Last proposals / install issue on conda / questions

[hanspi42]

I looked through everything a last time, and have found a few things:

• 01-Introduction-Solutions has from sympy import * quite far down the line
• In 04-Calculus, 05-Matrices, 05-Matrices-Solutions, and many more, the initial init_printing() consistently hangs my kernel. Without it, no problems.
• Nice to see Dominik Gruntz's name in 04-Calculus; he's a colleague of mine :)

One problem may come with the installation in conda. SymPy 1.12 is only on conda-forge, as far as I can see, so your recommendation may not work for all anaconda users. It will depend on the channel order they have set.

I recommend to disambiguate this by creating a file requirements.yml

with the contents

name: sympy-tutorial
channels:
  - conda-forge
dependencies:
  - python
  - sympy=1.12
  - jupyter
  - numpy
  - matplotlib
  - scipy

and let users install it with

conda-env create --name sympy-tutorial --file requirements.yml

All the best for tomorrow! I wish I could be there :)

[hanspi42]

This is the yml file I have tested on my machine:

requirements.zip

[asmeurer]

the initial init_printing() consistently hangs my kernel. Without it, no problems.

That's not good. Could you open a SymPy issue about that? Note that init_printing() isn't really needed, so we could remove it. All it does is make it so that non-SymPy objects print with latex, including lists of sympy objects (important for solve() output), and Python ints and floats.

Nice to see Dominik Gruntz's name in 04-Calculus; he's a colleague of mine :)

Wow. Thank him for his great thesis. The Gruntz algorithm was the first thing Ondřej implemented in SymPy back in ~2005 and having a sophisticated nontrivial algorithm implemented from the start really helped make SymPy useful for people.

One problem may come with the installation in conda. SymPy 1.12 is only on conda-forge, as far as I can see, so your recommendation may not work for all anaconda users. It will depend on the channel order they have set.

Thanks, I didn't realize it wasn't on the defaults channel yet. Everything should work with older SymPy, but we've only tested the notebooks with 1.12 so it's best if everyone uses that if possible.

[hanspi42]

the initial init_printing() consistently hangs my kernel. Without it, no problems.

That's not good. Could you open a SymPy issue about that?

Will do.

Nice to see Dominik Gruntz's name in 04-Calculus; he's a colleague of mine :)

Wow. Thank him for his great thesis. The Gruntz algorithm was the first thing Ondřej implemented in SymPy back in ~2005 and having a sophisticated nontrivial algorithm implemented from the start really helped make SymPy useful for people.

Have done immediately ;)

[hanspi42]

Dominik sends greetings back. He says he's also proud that his algorithm is implemented in SymPy, after all it does not happen a lot that the results of a PhD dissertation are actually used ;)