mptensor

"mptensor" is parallel C++ libarary for tensor calculations. It provides similar interfaces as Numpy and Scipy in Python.

Prerequisites

• C++11 compiler
• CMake (>= 3.6)
• LAPACK

For parallel computing

• MPI Library
• ScaLAPACK

For document generation

• Doxygen (>= 1.9.1)

How to Use

1. Build and install mptensor library

    # Modern CMake (>= 3.15)
    cmake -B build
    cmake --build build
    sudo cmake --install build
   
    # Traditional way
    mkdir build
    cd build
    cmake ../
    make
    sudo cmake install
   

2. Include the header file mptensor/mptensor.hpp in your codes.
3. Complie your applications with link option -lmptensor .

The default install directory is /usr/local. It can be changed by -DCMAKE_INSTALL_PREFIX option.

# Modern CMake
cmake --install build --prefix your_install_prefix
# Traditional way
cmake -DCMAKE_INSTALL_PREFIX=your_install_path ../

See also the CMake documentation.

Documents

The HTML documents are available in here.

Examples

#include <mptensor.hpp>
using namespace mptensor;
typedef Tensor<scalapack::Matrix,double> ptensor;
ptensor A(Shape(3,4,5));

Example codes of TRG and HOTRG for the 2D Ising model are in examples/Ising_2D.

License

GNU Lesser General Public License v3.0 (see LICENSE)

Links

• TeNeS: Parallel tensor network solver for 2D quantum lattice systems
• Tensordot: Code generator for tensor contraction
• cuscalapack: pdgemm and pzgemm with cuBLAS