FIX np.NaN is not available anymore in numpy2.0

[tomMoral]

Just notived a failure with numpy 2.0.
This is a quick fix for the failure I observed.

[tomMoral]

Not sure how to cope with the missing AttributeError: scipy.optimize.optimizehas no attributeMemoizeJac`

[arokem]

This seems relevant: scipy/scipy#17572 (comment)

[arokem]

Looks like this works:

from scipy.optimize._optimize import MemoizeJac

[arokem]

Looks good! Unless anyone objects, I can go ahead and merge it in a couple of days.

[tomMoral]

Great :)

Note that we saw this failure in the test of our benchmark on logreg l2, where we benchmark copt against other solvers :)
https://github.com/benchopt/benchmark_logreg_l2

It has been a while since we updated the copt solver so if you think we can improve its performances, feel free to contribute updates to the copt solver

[fabianp]

thanks all!

[arokem]

Hey @fabianp : any chance to put out a pypi release (0.9.2?) that includes this fix? Thanks!

[fabianp]

should be done now, sorry for the delay

…

On Wed, Sep 11, 2024 at 11:20 PM Ariel Rokem ***@***.***> wrote: Hey @fabianp <https://github.com/fabianp> : any chance to put out a pypi release (0.9.2?) that includes this fix? Thanks! — Reply to this email directly, view it on GitHub <#111 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AACDZBZ73RYJBI55U3QFWATZWCX2NAVCNFSM6AAAAABKD7BU3OVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDGNBUG4ZDEMRWGI> . You are receiving this because you were mentioned.Message ID: ***@***.***>