All electron

[Lesheng Li]

Hi Lucas, 

I saw that one can do all electron calculation with QWalk in the Github site, but I haven't found any reference about it.

Could you let me know how this works? Any document available? 

Thanks,

Lesheng

[Lucas Wagner]

Hi Lesheng,

All-electron is possible; you already found the tutorials on the wiki, I assume? 

https://github.com/QWalk/mainline/wiki/All-electron

However, it is very expensive and our implementation definitely has not focused on doing all-electron calculations. What are you trying to do? 

Best,

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+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Lesheng Li]

Yes, I found the tutorial and that's where I know it's possible to do all electron. 

I think my question is how this works (what's the difference between normal DMC and the one with key word "dynamics {UNR}"), if there is any document about how it work, that would be great. 

Thanks,

Lesheng

On Thursday, May 25, 2017 at 11:24:43 AM UTC-4, Lucas wrote:

Hi Lesheng,

All-electron is possible; you already found the tutorials on the wiki, I assume? 

https://github.com/QWalk/mainline/wiki/All-electron

However, it is very expensive and our implementation definitely has not focused on doing all-electron calculations. What are you trying to do? 

Best,

Lucas

On Thu, May 25, 2017 at 9:50 AM, Lesheng Li <leng...@gmail.com> wrote:

Hi Lucas, 

I saw that one can do all electron calculation with QWalk in the Github site, but I haven't found any reference about it.

Could you let me know how this works? Any document available? 

Thanks,

Lesheng

--
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.

[Lucas Wagner]

Hi Lesheng,

dynamics { } controls the method of imaginary time integration in DMC. All the options will give the same answer in the zero timestep limit, but in different situations, some may give smaller time step errors. 

The default dynamics section: dynamics { split } is optimized for pseudopotential calculations. dynamics { UNR } uses the algorithm from Umrigar, Nightingale, and Runge, which is more optimized for all-electron calculations.

Cheers,

Lucas

To unsubscribe from this group and stop receiving emails from it, send an email to qwalk-users+unsubscribe@googlegroups.com.

[Lesheng Li]

Hi Lucas,

Thanks for the above information. I guess now I could find more details in their JCP (J. Chem. Phys. 1993, 99 (4), 15) paper. 

Best,

Lesheng

[Lucas Wagner]

Yes, that's right. Just to reiterate, I wouldn't suggest doing all-electron calculations for heavier elements than Be or so. There are good pseudopotentials beyond that, and for most problems, the (potential, but not guaranteed) additional accuracy is not worth it in my opinion.

If your stochastic errors are larger than the pseudopotential error, then you should have used a pseudopotential!

Best,

Lucas

To unsubscribe from this group and stop receiving emails from it, send an email to qwalk-users+unsubscribe@googlegroups.com.

[Lesheng Li]

Thanks for these helpful information! 

Best,

Lesheng