Crunch is a direct method program developed for solving difficult small and medium sized structures ab initio. Datasets should be of atomic resolution, better than 1.1 Ångstrom. The system has been designed to run under the Unix operating system. So far successful installation has been effected on the Silicon Graphics, True 64, IBM Risc 6000, HP UX, SUN and Linux platforms, both Intel and Alpha.
Although in principle aimed at difficult equal-atom problems, Crunch may be used effectively for routine structure determination, of heavy-atom structures too. The program needs about 5 Mb disk space. A minimum of 16 Mb of Ram is needed for Crunch to operate successfully. More is better here!
The system consists of two main sections and assorted utilities. The first section, Deter, determines the phases. The second section, Autofour, evaluates the results of each Deter cycle, trying to find the complete model based on the results obtained by Deter. The contents of the unit cell should be given as accurately as possible, as especially the second principal program in the system, the section which tries to find the complete model based on the results of the first section, depends on the availability of good estimates of B-overall and the scale.
Most matrix and vector manipulations are done using the Blas and Lapack routines. It is of great advantage to use processor specific optimized versions of this software. A makefile for doing this is supplied for the program Deter. An include file 'flags' which defines the flags used in the makefiles is prepared during installation. This file may be adapted to the local situation.
Crunch supports just about any input of structure factors you can think of as long as your files contain hkl, F or F**2 and sigma(F or F**2), one reflection pro record and nothing else. The entries in the file must be separated by blanks, comma's or plus or minus signs. A record such as 12 3 4 1.2340.566 will not do.
Crunch_1.1 no longer requires the presence of the Dirdif system (inquiries: ptb@sci.kun.nl or rdg@sci.kun.nl). In earlier versions of Crunch Dirdif was used to handle the crystallographic data such as cell contents, unit cell dimensions, symmetry etc. Crunch needs a reflection file containing the asymmetric unit in reciprocal space ONLY. However, your reflection file is cleaned of redundancies automatically. Installation of the public domain graphics program Rasmol is optional but recommended. Atomic coordinates are produced in the .pdb and .spf formats. Visual inspection of the results is therefore straightforward, using Rasmol or the Pluton/Platon graphic suite written by Ton Spek (email: spea@ xray1.chem.uu.nl). Unix scripts are written to be compatible with the Bourne/Korn shells. The scripts provided do not allow running more than one crunch-job from within one directory. However, multiple runs of Crunch from different directories and/or by different users are allowed.
In the opinion of the authors Crunch and Dirdif are complementary systems: Use Crunch for equal atom problems and Dirdif for heavy atom structures such as coordination compounds containing metal clusters e.d. An area of overlap exists, of course. Simple coordination compounds are solved routinely by both systems.
A fundamental difference between Dirdif and Crunch is that Dirdif uses known information e.g. partial structures. This information may consist of just the knowledge of the presence of one or more heavy atom(s), when those particular coordinates are determined by Patterson search, or a priori knowledge of the conformation of a fragment of the compound under consideration.
Crunch was developed for the ab initio solution of the phase problem, which also covers the situation where a crystallographer doesn't know anything about his compound except that it is organic as well as its approximate atomic weight. This has proven to be extremely useful for the identification of unknown natural compounds.
If you've used Crunch in any resulting paper please refer to:
Automatic Determination of Crystal Structures using Karle-Hauptman Matrices.
Acta Crystallographica A49 (1993), 287-293,
R. de Gelder, R.A.G. De Graaff & H. Schenk.