v0.13.10

MPSKit v0.13.10

Diff since v0.13.9

Added

• expectation_value for local MPO tensors
  (#327)
• Base.copy for MPS types now performs a deep copy
  (#387)

Changed

• entropy can now also be called directly on a spectrum (singular value vector)
  (#377)
• Updated compat bounds to remove broken package versions

Fixed

• Fixed Adapt extension for GPU support
  (#389)

Full changelog: v0.13.9...v0.13.10

Merged pull requests:

• Expectation values of local MPO tensors (#327) (@borisdevos)
• Refactor entropy function to also use spectrum directly (#377) (@AFeuerpfeil)
• Bump v0.13.10 (#379) (@lkdvos)
• Modernize test workflow (#382) (@lkdvos)
• Bump actions/checkout from 4 to 6 (#383) (@dependabot[bot])
• Adapt extension (#384) (@lkdvos)
• Deep copying MPS with copy (#387) (@20akshay00)
• Skeletal tests for CuArray-backed MPSKit (#388) (@kshyatt)
• Fix Adapt extension (#389) (@lkdvos)
• Update pipeline.yml (#390) (@kshyatt)

Closed issues:

• Printing Symmetric Hamiltonians with identity (#358)
• FiniteMPOHamiltonian constructor fails with storagetype MethodError on Union{BraidingTensor, TensorMap} (#376)
• Console output of MPOs is not the same as in the documentation (#378)
• Unpredictable normalization of DMRG ground states (#380)
• DMRG calculations are staggeringly slower than in previous versions (#381)
• Copying an MPS currently only makes a shallow copy (#386)

v0.13.9

MPSKit v0.13.9

Diff since v0.13.8

This release brings significant improvements to MPSKit.jl, including enhanced compatibility with multifusion categories, refactored algorithms, and numerous bug fixes.
Key improvements include a complete IDMRG refactoring, introduction of new trait systems for geometry and operator types, and enhanced performance with reduced allocations.

Key Features

🎯 New Features

Multifusion Categories

• Added full compatibility with multifusion categories (#297)

Trait System for Dispatching

• Introduced GeometryStyle and OperatorStyle traits for better type-based dispatching on finite/infinite geometries and operator types, improving code organization and extensibility (#352, #354)

Examples and Documentation

• Added comprehensive Bose-Hubbard example demonstrating strongly correlated bosonic systems (#342)
• Updated WindowMPS example with improved documentation and clearer explanations (#350)

🔧 Infrastructure Improvements

Algorithm Refactoring

• Complete refactoring of IDMRG to follow the IterativeSolver interface, improving code maintainability and enabling better code sharing between IDMRG and IDMRG2 implementations (#348)
• Refactored orthogonalization methods: removed _left_orth and _right_orth workarounds in favor of new, cleaner orthogonalization algorithms (#373)

Performance Optimization

• Optimized changebonds(state, ::RandExpand) algorithm to sample spaces directly instead of performing truncated SVD, improving performance while maintaining mathematical equivalence (#335)
• Reduced memory allocations while computing Galerkin error, leading to better performance in iterative algorithms (#366)

Dependency Updates & Compatibility

• Bumped compatibility for TensorKit 0.16 and MatrixAlgebraKit 0.6, ensuring compatibility with the latest versions of core dependencies (#365)
• Added LocalPreferences.toml file to ensure proper TensorOperations precompilation on testing infrastructure

User Experience

• Updated show methods to reflect new TensorKit printing conventions, providing clearer and more consistent output formatting (#341)
• More informative error messages for finite MPS operations, helping users diagnose issues more quickly (#367)
• Minor documentation and docstring improvements throughout the codebase (#363, #372, #371)

🐛 Bug Fixes

Numerical Stability

• Fixed dynamic tolerances yielding NaN during initialization due to 1 / sqrt(iter) where iter = 0, improving algorithm stability at the first iteration (#335)
• Fixed InfiniteMPOHamiltonian environments with low bond dimension and high Krylov dimension to be properly clamped, preventing numerical instabilities (#335)

Correctness Issues

• Fixed logical operator precedence bug in getproperty function that could lead to incorrect behavior (#346)
• Fixed domain/codomain mismatch in MPODerivativeOperator, ensuring correct tensor operations (#370)
• Corrected orthogonalization algorithms to use proper methods, ensuring mathematical correctness (#373)
• Improved handling of in-place operations to avoid unintended mutation errors (#337)

Minor Fixes

• Fixed typo in VUMPSSvdCut algorithm name (#361)
• Fixed typo in time formatting for log messages (#336)

Merged pull requests:

• Compatibility with multifusion categories (#297) (@borisdevos)
• Various small fixes and improvements to RandExpand (#335) (@lkdvos)
• Fix typo in time formatting for logs (#336) (@leburgel)
• Don't assume transpose makes a copy (#337) (@lkdvos)
• [Feature] Update show methods to reflect new TensorKit printing (#341) (@lkdvos)
• Add Bose-Hubbard example (#342) (@20akshay00)
• Update list of works that use MPSKit (#344) (@lkdvos)
• Fix getproperty function in iterativesolvers.jl (#346) (@AFeuerpfeil)
• add Base.isfinite to mps types (#347) (@AFeuerpfeil)
• Refactor of IDMRG with IterativeSolver (#348) (@AFeuerpfeil)
• Update WindowMPS example (#350) (@VicVanderLinden)
• Add OperatorStyle and GeometryStyle (#352) (@AFeuerpfeil)
• Define GeometryStyle and OperatorStyle for multiple arguments (#354) (@AFeuerpfeil)
• Bump actions/checkout from 5 to 6 (#357) (@dependabot[bot])
• Fix typo in VUMPSSvdCut (#361) (@benjamints)
• Add more links to algo section and some language cleanup (#363) (@kshyatt)
• Bump compat for TensorKit 0.16 and MatrixAlgebraKit 0.6 (#365) (@lkdvos)
• Reduce allocation when computing Galerkin error (#366) (@20akshay00)
• More informative errors for finite mps (#367) (@borisdevos)
• Fix domain/codomain of MPODerivativeOperator (#370) (@20akshay00)
• entanglement_spectrum docstring (#371) (@borisdevos)
• Minor docs and docstring changes (#372) (@borisdevos)
• Use correct orth algorithms in makefullrank (#373) (@lkdvos)
• v0.13.9 (#374) (@lkdvos)

Closed issues:

• Infinite environments do not converge when supplying product state (#332)
• similar(InfiniteMPOHamiltonian,::Type) is not an InfiniteMPOHamiltonian (#349)
• Applying an operator to a WindowMPS (#359)
• Domain/Codomain of MPODerivativeOperator is broken (#369)

v0.13.8

MPSKit v0.13.8

Diff since v0.13.7

Update TensorKit.jl compat to 0.15

Merged pull requests:

• [WIP] Updates for Tensorkit v0.15 (#331) (@lkdvos)

v0.13.7

MPSKit v0.13.7

Diff since v0.13.6

Merged pull requests:

• Modify _totuple (#328) (@Yue-Zhengyuan)

v0.13.6

MPSKit v0.13.6

Diff since v0.13.5

This release brings a combination of new features, improvements, fixes, and internal refactoring to MPSKit.jl. Below is a summary of the most important changes since v0.13.5.

---

🚀 New Features and Improvements

• Support for Non-Sorted and Overlapping Indices in MPOHamiltonian Constructors
  You can now construct MPOHamiltonian objects using terms with non-sorted and/or overlapping indices, which would previously either error or give wrong results.
  #322

• Finite temperature utilites
  There is now more extended support for working with finite-temperature systems (Matrix Product Density Operators), in particular for working with tensors with real entries.
  #324

• Objective Printing for IDMRG
  IDMRG logging previously reported the total energy (over a number of sites that grows with the iterations), which is now changed to report the current local energy density.
  #323

📝 Miscellaneous

• Documentation and formatting updates for improved maintainability.
• Improved test coverage

---

🔗 Contributors

Many thanks to everyone who contributed code, bug reports, and suggestions to this release!

---

For a full list of merged pull requests and commits in this release, see the GitHub compare view.

Merged pull requests:

• Add support for non-sorted and overlapping indices in MPOHamiltonian constructors (#322) (@lkdvos)
• Update objective printing for IDMRG (#323) (@lkdvos)
• Finite-T improvements and utility (#324) (@lkdvos)
• Small fixes (#325) (@VictorVanthilt)
• Add pre-commit hook for Runic (#326) (@lkdvos)

v0.13.5

MPSKit v0.13.5

Diff since v0.13.4

Merged pull requests:

• [Feature] FiniteMPOHamiltonian compression (#312) (@lkdvos)
• Try to fix README badges (#313) (@kshyatt)
• Bump actions/checkout from 4 to 5 (#314) (@dependabot[bot])
• Update TensorOperations docs link (#316) (@lkdvos)
• Refactor quasiparticle regularization (#318) (@lkdvos)
• Update TensorKit links (#321) (@lkdvos)

v0.13.4

MPSKit v0.13.4

Diff since v0.13.3

Merged pull requests:

• IDMRG updates (#294) (@lkdvos)
• fix typo in InfiniteMPO constructor errormessage (#302) (@VictorVanthilt)
• Runic formatter (#303) (@lkdvos)
• Fix translation operator in Ising CFT example (#304) (@leburgel)
• Set up optional color inversion for docs images (#305) (@leburgel)
• Use DMRG2 SVD alg in finite approximate DMRG2 implementation (#308) (@VictorVanthilt)
• CompatHelper: bump compat for KrylovKit to 0.10, (keep existing compat) (#309) (@github-actions[bot])
• [Utility] Be more consistent about what space functions work for what arguments (#310) (@lkdvos)

Closed issues:

• Wrong correlation length (#299)
• error in InfiniteMPS with CUDA backend (#301)
• quantum number conservation in make_time_mpo (#306)
• MethodError for BraidingTensor in rightenv (#307)

v0.13.3

MPSKit v0.13.3

Diff since v0.13.2

Merged pull requests:

• Fix derivatives at the edges of finite operators (#296) (@lkdvos)

Closed issues:

• Memory usage (VSZ) keeps growing during long runs (#291)
• In MPSKit v0.13.2, timestep! yields unphysical results, whereas in v0.12.6, it does not (#295)

v0.13.2

MPSKit v0.13.2

Diff since v0.13.1

Merged pull requests:

• Bump julia-actions/julia-format from 3 to 4 (#282) (@dependabot[bot])
• Make changebonds truncation use relative truncation schemes (#283) (@lkdvos)
• Fix O_xzx in operators manual (#284) (@Yue-Zhengyuan)
• Fix_notebook.jl (#286) (@dartsushi)
• Add utility functions for physical and virtual space retrieval (#287) (@lkdvos)
• Fix MPO copy issue of #288 (#289) (@VictorVanthilt)
• Fix scaling dimensions in Ising CFT example (#290) (@dartsushi)
• fix braille and add small test (#293) (@VictorVanthilt)

Closed issues:

• Multiplication of InfiniteMPOHamiltonian with InfiniteMPS (#278)
• max_virtualspaces throw ArgumentError: Sector has negative dimension (#285)
• Copying InfiniteMPOs returns FiniteMPOs (#288)

v0.13.1

MPSKit v0.13.1

Diff since v0.13.0

Merged pull requests:

• [Fix] correctly handle eachsite for Multiline objects (#281) (@lkdvos)