Previous: APPENDIX: Data Exchange Proposal
Up: APPENDIX: Data Exchange Proposal
Next: Implementation decisions
Previous Page: APPENDIX: Data Exchange Proposal
Next Page: Implementation decisions

Why a new format, and which?

It it important that there is a common data format with which particle data (N-body, SPH, etc.) can be exchanged and archived. There will always be different analysis packages around, and as time progresses, so will their formats. With this in mind, the following the requirements for such an interchange/archive format are readily defined.

First of all, the data must be stored in a machine independent format, well-documented and to some extend be self-describing. The dataformat must be flexible in order to store a wide variety of data structures, in other words it must be extendible. And lastly, the data must be stored efficiently. This eliminates virtually any ASCII based format, particularly if precision is important.

Within the various branches of observational 2- and 3D-imaging astronomy this had been realized a long time ago, and resulted in the creation of the FITS (Wells et al. 1981) format that is now an accepted standard (not only astronomical). It can store data in binary as well as ascii format, but is mostly known for interchanging and archiving images. Recently binary tables have been added as a FITS extension (FITS is an extendible data format description), which has greatly improved its efficiency. The rows in binary tables may be a mixture of a number of numerical, logical and character data entries. In addition, each entry is allowed to be a single dimensioned array (with obvious extensions to higher order arrays). Numeric data are kept in IEEE formats, and hence guarantee machine independency. These binary tables are an excellent way to store N-body data.

We propose to use this BINTABLE extension to implement our particular NBODY format. There are several reasons why we want to use the BINTABLE format. First of all, the BINTABLE extension has now been properly defined (Cotton and Tody, 1992) and ratified by all three regional FITS committees, although it still needs to be ratified by the IAU. Secondly, the FITSIO library (Pence, 1992) is available for a wide number of machines. It is implemented in FORTRAN, but an experimental C calling interface is also available. Thirdly, various databanks are becoming available to the astronomical community. CD-ROM's already exist with collections of standard astronomical images and tabular material from the literature (e.g. ADC, AAS). Theory should not lag behind. Important simulations should be made available to the astronomical community in a well described format. Certain NASA divisions even require their public data to be stored in FITS format.

___________________________________________________
Previous: APPENDIX: Data Exchange Proposal
Up: APPENDIX: Data Exchange Proposal
Next: Implementation decisions
Previous Page: APPENDIX: Data Exchange Proposal
Next Page: Implementation decisions