Re: [sympy] sympy 1.1 performance

2017-07-28 Thread Andrew Corrigan
Thanks for trying it out. I'm away from my computer for the weekend. I'll put together a better example once I get back. On Jul 28, 2017 3:38 PM, "Aaron Meurer" wrote: > Hi. > > This example is difficult to test, because the times are so similar. > Sometimes when I run the scripts, the 1.0 vers

Re: [sympy] sympy 1.1 performance

2017-07-28 Thread Nathan Goldbaum
It might also help to run the test under Victor Stinner's perf module, which is supposed to eliminate a lot of sources of variance when testing python microbenchmarks: https://github.com/haypo/perf On Fri, Jul 28, 2017 at 2:38 PM, Aaron Meurer wrote: > Hi. > > This example is difficult to test,

Re: [sympy] sympy 1.1 performance

2017-07-28 Thread Aaron Meurer
Hi. This example is difficult to test, because the times are so similar. Sometimes when I run the scripts, the 1.0 version runs faster, and sometimes the 1.1 version runs faster. Do you have an example where the slowdown is more dramatic? Aaron Meurer On Thu, Jul 27, 2017 at 3:08 PM, Andrew Co

Re: [sympy] sympy 1.1 performance

2017-07-28 Thread Nathan Goldbaum
Has this been reported as an issue? If not it probably should. Has anyone tried profiling before and after to see where the extra time is going? On Fri, Jul 28, 2017 at 12:46 PM, Richard Fateman wrote: > Sometimes programs get slower because they are getting smarter. It often > happens > in th

[sympy] Re: PyCon India - 2017

2017-07-28 Thread Shekhar Prasad Rajak
Hello, The material for the workshop will be available in this github repo : https://github.com/Shekharrajak/PyCon-SymPy-SymEngine I am modifying/adding new examples for SymPy and improving other things in the repo. Shikhar please add few examples/tutorial and some lines about SymEngine in

Re: [sympy] sympy 1.1 performance

2017-07-28 Thread Richard Fateman
Sometimes programs get slower because they are getting smarter. It often happens in the other direction --- especially in symbolic mathematics systems. That is, someone writes a new program that does "the same thing" as an existing program, but much faster. The reality may be that the new progra

Re: [sympy] sympy 1.1 performance

2017-07-28 Thread Andrew Corrigan
. On Fri, Jul 28, 2017 at 12:49 AM, Francesco Bonazzi wrote: > Try to add *return obj* after > > https://github.com/sympy/sympy/blob/a3389a25ec84d36f5cf04a4f2562d8 > 20f131db64/sympy/core/basic.py#L1636 > > (just to make sure that it's not *_exec_constructor_postprocessors* > that's slowing down