Add WindowMPOHamiltonian enabling propagators and dynamical DMRG on WindowMPS#417
Merged
Conversation
Contributor
|
Your PR no longer requires formatting changes. Thank you for your contribution! |
Codecov Report❌ Patch coverage is
... and 2 files with indirect coverage changes 🚀 New features to boost your workflow:
|
maartenvd
marked this pull request as ready for review
April 29, 2026 08:08
lkdvos
reviewed
Apr 29, 2026
lkdvos
left a comment
lkdvos
left a comment
Member
There was a problem hiding this comment.
Looks good to me! I left some minor comments but otherwise I'm happy to merge this and go from there
Collaborator
Author
|
I added a windowmps constructor that takes in a unitrange, similar to windowmpohamiltonian. |
Member
|
Looks good to me! I rebased on the latest main to hopefully resolve the errors in the tests, but if it works I'll just merge this so it is included in the next release, which I hope to get out today. |
Collaborator
Author
|
I forgot that I still want to write some docstring referencing exactly what propagator returns, I wouldn't merge just yet |
lkdvos
force-pushed
the
mvd/windowham
branch
2 times, most recently
from
July 1, 2026 14:43
2f5c16d to
161c540
Compare
The WindowMPOHamiltonian environments method must call initialize_environments on the finite Hamiltonian (there is no environments(::WindowMPS, ::FiniteMPOHamiltonian, ...) method), and expose leftstart/rightstart as keywords so squaredenvs (Jeckelmann flavour) can pass squared boundary environments through. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
- Fix `+`/`-` on WindowMPOHamiltonian: the finite window carries full Jordan virtual spaces on its boundaries, so summands must be block-diagonalized at every site (including edges), like InfiniteMPOHamiltonian addition. Add a dedicated `_add_finite_window` helper and a `VectorInterface.scale` method so unary `-`, scalar `*`/`/` work through the AbstractMPO fallbacks. - Avoid spurious non-convergent fixed-point solves in `squaredenvs`: inline the `leftstart`/`rightstart` defaults in the WindowMPOHamiltonian environments method so the infinite environment of e.g. `conj(H) * H` is only computed when no explicit boundaries are passed. - Attach and clean up the WindowMPOHamiltonian docstring (was silently detached), document the interval constructors of WindowMPOHamiltonian and WindowMPS, and fix the propagator/DynamicalDMRG docstrings (return value + LaTeX). - Add test/operators/windowhamiltonian.jl covering non-trivial intervals (circshift alignment via offset invariance) and the linear algebra. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Add two Hamiltonians with different Jordan bond dimensions (Ising ZZ string vs XY hopping) so the boundary/bulk virtual spaces of the summands differ and must be block-diagonalized consistently, checking expectation-value additivity in both orders. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
lkdvos
changed the title
lkdvos
enabled auto-merge (squash)
July 1, 2026 20:10
lkdvos
approved these changes
Jul 1, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Adds a
WindowMPOHamiltonian: the operator counterpart of aWindowMPS, a finite window embedded between infinite left/right "bath" Hamiltonians.This resolves the pain points from #411, where
propagator/dynamical DMRG could not run on aWindowMPS.What was wrong
A
WindowMPShas a finite window but infinite environments.Slicing an
InfiniteMPOHamiltonianinto aFiniteMPOHamiltoniandoes not impose trivial finite boundary conditions, soenvironments/squaredenvshit aSpaceMismatch.There was also no finite energy reference for
z(the true ground-state energy is extensive).What this adds
WindowMPOHamiltonian(H::InfiniteMPOHamiltonian, interval)— carves a finite window out of an infinite Hamiltonian, keeping the boundary Hamiltonians and circshifting their unit cells to align with the window.environments(::WindowMPS, ::WindowMPOHamiltonian)— builds finite env caches with boundaries from the infinite fixed-points; explicitleftstart/rightstartavoid the fixed-point solve forconj(H)*H.propagator/squaredenvsaccept aWindowMPS+WindowMPOHamiltoniandirectly; docstrings corrected to the1/(z-H)convention and Jeckelmann functional (14).expectation_value,dot, and linear algebra (+,-,*,scale,conj) forWindowMPOHamiltonian;WindowMPS(ψ, interval)constructor with aligned environments.AC_hamiltonian/AC2_hamiltonian) forward to the finite window so DMRG/TDVP/propagator use the Jordan-optimized operators.WindowMPS(both flavours, vs analytic result), window-Hamiltonian linear algebra + circshift-offset invariance, andfind_groundstate/timestepon windows.Remaining work / limitations
AbstractMPOinterface coverage (only the pieces used here are wired).InfiniteMPOHamiltonianwindows (no dense-MPO windows yet).propagator.Closes #411