[FIXED] Retrieve approximate Hessian inverse from L-BFGS-B

Issue

With the L-BFGS-B minimizer in scipy, is it possible to retrieve the approximate inverse Hessian that's calculated internally?

Having it in the implicit factored form, so that it's possible to compute arbitrary inverse Hessian matrix - vector products, would be fine.

Solution

There's now a way to do this in SciPy.

Example:

from scipy.optimize import minimize, rosen
result = minimize(rosen, x0=[0] * 10, method='L-BFGS-B')
print(result.hess_inv.todense())

Seems to have been added in SciPy 0.16.0. See the PR for this feature.

Answered By - Nick ODell