deltaphi=np.random.uniform(-np.pi,np.pi,N)\n",
    "    r=[1000,10]\n",
    "    \n",
    "    return  precession.inspiral_precav(theta1=theta1,theta2=theta2,deltaphi=deltaphi,r=precession.tiler(r,q),q=q,chi1=chi1,chi2=chi2)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'theta1': array([[2.94488836, 2.98059893],\n",
       "        [1.27728648, 1.27937373],\n",
       "        [1.41167124, 1.41227667],\n",
       "        [1.34513319, 1.32918237],\n",
       "        [1.23039933, 1.29841712],\n",
       "        [0.91957826, 0.81491039],\n",
       "        [0.82251795, 0.80263049],\n",
       "        [2.53998017, 2.55169422],\n",
       "        [1.94585467, 1.9448505 ],\n",
       "        [1.32668421, 1.32225819],\n",
       "        [0.66553191, 0.65025305],\n",
       "        [2.25564358, 2.2110731 ],\n",
       "        [1.0671711 , 1.03377169],\n",
       "        [1.00276138, 1.03826409],\n",
       "        [1.99342059, 2.25608523],\n",
       "        [2.94665014, 2.85932274],\n",
       "        [2.88216242, 2.79159673],\n",
       "        [0.31032776, 0.87939436],\n",
       "        [1.68350625, 1.98646191],\n",
       "        [1.85114885, 1.9557549 ],\n",
       "        [2.44996007, 2.47873371],\n",
       "        [0.29332947, 0.29452643],\n",
       "        [2.62437531, 2.40373832],\n",
       "        [2.47040882, 2.47983197],\n",
       "        [2.26340111, 2.31282969],\n",
       "        [0.51075553, 0.51959353],\n",
       "        [0.93969824, 0.61359072],\n",
       "        [2.35984278, 2.4139208 ],\n",
       "        [2.27303642, 2.30868634],\n",
       "        [1.68508016, 1.68570207],\n",
       "        [1.09802179, 1.09965191],\n",
       "        [0.82574764, 0.87490039],\n",
       "        [1.41575626, 1.34192984],\n",
       "        [1.57162995, 1.76203419],\n",
       "        [2.07707965, 2.04475525],\n",
       "        [2.7649725 , 2.78067146],\n",
       "        [1.47952062, 1.03916805],\n",
       "        [0.78926405, 0.81857943],\n",
       "        [1.88841232, 1.89734955],\n",
       "        [0.67565655, 0.93469658],\n",
       "        [0.66513661, 0.91480707],\n",
       "        [0.72891921, 0.70917001],\n",
       "        [2.75796486, 2.91659433],\n",
       "        [2.61354207, 2.73599712],\n",
       "        [1.72849483, 1.92051685],\n",
       "        [1.79978441, 1.36557361],\n",
       "        [2.7548328 , 2.71882301],\n",
       "        [1.8355719 , 1.97815812],\n",
       "        [2.8451487 , 2.7019106 ],\n",
       "        [2.71662208, 2.4455714 ],\n",
       "        [1.12916314, 1.20626866],\n",
       "        [1.19304409, 1.21780342],\n",
       "        [1.65905535, 1.67499062],\n",
       "        [0.35116022, 0.38558715],\n",
       "        [0.34274275, 0.32600652],\n",
       "        [2.61270736, 2.39530749],\n",
       "        [2.24065274, 2.26235706],\n",
       "        [0.95393631, 0.99062648],\n",
       "        [0.99680379, 1.01844483],\n",
       "        [1.07502381, 1.04943871],\n",
       "        [1.09400667, 1.02124846],\n",
       "        [0.2418988 , 0.4214667 ],\n",
       "        [2.57041344, 2.55671918],\n",
       "        [1.75639654, 1.72338819],\n",
       "        [1.22504608, 1.23580113],\n",
       "        [2.37356768, 2.3618541 ],\n",
       "        [2.08841838, 2.10399673],\n",
       "        [1.38351711, 1.31062202],\n",
       "        [0.91206773, 0.90227566],\n",
       "        [2.30537368, 2.26528162],\n",
       "        [1.98157566, 2.19785436],\n",
       "        [1.6337814 , 0.96617254],\n",
       "        [2.28969524, 2.27702402],\n",
       "        [0.25671274, 0.83324376],\n",
       "        [1.01049476, 1.01017984],\n",
       "        [1.25208774, 1.23872862],\n",
       "        [1.34278611, 1.30740659],\n",
       "        [1.23608532, 1.25837702],\n",
       "        [1.67255867, 1.49646698],\n",
       "        [2.31038879, 2.24715358],\n",
       "        [1.80274217, 2.040379  ],\n",
       "        [2.44785676, 2.47810282],\n",
       "        [1.15705419, 1.21466661],\n",
       "        [0.58699963, 0.81084249],\n",
       "        [0.12331429, 0.86021682],\n",
       "        [0.87663151, 0.71387763],\n",
       "        [1.16558737, 1.110167  ],\n",
       "        [2.20975293, 2.17064945],\n",
       "        [1.02773797, 1.02871309],\n",
       "        [2.22636156, 2.2816828 ],\n",
       "        [1.50446673, 1.54071466],\n",
       "        [0.4159308 , 0.38422822],\n",
       "        [1.15172257, 1.08591865],\n",
       "        [1.06154716, 1.04691143],\n",
       "        [0.10704752, 0.19578697],\n",
       "        [0.31152706, 0.22229927],\n",
       "        [0.43784435, 0.49689336],\n",
       "        [1.85536342, 1.83005375],\n",
       "        [1.54436786, 1.61662239],\n",
       "        [1.06427884, 1.1219122 ]]),\n",
       " 'theta2': array([[1.36082546, 1.34754764],\n",
       "        [0.4739515 , 0.44673824],\n",
       "        [3.07624785, 2.85518177],\n",
       "        [0.10985605, 0.39903896],\n",
       "        [1.83320455, 1.5895845 ],\n",
       "        [1.20837708, 1.3259848 ],\n",
       "        [1.30595612, 1.3978502 ],\n",
       "        [1.10819408, 1.05869421],\n",
       "        [0.15430298, 0.32887147],\n",
       "        [2.03523781, 2.24876281],\n",
       "        [1.05039423, 1.05232474],\n",
       "        [1.35528998, 1.42855229],\n",
       "        [1.2519148 , 1.46935901],\n",
       "        [1.52843877, 1.37617055],\n",
       "        [2.48626247, 2.17004184],\n",
       "        [2.30882426, 2.37259057],\n",
       "        [2.30420526, 2.38586274],\n",
       "        [2.85677592, 1.13042289],\n",
       "        [1.54930265, 1.14842354],\n",
       "        [2.35224225, 2.18862717],\n",
       "        [2.27471112, 2.23201751],\n",
       "        [2.41803531, 2.41540051],\n",
       "        [2.43551788, 2.55208198],\n",
       "        [1.72945805, 1.70995396],\n",
       "        [1.5898712 , 1.55639007],\n",
       "        [1.27998596, 1.25745428],\n",
       "        [0.79144399, 1.92592486],\n",
       "        [1.10422124, 0.89517196],\n",
       "        [0.82664326, 0.67797564],\n",
       "        [1.60813094, 1.60749393],\n",
       "        [0.838558  , 0.83749605],\n",
       "        [1.5068054 , 1.47046463],\n",
       "        [1.1346207 , 1.46792568],\n",
       "        [2.96001239, 2.22021755],\n",
       "        [2.44839299, 2.84110824],\n",
       "        [2.43789826, 2.42196865],\n",
       "        [0.86464247, 1.40818632],\n",
       "        [1.4421025 , 1.39519876],\n",
       "        [1.095153  , 1.0882888 ],\n",
       "        [2.83448564, 1.8584865 ],\n",
       "        [2.33991565, 2.15334695],\n",
       "        [1.75352059, 1.86819893],\n",
       "        [2.37594129, 2.27492097],\n",
       "        [2.12481583, 2.06963844],\n",
       "        [1.63081463, 1.0985033 ],\n",
       "        [1.02554194, 1.89521202],\n",
       "        [0.58165652, 0.61473649],\n",
       "        [2.62382856, 2.03969402],\n",
       "        [2.29839622, 2.35266806],\n",
       "        [2.25247676, 2.39879655],\n",
       "        [2.33932007, 2.20334759],\n",
       "        [2.21097277, 2.16216415],\n",
       "        [2.3665982 , 2.33318421],\n",
       "        [0.5516106 , 0.42263381],\n",
       "        [1.43481685, 1.49278643],\n",
       "        [2.32639674, 2.50483993],\n",
       "        [1.33896126, 1.22546601],\n",
       "        [1.45108601, 1.111864  ],\n",
       "        [2.72956425, 2.701991  ],\n",
       "        [0.61477255, 0.69881532],\n",
       "        [0.37025656, 0.71890092],\n",
       "        [1.19710479, 1.13619418],\n",
       "        [2.35848358, 2.44845107],\n",
       "        [1.25611883, 1.42954272],\n",
       "        [1.50540405, 1.14585292],\n",
       "        [2.72694618, 2.75257514],\n",
       "        [1.96982669, 1.93140212],\n",
       "        [2.23359766, 2.2973559 ],\n",
       "        [2.75422009, 3.0561386 ],\n",
       "        [2.41176882, 2.4889056 ],\n",
       "        [1.47975138, 1.42589038],\n",
       "        [1.75303086, 2.43933533],\n",
       "        [1.87099026, 1.88578335],\n",
       "        [1.79680455, 1.4015198 ],\n",
       "        [0.45080407, 0.47442928],\n",
       "        [1.54998879, 1.63508075],\n",
       "        [1.97743659, 2.08346872],\n",
       "        [0.91924777, 0.86528413],\n",
       "        [1.96262495, 2.1207539 ],\n",
       "        [0.98663886, 1.14421465],\n",
       "        [1.93504299, 1.45321925],\n",
       "        [0.79287239, 0.66039247],\n",
       "        [1.54969775, 0.78401818],\n",
       "        [2.52270876, 2.3099163 ],\n",
       "        [2.32819778, 1.28222851],\n",
       "        [1.63593369, 2.07975427],\n",
       "        [0.61135388, 0.95053076],\n",
       "        [0.82029042, 0.89444414],\n",
       "        [1.46803207, 1.46494461],\n",
       "        [1.67752551, 1.61272682],\n",
       "        [1.45934688, 1.4391622 ],\n",
       "        [0.91091209, 0.94320713],\n",
       "        [0.66298454, 0.70293584],\n",
       "        [0.58177489, 0.60708264],\n",
       "        [1.21669737, 1.16927668],\n",
       "        [1.57241411, 1.59509848],\n",
       "        [0.63875725, 0.55568664],\n",
       "        [2.04850776, 2.11820081],\n",
       "        [1.92238057, 1.86305705],\n",
       "        [1.65713404, 1.58761959]]),\n",
       " 'deltaphi': array([[-0.05103117, -1.47707862],\n",
       "        [ 0.22644287, -1.57423232],\n",
       "        [ 0.96771608, -0.09164498],\n",
       "        [ 2.6517007 , -2.95232068],\n",
       "        [-1.23380578,  0.09346297],\n",
       "        [-3.04568903, -2.84780302],\n",
       "        [ 0.02754501,  1.69915022],\n",
       "        [-2.52106293,  1.51257784],\n",
       "        [ 0.39973095,  2.54397587],\n",
       "        [ 2.46036693,  2.74438899],\n",
       "        [ 2.20991406,  1.73960985],\n",
       "        [-0.79468044,  2.50945241],\n",
       "        [ 1.16006005,  2.9860068 ],\n",
       "        [ 2.91845063, -0.30873573],\n",
       "        [-2.56558673, -0.65650693],\n",
       "        [ 2.03374145,  2.56552416],\n",
       "        [ 0.38106152,  2.75979334],\n",
       "        [ 1.30748749, -0.27079656],\n",
       "        [-1.59314012, -1.25269073],\n",
       "        [ 0.79396517, -0.97884368],\n",
       "        [ 0.16693904,  1.08420287],\n",
       "        [ 2.11708491,  1.43071423],\n",
       "        [-2.65110124,  2.67653872],\n",
       "        [ 1.29982095, -1.35864411],\n",
       "        [ 0.71294379, -0.90132365],\n",
       "        [ 2.14514484, -1.31744323],\n",
       "        [-0.47286417, -2.88073583],\n",
       "        [-1.40927869,  0.29762454],\n",
       "        [-0.97816014, -1.43288874],\n",
       "        [-2.45443196,  1.65183405],\n",
       "        [-0.8855572 , -1.50267584],\n",
       "        [-1.14942427, -0.95229513],\n",
       "        [ 2.50132426,  2.88522847],\n",
       "        [ 0.34472659, -0.09996126],\n",
       "        [ 2.03327391,  2.63843862],\n",
       "        [-0.19697709, -1.45185981],\n",
       "        [-0.30847416,  2.84366299],\n",
       "        [ 1.75102925, -0.51930203],\n",
       "        [ 1.84128948,  1.56431799],\n",
       "        [ 1.87360965,  0.34446036],\n",
       "        [ 1.28206289, -0.96776992],\n",
       "        [ 2.22466978,  2.51141667],\n",
       "        [-3.1272458 , -0.78950549],\n",
       "        [ 1.07912394, -0.23240142],\n",
       "        [ 2.29658414,  0.05729516],\n",
       "        [ 0.54174643,  2.55277476],\n",
       "        [ 1.39997096,  2.06256012],\n",
       "        [ 2.17160399,  0.70696395],\n",
       "        [ 1.55407908, -2.43999128],\n",
       "        [ 1.30846014, -3.07204207],\n",
       "        [ 1.29778812,  0.72026649],\n",
       "        [-0.96529593,  1.25071908],\n",
       "        [-3.04860516, -1.29285296],\n",
       "        [ 2.31996522, -1.41747754],\n",
       "        [-0.81535384,  1.56826204],\n",
       "        [ 3.05422377, -2.55277033],\n",
       "        [ 0.93378598, -0.15368093],\n",
       "        [-1.86831155, -0.59131997],\n",
       "        [ 2.32425702,  1.03217139],\n",
       "        [ 1.69686587, -2.70311988],\n",
       "        [ 2.79831991,  3.11893464],\n",
       "        [ 3.1162714 ,  0.7652615 ],\n",
       "        [ 0.46263748,  2.15913418],\n",
       "        [ 1.64246197, -2.43623584],\n",
       "        [ 2.5096982 , -0.33718076],\n",
       "        [ 1.15268567, -1.84805892],\n",
       "        [-0.28495587,  0.67810603],\n",
       "        [-0.30350588, -1.75841458],\n",
       "        [ 0.24785   , -2.48429251],\n",
       "        [ 2.91636999, -1.80517744],\n",
       "        [ 0.41529777,  0.55649011],\n",
       "        [ 1.72225572,  2.7417949 ],\n",
       "        [ 0.65473279,  1.75490771],\n",
       "        [ 1.87366191, -0.08722486],\n",
       "        [-3.131754  ,  1.96933124],\n",
       "        [-0.7173602 , -1.70326938],\n",
       "        [-2.71731551, -1.6383356 ],\n",
       "        [-2.16429211, -0.24168729],\n",
       "        [ 1.14295845, -1.79663376],\n",
       "        [ 1.29441776, -2.2661924 ],\n",
       "        [ 1.05310704, -0.85964593],\n",
       "        [-2.4426164 ,  0.6059895 ],\n",
       "        [ 0.83033537, -0.2882717 ],\n",
       "        [-1.81424859, -0.60097657],\n",
       "        [-2.47622421,  0.05157742],\n",
       "        [-2.564333  , -2.76250446],\n",
       "        [ 0.59538188,  2.3322212 ],\n",
       "        [-2.61080589,  2.2866226 ],\n",
       "        [-2.00010963,  1.59070463],\n",
       "        [ 2.50017003,  0.76892987],\n",
       "        [ 1.28358708, -1.16511826],\n",
       "        [-1.92993258,  1.30199686],\n",
       "        [-2.00943291, -1.8998927 ],\n",
       "        [ 2.40650209,  2.17142084],\n",
       "        [-1.1447002 , -0.53315233],\n",
       "        [ 1.00019128,  3.00603742],\n",
       "        [-1.3248395 , -0.08881492],\n",
       "        [-2.41504665,  1.73694325],\n",
       "        [-1.72723203,  0.76725002],\n",
       "        [ 0.94763575,  0.8011545 ]]),\n",
       " 'deltachi': array([[-0.52043326, -0.52653959],\n",
       "        [ 0.07781534,  0.07562822],\n",
       "        [ 0.13770485,  0.13675813],\n",
       "        [ 0.00456539,  0.01730651],\n",
       "        [ 0.08740176,  0.0593311 ],\n",
       "        [ 0.09408865,  0.13649818],\n",
       "        [ 0.20552952,  0.21482929],\n",
       "        [-0.63147694, -0.64078881],\n",
       "        [-0.31024257, -0.30874983],\n",
       "        [ 0.15185224,  0.15701677],\n",
       "        [-0.15565491, -0.15418993],\n",
       "        [-0.18790823, -0.16998594],\n",
       "        [ 0.19532284,  0.22106202],\n",
       "        [ 0.17129149,  0.15171319],\n",
       "        [ 0.05635929, -0.01307297],\n",
       "        [-0.26805121, -0.25173099],\n",
       "        [-0.3263147 , -0.29925105],\n",
       "        [ 0.39404823,  0.18214276],\n",
       "        [-0.04980847, -0.27549492],\n",
       "        [ 0.11564418,  0.0340012 ],\n",
       "        [-0.14654627, -0.15919083],\n",
       "        [ 0.10905414,  0.10898575],\n",
       "        [ 0.07754694,  0.11536579],\n",
       "        [-0.14322691, -0.14549618],\n",
       "        [-0.24405369, -0.27350191],\n",
       "        [ 0.27260889,  0.26969354],\n",
       "        [ 0.24450036,  0.49866983],\n",
       "        [-0.45288919, -0.49459057],\n",
       "        [-0.16157369, -0.17207348],\n",
       "        [-0.01430537, -0.01453256],\n",
       "        [-0.12234405, -0.12280287],\n",
       "        [ 0.27395198,  0.24092865],\n",
       "        [ 0.04059143,  0.14079738],\n",
       "        [ 0.2590282 ,  0.0591512 ],\n",
       "        [-0.12282211, -0.10373359],\n",
       "        [-0.28082626, -0.28699574],\n",
       "        [-0.05641984,  0.0449453 ],\n",
       "        [ 0.11776939,  0.11007206],\n",
       "        [-0.12852251, -0.13081732],\n",
       "        [ 0.51776654,  0.33330102],\n",
       "        [ 0.42699315,  0.3345032 ],\n",
       "        [ 0.3041486 ,  0.31446331],\n",
       "        [-0.17186198, -0.20718844],\n",
       "        [-0.05171373, -0.07393695],\n",
       "        [-0.05378611, -0.20120418],\n",
       "        [-0.17752975,  0.13221869],\n",
       "        [-0.40740632, -0.40001016],\n",
       "        [ 0.00581542, -0.07852463],\n",
       "        [-0.01877478,  0.00270991],\n",
       "        [-0.00816692,  0.03029824],\n",
       "        [ 0.17937927,  0.15117382],\n",
       "        [ 0.35599001,  0.33298087],\n",
       "        [ 0.08389266,  0.07704225],\n",
       "        [ 0.50068676,  0.48443814],\n",
       "        [ 0.49229789,  0.49811951],\n",
       "        [-0.03186704,  0.03986095],\n",
       "        [-0.28574099, -0.30042316],\n",
       "        [ 0.24730213,  0.22087502],\n",
       "        [ 0.33147659,  0.32545679],\n",
       "        [ 0.02090701,  0.02870263],\n",
       "        [ 0.08801721,  0.17371294],\n",
       "        [ 0.09656479,  0.0774558 ],\n",
       "        [-0.5490946 , -0.53819187],\n",
       "        [-0.09612275, -0.070511  ],\n",
       "        [ 0.23840879,  0.22406437],\n",
       "        [ 0.00555265,  0.00876843],\n",
       "        [-0.05858936, -0.06312631],\n",
       "        [ 0.18666757,  0.21125453],\n",
       "        [ 0.28944483,  0.29570337],\n",
       "        [-0.08786402, -0.06266506],\n",
       "        [-0.04398319, -0.0669482 ],\n",
       "        [ 0.0480188 ,  0.50185381],\n",
       "        [-0.03989718, -0.03822471],\n",
       "        [ 0.17530774,  0.08429444],\n",
       "        [ 0.219098  ,  0.21932779],\n",
       "        [ 0.23964247,  0.25919744],\n",
       "        [ 0.12940165,  0.15347328],\n",
       "        [ 0.00316349, -0.00255485],\n",
       "        [ 0.02580584,  0.04997934],\n",
       "        [-0.616333  , -0.54799011],\n",
       "        [-0.00953582, -0.07760361],\n",
       "        [-0.33594548, -0.34980527],\n",
       "        [ 0.22746937,  0.16680795],\n",
       "        [ 0.57023396,  0.47170261],\n",
       "        [ 0.69892973,  0.31329262],\n",
       "        [ 0.40995059,  0.55467698],\n",
       "        [ 0.10626952,  0.15467411],\n",
       "        [-0.36950032, -0.34558296],\n",
       "        [ 0.15366495,  0.15313996],\n",
       "        [-0.12119818, -0.14049465],\n",
       "        [-0.01440926, -0.02215548],\n",
       "        [ 0.32023582,  0.33295594],\n",
       "        [-0.24246399, -0.22257925],\n",
       "        [-0.12303708, -0.11111523],\n",
       "        [ 0.28725636,  0.27860893],\n",
       "        [ 0.15265138,  0.16018402],\n",
       "        [ 0.1188237 ,  0.10472558],\n",
       "        [-0.0541209 , -0.02668077],\n",
       "        [ 0.07086136,  0.04901182],\n",
       "        [ 0.06648834,  0.05408656]]),\n",
       " 'kappa': array([[-0.31093708, -0.23595317],\n",
       "        [ 0.18288536,  0.48404343],\n",
       "        [ 0.18120072,  0.8325255 ],\n",
       "        [ 0.09536163,  0.14005745],\n",
       "        [ 0.04216442,  0.05406602],\n",
       "        [ 0.13997531,  0.17461374],\n",
       "        [ 0.13840681,  0.16075199],\n",
       "        [-0.39198089, -0.21099635],\n",
       "        [-0.18222007,  0.46926144],\n",
       "        [ 0.17323662,  0.57934459],\n",
       "        [ 0.14201885,  0.17290159],\n",
       "        [-0.12668717, -0.07578277],\n",
       "        [ 0.19216995,  0.31631983],\n",
       "        [ 0.1508368 ,  0.22028596],\n",
       "        [-0.08838682, -0.07865596],\n",
       "        [-0.27822176, -0.22811562],\n",
       "        [-0.38168898, -0.25899368],\n",
       "        [ 0.13693665,  0.15789351],\n",
       "        [-0.01726503,  0.01078959],\n",
       "        [-0.1616042 , -0.09876115],\n",
       "        [-0.22732865, -0.15842392],\n",
       "        [ 0.07392708,  0.07933129],\n",
       "        [-0.15694174, -0.13889968],\n",
       "        [-0.1046001 , -0.09279054],\n",
       "        [-0.12233118, -0.05136403],\n",
       "        [ 0.26994691,  0.38223292],\n",
       "        [ 0.23542539,  0.29869819],\n",
       "        [-0.25207161, -0.15350289],\n",
       "        [-0.04887871, -0.04421421],\n",
       "        [-0.01658607,  0.00198083],\n",
       "        [ 0.1121223 ,  0.1293858 ],\n",
       "        [ 0.17649974,  0.26476585],\n",
       "        [ 0.12245337,  0.36307982],\n",
       "        [-0.10589581, -0.02112015],\n",
       "        [-0.10775403, -0.08328973],\n",
       "        [-0.38108875, -0.27798244],\n",
       "        [ 0.03848509,  0.04179186],\n",
       "        [ 0.12284104,  0.17027932],\n",
       "        [-0.00517222,  0.00205734],\n",
       "        [ 0.16671149,  0.21586483],\n",
       "        [ 0.03835302,  0.05514223],\n",
       "        [ 0.25285785,  0.35581196],\n",
       "        [-0.2747803 , -0.22160216],\n",
       "        [-0.15719481, -0.13329557],\n",
       "        [-0.03793669, -0.00363966],\n",
       "        [ 0.00477989,  0.02269635],\n",
       "        [-0.07722211, -0.07111143],\n",
       "        [-0.08288123, -0.06036779],\n",
       "        [-0.18952183, -0.16396032],\n",
       "        [-0.11721305, -0.10704458],\n",
       "        [ 0.02745853,  0.03855738],\n",
       "        [ 0.05585479,  0.11904433],\n",
       "        [-0.02843552,  0.00952383],\n",
       "        [ 0.46982266,  0.6185448 ],\n",
       "        [ 0.33087061,  0.40371332],\n",
       "        [-0.22284816, -0.18686464],\n",
       "        [-0.21414204, -0.00335539],\n",
       "        [ 0.1785124 ,  0.23559301],\n",
       "        [-0.04365145, -0.03480194],\n",
       "        [ 0.0795122 ,  0.09448107],\n",
       "        [ 0.3090116 ,  0.50875397],\n",
       "        [ 0.11875033,  0.13118234],\n",
       "        [-0.48083668,  0.20251463],\n",
       "        [-0.04092972,  0.0221695 ],\n",
       "        [ 0.23506718,  0.57118531],\n",
       "        [-0.13876478, -0.10780461],\n",
       "        [-0.07301793, -0.03728337],\n",
       "        [-0.02653564, -0.01421967],\n",
       "        [ 0.1722249 ,  0.22878773],\n",
       "        [-0.23373047, -0.18318229],\n",
       "        [-0.0064251 , -0.00178291],\n",
       "        [-0.03906564, -0.00419669],\n",
       "        [-0.04391881, -0.04119088],\n",
       "        [ 0.07027096,  0.07664297],\n",
       "        [ 0.23786428,  0.49979756],\n",
       "        [ 0.23773716,  0.59510122],\n",
       "        [ 0.02355396,  0.04760268],\n",
       "        [ 0.04627282,  0.06707598],\n",
       "        [-0.0191777 , -0.01757168],\n",
       "        [-0.29174943, -0.1177515 ],\n",
       "        [-0.02988041, -0.02517865],\n",
       "        [-0.17728844, -0.13646769],\n",
       "        [ 0.14783054,  0.21933857],\n",
       "        [ 0.04584002,  0.07578642],\n",
       "        [ 0.27952995,  0.34990346],\n",
       "        [ 0.28553492,  0.41333714],\n",
       "        [ 0.15734069,  0.21451645],\n",
       "        [-0.13183996, -0.07568366],\n",
       "        [ 0.12886328,  0.17230759],\n",
       "        [-0.11275453, -0.07752088],\n",
       "        [ 0.01097623,  0.01902757],\n",
       "        [ 0.33476364,  0.41186129],\n",
       "        [ 0.17695264,  0.203961  ],\n",
       "        [ 0.28906635,  0.35748152],\n",
       "        [ 0.22350061,  0.2552577 ],\n",
       "        [ 0.12870788,  0.14768448],\n",
       "        [ 0.21264202,  0.24975884],\n",
       "        [-0.1314252 , -0.03610651],\n",
       "        [-0.00845711,  0.00618692],\n",
       "        [ 0.03707814,  0.04170666]]),\n",
       " 'r': array([[1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10],\n",
       "        [1000,   10]]),\n",
       " 'u': array([[0.07671204, 0.76712035],\n",
       "        [0.14478858, 1.44788582],\n",
       "        [0.14920046, 1.49200458],\n",
       "        [0.0692823 , 0.692823  ],\n",
       "        [0.0685592 , 0.68559196],\n",
       "        [0.08368288, 0.83682876],\n",
       "        [0.06552814, 0.65528139],\n",
       "        [0.07957702, 0.79577022],\n",
       "        [0.14766042, 1.47660419],\n",
       "        [0.15834239, 1.58342392],\n",
       "        [0.06324971, 0.63249706],\n",
       "        [0.12315159, 1.23151592],\n",
       "        [0.10457483, 1.04574831],\n",
       "        [0.10841504, 1.08415036],\n",
       "        [0.06399729, 0.63997286],\n",
       "        [0.0665321 , 0.66532096],\n",
       "        [0.08179823, 0.81798231],\n",
       "        [0.06347706, 0.63477059],\n",
       "        [0.06389513, 0.6389513 ],\n",
       "        [0.06343041, 0.63430405],\n",
       "        [0.08858874, 0.88588743],\n",
       "        [0.10149636, 1.01496363],\n",
       "        [0.06551306, 0.65513055],\n",
       "        [0.0720781 , 0.72078102],\n",
       "        [0.06324767, 0.6324767 ],\n",
       "        [0.1434727 , 1.43472699],\n",
       "        [0.06485082, 0.64850819],\n",
       "        [0.07522007, 0.75220075],\n",
       "        [0.06338532, 0.63385319],\n",
       "        [0.09311326, 0.93113258],\n",
       "        [0.07423044, 0.7423044 ],\n",
       "        [0.06325541, 0.63255414],\n",
       "        [0.09201967, 0.92019674],\n",
       "        [0.06431996, 0.64319958],\n",
       "        [0.06475536, 0.64755357],\n",
       "        [0.06573385, 0.65733848],\n",
       "        [0.06344069, 0.63440688],\n",
       "        [0.18999985, 1.89999853],\n",
       "        [0.077835  , 0.77835004],\n",
       "        [0.06446024, 0.64460244],\n",
       "        [0.06612518, 0.66125175],\n",
       "        [0.10817581, 1.08175811],\n",
       "        [0.06728992, 0.67289917],\n",
       "        [0.0798978 , 0.798978  ],\n",
       "        [0.0657108 , 0.65710796],\n",
       "        [0.0635824 , 0.63582399],\n",
       "        [0.06580926, 0.65809262],\n",
       "        [0.06354048, 0.63540476],\n",
       "        [0.06765127, 0.67651273],\n",
       "        [0.06458927, 0.64589273],\n",
       "        [0.07171493, 0.7171493 ],\n",
       "        [0.06702422, 0.67024217],\n",
       "        [0.12152418, 1.21524179],\n",
       "        [0.07228743, 0.72287433],\n",
       "        [0.06874284, 0.68742844],\n",
       "        [0.06466567, 0.64665674],\n",
       "        [0.15799625, 1.57996249],\n",
       "        [0.07230508, 0.72305079],\n",
       "        [0.06575741, 0.65757413],\n",
       "        [0.08556086, 0.85560861],\n",
       "        [0.08225994, 0.82259945],\n",
       "        [0.0646636 , 0.646636  ],\n",
       "        [0.17396903, 1.73969027],\n",
       "        [0.08279354, 0.82793535],\n",
       "        [0.1095765 , 1.09576504],\n",
       "        [0.10452244, 1.04522441],\n",
       "        [0.17095059, 1.70950591],\n",
       "        [0.07183467, 0.71834671],\n",
       "        [0.0776019 , 0.776019  ],\n",
       "        [0.06351951, 0.63519514],\n",
       "        [0.06647236, 0.66472362],\n",
       "        [0.06472468, 0.64724677],\n",
       "        [0.07164105, 0.71641045],\n",
       "        [0.06379413, 0.63794135],\n",
       "        [0.18975655, 1.89756551],\n",
       "        [0.10186719, 1.01867187],\n",
       "        [0.06412625, 0.64126246],\n",
       "        [0.15649002, 1.56490024],\n",
       "        [0.06331933, 0.63319326],\n",
       "        [0.08004124, 0.80041239],\n",
       "        [0.06332277, 0.63322773],\n",
       "        [0.0760314 , 0.76031401],\n",
       "        [0.06637926, 0.66379255],\n",
       "        [0.06482127, 0.64821267],\n",
       "        [0.06526012, 0.65260115],\n",
       "        [0.07455741, 0.74557405],\n",
       "        [0.06753964, 0.67539638],\n",
       "        [0.08782316, 0.87823157],\n",
       "        [0.08480559, 0.84805591],\n",
       "        [0.11004392, 1.10043922],\n",
       "        [0.10083838, 1.00838378],\n",
       "        [0.06388298, 0.63882978],\n",
       "        [0.06426788, 0.64267879],\n",
       "        [0.06427174, 0.64271736],\n",
       "        [0.06926882, 0.6926882 ],\n",
       "        [0.11228041, 1.12280406],\n",
       "        [0.08106011, 0.81060107],\n",
       "        [0.06990851, 0.69908514],\n",
       "        [0.1000959 , 1.00095905],\n",
       "        [0.07079597, 0.70795966]]),\n",
       " 'deltachiminus': array([[-0.52043702, -0.53521307],\n",
       "        [ 0.07777057,  0.06449587],\n",
       "        [ 0.13766501,  0.13594686],\n",
       "        [ 0.0035961 ,  0.01311457],\n",
       "        [ 0.0856216 ,  0.0591986 ],\n",
       "        [ 0.08446188,  0.03942765],\n",
       "        [ 0.20552795,  0.160261  ],\n",
       "        [-0.64052532, -0.66854989],\n",
       "        [-0.31025367, -0.31053603],\n",
       "        [ 0.15074267,  0.14459043],\n",
       "        [-0.30431336, -0.33993971],\n",
       "        [-0.1885501 , -0.20585321],\n",
       "        [ 0.19396278,  0.17252868],\n",
       "        [ 0.16784129,  0.15083516],\n",
       "        [ 0.04151752, -0.02876465],\n",
       "        [-0.27172589, -0.28175136],\n",
       "        [-0.32651303, -0.34409487],\n",
       "        [ 0.39113829,  0.1691043 ],\n",
       "        [-0.11805656, -0.40409423],\n",
       "        [ 0.07428654, -0.23934749],\n",
       "        [-0.14659935, -0.17728826],\n",
       "        [ 0.10899575,  0.10862161],\n",
       "        [ 0.06971068,  0.0439542 ],\n",
       "        [-0.14414148, -0.15382129],\n",
       "        [-0.31767967, -0.38576819],\n",
       "        [ 0.27120219,  0.26186423],\n",
       "        [ 0.24283686,  0.1758516 ],\n",
       "        [-0.45816386, -0.49600105],\n",
       "        [-0.16433109, -0.18595102],\n",
       "        [-0.02074184, -0.05243822],\n",
       "        [-0.12415827, -0.16282817],\n",
       "        [-0.23913298, -0.38596044],\n",
       "        [ 0.02114435, -0.06020216],\n",
       "        [ 0.25644353,  0.03205103],\n",
       "        [-0.12878964, -0.15951455],\n",
       "        [-0.28104884, -0.33221839],\n",
       "        [-0.05689561, -0.10824414],\n",
       "        [ 0.11678949,  0.10888708],\n",
       "        [-0.13243566, -0.15657997],\n",
       "        [ 0.50640846,  0.21113716],\n",
       "        [ 0.41779232,  0.11064128],\n",
       "        [ 0.30221192,  0.28929863],\n",
       "        [-0.18310271, -0.21220993],\n",
       "        [-0.05315321, -0.07438524],\n",
       "        [-0.08378596, -0.2013527 ],\n",
       "        [-0.18332498, -0.32579367],\n",
       "        [-0.40992554, -0.42464433],\n",
       "        [-0.02100905, -0.12599719],\n",
       "        [-0.02155931, -0.03606691],\n",
       "        [-0.01053275, -0.02887437],\n",
       "        [ 0.17693088,  0.13809525],\n",
       "        [ 0.3422856 , -0.02825845],\n",
       "        [ 0.07988782,  0.06034796],\n",
       "        [ 0.49577452,  0.46697466],\n",
       "        [ 0.49162559,  0.45202839],\n",
       "        [-0.05631437, -0.10421021],\n",
       "        [-0.28647034, -0.3005949 ],\n",
       "        [ 0.24390302,  0.21664305],\n",
       "        [ 0.32249442,  0.18709617],\n",
       "        [ 0.01997976,  0.01297277],\n",
       "        [ 0.07605837,  0.0564486 ],\n",
       "        [ 0.09041559,  0.0462014 ],\n",
       "        [-0.54930113, -0.56401362],\n",
       "        [-0.09995603, -0.12811029],\n",
       "        [ 0.23597417,  0.22340073],\n",
       "        [ 0.00501895, -0.00335914],\n",
       "        [-0.05861686, -0.06461909],\n",
       "        [ 0.18637694,  0.11649186],\n",
       "        [ 0.28941854,  0.27744563],\n",
       "        [-0.18383066, -0.30667607],\n",
       "        [-0.04427033, -0.07035948],\n",
       "        [-0.01849395, -0.35234375],\n",
       "        [-0.04003834, -0.04587441],\n",
       "        [ 0.17044195,  0.0835045 ],\n",
       "        [ 0.21883534,  0.21778975],\n",
       "        [ 0.23747948,  0.15553659],\n",
       "        [ 0.08550014, -0.09106555],\n",
       "        [ 0.00228275, -0.00270771],\n",
       "        [ 0.01970027, -0.05159118],\n",
       "        [-0.62643946, -0.70428969],\n",
       "        [-0.01883897, -0.10278625],\n",
       "        [-0.33934472, -0.35173658],\n",
       "        [ 0.2252201 ,  0.16550929],\n",
       "        [ 0.54516831, -0.02415251],\n",
       "        [ 0.68952733,  0.31224561],\n",
       "        [ 0.38435314,  0.225172  ],\n",
       "        [ 0.10522048,  0.07025746],\n",
       "        [-0.37950859, -0.414793  ],\n",
       "        [ 0.14955566,  0.12129262],\n",
       "        [-0.12577354, -0.14663223],\n",
       "        [-0.01593053, -0.03471746],\n",
       "        [ 0.27945025,  0.14056064],\n",
       "        [-0.26680492, -0.33876708],\n",
       "        [-0.19130884, -0.25979397],\n",
       "        [ 0.2868029 ,  0.27296216],\n",
       "        [ 0.15228101,  0.14369859],\n",
       "        [ 0.11778079,  0.10467031],\n",
       "        [-0.09335487, -0.25911567],\n",
       "        [ 0.06751904,  0.04046514],\n",
       "        [ 0.06571856,  0.04777941]]),\n",
       " 'deltachiplus': array([[-0.51361494, -0.48378501],\n",
       "        [ 0.0815775 ,  0.10269728],\n",
       "        [ 0.13775167,  0.13681096],\n",
       "        [ 0.00472085,  0.02353998],\n",
       "        [ 0.09088083,  0.11519485],\n",
       "        [ 0.09411068,  0.1384575 ],\n",
       "        [ 0.21383592,  0.26398364],\n",
       "        [-0.63050278, -0.59634079],\n",
       "        [-0.30985431, -0.30707518],\n",
       "        [ 0.15198836,  0.15738396],\n",
       "        [-0.1403473 , -0.12473859],\n",
       "        [-0.18421658, -0.16567217],\n",
       "        [ 0.19852198,  0.22137722],\n",
       "        [ 0.17133468,  0.1881881 ],\n",
       "        [ 0.05756571,  0.10871498],\n",
       "        [-0.26638702, -0.24518809],\n",
       "        [-0.32061286, -0.29674304],\n",
       "        [ 0.39687057,  0.23797072],\n",
       "        [ 0.0185254 ,  0.4379656 ],\n",
       "        [ 0.26460637,  0.25361455],\n",
       "        [-0.1389803 , -0.10849352],\n",
       "        [ 0.10907245,  0.10943021],\n",
       "        [ 0.07807983,  0.12206113],\n",
       "        [-0.14162844, -0.13143847],\n",
       "        [ 0.50461886,  0.43926809],\n",
       "        [ 0.27302285,  0.28207593],\n",
       "        [ 0.27770742,  0.50742607],\n",
       "        [-0.44524018, -0.39849484],\n",
       "        [-0.14964584, -0.10787655],\n",
       "        [-0.01347996,  0.01887703],\n",
       "        [-0.11425532, -0.07467534],\n",
       "        [ 0.51856322,  0.40774406],\n",
       "        [ 0.04280677,  0.14509291],\n",
       "        [ 0.26650443,  0.08347536],\n",
       "        [-0.12081499, -0.10302538],\n",
       "        [-0.25649485, -0.19374743],\n",
       "        [-0.03460496,  0.05039772],\n",
       "        [ 0.11844797,  0.12614216],\n",
       "        [-0.12621892, -0.09954536],\n",
       "        [ 0.52143931,  0.37074948],\n",
       "        [ 0.44130539,  0.52668414],\n",
       "        [ 0.30461574,  0.31682934],\n",
       "        [-0.17186135, -0.12975356],\n",
       "        [-0.04754452, -0.02143819],\n",
       "        [-0.04768817,  0.08294109],\n",
       "        [-0.08621592,  0.21001421],\n",
       "        [-0.40337827, -0.3791999 ],\n",
       "        [ 0.01069521,  0.02703197],\n",
       "        [-0.01542715,  0.0179014 ],\n",
       "        [-0.00365536,  0.03043408],\n",
       "        [ 0.18346447,  0.21410283],\n",
       "        [ 0.4001987 ,  0.58846911],\n",
       "        [ 0.08390117,  0.10155996],\n",
       "        [ 0.50165203,  0.54173952],\n",
       "        [ 0.49580387,  0.52823472],\n",
       "        [-0.03181432,  0.06176901],\n",
       "        [-0.28283969, -0.26791028],\n",
       "        [ 0.24916596,  0.27762827],\n",
       "        [ 0.33295758,  0.37519123],\n",
       "        [ 0.0216447 ,  0.02967664],\n",
       "        [ 0.08846128,  0.17377221],\n",
       "        [ 0.09656555,  0.10482149],\n",
       "        [-0.54545827, -0.5322854 ],\n",
       "        [-0.09274035, -0.06149846],\n",
       "        [ 0.23867013,  0.25170259],\n",
       "        [ 0.00678536,  0.01411401],\n",
       "        [-0.05725613, -0.05132137],\n",
       "        [ 0.19870645,  0.25828647],\n",
       "        [ 0.29094162,  0.29580161],\n",
       "        [-0.08673335,  0.02764271],\n",
       "        [-0.03733432, -0.00686377],\n",
       "        [ 0.09691731,  0.52522296],\n",
       "        [-0.03866641, -0.03264687],\n",
       "        [ 0.17723879,  0.16601475],\n",
       "        [ 0.21909801,  0.22060059],\n",
       "        [ 0.25498582,  0.33709735],\n",
       "        [ 0.13128127,  0.27132187],\n",
       "        [ 0.00341406,  0.00859295],\n",
       "        [ 0.04009367,  0.11769338],\n",
       "        [-0.59732942, -0.49083587],\n",
       "        [ 0.01612796,  0.07992132],\n",
       "        [-0.3354731 , -0.32014379],\n",
       "        [ 0.23898115,  0.29811141],\n",
       "        [ 0.58208789,  0.51705295],\n",
       "        [ 0.69963195,  0.46697145],\n",
       "        [ 0.41214872,  0.56270996],\n",
       "        [ 0.11917373,  0.20460069],\n",
       "        [-0.36872521, -0.32421339],\n",
       "        [ 0.15535424,  0.18353115],\n",
       "        [-0.12068854, -0.09865027],\n",
       "        [-0.0116855 ,  0.00719551],\n",
       "        [ 0.33686889,  0.40228124],\n",
       "        [-0.23213649, -0.14320908],\n",
       "        [-0.1096493 ,  0.15591798],\n",
       "        [ 0.28815949,  0.29058916],\n",
       "        [ 0.15383126,  0.16022434],\n",
       "        [ 0.1205283 ,  0.13327506],\n",
       "        [-0.04866943,  0.09810933],\n",
       "        [ 0.0732877 ,  0.09964116],\n",
       "        [ 0.06938286,  0.08635029]]),\n",
       " 'deltachi3': array([[8.14021138e+00, 1.18034202e+00],\n",
       "        [2.02920753e+01, 2.07667503e+00],\n",
       "        [2.06174772e+01, 2.18038910e+00],\n",
       "        [4.26215407e+00, 4.97448994e-01],\n",
       "        [3.79829204e+00, 3.70100743e-01],\n",
       "        [1.02837918e+01, 1.02972381e+00],\n",
       "        [1.79654987e+00, 1.77001667e-01],\n",
       "        [9.34897164e+00, 1.48090197e+00],\n",
       "        [2.08786270e+01, 2.54857602e+00],\n",
       "        [2.12803877e+01, 2.13716959e+00],\n",
       "        [2.14439215e-02, 1.42273708e-01],\n",
       "        [1.82802017e+01, 1.99091383e+00],\n",
       "        [1.52038492e+01, 1.46273227e+00],\n",
       "        [1.58850043e+01, 1.51082816e+00],\n",
       "        [6.64223415e-01, 9.91278205e-02],\n",
       "        [2.66839133e+00, 4.91579051e-01],\n",
       "        [9.94512044e+00, 1.36752034e+00],\n",
       "        [1.77026520e-01, 9.62286441e-02],\n",
       "        [6.06339734e-01, 1.48765094e-01],\n",
       "        [1.79531472e-01, 2.89276967e-01],\n",
       "        [1.18969528e+01, 1.37193275e+00],\n",
       "        [1.46861145e+01, 1.39682247e+00],\n",
       "        [1.83020251e+00, 2.30446390e-01],\n",
       "        [5.81680363e+00, 6.68485273e-01],\n",
       "        [3.21304423e-02, 3.14236881e-01],\n",
       "        [2.00063821e+01, 1.83733820e+00],\n",
       "        [1.30233649e+00, 2.04014746e-01],\n",
       "        [7.46742059e+00, 1.09916019e+00],\n",
       "        [1.49202542e-01, 4.34078152e-02],\n",
       "        [1.29667619e+01, 1.33342142e+00],\n",
       "        [6.83167557e+00, 8.04263141e-01],\n",
       "        [4.38148696e-02, 3.90156950e-01],\n",
       "        [1.26817708e+01, 1.43209443e+00],\n",
       "        [8.97314671e-01, 4.15175050e-01],\n",
       "        [1.31937787e+00, 2.36769381e-01],\n",
       "        [2.12430959e+00, 6.07280580e-01],\n",
       "        [1.90725625e-01, 3.35127464e-02],\n",
       "        [2.31197507e+01, 2.23185366e+00],\n",
       "        [8.35321177e+00, 9.23541292e-01],\n",
       "        [9.33431370e-01, 2.34046978e-01],\n",
       "        [2.13469868e+00, 2.15497236e-01],\n",
       "        [1.57420758e+01, 1.41981945e+00],\n",
       "        [3.13861867e+00, 5.26054254e-01],\n",
       "        [9.08751275e+00, 9.91658569e-01],\n",
       "        [2.01851511e+00, 3.07266570e-01],\n",
       "        [3.39121066e-01, 9.56599252e-02],\n",
       "        [2.19314829e+00, 3.48789548e-01],\n",
       "        [2.83154053e-01, 1.24460544e-01],\n",
       "        [3.29838048e+00, 4.19013809e-01],\n",
       "        [1.15597566e+00, 1.53721339e-01],\n",
       "        [5.46867111e+00, 4.89391826e-01],\n",
       "        [2.75304628e+00, 3.77759461e-01],\n",
       "        [1.78560200e+01, 1.75402850e+00],\n",
       "        [5.61097378e+00, 5.91231818e-01],\n",
       "        [3.73933590e+00, 3.23402124e-01],\n",
       "        [1.23559956e+00, 2.58919935e-01],\n",
       "        [2.16304220e+01, 2.45115799e+00],\n",
       "        [5.73288372e+00, 5.46263060e-01],\n",
       "        [1.91677375e+00, 1.62626953e-01],\n",
       "        [1.09068010e+01, 1.09624449e+00],\n",
       "        [9.84663773e+00, 1.14691343e+00],\n",
       "        [1.18821186e+00, 1.52663114e-01],\n",
       "        [2.28979270e+01, 2.94454394e+00],\n",
       "        [1.01191741e+01, 1.13173944e+00],\n",
       "        [1.60366972e+01, 1.62521749e+00],\n",
       "        [1.53373052e+01, 1.57153254e+00],\n",
       "        [2.22669307e+01, 2.28568115e+00],\n",
       "        [5.52407704e+00, 5.09083049e-01],\n",
       "        [8.01338374e+00, 7.14162199e-01],\n",
       "        [2.92876483e-01, 2.44606934e-01],\n",
       "        [2.53311668e+00, 2.89253005e-01],\n",
       "        [1.26087809e+00, 2.24241627e-01],\n",
       "        [5.54217302e+00, 5.78277496e-01],\n",
       "        [4.70009791e-01, 6.07058914e-02],\n",
       "        [2.30222616e+01, 2.19000476e+00],\n",
       "        [1.46769010e+01, 1.53475484e+00],\n",
       "        [7.62472287e-01, 1.41579283e-01],\n",
       "        [2.12663410e+01, 2.13355385e+00],\n",
       "        [7.00546253e-02, 1.18046898e-02],\n",
       "        [9.50508767e+00, 1.45374952e+00],\n",
       "        [7.65202878e-02, 2.82437671e-02],\n",
       "        [7.74369571e+00, 9.96098383e-01],\n",
       "        [2.38681383e+00, 3.30996168e-01],\n",
       "        [1.18298377e+00, 2.55569249e-01],\n",
       "        [1.46461265e+00, 3.31166528e-01],\n",
       "        [6.70139298e+00, 6.62581593e-01],\n",
       "        [3.17974175e+00, 3.82795625e-01],\n",
       "        [1.18395014e+01, 1.45556923e+00],\n",
       "        [1.05916103e+01, 1.01269826e+00],\n",
       "        [1.63801001e+01, 1.75330764e+00],\n",
       "        [1.46519560e+01, 1.49360928e+00],\n",
       "        [5.45258448e-01, 2.97833506e-01],\n",
       "        [9.61765392e-01, 2.63342479e-01],\n",
       "        [9.59826294e-01, 3.39687249e-01],\n",
       "        [4.12284639e+00, 3.67276421e-01],\n",
       "        [1.65105873e+01, 1.56400637e+00],\n",
       "        [9.39361173e+00, 9.35679118e-01],\n",
       "        [4.65895479e+00, 7.01320668e-01],\n",
       "        [1.44404511e+01, 1.42039015e+00],\n",
       "        [5.05242740e+00, 4.86255415e-01]]),\n",
       " 'chieff': array([-4.22283436e-01,  2.38787390e-01,  1.13231899e-01,  1.79156330e-01,\n",
       "         5.71394023e-02,  2.28140385e-01,  2.32676496e-01, -5.36451618e-01,\n",
       "        -2.74880616e-01,  1.38949367e-01,  2.79002273e-01, -1.34735549e-01,\n",
       "         2.33342570e-01,  1.76779849e-01, -1.84211570e-01, -5.07937807e-01,\n",
       "        -6.36715182e-01,  2.47316723e-01, -3.73212470e-02, -3.44228618e-01,\n",
       "        -3.93593551e-01,  7.97180240e-02, -3.31807921e-01, -1.61815613e-01,\n",
       "        -2.60830741e-01,  3.11468744e-01,  4.14543233e-01, -3.46036164e-01,\n",
       "        -9.15011213e-02, -2.76263914e-02,  2.66244003e-01,  3.32261573e-01,\n",
       "         1.72975299e-01, -2.59908672e-01, -2.01768058e-01, -7.32983994e-01,\n",
       "         7.86410705e-02,  1.39079053e-01,  4.39852617e-02,  2.55120392e-01,\n",
       "        -1.55820847e-02,  2.87313073e-01, -5.18062240e-01, -2.96822592e-01,\n",
       "        -7.09588948e-02,  1.43121080e-02, -7.56581420e-02, -1.69918592e-01,\n",
       "        -3.80251130e-01, -2.35782523e-01, -9.36467694e-03,  1.10037536e-02,\n",
       "        -1.21961633e-01,  7.30068384e-01,  5.05991856e-01, -4.47620797e-01,\n",
       "        -2.54719104e-01,  2.57056065e-01, -1.52845717e-01,  1.45045414e-01,\n",
       "         5.32572019e-01,  2.21144484e-01, -6.76774068e-01, -4.91405927e-02,\n",
       "         2.41110675e-01, -2.88088458e-01, -1.07987469e-01, -1.22248933e-01,\n",
       "         2.07186265e-01, -4.69871242e-01, -4.94140142e-03, -9.32613025e-02,\n",
       "        -7.49754337e-02,  1.23353459e-01,  2.38714989e-01,  2.44426477e-01,\n",
       "         2.91620961e-02,  8.57173324e-02, -3.97708086e-02, -3.42686210e-01,\n",
       "        -6.07421294e-02, -2.25349889e-01,  2.29332252e-01,  5.89939217e-04,\n",
       "         4.26243135e-01,  3.87636618e-01,  2.73078232e-01, -7.85105993e-02,\n",
       "         1.71203437e-01, -1.53014270e-01,  2.85908142e-02,  6.22119447e-01,\n",
       "         3.79176136e-01,  5.79754579e-01,  3.55291482e-01,  1.52114121e-01,\n",
       "         3.61183859e-01, -2.61870679e-01, -6.28682907e-02,  5.10807353e-02]),\n",
       " 'q': array([0.40946118, 0.14255827, 0.13700009, 0.54418415, 0.56446022,\n",
       "        0.33851812, 0.68543426, 0.37644216, 0.13888936, 0.12678012,\n",
       "        0.98392411, 0.17823548, 0.22800394, 0.21545629, 0.80443458,\n",
       "        0.63631805, 0.35478775, 0.88609665, 0.81681421, 0.8975624 ,\n",
       "        0.30304913, 0.23923693, 0.68628158, 0.48141665, 0.98849549,\n",
       "        0.14430631, 0.72811329, 0.4296026 , 0.91029684, 0.27684293,\n",
       "        0.44437041, 0.97533706, 0.28271594, 0.77109608, 0.73506562,\n",
       "        0.67425486, 0.89490767, 0.10084918, 0.39572769, 0.75859184,\n",
       "        0.65469539, 0.21619548, 0.60621891, 0.37312685, 0.67547419,\n",
       "        0.86430492, 0.67032134, 0.87243385, 0.5933792 , 0.74789088,\n",
       "        0.48847884, 0.61624042, 0.18167998, 0.47746519, 0.55910437,\n",
       "        0.74186845, 0.12713877, 0.47713594, 0.67301666, 0.32389164,\n",
       "        0.3506389 , 0.74202906, 0.11248233, 0.34597967, 0.21194141,\n",
       "        0.22818579, 0.11498375, 0.48612054, 0.39848777, 0.87674728,\n",
       "        0.63890609, 0.73736194, 0.48994907, 0.83027473, 0.10100755,\n",
       "        0.23782205, 0.79020413, 0.12872375, 0.93398619, 0.37166621,\n",
       "        0.93251512, 0.41835955, 0.64301214, 0.73023534, 0.70111738,\n",
       "        0.43935367, 0.59725499, 0.3080345 , 0.3296005 , 0.21056053,\n",
       "        0.24179195, 0.81836366, 0.77600334, 0.77563455, 0.54454317,\n",
       "        0.20420536, 0.36165949, 0.5282076 , 0.2447456 , 0.50764343]),\n",
       " 'chi1': array([0.67742474, 0.62516399, 0.90030665, 0.63395909, 0.33865691,\n",
       "        0.35577447, 0.54276253, 0.97497524, 0.90955744, 0.67785775,\n",
       "        0.15555264, 0.30048811, 0.5453778 , 0.39320063, 0.28123628,\n",
       "        0.64714002, 0.67493641, 0.63517837, 0.70372754, 0.78381274,\n",
       "        0.45690953, 0.12218569, 0.2466385 , 0.28853478, 0.78613258,\n",
       "        0.38307116, 0.96511936, 0.80468864, 0.37422668, 0.23475324,\n",
       "        0.22822124, 0.88308456, 0.88701595, 0.93532033, 0.58068694,\n",
       "        0.91265353, 0.23097989, 0.20071286, 0.18890485, 0.87094627,\n",
       "        0.43259525, 0.48219474, 0.59751639, 0.27702571, 0.66543384,\n",
       "        0.67025989, 0.43561249, 0.58708624, 0.33239858, 0.23401383,\n",
       "        0.29603807, 0.80409235, 0.25517954, 0.96828954, 0.82627823,\n",
       "        0.48368874, 0.49057786, 0.6439374 , 0.27519195, 0.23092055,\n",
       "        0.9132039 , 0.28502802, 0.81054054, 0.52976383, 0.85739896,\n",
       "        0.24121653, 0.18767663, 0.25709147, 0.56732474, 0.78082158,\n",
       "        0.1003982 , 0.62439074, 0.12994659, 0.28257608, 0.47423412,\n",
       "        0.95613451, 0.62790183, 0.1526987 , 0.13293096, 0.97587524,\n",
       "        0.29541089, 0.51772491, 0.93340634, 0.59311331, 0.96434841,\n",
       "        0.89724213, 0.76851658, 0.49132696, 0.41793857, 0.27226603,\n",
       "        0.1328476 , 0.93662865, 0.29834425, 0.83172008, 0.49907823,\n",
       "        0.19277924, 0.36084251, 0.86004369, 0.18825338, 0.18268344]),\n",
       " 'chi2': array([0.81046959, 0.724983  , 0.10177097, 0.24921307, 0.16166726,\n",
       "        0.74751993, 0.12750948, 0.38928102, 0.14672684, 0.12800901,\n",
       "        0.88130011, 0.82187031, 0.32658213, 0.36558762, 0.34030959,\n",
       "        0.4584204 , 0.88533051, 0.16271769, 0.64616832, 0.69020385,\n",
       "        0.82064372, 0.10138015, 0.66093961, 0.18102007, 0.88470878,\n",
       "        0.53734966, 0.28711659, 0.39523623, 0.10854839, 0.82300168,\n",
       "        0.94464396, 0.92336745, 0.71085819, 0.60592308, 0.12112878,\n",
       "        0.73627849, 0.22035533, 0.90624117, 0.6643583 , 0.31938049,\n",
       "        0.80414997, 0.26060134, 0.63617835, 0.8572487 , 0.35507239,\n",
       "        0.39893395, 0.49467637, 0.21702813, 0.72973192, 0.42212343,\n",
       "        0.41372189, 0.75737018, 0.93705733, 0.41670328, 0.1408469 ,\n",
       "        0.71177731, 0.59848852, 0.12642542, 0.65695605, 0.31056856,\n",
       "        0.91843682, 0.40058422, 0.89088776, 0.29526802, 0.1182184 ,\n",
       "        0.86341268, 0.61644355, 0.76738777, 0.15589368, 0.54858833,\n",
       "        0.55075832, 0.91840948, 0.18037201, 0.25554367, 0.11878169,\n",
       "        0.59838168, 0.28707644, 0.59684817, 0.17778955, 0.91561472,\n",
       "        0.14893992, 0.26713443, 0.11281244, 0.82853467, 0.481502  ,\n",
       "        0.56153549, 0.27238762, 0.90589421, 0.34484029, 0.85855474,\n",
       "        0.99281371, 0.54709847, 0.90255875, 0.96283637, 0.27826605,\n",
       "        0.97919519, 0.56829306, 0.65368484, 0.98745765, 0.26532865])}"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "blackholes()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Who cares what that is, let's profile it."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "with cProfile.Profile() as pr:\n",
    "    blackholes()  \n",
    "\n",
    "pr.dump_stats('example.prof')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "And now let's have a look where I'm being silly...\n",
    "\n",
    "```snakeviz example.prof```\n",
    "\n",
    "I reported the results of a similar profiling exercise in Sec 6.C (page 19-20) of the paper that describes this code https://arxiv.org/pdf/2304.04801.pdf\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "\n",
    "\n",
    "# Exercises\n",
    "\n",
    "For the exams work on at least two of the these three exercises\n",
    "\n",
    "## Q1: I love pip \n",
    "\n",
    "- Take a piece of python code you wrote (for instance pick one of the exercises you've done for this class). \n",
    "- Turn it into a module\n",
    "- Install it locally\n",
    "- Deploy on pypi (only using the test-pypi server!)\n",
    "\n",
    "<div class=\"alert alert-block alert-warning\">\n",
    "<span class=\"fa fa-flash\"></span> VERY IMPORTANT\n",
    "    \n",
    "Both the pypi and test-pypi server are public on the web! Don't compromise your research by putting up something your competitors can use and steal your idea before you've published a paper on it!!! \n",
    "\n",
    "</div>\n",
    "\n",
    "\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Q2: My own test\n",
    "\n",
    "- Pick a piece of python code that you like (your own PhD project, or take one of the exercises from this class). Implement a unit test and a regression test. \n",
    "- Put it under git version control, and write a github action that runs the tests at every commit.\n",
    "- Edit the github options to make sure the code *cannot* be committed if the tests fail (that's a common thing for big projects with many collaborations, nobody is allowed to break the code)\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Q3: How do professionals do it?\n",
    "\n",
    "Pick a big git repo, say [scipy](https://github.com/scipy/scipy) or [numpy](https://github.com/numpy/numpy),  and have a look at their development workflow, including their testing strategies."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Q4: It's over Anakin, I have the high ground. I hate you! \n",
    "\n",
    "Pick a piece of your code that your really hate.  You hate it because it's soooooo slooooow. \n",
    "- Profile it.\n",
    "- Find the culprit\n",
    "- Rewrite that bottleneck making a better use of numpy arrays, or jitting it with Numba.\n",
    "- Enjoy your faster code\n",
    "\n",
    "**You underestimate my power**."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Huge thanks all!"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "... and we're done. Hope you enjoyed this, thanks for showing up. Hope this material is helpful for your PhD. Looking forward to reading your amazing papers on the results you got with some amazing code!\n",
    "\n",
    "\n",
    "I've very available to chat coding and science with you (my office is U2-2007). \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.11.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}