MPS as Square Matrices

How do I use this algorithm? What does that parameter do?
Post Reply
dorf
Posts: 32
Joined: 16 Feb 2021, 16:34

MPS as Square Matrices

Post by dorf »

Hey,
is there a way I tell tenpy to compute a MPS that consists only of square matrices?
Thanks!
User avatar
Johannes
Site Admin
Posts: 413
Joined: 21 Jul 2018, 12:52
Location: TU Munich

Re: MPS as Square Matrices

Post by Johannes »

For a finite system, this doesn't make sense, so I guess you mean infinite and have some translation invariance by a single site?
You can just use truncation parameters with the same chi_min=chi_max, which ultimately fixes the bond dimension exactly. I'd do that only once you have your final MPS, though, not during DMRG/TEBD/..., since it doesn't allow to dynamically change chi.
dorf
Posts: 32
Joined: 16 Feb 2021, 16:34

Re: MPS as Square Matrices

Post by dorf »

Hey Johannes, thanks for your reply.
You are right. I am looking at an infinite system. Because even for iDMRG L>=2, I would like to construct a translationally invariant MPS for a single site. I calculate ground states over a range of parameters and would like to fix the form of the output MPS.

So do you mean I run the calculation with a variable chi until convergence is reached and then I run iDMRG again using the ground state as initial state but with chi_min=chi_max once? Thanks for clarifying.
dorf
Posts: 32
Joined: 16 Feb 2021, 16:34

Re: MPS as Square Matrices

Post by dorf »

So doing the above doesn't seem to work for me. I converged an MPS with variable bond dimension, and then fix chi_min=chi_max for a new run, using the MPS as initial state. However, the bond dimension won't change, because it immediately is converged without the bond dimension being changed (increased). Should I perturb the MPS a little bit before rerunning?
Post Reply