Possible speed improvements to transit model fitting

[hrwakeford]

Some possible speed improvements to occultnl the transit model fitting. Suggested by Lavender Elle Hanson (JHU).

Insert the following two lines immediately before the occultuniform function:

import numba
@numba.jit(numba.float64[:](numba.float64[:], numba.float64))

Then the start of the occultuniform function will look like this:

import numba
@numba.jit(numba.float64[:](numba.float64[:], numba.float64))
def occultuniform(b0, w):
"""
Compute the lightcurve for occultation of a uniform source without microlensing (Mandel & Agol 2002).

:param b0: array; impact parameter in units of stellar radii
:param w: array; occulting star size in units of stellar radius
:return: muo1: float; fraction of flux at each b0 for a uniform source
"""

if np.abs(w - 0.5) < 1.0e-3:
    w = 0.5

This trick requires the numba library, it can be installed it by adding the following in a Jupyter cell at the top of the notebook:

import sys
!conda install --yes --prefix {sys.prefix} numba

or (if that doesn’t work)

import sys
!{sys.executable} -m pip install numba

[ivalaginja]

This is super interesting stuff, I have just recently started looking at numba and what it can do.

@hrwakeford do you want me to try the implementation suggested here by Lavender Elle Hanson or were they thinking of giving it a go themselves?

[hrwakeford]

Give it a go if you have time. I was passed this information second hand but wanted to attribute credit here. It might be a good test for anything else you want to do with numba

[hrwakeford]

I will try to find her on GitHub and tag her here