Re: [PATCH] Add a new Matrix.QRsolve method.

2009-05-14 Thread Riccardo Gori
Hello, thanks for the patch! Just a comment, see below: On Wednesday 13 May 2009 19:54:18 Jochen Voss wrote: The new method is slower than Matrix.LUsolve, but should be more stable. The commit also fixes a closeby comment and adds a new docstring. This fixes issue #907 from the issue

Re: [PATCH] Add a new Matrix.QRsolve method.

2009-05-14 Thread Jochen Voß
Hi Riccardo, 2009/5/14 Riccardo Gori goricca...@gmail.com: + # back substitution to solve R*x = y: + # We build up the result backwards in the vector 'x' and reverse it + # only in the end. + x = [] + n = R.lines + for j in range(n-1, -1, -1): + tmp = y[j,:] Why do you use y[j,:]

[PATCH] Add test for issue 1406

2009-05-14 Thread Priit Laes
Issue #1406 has been fixed in tree meanwhile (probably thanks to Mateusz) so lets just add a testcase for future. Can be seen at: https://code.plaes.org/git/sympy.git/commit/?id=b33806aa5e5e6cc14320dfe991f82cec2c70ad4b --~--~-~--~~~---~--~~ You received this

Re: [PATCH] Rename Basic.is_fraction to Basic.is_rational_function

2009-05-14 Thread Mateusz Paprocki
Hi, On Thu, May 14, 2009 at 06:37:15PM +0300, Priit Laes wrote: Renames Basic.is_fraction to Basic.is_rational_function and adds a docstring with doctests. Can be viewed here: https://code.plaes.org/git/sympy.git/commit/?h=patches-0514id=5d27fd77f72bc3887fae02ad568b00838227e3b8 or as an

[PATCH] Fix corner case in bernoulli ode solver

2009-05-14 Thread Priit Laes
Bernoulli ODE has one case where it should be solved using different algorithm. Currently we don't detect it and hilarity ensues in the form of divide by zero :) https://code.plaes.org/git/sympy.git/commit/?h=patches-0514id=56524bf458074ac9635f537ef6f83c4727ea7a56

Re: [PATCH] Fix corner case in bernoulli ode solver

2009-05-14 Thread Aaron S. Meurer
I thought I already sent in a patch to fix this. Aaron Meurer Sent from my iPod Touch. On May 14, 2009, at 9:53 AM, Priit Laes plaes...@gmail.com wrote: Bernoulli ODE has one case where it should be solved using different algorithm. Currently we don't detect it and hilarity ensues in the

[PATCH] Allow Function.eval to return non-instances of Basic.

2009-05-14 Thread Fabian Pedregosa
Specially annoying was the fact that you could not return list or tuples. I don't know the reason of this behaviour, fortunately changing it did not break any tests. --- sympy/core/function.py | 17 ++--- sympy/core/tests/test_functions.py |7 +++ 2 files

Re: [PATCH] Small cleanups for ODE tests

2009-05-14 Thread Vinzent Steinberg
2009/5/14 Priit Laes plaes...@gmail.com Spotted some low-hanging fruit in ODE solver and tests https://code.plaes.org/git/sympy.git/commit/?id=db94caf3f64b7bfa4dd40266c0eeafc774f7233d x, C1 = map(Symbol, ['x', 'C1']) Please note that you can use x, C1 = symbols('x C1') Vinzent

Re: [PATCH] Add a new Matrix.QRsolve method.

2009-05-14 Thread Jochen Voß
Hi, I'll be travelling for a few days and will be able to send a new patch only sometimes next week. What I suggest to do is: 1) Remove the more stable wording from the QRsolve docstring. 2) Add warnings to the both the LUsolve and the QRsolve docstrings, stating that these functions should

Re: Added support for the integration of 1/sin(x)**(2k) and 1/cos(x)**(2k), fixes issue #1188

2009-05-14 Thread Akshay Srinivasan
Can this be passed in now? I didn't see any more comments on the issues page. On Mar 30, 11:08 am, Ondrej Certik ond...@certik.cz wrote: Thanks! On Sun, Mar 29, 2009 at 11:01 PM,AkshaySrinivasan akshaysriniva...@gmail.com wrote: This patch adds support for the integration of functions of