Introduction

[Jared Lumpe]

Hi all,

My name is Jared Lumpe and I am a PhD student in bioinformatics at the University of California in San Francisco. One of my primary interests is in building mathematical models for biological systems.

Most of my math and algorithmic interests and knowledge are based in machine learning, nonlinear optimization, and dynamical systems. Beyond calculus/differential equations I am mostly self-taught. Python has been my primary language for 3+ years and I think I could be considered to be at an advanced level. I have been using Sympy extensively for the last few months but have little experience with computer algebra systems before that.

The main contributions I would like to make to Sympy are in the code generation module.

Thank you!

[Björn Dahlgren]

Hi Jared,

Glad to have you! -And thanks for your fixes and feedback so far. I've been using SymPy in my own PhD project (modeling of chemical systems)
for the past couple of years and have found it to be very useful. Looking forward to continue working with you.

Since you mention that you work with both dynamical systems and nonlinear optimization (I'm guessing your looking at them using numerical methods?)
I thought I could mention that I've written the following packages for using SymPy expressions together with numerical solvers:

- pyodesys
- pyneqsys

if nothing else they might serve as a reference / source of ideas.

Best regards,
Björn Dahlgren (@bjodah)

[Jared Lumpe]

Hi Bjorn,

I've come across the pyodesys and pyneqsys packages before (I think from one of Sympy's wiki pages on code generation) and definitely plan to be using them. I think the unified frontend/flexible backend approach is great and something that is really needed here.

My trouble with Sympy so far has been that my model has a lot of intermediate variables and so isn't represented well as a single expression. I can flatten it into one of course, but it makes more sense to generate code that contains multiple assignments. This is why I've been very interested in your additions to the code generation module.

To get around current limitations I've built a class that manages a DAG of expressions assigned to symbols and supports several optimization and editing routines (like pulling out common sub-expressions and automatic differentiation) that I'm using to generate function definitions. If you think it's a good idea I could try to make a PR for it.

Best,

Jared

--
You received this message because you are subscribed to a topic in the Google Groups "sympy" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/sympy/d-tcveQN8uQ/unsubscribe.
To unsubscribe from this group and all its topics, send an email to sympy+un...@googlegroups.com.
To post to this group, send email to sy...@googlegroups.com.
Visit this group at https://groups.google.com/group/sympy.
To view this discussion on the web visit https://groups.google.com/d/msgid/sympy/421ac2d8-f2cd-4537-994c-0c6cb3593ba8%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.