Profiling functions

[Shri Keshavinee]

Hi,

I'm writing a proposal on the following project "Classical Mechanics: Efficient Equation of Motion Generation with Python". As a first step, I would like to contribute to it by finding where the functions are getting slower through python profiling and add this as a pull request. But I'm confused about where to implement it. Either adding this file ( profiling functions) in the test folder(sympy/sympy/physics/mechanics/tests/) or in  examples folder(sympy/examples/). Please guide me in choosing the path to implement profiling.

Regards,

Keshav

[Oscar Benjamin]

On Fri, 9 Apr 2021 at 13:35, Shri Keshavinee <shrikes...@gmail.com> wrote:
>
> I'm writing a proposal on the following project "Classical Mechanics: Efficient Equation of Motion Generation with Python". As a first step, I would like to contribute to it by finding where the functions are getting slower through python profiling and add this as a pull request. But I'm confused about where to implement it. Either adding this file ( profiling functions) in the test folder(sympy/sympy/physics/mechanics/tests/) or in examples folder(sympy/examples/). Please guide me in choosing the path to implement profiling.

The simplest way to profile something in Python is to make a .py file
that does the thing you want to profile and then run it as:

$ python -m cProfile -s cumulative myscript.py

From ipython you can also use the %prun magic command:

In [1]: M = Matrix([[1, 2], [3, 4]])

In [2]: %prun -s cumulative M**2

%prun has the advantage that it profiles the operation itself rather
than the whole time taken to import sympy and set up variables etc.


Oscar