In order to run MOLPRO you have first to write an input file. This can be done in any directory of your choice. Input files can have any name, but it is best to avoid using a file name with extension .out. It is often convenient to name the input file as h2o.inp, benzene.inp, depending on the molecule, or whatever you like.
Simple examples for input files will be given in the following sections. Once the input file is ready, a MOLPRO calculation can be started using the molpro command. Assume the input file is called h2o.inp. The command to run MOLPRO is then simply
molpro h2o.inp &
This will create an output file h2o.out, i.e. the extension .inp is replaced by .out. The same would happen with any other extension, e.g., h2o.com, h2o.input, h2o.test would all produce the output in h2o.out. If h2o.out already exists, the old output will be moved to a backup directory that is, by default, created with the name h2o.d. In this way old outputs are not lost. This mechanism can be disabled using the -s option:
molpro -s h2o.inp &
would not save an existing output file but simply overwrite it. If you want to give your output a different name than the default one, you can use the -o option, for instance
molpro -o water.output h2o.inp &
As well as producing the .out file, a structured XML file (which is actually a parent of the .out file) is created with suffix .xml. This file can be useful for automated post-processing of results, for example graphical rendering by the MolproView program.
There are many other options for the molpro command, most of which, however, do not often need to be specified. You can find a full description in the MOLPRO reference manual. Some of the more important ones are
The amount of memory MOLPRO is allowed to use can be specified using the -m option, e.g.,
molpro -m 4M h2o.inp &
This means that 4 megawords (MW) of memory are allocated by molpro (see also memory input card, section 19.1.
molpro -n 8 h2o.inp &
will run 8 cooperative processes. Depending on how the program has been built, this will result in 7- or 8-way parallel execution.
Optionally, but not by default, it is possible so specify one or more named files in the input, and these are used by the program to store intermediate parameters and results, that can be used to restart (see section 19.2). If nothing is given in the input, these files are still present, but they are temporary, and disappear at the end of the job. For performance reasons, the program runs in scratch directory, for example a high-performance file system. By default, the location of this directory is /tmp/$USER, or taken from the environment variable TMPDIR (if set), but it can be controlled using the -d option, for example
molpro -d /scratch/$USER h2o.inp &
At the end of the job, the named files are copied to a permanent location. The most important file for a restart is file 2, which contains the wave function information (cf. section 19.2). By default, this file is copied into the directory $HOME/wfu, but this directory can be changed with the -W option. At the beginning of the job, if appropriate files exist in the permanent location, they are copied in to the scratch directory.