The last installation step applies to Unix and non-Unix installers. It is the execution of the supplied test files. Please consider this to be an important and critical step. [We have been less than amused in the past, following a considerable investment of our time checking out a problem, to discover that the initial Xtal installation was not tested!] Use the following approach.

A. To apply the Xtal module generated from the linking process, it is necessary to ensure that the standard line input and line output files are correctly attached. All other files are opened automatically by the Xtal nucleus as required (see B below). The assignment of the line input and line output files is simple, but it does require knowledge of how the macros unitinp: and unitprt: have been defined in the XMACRO file. For Unix machines these will be set as stdin and stdout, respectively. In this case the input line file (named, say, test.inp) and the line output file (named, say, test.prt) would be attached to the execuatable file (named, say, xtal.x) as

xtal.x < test.inp > test.prt

For non-Unix machines the assignment of stdin and stdout will be done within a command procedure.

B. All line and binary files, other than the standard line input and line output files, are opened by the Xtal nucleus using filenames based on the construction defined by macro filename:. These filenames will normally be of the form <compid>.<ext>.

C. You are now ready to run one of the test decks supplied on the Xtal tape. Rename an input file (e.g. p6122.d) as the line input file (e.g. p6122.inp).

D. Run Xtal with the p6122.inp file as input. When the run is complete, use your editor to check the output line file (say, p6122.prt) against the supplied p6122.l list file. Note that lines in p6122.l are truncated to 80 characters whereas your list file will probably have variable length lines which will extend to 132 characters. There will also be differences between these two files due to filenames, dates and times. However, the numerical results of all calculations must be equivalent apart from small differences due to rounding-off errors (the supplied list files are from a 32-bit Unix machine applying the IEEE fp standard).

E. If the test run fails, either through an Xtal error (i.e. a controlled Xtal exit) or an operating system error (i.e. Fortran error or crash), you will need to establish the precise cause. The Xtal error message will contain the program name, subroutine number, and error number. If the errfile: macro has been correctly defined, this information will be accompanied by a descriptive message. The supplied test files have been run on a wide variety of machines, so the most likely cause of the error will be incorrect file assignments, or inappropriate macro definitions. Operating system errors can arise for the same reasons. Further advice about errors is given in H below.

F. The p6122.d test file applies a variety of the small molecule calculations available in the Xtal system. If you have not installed all of the programs applied in this test, then the test execution will exit when the system attempts to call the first unavailable subroutine. Either remove the input lines requesting the uninstalled program, or re-link the system including these programs. Repeat the test run until it is error free. Try the other test files (saly.d, diam.d, ags4.d and lac1.d) as well. Each provides you with more information on how the system works and gives you experience for your own calculations. Run the test file exampl.d if you have installed PIG.

G. When the test files execute successfully, try your own input data file. To prepare your first input data file read the Primer Section and relevant program descriptions carefully. Refer to the test input files as examples. Note that most input parameters have default values. After a run look closely at the print file and understand how the archive files are interchanged between calculations, and how the master files are stored. Use your editor to examine the aaa file and compare this with Archive File description in the back of this manual.

H. If you encounter an error in a calculation which does not appear to be related to local conditions, please do the following:

* Carefully check your input file and the documented requirements for this calculation. Our experience is that most reported errors are because the user has not read the documentation and inputs data based on experience with other software.

* Send (by email if possible) a copy of all relevant information to the Xtal Coordinator. This should include

- an input line file that will enable the error to be recreated as an ab initio run. That is, from STARTX onwards.

- a copy of the output print file of the erroneous run.

- an assessment of the error, and any evidence that it is an error.

- any supplementary data about a possible cause of the error.

A final word of advice about errors. Most 'problems' reported to us are due to user input errors during the calculation where the 'problem' was detected, or during some prior calculation. A typical mistake is that some required quantity is missing from the archive file and a preliminary calculation needed to run. We appreciate your diligence in keeping 'ghost' errors to a minimum!