Introduction

PyMGRIT is a package for the Multigrid-Reduction-in-Time (MGRIT) algorithm in Python.

PyMGRIT is currently developed by Jens Hahne and Stephanie Friedhoff.

What is MGRIT?

The MGRIT algorithm is a reduction-based time-multigrid method for solving time-dependent problems. In contrast to solving sequentially for one time step after the other, the MGRIT algorithm is an iterative method that allows calculating multiple time steps simultaneously by using a time-grid hierarchy. The MGRIT method is a non-intrusive approach that essentially uses the same time integrator as a traditional time-stepping algorithm. Therefore, it is particularly well suited for introducing time parallelism in simulations using existing application codes.

PyMGRIT Features

PyMGRIT features:

• Classical Multigrid-Reduction-in-Time (MGRIT) for solving evolutionary systems of equations
  • Non-intrusive approach
  • Optimal time-multigrid algorithm
  • A variety of cycling strategies, relaxation schemes, and coarsening strategies
• Time parallelism
• Specific to space-time problems
  • Space & time parallelism
  • Additional coarsening in space

Citing

@MISC{PyMGRIT,
  author = "Hahne, J. and Friedhoff, S.",
  title = "{PyMGRIT}: Multigrid-Reduction-in-Time in {Python} v1.0",
  year = "2020",
  url = "https://github.com/pymgrit/pymgrit",
  note = "Release 1.0"
  }

Installation

PyMGRIT requires mpicc (from openmpi or mpich)

>>> pip3 install pymgrit

or

>>> git clone https://github.com/pymgrit/pymgrit.git
>>> cd pymgrit
>>> pip3 install .

Example Usage

PyMGRIT is easy to use! The following code generates a discrete Dahlquist test problem and solves the resulting linear system using a two-level MGRIT algorithm.:

# Import PyMGRIT
from pymgrit import *

# Create Dahlquist's test problem with 101 time steps in the interval [0, 5]
dahlquist = Dahlquist(t_start=0, t_stop=5, nt=101)

# Construct a two-level multigrid hierarchy for the test problem using a coarsening factor of 2
dahlquist_multilevel_structure = simple_setup_problem(problem=dahlquist, level=2, coarsening=2)

# Set up the MGRIT solver for the test problem and set the solver tolerance to 1e-10
mgrit = Mgrit(problem=dahlquist_multilevel_structure, tol=1e-10)

# Solve the test problem
info = mgrit.solve()

Program output:

INFO - 21-02-20 16:18:43 - Start setup
INFO - 21-02-20 16:18:43 - Setup took 0.009232759475708008 s
INFO - 21-02-20 16:18:43 - Start solve
INFO - 21-02-20 16:18:43 - iter 1  | conv: 7.186185937031941e-05  | conv factor: -                       | runtime: 0.013237237930297852 s
INFO - 21-02-20 16:18:43 - iter 2  | conv: 1.2461067076355103e-06 | conv factor: 0.017340307063501627    | runtime: 0.010195493698120117 s
INFO - 21-02-20 16:18:43 - iter 3  | conv: 2.1015566145245807e-08 | conv factor: 0.016864981158092696    | runtime: 0.008922338485717773 s
INFO - 21-02-20 16:18:43 - iter 4  | conv: 3.144127445017594e-10  | conv factor: 0.014960945726074891    | runtime: 0.0062139034271240234 s
INFO - 21-02-20 16:18:43 - iter 5  | conv: 3.975214076032893e-12  | conv factor: 0.01264329816633959     | runtime: 0.006150722503662109 s
INFO - 21-02-20 16:18:43 - Solve took 0.05394101142883301 s
INFO - 21-02-20 16:18:43 - Run parameter overview
  time interval             : [0.0, 5.0]
  number of time points     : 101
  max dt                    : 0.05000000000000071
  number of levels          : 2
  coarsening factors        : [2]
  cf_iter                   : 1
  nested iteration          : True
  cycle type                : V
  stopping tolerance        : 1e-10
  time communicator size    : 1
  space communicator size   : 1

Getting Help

For documentation see https://pymgrit.github.io/pymgrit/

Create an issue.

Look at the Quickstart, Tutorial or the Examples.