sympy/sympy

A computer algebra system written in pure Python http://sympy.org/ . To get started to with contributing https://github.com/sympy/sympy/wiki/Introduction-to-contributing

Craig Russell

@ctr26

Currently I'm lambdifying it using mpmath as it returns nan with numpy

But I can't get dask to serialise the lambda function to make it parallel

Aaron Meurer

@asmeurer

There can be issues parallelizing things with SymPy because of serialization bugs.

jicky94

@jicky94

Hi everybody, i have a question regarding `nonlinsolve'. When i used it for my system of equations (some of which are non linear), i get solutions in terms of other unknown and not only in terms of the parameters. I checked the documentation but did not see the relevant options. What am I doing wrong? If anyone could help me, that would be fantastic. In other words i am not able to get explicit solutions. I could copy paste my code but it does not seem the right place to do that :) many thanks and have a nice day

Craig Russell

@ctr26

There can be issues parallelizing things with SymPy because of serialization bugs.
@asmeurer K so, the best way to do this is by serialising a function that wraps lambidfy rather than creating your lambidfied function first then trying to serialise that. Thi is working for me in task nicely

def numba_func(consts):
    model = sp.lambdify(tuple(consts.keys()),integral,"mpmath")
    return model(*tuple(consts.values()))

out = client.map(numba_func, consts_list)
result = client.gather(out)

Aaron Meurer

@asmeurer

I don't know how well lambdify functions serialize because they are dynamically created

But other than that, they do not involve SymPy expressions, so they might actually serialize better

Craig Russell

@ctr26

The above code works well with dask as the lambification runs on the workerrs

Aaron Stiff

@AaronStiff

I'm investigating #20762, and I'm pretty sure it's an issue with ANTLR. I've been looking at LaTeX.g4 to see how it handles subscripts, and everything seems correct. The only potential problem I've found is this:

subeq: UNDERSCORE L_BRACE equality R_BRACE;
supeq: UNDERSCORE L_BRACE equality R_BRACE;

Not sure if this could be causing the problem, but could anyone familiar with ANTLR comment on whether or not supeq should be using CARET instead?

Sayandip Halder

@sayandip18

I was going through plot.py and came across the check_arguments() function here. Though I sort of understand how it works, I still don't understand it's purpose. The docstrings aren't much of a help either. Can someone tell me why the function behaves that way?

10mayje

@10mayje

hello, I am Tanmay Mondal a second-year student of nit Durgapur, West Bengal, India. I am expressed in python, Django. How can I get started?

Aaron Meurer

@asmeurer

@AaronStiff I think that issue is unrelated to parse_latex. It is parsing the expression right. python() just doesn't print it right.

1 reply

Craig Russell

@ctr26

Anyone know why this simplfies to 0, it doesn't in mathematica?

import sympy as sp
from sympy.concrete.summations import Sum

sp.Derivative(Sum(sp.exp(x*sp.sin(n/m)),(n,1,m)),x).simplify()

Craig Russell

@ctr26

This works fine:
Sum(sp.Derivative(sp.exp(x*sp.sin(n/m)),x),(n,1,m)).doit()

Aaron Meurer

@asmeurer

Looks like it is a bug

Something weird is happening here

>>> Sum(exp(x*sin(n/m)),(n,1,m))
Sum(exp(x*sin(n/m)), (n, 1, m))
>>> Sum(exp(x*sin(n/m)),(n,1,m)).doit()
Piecewise((m*exp(c3_), Eq(exp(c2_), 1)), ((exp(c2_) - exp(c2_*(m + 1)))*exp(c3_)/(1 - exp(c2_)), True))

Alex Nguyen

@rxng8

Hi I'm new to sympy. I hope I can go with sympy and contribute from now on

I have a few questions: when we rebuild the repo after making changes, does the repo get re-install again or only the modified part get installed? How do we run test on a single test case? How do we log the info from the test case when testing (is there any params when build that has to be added?). Sorry for the long questions!

Aaron Meurer

@asmeurer

You don't have to build SymPy because it is pure Python. If you installed from git, you will have to install again unless you did a pip install -e or setup.py develop. Although I personally just run Python directly from the sympy directory. That does not require reinstalling, as it will just import the development version of SymPy directly.

To run the tests on a single case, you can pass in the file name like ./bin/test test_solvers.py, and use -k to select a specific test like ./bin/test -k test_whatever

I'm not sure about the log question. The output is printed to the screen. You don't have to do anything special to run the tests.

Alex Nguyen

@rxng8

Well that's great helps, thanks a lot !

Harshit Yadav

@hyadav2k

Hi, can someone tell me how can we add a new function in sympy, beacuse on directly adding, and then calling in terminal gives NameError and says that function called is not defined. Thanks!

Naveen Sai

@naveensaigit

@hyadav2k If you open python in the same working directory as the repo you clone, you can import the sympy you made changes to without any problems

Craig Russell

@ctr26

Looks like it is a bug

Thanks aaron!

Nisarg Chaudhari

@Nisarg-Chaudhari

Hello everyone, I'm Nisarg Chaudhari, a second year computer science undergrad, studying at BITS Pilani Hyderabad Campus, India. I am familiar with python and made some personal projects with it (mainly small games with pygame). I would like to contribute to SymPy. This is my first time contributing to open source, currently going through "Introduction to contributing" in sympy wiki. Is there anything else I need to know to get started?

Harshit Yadav

@hyadav2k

@hyadav2k If you open python in the same working directory as the repo you clone, you can import the sympy you made changes to without any problems

Thanks for sharing @naveensaigit!

Naveen Sai

@naveensaigit

@hyadav2k If you open python in the same working directory as the repo you clone, you can import the sympy you made changes to without any problems

Thanks for sharing @naveensaigit!

No problem!

Aaron Stiff

@AaronStiff

I'm just running some tests in the printing module, but even without making any changes to any files, pytest is reporting 8 failed tests. Is this normal?

Thomas Aarholt

@thomasaarholt

What are the failures?

I'm not familiar with the sympy ones, but over at hyperspy we often have test failures due to mismatches on matplotlib versions when testing generated images

That said, if you go to the github master, you can see that the red cross implies that the master branch is currently not passing tests

Aaron Stiff

@AaronStiff

Thomas Aarholt

@thomasaarholt

Are you getting something ala "SyntaxError: invalid escape sequence \e"?

None of those are failing :)

Aaron Stiff

@AaronStiff

They aren't?

Thomas Aarholt

@thomasaarholt

Unless you've done something to your colorscheme

Red = failures, yellow is just that it output a warning.

Aaron Stiff

@AaronStiff

I get this at the end: 844 passed, 61 skipped, 8 xfailed, 1 xpassed, 1 warning in 7.50s

Thomas Aarholt

@thomasaarholt

xfailed are "allowed/expected failures"

Aaron Stiff

@AaronStiff

Oh! That makes sense. Thanks.

Thomas Aarholt

@thomasaarholt

You're welcome!

Zachariah Etienne

@zachetienne

Sympy Travis is failing atm; sympy-dev appears to be pretty badly broken (AssertionError: {'sympy.polys.matrices'}). Any news on when it'll be fixed?

Sayandip Halder

@sayandip18

>>> sympify(0.0) == S.false
True
>>> S.false==sympify(0.0)
False

What's the reason for this behaviour? Shouldn't the latter be True too?

12 replies

Thomas Aarholt

@thomasaarholt

What does sympify(0.0) evaluate to?

I agree that it seems odd

(and what does S.false evaluate to?)

Sayandip Halder

@sayandip18

sympify(0.0) is sympy.core.numbers.Float whereas S.false is sympy.logic.boolalg.BooleanFalse

Where communities thrive

People

Repo info

Activity