This web-page gives links for downloading the CRUNCH programs for carrying out
ab initio
valence bond
calculations
as well as RHF, ROHF, MCSCF, and MOCI treatments of molecular systems. These
are in the source code form and have been successfully compiled and run on a
number of systems. (See below) The package is in the form of a gzipped tar
archive, and when it is unpacked, a crunchN.MM/README file will be
produced. This is reproduced below for the convenience of user.
The unpacked files will also contain both a postscript and .pdf versions of the user's manual. Separate copies of these are linked below, so that the user may examine it before dealing with the programs, should that be the wish.
The ``README'' file:
April 15, 2008 These are C versions of the CRUNCH programs. The compilation should be fairly easy. The programs do not use resources from any obscure corners of gcc and its friends. It is assumed that the file, crunchN.MM.tgz, is unpacked in the user's home directory. At present, the system is not set up to change this easily. The compilation and installation will produce a $HOME/cfg directory and will use, if already there, or produce a $HOME/bin directory. If these conflict in any way with the user's setup, it is probably best, at present, to set up a separate user and $HOME directory for these programs. Use the command prompt> tar zxf crunchN.MM.tgz <enter> This will create a directory, crunchN.MM/. The following sequence of operations should work. prompt> cd crunchN.MM/ <enter> prompt> compall >&! compall.log & <enter> The compilation requires about 1 minute on a 1.35 GHz Athlon system. You may ignore the warnings about unused variables that occur in the .log file. I strongly recommend sending all of the output to the the compall.log file, as shown. All of the executables go into a directory $HOME/bin and to use these easily, this should be in your path in .tschrc or .profile or whatever system you use. The programs, lobeWat, gscfnn, dipole, symgenn, trannn, matrix, omatrix, and neweig, are, I think, about as bug-free as anything ever gets. For some years the only problems I have had with them have been cured by correcting the input data. The exception is with convergence in the SCF, but of course, this is not a bug in the classic sense. The remainder of the modules have not been tested so thoroughly. The manual for using crunch is in the docs/ directory in both postscript and .pdf formats. There are example input files for a number of calculations in the examples/ directory. There is also a copy of what I think the concatenated output should look like for each case. This has the suffix ".sve" attached. Some of these modules are the output of a rather precarious FORTRAN to C translator I once had, long before GNU's f2c was available. That is the reason there is a header file called 'fortran.h' in some of the directories. It needs to be there. That translator produced ugly C code, but it did it right, mostly. The only portability problem I have ever had concerns the 'time.c' function used in some places. It has worked on LINUX, AIX, IRIS, and HPUX, but not XENIX or SCO-UNIX. It may need to be kludged for some unices. These programs have never been tested in Bourne-shell-like environ- ments. I have only used them with csh and tcsh. If you have any questions or comments, please e-mail them to me. Good Luck! G. A. Gallup ggallup1@unl.edu http://www.unl.edu/ggallup
Click here crnchman.pdf to examine the CRUNCH manual in PDF form. A ``right-click'' will allow you to download it or the Postscript form of the manual crunch.man.ps. A version of the manual in HTML is currently being reconstructed.
Right-click here crnch301.tgz in order to download the gzipped tar archive of the program and documents.
This page last modified 15 April 2008.