Hessian preconditioning in NLOPT_LD_CCSAQ

[pata-eth]

Hi, in the nlopt website the author describes a way to provide a user-supplied hessian with the NLOPT_LD_CCSAQ. How we can do that in R with your package?

http://ab-initio.mit.edu/wiki/index.php/NLopt_Reference#Preconditioning_with_approximate_Hessians

Thank you

[jyypma]

This feature is not implemented in the R package.

Best wishes, Jelmer

On Fri, Aug 5, 2016 at 8:49 PM, hherrada notifications@github.com wrote:

Hi, in the nlopt website the author describes a way to provide a user-supplied hessian with the NLOPT_LD_CCSAQ. How we can do that in R with your package?

http://ab-initio.mit.edu/wiki/index.php/NLopt_Reference# Preconditioning_with_approximate_Hessians

Thank you

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/jyypma/nloptr/issues/23, or mute the thread https://github.com/notifications/unsubscribe-auth/ADlPAHX5BrZDnzZ-cqm2xYX47alj6vRtks5qc4XXgaJpZM4Jd9ag .

[aadler]

@astamm this should probably be closed based on @jyypma email.

[astamm]

Which email? It might be useful to implement it. Did @jyypma advise against that?

[eddelbuettel]

@astamm Presumably https://github.com/astamm/nloptr/issues/23#issuecomment-237933210 two comments above yours.

[aadler]

https://github.com/astamm/nloptr/issues/23#issuecomment-237933210

[jyypma]

This feature could be implemented in the R package. I've never implemented it, because I never felt the need to use this. Also, the NLopt documentation mentions "Currently, support for preconditioners in NLopt is somewhat experimental, and is only used in the NLOPT_LD_CCSAQ algorithm." https://nlopt.readthedocs.io/en/latest/NLopt_Reference/#preconditioning-with-approximate-hessians

On Thu, Aug 17, 2023 at 6:54 PM Avraham Adler @.***> wrote:

23 (comment)

https://github.com/astamm/nloptr/issues/23#issuecomment-237933210

— Reply to this email directly, view it on GitHub https://github.com/astamm/nloptr/issues/23#issuecomment-1682635472, or unsubscribe https://github.com/notifications/unsubscribe-auth/AA4U6ADRYA6B6UBZF4QL7TTXVZEFBANCNFSM4CLX22QA . You are receiving this because you were mentioned.Message ID: @.***>

[astamm]

Ok. It does not seem like an impossible thing to do. The fact that it is still experimental however makes me think we should wait. @pata-eth is this something you really want to try out right now ?

[pata-eth]

Hey folks. It's been a while. I'm no longer in need of this feature and given the lack of plus ones in 7 years I'd say let's close this one. Cheers

[astamm]

Thanks for the quick reply.