DDASSL 91-06-25 Replaced 83-03-15 version with 91-06-24 version. The
details of the changes are described in the leading
comments after "REVISION HISTORY" in the DDASSL routine.
SDASSL 91-06-25 Replaced 83-03-15 version with 91-06-24 version. The
details of the changes are described in the leading
comments after "REVISION HISTORY" in the SDASSL routine.
Thu Jun 27 14:40:58 EDT 1991 ehg
Installed VODE and VODPK codes from ERENJ.BITNET!GDBYRNE.
6 Sep 1991 wmc
single precision files merged into this directory.
7 Sep 1991 wmc
initial installation of rkc from cwi.nl!bsom
15 Jul 1992 ehg
vodpk.f changed R1MACH to D1MACH, since this is double precision
2 Jul 2001 ehg based on mail from "N. Shamsundar" <shamsundar@UH.EDU>
acdc.f Save Nvold
Date: Wed, 23 Jul 2003 13:17:49 -0700 (PDT)
From: Alan Hindmarsh <alanh@llnl.gov>
Subject: DASSL family upgrade
...
DASKR is a relatively new member of the family; it is DASPK with
rootfinding added. We now are encouraging users of DASSL/DASRT
to switch to DASPK/DASKR, but are leaving the older codes for the
sake of users who don't need the newer features.
...
The only changes in DASSL and DASPK are the fix to the unit roundoff
routines. For DASRT, there were also some minor corrections needed.
Date: Tue, 17 May 2005 14:33:20 -0700 (PDT)
From: Alan Hindmarsh <alanh@llnl.gov>
The changes have to do with fixes to bugs that are relatavily minor,
which is why I waited till I had several of them before doing this
request. In summary, the bugs are:
1) In the interpolation routine (all ODEPACK and VOD* solvers), the
roundoff perturbation for testing the input t for legality is
replaced with a more robust expression.
2) In the rootfinding routines *RCHEK (in LSODAR, LSODKR, and DASKR),
imposed a better minimum on the small t increment used to check
for two nearby zeros.
3) In the *ROOTS routines (rootfinding in LSODAR, LSODKR, DASKR), used
a more robust algorithm for adjusting the new guess X2 in case it
is too close to the endpoints of the current bracketing interval.
4) In LSODKR demo program, corrected uninitialized workspace
variables to handle output at tout = 0.
5) Minor modifications in the DASKR package -- in the main driver
(rewrote stopping tests), the example and demo programs and their
outputs, and in preconditioner files.
From: "Severiano Gonzalez Pinto" <spinto@ull.es>
To: "Eric Grosse" <ehg@lucent.com>
Subject: Re: Software for second order problems (this-is-not-spam)
Date: Wed, 8 Mar 2006 19:11:52 +0100
Dear Eric,
I have followed your three suggestions.
1) FCN has been changed to FNC everywhere.
2) All Tabs have been deleted and they have been replaced by spaces.
3) The version of Fortran used is FORTRAN 90 (standard), supplied by Compaq
Visual Fortran, Professional Edition 6.6.0. Sorry, because firstly I thought
it was FORTRAN 77.
For any question or help, do not hesitate to contact me.
Thanks in advance for everything.
Best wishes.
Severiano Gonzalez Pinto
Dpto Analisis Matematico
Universidad de La Laguna
phone: 922 31 82 01
2006-07-19
In response to two user bug reports, I have made some minor corrections
in the DASKR package (which you installed for me in 2003, and replaced
with an upgraded version in 2005).
There were two minor bugs -- (a) in the ilut and ilutp routines in the
SPARSKIT package we use for preconditioning, and (b) in the increment
formula for difference quotient Jacobians. Both bugs were easy to fix.
I have discussed the changes with Linda Petzold, and she approves.
There are changes in 6 files. Along with four new source files (both
single and double precision), there are two updated output files for
examples that I ran to validate the code.
Alan Hindmarsh