_____ ________________
|_ _| /\ / ______________/
| | __ _ / \ | (___ | ____/
| | / _' | / /\ \ \___ \| |__
_| || (_| |/ ____ \ ) | __|
|_____\__, /_/ _\_\____) | |
__/ | /__________/|_|
|___/
--------------------------------------------------------------------------
Isogeometric-Assembling-by-Sum-Factorization
--------------------------------------------------------------------------
This is a small library that assembles the system matrix of an Isogeometric Analysis discretization of PDEs based on tensor product splines. It implements the algorithms described in the research paper
A. Bressan, S. Takacs. Sum-factorization techniques in Isogeometric Analysis. Computer Methods in Applied Mechanics and Engineering, vol. 352, p. 437 - 460, 2019. journal version or arXiv:1809.05471
It comes with some test programs that can assemble the system matrix of scalar second order PDEs with constant coefficients.
This work has been possible thanks to the support of the European Research Council under the European Union's Seventh Framework Programme (FP7/2007-2013) / ERC grant agreement 339643 and of the Austrian Science Fund (FWF) under grant NFN S117-03.
Use the makefile:
...
Type
make all ... to make executables of this library (debug)
make all-release ... to make executables of this library (release)
make so ... to make this library as so-file (debug)
make so-release ... to make this library as so-file (release)
make unittest ... to make and run unittests (debug)
make unittest-release ... to make and run unittests (release)
make clean ... to clean the build environment (debug)
make clean-release ... to clean the build environment (release)
make echoall ... to get the contents of the given variables
Use the files
config/debug and config/release
to specify the corresponding configuration. If those files are deleted,
they will be restored based on the corresponding templates.
Invoke after 'make all'
./generateTest-debug ... to create a configuration file for assembling
the problem of interst
./runTest-debug test ... to assemble a problem specified by such a
configuration file test, like with
./runTest-debug tests/QuarterAnnulus_stiff_128/deg03 -o mat
to assemble and store result in file mat using
a binary format
./echoMatrix-debug mat ... to echo the matrix data from file mat
./compareMatrices-debug mat1 mat2
... to compare such matrix files
./runTestsAndCompare-debug test[s]
... to assemble the problem with all methods and
compare the results and write results to
log.txt and compare.txt, like with
./runTestsAndCompare-debug tests/QuarterAnnulus_stiff_128/*
To extend the library to different PDEs (not second order with constant coefficients) it is necessary to code a description of the PDE in c++.
The description should be an object that derives from Model and implements the initParts method.
The bilinear form should be decomposed in terms, each involving a single pair of partial derivatives of the test and trial functions. The initParts method should allocate an array containing the values of weight for each derivative pair.
See assembling/model.h and assembling/second_order.h for more details.
