v0.12.4

MPSKit v0.12.4

Diff since v0.12.3

Merged pull requests:

• Relax MPO element type restrictions (#244) (@leburgel)

v0.12.3

MPSKit v0.12.3

Diff since v0.12.2

Bump OptimKit compatibility and refactor FiniteMPS gauging to be iterative instead of recursive

Merged pull requests:

• CompatHelper: bump compat for OptimKit to 0.4, (keep existing compat) (#239) (@github-actions[bot])
• Rewrite orthogonalization from recursive to iterative (#241) (@lkdvos)
• Enable secret key in compathelper runs (#242) (@lkdvos)

Closed issues:

• StackOverflow when computing expectation value of large FiniteMPS (#240)

v0.12.2

MPSKit v0.12.2

Diff since v0.12.1

Merged pull requests:

• Fix entropy (#238) (@lkdvos)

Closed issues:

• Potential error in calculating entropy in v0.12 (#237)

v0.12.1

MPSKit v0.12.1

Diff since v0.12.0

Fix an issue in the constructors of MPOHamiltonians with multiple terms with the same indices.

Merged pull requests:

• Fix MPOHamiltonian constructors for adding multiple terms on the same site (#235) (@lkdvos)

v0.12.0

MPSKit v0.12.0

Diff since v0.11.6

v0.12 of MPSKit boasts a wide variety of changes, most notably related to the usage of BlockTensorKit for the MPOs and environments. This enables a wide range of simplifications behind the screens, that should hopefully both simplify maintenance, as well as lower the bar for contributions and future developments.
Various things have been refactored to make use of this new framework, and the documentation pages have seen some improvements (but are still WIP).
As such, the main breaking changes refer to the internals of the operator structs (FiniteMPO, InfiniteMPO, FiniteMPOHamiltonian and InfiniteMPOHamiltonian), as well as the internal structure of the environments. Furthermore, some algorithms have gotten updates to their fields to streamline the interface a bit.
Additionally, this release ensures compatibility with the new versions of TensorKit and TensorOperations, and thus some performance gains are to be expected because of that.
Finally, the interface to work with parallelization has been reworked to now use OhMyThreads.jl, which should ease the procedure and yield a bit more flexibility.

As a warning, for users that were relying on the infinite environment managers to automatically recompute their environments, this feature has been disabled, so recalculate! has to be called manually now.

Merged pull requests:

• BlockTensorMap (#199) (@lkdvos)
• update left and right_virtualspace (#205) (@lkdvos)
• Type specifications for GradientGrassmann (#209) (@lkdvos)
• Renaming MPSMultiline and MPOMultiline to MultilineMPS and MultilineMPO (#210) (@VictorVanthilt)
• Make Jordan MPO matrix render properly in MPOHamiltonian docstring (#212) (@VictorVanthilt)
• Remove obsolete iszero checks (#213) (@lkdvos)
• Enforce InfiniteMPOHamiltonian to be square (#215) (@lkdvos)
• Rewrite MPS types to use C instead of CR (and CL) and add ψ.center functionality. (#216) (@VictorVanthilt)
• CompatHelper: bump compat for TensorKit to 0.14, (keep existing compat) (#217) (@github-actions[bot])
• Add FiniteMPS getindex (#218) (@VictorVanthilt)
• Parallelization through OhMyThreads.jl (#219) (@Gertian)
• Revert TaylorCluster Optimizations of #199 (#220) (@VictorVanthilt)
• Pass appropriate tol_min in default dynamic tolerances (#221) (@leburgel)
• make tol a kwarg in remove_orphans!(::FiniteMPO) (#222) (@VictorVanthilt)
• Handle diagonal C - ensure TensorKit v0.14 compatibility (#223) (@lkdvos)
• Refactor InfiniteEnvironments and GradientGrassmann (#224) (@lkdvos)
• Minor improvements (#225) (@lkdvos)
• Documentation improvements (#226) (@lkdvos)
• Set default tol_max in gauging eigsolver (#227) (@leburgel)
• CompatHelper: bump compat for KrylovKit to 0.9, (keep existing compat) (#228) (@github-actions[bot])
• Avoid adding adding exact zeros in TaylorCluster implementation (#229) (@leburgel)
• Last updates before v0.12 tag (#230) (@lkdvos)
• Fix typo in IDMRG struct (#231) (@VictorVanthilt)
• Various (hopefully final) fixes (again) (#233) (@lkdvos)
• Final final fixes (#234) (@lkdvos)

Closed issues:

• @show for sparse_MPO does not work. (#169)
• Inconsistent type enforcement for finalize argument in ground state solvers (#207)
• Failure in addition of InifiniteMPOHamiltonian (#211)
• Incorrect indexing of Hamiltonian (#214)
• [BUG] periodic_boundary_conditions gives invalid results (#232)

v0.11.6

MPSKit v0.11.6

Diff since v0.11.5

Merged pull requests:

• [Bugfix] Ensure that projection environments have correct above (#204) (@lkdvos)

Closed issues:

• FiniteExcited() throws SpaceMismatch() error (#203)

v0.11.5

MPSKit v0.11.5

Diff since v0.11.4

Small fixes and improvements

Merged pull requests:

• CompatHelper: bump compat for VectorInterface to 0.5, (keep existing compat) (#190) (@github-actions[bot])
• Bump codecov/codecov-action from 4 to 5 (#191) (@dependabot[bot])
• Actions update (#192) (@lkdvos)
• Add Hubbard excitations example (#193) (@lkdvos)
• Macroexpand generated expressions (#197) (@lkdvos)
• Ensure expval returns normalized values (#200) (@lkdvos)
• update normalization convention environments (#201) (@lkdvos)

Closed issues:

• Inconsistent normalization when computing expectation value (#194)
• VUMPS for real-valued iMPS (#198)

v0.11.4

MPSKit v0.11.4

Diff since v0.11.3

Merged pull requests:

• excise RecursiveVec (#189) (@lkdvos)

v0.11.3

MPSKit v0.11.3

Diff since v0.11.2

Merged pull requests:

• 3 site tests (#180) (@Gertian)
• Fixes an issue with missing duals for spaces in fusers. (#184) (@lkdvos)
• Temporary fix for RecursiveVec change (#187) (@lkdvos)

Closed issues:

• How to apply a single creation/annihilation operator to a finite MPS ? (#160)
• Possible bug for the multiplication between a FiniteMPO and FiniteMPS (#183)

v0.11.2

MPSKit v0.11.2

Diff since v0.11.1

Merged pull requests:

• Documentation update: changebonds improvements (#168) (@Gertian)
• Add documentation for some exported methods (#171) (@leburgel)
• Some MPOHamiltonian fixes (#172) (@leburgel)
• Fix issue where SVDCut produces InfiniteMPS that is not full rank (#174) (@lkdvos)
• CompatHelper: bump compat for TensorKitManifolds to 0.7, (keep existing compat) (#177) (@github-actions[bot])

Closed issues:

• SVD using the divide and conquer algorithm might fail for certain tensors. (#109)
• Exact diagonalization implementation (#170)