MOLSCAT v. 14

Bugfixes

The following problems have been identified since the original release of MOLSCAT, version 14, in Aug 1994. Affected calculations are briefly described, along with the action taken to correct the problem. The "version date" (PDATE) reported by the program has not been changed unless indicated.

1 Nov 94: base.f
ITYPE=4 cases incorrectly calculated parity for assignment to parity subblocks.
Corrected and replaced base.f in v14.f and v14.f.Z
10 Mar 95: gaushp.f
Gauss-Hermite routine provided incorrect weight if only a single integration point was requested
Corrected and replaced gaushp.f in v14.f and v14.f.Z
21 Mar 95: EUNITS
Allowing the NAMELIST variable EUNITS to accept either an integer or a character string value, which worked on some earlier machines, rarely works on modern computers. Subroutines DRIVER, BASE, ECNV, ECNVX, and STSRCH have been modified so that EUNITS is restricted to integer values, and a new CHARACTER*4 NAMELIST variable EUNITC has been added, with a default value of blanks. A nonblank input value for EUNITC will override any value input in EUNITS. The version date (PDATE) has been changed.
Modified routines DRIVER, BASE, ECNV, ECVNX, and STSRCH have been put into v14.f and v14.f.Z and are also available separately here.
Section 2 and Section 3 of the documentation have been modified.
28 Mar 95: findrm.f
In trying to estimate a starting distance for the propagation findrm.f sometimes encounters difficulties. Often these are due to an interaction potential which is poorly behaved in the classically forbidden region. In cases where a classically allowed region was encountered and an inner turning point could not be located or where the origin was reached before a suitable staring point was found, the original algorithm chose a starting value of R=0. but this is rarely desirable. The new routine chooses 0.8 times the input RMIN value for these cases which generally is reasonable.
Corrected and replaced findrm.f in v14.f and v14.f.Z
31 Jul 95: thrj.f
Problems with the S-matrix post-processing program for differential scattering cross sections were traced to a faulty initialization of the factorials in thrj.f.
Function thrj.f has been corrected and replaced in v14.f and v14.f.Z
3 Aug 95: storag.f
A bug was found in allocating storage for the WKB integrator in cases which use numerical derivatives.
Subroutine storag.f has been corrected and replaced in v14.f and v14.f.Z
30 Aug 95: Driver.f, output.f, restrt.f
This was not a bug, but these routines were modified so that the program would give the number of channels and number of open channels as part of the summary line printed for each set of coupled equations.
The modified routines are available separately and have also been replaced in v14.f and v14.f.Z
5 Sept 95: couple.f, restrt.f, sig6.f
Corrected bugs associated with unitialized variables in some error condition write statements. None of these should have caused erroneous results.
The modified routines have been replaced in v14.f and v14.f.Z
21 Sept 95: plm.f
It was pointed out by Dr. Marie-Lise Dubernet <mld@ipp-garching.mpg.de> that the algorithm used to obtain associated Legendre polynomials was inaccurate and caused problems for high values of the indices (higher than typically used by MOLSCAT). She suggested an alternate routine which has been tested and incorporated into the code.
NOTE: the version posted 28 Jun 95 had bugs. If you don't have a version dated 21 Sept 91, you should update.
New plm.f has been incorporated in v14.f and v14.f.Z
19 Oct 95: potenl.f
Some changes were made to get the print out to reflect correctly whether and how the projection mechanism was used. It is believed that actual calculations were not affected by these bugs, but only the print out. New potenl.f has been incorporated in v14.f and v14.f.Z

Return to MOLSCAT Homepage