Tutorial

using FunnyTN.TensorNetworks
using LinearAlgebra

Matrix Product State

isnormalized(mps::MPS) = norm(mps) ≈ 1

isnormalized (generic function with 1 method)

Example: compressing

Let's prepair an MPS and two copies of them

mps0 = rand_mps(2, [1,5,8, 16, 64,100, 400, 400, 200, 200, 100, 80, 40, 20, 6, 1], l=1)
mps0 |> normalize!
@assert mps0 |> isnormalized
mps1 = copy(mps0)
mps2 = copy(mps0);

Now, let's compress mps1 and mps2 in two different ways, sweep and direct. We will see sweep has higher fidelity than direct compress.

compress!(mps1, 20)
canomove!(mps2, nsite(mps2)-1, D=20)
@show mps1'*mps0
@show mps2'*mps0;

This page was generated using Literate.jl.