[ITensors] Fix eigen_perturbation rank
#1302
Conversation
Codecov ReportAll modified and coverable lines are covered by tests ✅
❗ Your organization needs to install the Codecov GitHub app to enable full functionality. Additional details and impacted files@@ Coverage Diff @@
## main #1302 +/- ##
===========================================
- Coverage 84.14% 53.88% -30.26%
===========================================
Files 100 99 -1
Lines 8594 8490 -104
===========================================
- Hits 7231 4575 -2656
- Misses 1363 3915 +2552 ☔ View full report in Codecov by Sentry. |
|
[test ITensors mps] |
![github-actions[bot]](https://avatars.githubusercontent.com/in/15368?s=60&v=4){kind=small}
|
[test ITensors mps] |
|
Run ITensors mps tests from comment trigger: succeeded ✅ |
1 similar comment
|
Run ITensors mps tests from comment trigger: succeeded ✅ |
|
Run ITensors mps tests from comment trigger: succeeded ✅ |
1 similar comment
|
Run ITensors mps tests from comment trigger: succeeded ✅ |
It appears that #1214 may have caused an issue when non-trivial
eigen_perturbationis passed tofactorize, where it is too aggressive about setting an upper bound on the rank of the factorization. This appears to cause issues for the noise term in DMRG, see https://itensor.discourse.group/t/revert-back-to-older-version-of-itensors-jl/1366.Really this logic should not be in
factorizesince it adds this extra complexity and breaks the abstraction offactorizethat it is doing a factorization of the input tensor, but disentangling it will be left for future work.I need to add a minimal test once we confirm this fixes the issue with the noise term.