(Originally forked from https://github.com/orialb/TimeEvoMPS.jl)
Implementations of time-evolution algorithms for matrix-product states, based on the ITensor library:
- one-site time-dependent variational principle (TDVP1) in its standard version
as well as
- its variant with adaptive bond dimensions, and
- a non-unitary version (for vectorised mixed states);
- matrix-product states in the Vidal (or canonical) and inverse-canonical gauges,
- the time-evolving block-decimation algorithm (TEBD) for Vidal-form MPS, together with automatic 1st and 2nd order Suzuki-Trotter decompositions,
- the two-site time-dependent variational principle (TDVP2), in a serial and a parallel version.
See the package documentation for a complete list of features, references, a description of the available methods, and some tutorials.
This package is registered in the TensorNetworkSimulations registry. By first adding this registry, with
using Pkg
pkg"registry add https://github.com/phaerrax/TensorNetworkSimulations.git"(this must be done just once per Julia installation) the package can then be installed as a normal one:
using Pkg
pkg"add MPSTimeEvolution"Alternatively, straight installation from GitHub is also possible:
using Pkg
pkg "add https://github.com/phaerrax/MPSTimeEvolution.jl"