Order for spin-pairs in electron systems

[Vladimir_Konjkov]

Hello, dear Casino developers.

Recently I was discouraged by the strange fact that I always thought that in the file correlation.data the lines like
Spin dep (0->uu=dd=ud; 1->uu=dd/=ud; 2->uu/=dd/=ud) determines the order of coefficients in all Jastro and backflow terms.
However, looking closer (in correlation.out.0 for blank terms), I discovered, and even found in the documentation (page 69, 11-link),
that the order for the u-term is different, namely (0->uu=dd=ud; 1->uu=dd/=ud; 2->uu/=ud/=dd), and in my code, apparently by mistake,
I implemented just such order for all other terms (f-, eta-, phi-, theta-). I would like to know what is the actual order.
Having carried out the calculation for the helium atom (which only has antiparallel pairs), I can assume that it is just such (uu/ud/dd).

Best Vladimir.

[Neil Drummond]

Dear Vladimir,

The spin-dependence options, with their precise groupings and orderings, can be found in the out file at

Code: Select all

 Particle-pair groupings
 -----------------------
 Spin-pair dep. 0 : (1-1,1-2,2-2)
 Spin-pair dep. 1 : (1-1,2-2) (1-2)
 Spin-pair dep. 2 : (1-1) (1-2) (2-2)

The above is for an electron system. In a non-electron system there may be many more spin dependence options.

The comment lines in correlation.data are ignored by CASINO when correlation.data is read, and hence can contain anything you like. However, obviously it would be good if the comment lines written to correlation.out.x were helpful and not misleading. Do you think it would be better to make the comment line "Spin dep (0->uu=dd=ud; 1->uu=dd/=ud; 2->uu/=ud/=dd)" for electron systems and "Spin dep (see out file for particle groupings)" for non-electron systems?

Thanks,

Neil.

[Vladimir_Konjkov]

The forum works terribly, my authorization session is constantly interrupted. It is not possible to write a long message. This would be more intuitive if the comments reflected the order of the terms.

I'm sorry Neil it's absolutely impossible to write

[Neil Drummond]

I keep losing my replies and having to log out and log in again. I'll ask Mike about it.