Different results for dynamics with version 2.1 and 3/3.1

Numerical routines for variational matrix product state simulations.

Brought to you by: djaschke, lincolncarr, matjones, mlwall

Different results for dynamics with version 2.1 and 3/3.1

Forum: Technical

Creator: Sebastian Stumper

Created: 2020-10-20

Updated: 2020-11-03

Sebastian Stumper - 2020-10-20

Dear all,

I have trouble computing the dynamics of some correlation functions (Bose-Hubbard, Spin-1 Heisenberg), depending on whether some other Observables are defined or not. Attached is an example script for Quench dynamics in a Bose-Hubbard model, using the TDVP. The Observable 'Ostr' is only computed correctly, if the line

myObservables.AddObservable('corr',['dn','dn'],'nn')

is commented out. It seems that this problem did not exist in version 2.1, but in the newer versions 3 and 3.1.

Could you check whether you get the same beahaviour as in the attached plots? And, if so, is there maybe a simple fix?

Thanks a lot in advance,
Sebastian

EBHM_test.py

test_with_nn.pdf

test_without_nn.pdf
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
- Matthew Jones - 2020-10-23
  
  Hi Sebastian,
  
  Thanks for posting, and checking into correctness between major versions of OpenMPS. That saves me a lot of time!
  
  I will look into your issue and provide guidance early next week.
  
  If you would like to refer to this comment somewhere else in this project, copy and paste the following link:

Sebastian Stumper - 2020-11-03

Hi Matthew,

did you find time to look at the issue?

I just realized that, even if I only define 'Ostr', but set the system size to L=16 instead of 8, there are similar problems in Version 3.

Best,
Sebastian

If you would like to refer to this comment somewhere else in this project, copy and paste the following link:

Log in to post a comment.