Automatic building of matrix product operators (autoMPO)

Automatic building of matrix product operators (autoMPO)#

namespace autompo#

This approach is applicable when the number of term is not too big.

Typedefs

template<class T = double, int d = 2> using LocOp = typename arma::Mat<T>::template fixed<d, d>#

Functions

template<class T, int d> ProdOp<T, d> operator*(ProdOp<T, d> const &A, ProdOp<T, d> const &B)#

template<class T, int d> PolyOp<T, d> operator+(PolyOp<T, d> A, PolyOp<T, d> const &B)#

template<class T = double, int d = 2> struct ProdOp : public std::map<int, LocOp<double, 2>>#

template<class T = double, int d = 2> struct PolyOp#

#include <auto_mpo.h>

just a collection of ProdOp

Public Functions

inline T overlap(const TensorTrain<T> &mps) const#: compute the overlap with a given mps (the physical dimensions should match)

Public Members

int compressEvery = 20#

These parameters control the conversion to mps. They are also used when the size of the collection reached maxNTerm.

number of terms to form one mps.

int maxNTerm = 100000#: to protect against memory overflow, when this number is reached, the terms are compressed to tt.