New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Heuristics for initial choice of `lambda` #34

Open
blakejohnson opened this Issue Jan 23, 2017 · 1 comment

Comments

Projects
None yet
2 participants
@blakejohnson
Contributor

blakejohnson commented Jan 23, 2017

Madsen, Nielsen, Tingleff (2004) suggest choosing an initial lambda of:

lambda_0 = tau * maximum(J'*J)

where tau is supplied by the user, but they recommend values of 10^-6, 10^-3, or 1.

We need a better test of test cases to see if this improves convergence.

@pkofod

This comment has been minimized.

Show comment
Hide comment
@pkofod

pkofod Dec 3, 2017

Collaborator

Have you tried anything like this ? Could be interested to see the effects. CUTEst has a lot of least squares problems but I am actually not sure if you can get the vector objective or if it's just an R^N->R f(x) they provide. I'll have a look

Collaborator

pkofod commented Dec 3, 2017

Have you tried anything like this ? Could be interested to see the effects. CUTEst has a lot of least squares problems but I am actually not sure if you can get the vector objective or if it's just an R^N->R f(x) they provide. I'll have a look

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment