pypipre 1.0.4

Physics Informed Preconditioners and Solvers

Physics Informed Preconditioners

Introduction

Pipre, short for Physics Informed Preconditioners, is a large-scale linear equation solver package. It supports distributed heterogeneous parallel computing (MPI+CPU/GPU/DCU/...) and provides physics-informed preconditioners for solvers, especially physics-informed AMG preconditioners. It is dedicated to stably and efficiently solving large-scale linear algebraic equations derived from the discretization of mathematical and physical models.

Install

$ pip install pypipre

Usage

use the following command to get the help message:

$ python -m pypipre help

copy the example out

$ cp -r path_to_pypipre/examples .

then run the example:

$ python solver.py thermal1.mtx thermal1_b.mtx solver.json

The following python script shows the above solver.py. read matrix A and vector b from MatrixMarket files, using CUDA to solve Ax=b.

# solver.py
import pypipre as pipre
import sys, re, os
#import json
import commentjson as json

fnA = sys.argv[1]
fnb = sys.argv[2]
config = sys.argv[3]

params = json.load(open(config))

A = pipre.ParCSRMatrix()
A.loadFromFile(fnA)
b = pipre.ParMatrix()
b.loadFromFile(fnb)
if b.getSize() == 0:
    b.resize(A.getRows(), 1)
    b.fill(1)

# transfer the matrix and vector to gpu 0.
dev = pipre.Device("cuda:0")
A = A.toDevice(dev)
b = b.toDevice(dev)

# use gpu 0 to finish the computation.
precond = pipre.createPrecond(params["preconditioner"])
precond.setup(A)
solver = pipre.createSolver(params["solver"])
solver.setup(A)

out = solver.solve(precond, A, b)

Solvers and Preconditioners

use

$ python -m pypipre info

to show the builtin solvers, preconditioners, smoothers, level_transfers.

Compatibility

• OS version

Now support Linux, Ubuntu/Federa/... will be OK. Windows version will be available later.

• PYTHON version

the python version is >=3.8.

• GPU/CUDA version

If you want to use CUDA, ensure that CUDA Version >= 12.4.

• DCU version

please contact us to compile the specified version.

• DISTRIBUTED version

pypipre support MPI since pypipre-1.0.4.

License

This software is free software distributed under the Lesser General Public License or LGPL, version 3.0 or any later versions. This software distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Lesser General Public License for more details.

You should have received a copy of the GNU Lesser General Public License along with PIPRE. If not, see http://www.gnu.org/licenses/.

This software depends on GLOG, AMGCL, their license files are located in the package.