LINEAR optimizer and analytic parameter derivatives

[Alex Findlater]

Hello,

I am trying to optimize a Slater-Jastrow wavefunction with the LINEAR method (since the manual indicates that the OPTIMIZE2 method is depreciated, and recommends LINEAR) however, everytime my job gets to the LINEAR method, I get the following error message: Error   Wavefunction needs to supports analytic parameter derivatives.

Should I be doing something different with LINEAR as opposed to what I was doing with OPTIMZE2?

Thank you!
Alex

[Lucas Wagner]

Hi Alex,

See the tutorial here:
https://github.com/QWalk/mainline/wiki/Energy-optimization

The main reason you have to do this is that no one has coded up analytic derivatives for the parameters in the basis. This has been on my todo list for a while.

Lucas

--
You received this message because you are subscribed to the Google Groups "QWalk user's group" group.
To unsubscribe from this group and stop receiving emails from it, send an email to qwalk-users...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Alex Findlater]

Thanks Lucas!

Would you advise doing a run with OPTIMIZE to roughly optimize the shape of the Jastrow function and then switching over to LINEAR to optimize the energy (and maybe finish up with
OPTIMIZE again) to get the WF as optimized as possible? I'll be using this WF for the trial function for DMC (in which case I'd guess not optimizing the basis parameters wouldn't be
a huge deal), however I'd also like to report slater+jast2 and jast3 energies as well.

Thanks again,
Alex

[Lucas Wagner]

Hi Alex,

My suggestion is the procedure in the tutorial; that is, OPTIMIZE and then LINEAR. You don't want to OPTIMIZE again because then it'll just return to the variance-optimized parameters and undo all that work you did with LINEAR! OPTIMIZE will optimize the basis parameters for you.

Lucas

[Alex Findlater]

OK, thanks again for your help.

-Alex