on RMC

[Marc Segovia]

Hi Friends

Doing RMC you can adjust timestep and reptile size, the former affects mixed operators , the latter pure operators.

If this is right, then you must in some way find a combination of both who converges simultaneously, right ?

In example energy converges as well as dipole moment.

Please let me know if I misunderstand something.

Best Regards

Marc

[Lucas Wagner]

Hi Marc,

Yes, you need to converge the timestep (as it goes to zero) and reptile size (as it goes to infinity). This affects both mixed and pure operators.

A good place to start for simple elements with decent wave functions is a timestep of 0.01 or 0.02 and a reptile length of 2/timestep + 1.

Cheers,

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.

[Marc Segovia]

Thank you Lucas, by your quick and clear answer.

Have a nice day

Marc

[Marc Segovia]

Good Morning

Attached is a file with my trials to converge both , energy(mixed observable) and dipole(pure observable).

Which was done, try several #beads upto energy converges, then dipole aproach to a close to real value, this was done with fixed timestep.

Later I try differnts timestep, keep reptile length constant, then I didn't care in energy but in dipole convergence.

This is the right way to do ?

Best Regards

From a pretty cold Montevideo

Marc

On Thu, May 26, 2016 at 2:17 PM, Lucas Wagner <lucas....@gmail.com> wrote:

[Lucas Wagner]

Hi Marc,

You should always share the stochastic uncertainties (the number after +/- in gosling). It's a little hard to figure out whether the differences are significant or not without that. 

For changing the timestep, you should *increase* the number of beads to keep nbeads*timestep constant. Energy matters here too.

Cheers,

Lucas

[Marc Segovia]

Hi Lucas

 Thank you by your feedback, errors bar are now included in attached file.

 But I understand your point about keep constant #beads* timestep, this is my mistake.

 
So is this procedure fine, first fine an optimum #beads , with converging energy and then try several timestep (keeping product constant) , to find a better energy if possible .

Best Regards

Marc