Backtranformed Jastrow

General discussion of quantum Monte Carlo in electronic structure theory
Post Reply
Posts: 158
Joined: Wed Apr 15, 2015 3:14 pm

Backtranformed Jastrow

Post by Vladimir_Konjkov »

Hello everybody.

VMC is just Multiple integral brevity, written in bra–ket notation <φ(r)|H|φ(r)>/<φ(r)φ(r)> then backflow tranformation
(if it includes Jastrow factor) is change of variables and only requires the calculation of the determinant of backflow tranformation Jacobian matrix
opposed to when Jastrow factor is not backtranformed. Has anyone tried this kind of wfn? Are there any pitfalls here?

Best Vladimir
In Soviet Russia Casino plays you.
Neil Drummond
Posts: 109
Joined: Fri May 31, 2013 10:42 am
Location: Lancaster

Re: Backtranformed Jastrow

Post by Neil Drummond »

The obvious pitfall in evaluating a Jastrow factor at backflow-transformed quasiparticle coordinates is the need to ensure the Kato cusp conditions are still satisfied.

Evaluating a Jastrow factor at backflow transformed coordinates is used in, for example, recursive backflow wave functions after the first round of recursion. (In the recursive backflow method, quasiparticle coordinates are subjected to further backflow transformations and so forth; at each round of recursion another Jastrow factor in the transformed coordinates is introduced.)

My feeling is that evaluating additional Jastrow factors at backflow transformed coordinates is better viewed as a way of introducing n-body correlation terms into the Jastrow factor, and it is probably more efficient to implement n-body terms explicitly if they are needed.

Best wishes,

Post Reply