[SciPy-dev] scipy.optimize.nonlin rewrite

Pauli Virtanen pav@iki...
Mon Dec 1 11:22:24 CST 2008


Mon, 01 Dec 2008 14:48:58 +0200, Stéfan van der Walt wrote:
> I'm happy to postpone, but then we should accept that the API may change
> just after 0.7.  It concerns me that non-reviewed code becomes the
> standard quo, and then sets the API for the future.  Also, if Pauli is
> willing to write some tests, we can commit those so long.

To clarify: there are unit tests, cf. test_nonlin.py in the branch, old 
and a couple of new ones. The problem is that these currently only show 
that the algorithms (well, the quasi-Newton ones) find solutions to 
several test problems without  problems, but not really that they work 
optimally.

-- 
Pauli Virtanen



More information about the Scipy-dev mailing list