Probable population explosion occurs

General discussion of the Cambridge quantum Monte Carlo code CASINO; how to install and setup; how to use it; what it does; applications.
Post Reply
Tomohiro Ichibha
Posts: 10
Joined: Sun May 24, 2015 8:14 am

Probable population explosion occurs

Post by Tomohiro Ichibha »

Dear all,

I'm trying to make a success of a DMC calculation.
but I encounter 'probable population explosion'.
The calculation is all electron one, and its trial wave
function is CI and its basis set is gaussian basis set.

Could you tell me how to prevent this probrem?

Ichibha
Katharina Doblhoff
Posts: 84
Joined: Tue Jun 17, 2014 6:50 am

Re: Probable population explosion occurs

Post by Katharina Doblhoff »

have you plotted the dmc.hist? Does the population explode? My first guess would be that something went wrong with the cusps.
Tomohiro Ichibha
Posts: 10
Joined: Sun May 24, 2015 8:14 am

Re: Probable population explosion occurs

Post by Tomohiro Ichibha »

Dear Katharina,

Thank you very much for you supply.

I saw the picture of dmc.hist, and config increased suddenly.
My supervisor also said that something may have been wrong with cusp.
So I will brush up the Jastrow function and retry.

Ichibha
Neil Drummond
Posts: 113
Joined: Fri May 31, 2013 10:42 am
Location: Lancaster
Contact:

Re: Probable population explosion occurs

Post by Neil Drummond »

Dear Ichibha,

The electron-nucleus cusp conditions are usually imposed by cusp-correcting the orbitals rather than via the Jastrow factor. By default the Gaussian cusp-correction scheme is applied to the orbitals (unless you have disabled it by setting cusp_correction=F).

If you want to check that the cusp conditions are applied successfully, you can plot the local energy when one electron is moved through a nucleus using the qmc_plot block in input. If the cusp conditions are satisfied then the local energy should be nondivergent at the coalescence point (although it will in general be discontinuous at the coalescence point). Note that the local energy diverges as 1/x at nodes; these divergences are not dangerous. If you see a 1/|x| divergence at the coalescence point, this would be a problem, however.

Best wishes,

Neil.
Post Reply