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
StillingerWeber hessian gradient test failure #89
Comments
ouch, that's a bit embarrassing. I'll look at it as soon as I can. |
no idea, what I did, so I reverted to a slightly older code, and replaced the hessian with an AD hessian. This should now be ok. Would you be willing to pull the latest master and rerun your tests? |
@lifelemons - did you ever manage to check whether the hessian now works ok? |
So the above script now passes the test.
(Our newton linesearch method still gets stuck so its probably another issue now.) Thanks for looking into this! |
Can you put together and post a minimal failing example for the Newton linesearch code? |
So I'll close this, but if there is a line search problem please open another issue |
@lifelemons and I have been debugging our Newton optimiser, and tracked down a problem in the new StillingerWeber Hessian code, which appears to fail the gradient test for the ~400 atom Si configuration below.
@cortner Would you be able to take a look when you have time? We confirmed that the 2-atom bulk cell passes the test, but even adding
rattle!(at, 0.01)
to that test leads to fdtest errors of order 1e-5.Expected output:
Code and input file (inline):
The text was updated successfully, but these errors were encountered: