search

FRBs / pygedm

Python bindings for YMW16, NE2001 and YT2020 electron density models
https://pygedm.readthedocs.io/
42 stars 17 forks source link

compatibility with arrays and lists as arguments #21

Open MicheleRonchi opened 3 years ago

MicheleRonchi commented 3 years ago

Hi,

I'm trying to compute the DM values given the distance and the sky position for a lot of objects. I have saved the distances and the sky positions in galactic coordinates in numpy.arrays and I was wondering if there is a way to directly pass numpy.arrays as arguments to the methods pygedm.dist_to_dm.

At the moment if I run the following example:

l_gal = np.array([20, 30])
b_gal = np.array([-10, 30])
dist = np.array([1000, 2000])

dm, _ = pygedm.dist_to_dm(l_gal, b_gal, dist, method='ymw16')

The following error comes out:

TypeError: dmdtau(): incompatible function arguments. The following argument types are supported:
    1. (gl: float, gb: float, dordm: float, DM_Host: float, ndir: int, np: int, vbs: int, dirname: str, text: str) -> Dict[str, float]

Looping on the values is a solution but it is also slow for many objects and I'm trying to optimize the script I'm working on.

Thank you very much for the help,

Michele

telegraphic commented 3 years ago

Hey @MicheleRonchi ,

I did think about doing this but got exhausted of working on the code! Nevertheless you inspired me, so I had a stab at it. There's a first attempt on a new branch numpy-array-issue21.

Currently only works for YMW16, and you need to use the pybind shared object manually (import ymw16).

Here's an example:

import numpy as np
import ymw16
import os
from pkg_resources import resource_filename

DATAPATH = os.path.dirname(resource_filename("pygedm", "spiral.txt"))

gl = np.array([1,2,3,4], dtype='float64')
gb = np.array([1,2,3,4], dtype='float64')
dm = np.array([10, 15, 20, 25], dtype='float64')
out = ymw16.dmdtau_np(gl, gb, dm, 0., 2, 1, 0, DATAPATH, '')

# output is Nx2 array (DM, tau_sc), but needs to be reshaped
print(out.reshape((-1, 2)))

To use this you will need to checkout the branch with git, then install from source with pip install -e .. I don't have time at the moment to properly integrate this with pygedm, could me a month or two before I get around to it (If you're up for the challenge, pull requests always welcome 😄).

For reference to me in the future / anyone else interested, we would need to: