Scikit Tt Versions Save

updates in module tensor_train.py:

• diag: construct diagonal TT operators
• squeeze: contract cores with mode dimensions 1
• tensordot: index contractions of tensor trains
• canonical: tensor products of the canonical basis
• residual_error: error of systems of linear equations
• improved stability of ortho
• TT class now support complex tensor trains

• updates in module ode.py:

  • hod: higher-order differencing for quantum mechanics
  • lie_splitting, strang_splitting, yoshida_splitting, kahan_li_splitting: splitting schemes for ODEs with SLIM operators

• updates in module evp.py:

  • als now uses an integrated Wielandt deflation for eigenvalue shifting
  • stop iterations if convergence is detected

• new module quantum_computation.py:

  • methods for simulating quantum circuits in TT format

• updates in module models.py:

  • exciton_chain: chain of coupled excitons
  • iqft: inverse quantum Fourier transform
  • qfa: quantum full adder
  • qfan: quantum full adder network
  • qft: quantum Fourier transform
  • shor: oracle of Shor's algorithm
  • simon: final quantum state after applying a Simon's circuit

• new module tgedmd.py:

  • methods for approximating infinitesimal Koopman generators

• updates in module transform.py:

  • changed basis functions to classes
  • added Legendre

• updates in module utils.py:

  • truncated_svd: compute truncated SVD using relative/absolute threshold

• added new examples:

  • ala10_rank_test, ala10_tgedmd
  • lemon_slice_reversibel, lemon_slice_reweighting
  • qfa, qfan, qft, shor, simon

• added unit tests

• fixed docstrings

• use Github actions for automatic building and testing

• minor changes and bug fixes

• new functions in module tensor_train:
  • svd: compute global SVDs of tensor trains
  • conj: complex conjugate of a tensor train
  • TT class can now handle segments of tensor trains
• new function in module ode:
  • sod: second-order differencing for linear differential equations
• added module data_driven.transform:
  • methods for the construction of transformed data tensors in TT format
  • different basis functions
  • coordinate-major, function-major, and general basis decomposition
  • hocur: higher-order CUR decomposition
  • gram: compute Gram matrix of two transformed data tensors
• added module data_driven.tedmd:
  • AMUSEt algorithm using HOSVD and HOCUR
• added module data_driven.regression:
  • methods for solving regression problems in TT format
  • ARR, MANDy using coordinate-major and function-major approach, kernel-based MANDy
• added examples:
  • ala10: apply tEDMD to time series data of deca-alanine
  • mnist: tensor-based image classification of theMNIST and FMNIST data set
  • ntl9: apply tEDMD to time series data of NTL9
  • toll_station: compute distribution of cars at a toll station
• minor changes and bug fixes

• added module data_driven
  • added tDMD
  • renamed perron_frobenius to ulam
• new models
  • toll station
  • fractals
• new methods in solvers.evp
  • ALS can now be applied to generalized eigenvalue problems
  • added power method
• Python 2 compatibility
• removed Matplotlib dependency
• sync with Travis CI and Codecov
• minor changes in some examples

First version of scikit_tt