MIDAS Data Structures

Here we describe and discuss the various data entities (structures) that MIDAS recognizes. They are stored in an internal binary format, accessible only through MIDAS and fall into the following categories:

Images

  are a set of data of same physical significance in one to three dimensions. The data must be sampled with constant step size along all 1, 2 or 3 axes and are stored in different formates, e.g., as bytes, 16 bit integers, or 32 bit reals on disk. However, most MIDAS applications work on real data, so the image pixels are converted on the fly to real format if necessary. The default file type is .bdf .
Besides the internal binary format also images in FITS format are supported by all Midas applications.

Tables

  are a structure for handling data which can be arranged in rows and columns. The data may be of numerical or character type. Numerical data may be sampled in any arbitrary fashion. The default file type is .tbl .

Fit-files

  are ``degenerate'' image files with just descriptors and no pixels and used to store the parameters needed for the fitting functions. The default file type is .fit .

Descriptors

  are variables attached to the structures mentioned above (i.e. stored in the same file) and describe the structure of the tables, images and fit files. They can also store any other auxiliary information connected to the data such as histograms, coordinates, comments and so on. For fit files they contain the fitting parameters.
The Midas descriptors are the equivalent of the FITS header keywords.

Catalogs

  contain lists of either images or tables or fit files for the purpose of grouping data together within MIDAS. They are exceptional in the sense that they are implemented as ASCII files so you can list and edit them (with care!) outside MIDAS. The default file type is .cat .

Keywords

  are variables which can be used to pass information from one MIDAS program to the next or to temporarily store intermediate results (there are also reserved or system keywords that keep MIDAS system parameters). They are referred to by a name and can be easily manipulated from the terminal or MIDAS procedures.

  
The individual data points in an image are referred to as ``pixels''  and in a table they are called ``elements'' . The paragraphs below describe the structure of descriptors, and keywords, and the methods for specifying the individual data values in images and tables.

Note

There is no special syntax for file names in MIDAS. You can use any legal name of your host file system for images, tables and fit files. However, a name beginning with a digit or using any of the characters
+, -, *, /, `, !, |, ( and ),    should be avoided, because these symbols will cause problems in e.g. the COMPUTE/IMAGE command. If you do want to use a file name with these special characters in a COMPUTE/IMAGE command you have to enclose the full name (including the file type) in quotes, i.e.
Midas 456> COMPUTE/IMAGE res = 12 + "quasar01+12.bdf"

The length of these names is, in principle, limited to 60 characters for MIDAS applications (which used to be the size of keywords IN_A, OUT_A, employed in most procedures to store the image, table names...). However, via the SET/MIDAS dpath= command you can specify other directories which are scanned for data files besides the current directory.

Also file names like abc.bdf.mine will not be appreciated by all MIDAS applications.
As mentioned before, file names are case sensitive in MIDAS on Unix systems; names for descriptors and keywords are not. Thus, referring to a keyword with name KEYA may be done e.g. via keyA or Keya.