Abstract
BACOL and BACOLR are (Fortran 77) B-spline
adaptive collocation packages for the numerical solution of 1D parabolic
Partial Differential Equations (PDEs). The packages have been shown to
be superior to other similar packages, especially for problems
exhibiting sharp, moving spatial layer regions, where a stringent
tolerance is imposed. In addition to providing temporal error control
through the timestepping software, BACOL and BACOLR feature control of a
high-order estimate of the spatial error of the approximate solution,
obtained by computing a second approximate solution of one higher
order of accuracy; the cost is substantial—execution time and memory
usage are almost doubled.
In this article, we
discuss BACOLI, a new version of BACOL that computes only one
approximate solution and uses efficient interpolation-based schemes to
obtain a spatial error estimate. In previous studies these schemes have
been shown to provide spatial error estimates of comparable quality to
those of BACOL. We describe the substantial modification of BACOL needed
to obtain BACOLI, and provide numerical results showing that BACOLI is
significantly more efficient than BACOL, in some cases by as much as a
factor of 2. We also introduce a Fortran 95 wrapper for BACOLI (called
BACOLI95) and discuss its simplified user interface.
Original language | English |
---|---|
Article number | 25 |
Pages (from-to) | 1-17 |
Number of pages | 17 |
Journal | ACM Transactions on Mathematical Software |
Volume | 42 |
Issue number | 3 |
Early online date | 23 May 2016 |
DOIs | |
Publication status | Published - Jun 2016 |
User-Defined Keywords
- 1D parabolic PDEs
- Algorithms
- Collocation
- Efficiency
- G.1.8 [partial differential equations]: method of lines
- Interpolation
- Numerical software
- Performance
- Spatial error estimation