TeNeS
TeNeS copied to clipboard
Massively parallel tensor network solver
Branch | Build status | Documentation |
---|---|---|
master (latest stable) | ||
develop (latest) |
TeNeS
TeNeS (Tensor Network Solver) is a solver for 2D quantum lattice system based on a PEPS wave function and the CTM method. TeNeS can make use of many CPU/nodes through an OpenMP/MPI hybirid parallel tensor operation library, mptensor.
Online manual
Getting started
- Prerequisites and dependencies
-
Install
- Simplest way to build
- Install binaries and samples
- Specify compiler
- Disable MPI/ScaLAPACK parallelization
- Specify ScaLAPACK
- Use the pre-built mptensor
- Specify Python interpreter
-
Usage
- Use pre-defined model and lattice
- Calculate imaginary time evolution operators
- Perform
- Question or comment
- Contibution
- License
- Acknowledgement
Prerequisites and dependencies
The following tools are required for building TeNeS.
- C++11 compiler
- CMake (>=3.6.0)
- BLAS/LAPACK
TeNeS depends on the following libraries, but these are downloaded automatically through the build process.
TeNeS can be parallerized by using MPI and ScaLAPACK.
TeNeS tools (tenes_simple
, tenes_std
) are written in Python3.
The following external packages are required:
- numpy
- scipy
- toml
- typing (mandatory for python < 3.5)
Install
Simplest way to build
mkdir build
cd build
cmake ../
make
(NOTE: Some system (e.g. CentOS) provides CMake 3 as cmake3
)
The above commands makes an exectutable file tenes
in the build/src
directory.
Install binaries and samples
cmake -DCMAKE_INSTALL_PREFIX=<path to install to> ../
make
make install
Noted that the parallel building make -j <num_parallel>
can reduce the time to build.
The make install
command installs tenes
, tenes_std
, and tenes_simple
into the <path to install to>/bin
.
Samples will be also installed into the <path to install to>/share/tenes/<VERSION>/sample
.
The default value of the <path to install to>
is /usr/local
.
Specify compiler
CMake detects your compiler automatically but sometimes this is not what you want. In this case, you can specify the compiler by the following way,
cmake -DCMAKE_CXX_COMPILER=<path to your compiler> ../
Disable MPI/ScaLAPACK parallelization
To disable parallelization, pass the -DENABLE_MPI=OFF
option to cmake
commands.
If you use macos, MPI/ScaLAPACK parallelization is disabled by default because the combination of Apple Accelerate BLAS/LAPACK library with ScaLAPACK seems to have some troubles.
Specify ScaLAPACK
TeNeS finds ScaLAPACK automatically, but may fail.
In such a case, -DSCALAPACK_ROOT=<path>
option specifies the path to the ScaLAPACK library file, <path>/lib/libscalapack.so
.
Use the pre-built mptensor
TeNeS is based on the parallerized tensor library, mptensor (>= v0.3). The build system of TeNeS installs this automatically, but you can use the extra pre-built mptensor by the following way.
cmake -DMPTENSOR_ROOT=<path to mptensor> ../
Specify Python interpreter
TeNeS tools tenes_simple
and tenes_std
use python3
which can be found in PATH
via /usr/bin/env python3
.
Please make sure that python3
command invokes Python3 interpreter, for example, by using type python3
.
If you want to fix the interpreter to be used (or /usr/bin/env
does not exist), you can specify it by the following way,
cmake -DTENES_PYTHON_EXECUTABLE=<path to your interpreter> ../
Usage
Use pre-defined model and lattice
For example, the following file simple.toml
represents the transverse field Ising model on the square lattice.
[parameter]
[parameter.general]
is_real = true
[parameter.simple_update]
num_step = 1000
tau = 0.01
[parameter.full_update]
num_step = 0
tau = 0.01
[parameter.ctm]
iteration_max = 10
dimension = 10
[lattice]
type = "square lattice"
L = 2
W = 2
virtual_dim = 2
initial = "ferro"
[model]
type = "spin"
Jz = -1.0 # negative for FM interaction
Jx = 0.0
Jy = 0.0
hx = 1.0 # transverse field
tenes_simple
is a utility tool for converting this file to another file, std.toml
, denoting the operator tensors including bond hamiltonian.
tenes_simple simple.toml
Calculate imaginary time evolution operators
tenes_std
is another utility tool for calculating imaginary time evolution operators and converting std.toml
to the input file of tenes
, input.toml
.
tenes_std std.toml
By editing std.toml
, users can perform other models and lattices as ones like.
Perform
To perform simulation, pass input.toml
to tenes
as the following
tenes input.toml
Results can be found in output
directory.
For example, expectation values of operators per site are stored in output/densities.dat
as the following,
Sz = 2.97866964051826333e-01 0.00000000000000000e+00
Sx = 3.86024172907023511e-01 0.00000000000000000e+00
hamiltonian = -7.57303058659582140e-01 0.00000000000000000e+00
SzSz = 2.16869216589772901e-01 0.00000000000000000e+00
SxSx = 3.19350111777505108e-01 0.00000000000000000e+00
SySy = -4.77650003168152704e-02 0.00000000000000000e+00
The file format of input/output files is described in the manual page.
Question or comment
Feel free to ask any question through an issue (public) or an e-mail (private) (tenes-dev__at__issp.u-tokyo.ac.jp
, __at__ -> @
).
Contibution
Pull request is welcome (even for a small typo, of course!). Before send a PR, please make sure the following:
- Rebase (or merge)
develop
branch into your feature branch - Check
make
andctest
processes pass - Format Codes by using
clang-format
(C++) andblack
(Python)
(Incomplete) developer's document written in doxygen is available.
- Move to
docs/doxygen
- Invoke
doxygen
- Open
doxygen_out/html/index.html
in your browser
License
TeNeS is available under the GNU GPL v3.
Paper
When you publish the results by using TeNeS, we would appreciate if you cite the following paper:
Acknowledgement
TeNeS was supported by MEXT as "Exploratory Challenge on Post-K computer" (Frontiers of Basic Science: Challenging the Limits) and "Priority Issue on Post-K computer" (Creation of New Functional Devices and High-Performance Materials to Support Next-Generation Industries). We also would also like to express our thanks for the support of the "Project for advancement of software usability in materials science" of The Institute for Solid State Physics, The University of Tokyo, for the development of TeNeS.