How to use DOUBLE PRECISION in PYMANOPT

[Beñat Mencia]

I want use pymanopt to do optimization (https://pymanopt.github.io/). I need double precision, but I have not seen any option related to this in the documentation (https://pymanopt.github.io/). The following simple code illustrates my problem.

import pymanopt.manifolds
manifold = pymanopt.manifolds.Euclidean(1)
A = manifold.rand()
print('A=',A)

This outputs:

A= [-0.85062969]

This is single precision.

How can I make it use double precision?

Thank you.

[Nicolas Boumal]

Hello,

This is a question for the pymanopt team, which is different from the manopt team.

This being said (and not having much experience with Python), I do think they use double precision by default, and suspect that the issue may just be in how the 'print' command works. Perhaps look into the help for the print command to see how you can control the number of digits it shows?

Best,

Nicolas