FreeEOS is a software library for calculating an equation of state for conditions in stellar interiors. The latest stable version of this library source code, FreeEOS-2.0.0, has been released on 2006-01-15. The principal change to the software for this version of FreeEOS is a complete reorganization of the Newton-Raphson iteration scheme. The former scheme, which did an "outer" iteration to determine the degeneracy parameter, fl, and an "inner" iteration to determine auxiliary variables for fixed fl, has been replaced by a combined iteration to determine fl and auxiliary variables simultaneously. The new iteration scheme is significantly faster (a factor of two in some situations).

The argument list for the free_eos subroutine has remained the same so users do not need to change their programmes at all to use the new version of FreeEOS. However, I have changed the major version number of this release from 1 to 2 for this release because the subroutines called by the free_eos have been considerably changed. Because of these changes, I have done extensive comparisons of results for FreeEOS-2.0.0 versus FreeEOS-1.6.0 for the "PTEH", "GEFF", "SIREFF", "MDH", "EOS1", "EOS1a", "EOS2", "EOS3", and "EOS4" option suites. (See the README file for an explanation of these various option suites.) In all cases, the differences were negligible since for each option suite the same free-energy model is used for both the FreeEOS-1.6.0 and FreeEOS-2.0.0 calculations. The pressure differences for EOS1 are typical of the rest of the comparisons and are in the numerical noise as expected.

This comprehensive set of tests for all named option suites was done for the kif=2 case where ln rho and ln T are the independent variables of the EOS. Only spot checks of the kif=1 case (ln P and ln T as independent variables) and kif=0 case (ln f and ln T as independent variables) have been done. Thus, those using those modes of FreeEOS calculation are advised to do comprehensive comparisons of FreeEOS-2.0.0 versus FreeEOS-1.6.0 for themselves to make sure they get the same results (in less computer time with version 2.0.0).

Although the new version of FreeEOS is significantly faster, it's cold-start procedure is not as reliable at high densities and intermediate temperatures (e.g., log rho(SI) = 3, log T = 5) as FreeEOS-1.6.0. I plan to address this issue in the next release, but, in any case, this issue should only affect extreme LMS models that are calculated with too large a step size. For smaller step sizes, a Taylor series approach is used rather than a cold start, and in the above comparisons the Taylor series approach has proved to give reliable and easily converged starting solutions for FreeEOS-2.0.0 right up to the same high-density, low temperature calculational limit that was used for FreeEOS-1.6.0. For further discussion of this calculational limit, see Paper II .

Download the FreeEOS-2.0.0 tarball from here and have a look at the release notes and ChangeLog here.

---

Copyright © 2006 Alan W. Irwin

Webmaster for FreeEOS

Hosted by