Up: The SIMBAD astronomical database

4 SIMBAD structure and query management

SIMBAD query mechanism can be summarized by the following key features:

Database queries can be made mainly through:
- identifiers (names of astronomical objects) and lists of identifiers;
- sets of coordinates (retrieving one object by its position on the sky, or extracting all objects lying in a given direction), and;
- sampling criteria (or filters);
Data output is driven by formats. The user may define his/her own formats or modify existing ones. Output files can be saved and mailed to the user;
The user interface is adaptable to user preferences.

The database management system of SIMBAD has been developed by the CDS, using the concepts of object-oriented programming.

4.1 Object-oriented concepts

The command language is using the concepts of objects (or agents). Typical object classes are: astronomical object, object list, database, session, reference list, filter, format. Examples of methods are: display, describe, bye (quit).

This structure is only visible for the user of the command line interface. The WWW interface is rendered quite independent of the database structure.

4.2 Indexing

SIMBAD is organized for optimized access by identifier (through an index table of object names) and by position, through an index of small regions.

Identifiers:: A B-tree file contains all identifiers allowing a fast access to any of them. For each identifier, a record contains a pointer to the astronomical object itself in the main database.
Position:: Indexing by coordinates is done in two steps: the coordinates are mapped into a set of boxes. SIMBAD uses the spherical-cubic projection - a technique also used, e.g., for the Cosmic Background Explorer (COBE) data: the celestial sphere is projected onto the six faces of a cube, giving six boxes at the first level. By dividing each face into four parts, one obtains a partition at level two. Further levels are obtained by further divisions of each box into four sub-boxes. Through this mechanism one obtains 6144 boxes at the level 5 with an average size of 6 square degrees and an average number of objects of 500. Box #6145 contains all objects without recorded position.
In order to optimize access to objects in a coordinate box, all objects belonging to a box should be physically grouped in a common place in the database. This is done through a clustering mechanism placing objects from the same box in data blocks linked together in the database files.

When a set of criteria includes some limitations in coordinates, this generates the definition of a list of boxes including the requested area: all entries from these boxes are read and checked against the whole set of criteria.

When a set of criteria includes no limits in sky position, the complete database must be scanned - a long and somewhat expensive operation, which takes typically 15 minutes in the current hardware configuration.

4.3 Query by identifier

In principle any name found in the literature - provided it is given as a syntactically correct character string - can be submitted to the database in order to retrieve information known for this object.

The general syntax of an identifier is the abbreviated catalogue name (or acronym: generally one to four characters), followed by a number or a name (character string) within the catalogue.

Object names such as Vega and Altair, but also Barnard's star, Crab Nebula, Sgr A, HDFN, or HDFS are stored in the database in a specific catalog called " NAME'', while star names in constellations, such as $\alpha$ Lyrae, are stored in the catalogue "*'', and variable stars (such as RR Lyrae) in the catalogue " VAR'' (also called "V*'').

The user can generally type Vega, Altair, alf Lyrae (or alf Lyr): the sesame name resolving module (Sect. 8.2) is used for guessing the catalogue and making the internal conversion. There are however some difficult cases in which the NAME keyword remains necessary, such as in NAME SGR 1900+14 where SGR stands for Soft Gamma Repeater.

In addition the following hints can help the user understand the best way to submit an identifier to SIMBAD:

Case sensitivity:: SIMBAD is not case-sensitive at this level: ALF AQL or alf Aql are, for instance, both valid. There are some exceptions to the rule, such as the cases of the star cluster RMC 136a, or the star in a multiple system VdBH 25a A, for which case-sensitivity may be necessary for solving format ambiguities.
Greek letters:: should be abbreviated as three letters: alf, bet, for $\alpha$ and $\beta$ , but also mu. nu. and pi. (with a dot), for $\mu$ , $\nu$ and $\pi$ .
Constellations:: constellation names should be abbreviated with the usual three letters: alf Boo, del Sct, FG Sge, NOVA Her 1991. The full list is available on-line.
Multiple systems:: Identifiers of a multiple system may generate a list of the objects of the system. For instance, ADS 5423 calls for the four components, A to D, of the stellar system around Sirius. This is true only for some specific identifiers.
Stellar clusters:: Clusters which have no NGC or IC number are named under the generic appellation Cl followed by the cluster name and number: e.g., Cl Blanco 1 is the 1st stellar cluster named by Blanco. Stars in clusters may belong to a "main'' designation list, or to subsequent lists. NGC 5272 692 is star 692 in the list by Von Zeipel, considered as the main list. Subsequent lists have designations starting with Cl*. Examples: Cl* NGC 5272 AC 968 (list by Auriere & Cordoni); Cl* Melotte 25 VA 13 (13th star in the list by Van Altena for Melotte 25 - the Hyades cluster); Cl* Collinder 110 DI 1101 (list by Dawson & Ianna - there is no "main'' list for this cluster). More details are available in the on-line description.
Unknown name ?:: If the object name seems unknown to SIMBAD, the user is advised to enter the coordinates of the object: the object may actually exist in the database under a different designation. Submitting the identifier, or the name of the first author of the catalogue, to the Dictionary of Nomenclature may also give useful clues.

$\begin{figure}\includegraphics[width=9cm]{m81query.ps}\end{figure}$

Figure 3: Example of Simbad response page for a query concerning M 81 (only the first part of the response is visible here)

Figure 3 illustrates the response received from the database after submitting the identifier "M 81''. In the identifier list, the meaning of acronyms, such as [VDD93], is explained through a link to the on-line Dictionary of Nomenclature.

The user interface provides an option for querying around objects, with a radius set by default at 10 $^\prime$ . This is equivalent to a query by position using the object coordinates.

It is also possible to generate the list of 10 or 25 next objects following a given identifier, or to submit a list of object names, stored in a file with one identifier per line.

4.4 Query by coordinates

Query by coordinates can be used to retrieve all objects in a circular field defined by the coordinates of the center and a radius.

The coordinates can be replaced by the name of an object lying at the center of the field, in which case the coordinates are found through an internal query to SIMBAD. The radius can have any size (default value is 10 $^\prime$ ). Queries with a radius smaller than $1-2\hbox{$^\circ$ }$ are answered quite instantaneously.

4.5 Sampling

The sampling mode (also named filter) allows users to define criteria for selecting objects in SIMBAD.

The user may extract objects which satisfy one set of coordinate criteria, several physical criteria (using a simple syntax), objects which have specified identifiers or measurements, and, finally, objects having citations within a range of years.

The WWW interface provides an interactive form which presents all possible sampling options.

The resulting list may be ordered according to sort criteria and, furthermore, it is possible, through the command line mode, to define precisely the output format.

Note that reading the whole database for extracting a sample spread on the whole celestial sphere is possible, but quite time-consuming (as mentioned above). The user is thus encouraged to test the filter on a limited region of the sky, before applying it to the whole database.

4.6 Charts and sky maps

After a sampling by position the user can ask for the corresponding sky plot. This feature is only available through the WWW interface and is generally optimized for a radius range of 10-60 minutes.

The maps display the objects with different symbols according to object type; symbol size for stars varies with object magnitude (see Fig. 4). The maps are clickable and return the object in Simbad corresponding to cursor position.

$\begin{figure}\resizebox{\hsize}{!}{\includegraphics{m81chart.ps}}\end{figure}$

Figure 4: Example of finding chart around M 81 (radius of the circular field: $10\hbox {$^\prime $ }$ )

The WWW interface provides also direct access to the ALADIN interactive digitized atlas (Bonnarel et al. [2000]) as illustrated in Fig. 5.

$\begin{figure}\resizebox{\hsize}{!}{\rotatebox{+90}{\includegraphics{m81aladin.ps}}}\end{figure}$

Figure 5: Use of ALADIN for displaying SIMBAD entries (red diamonds) on top of a DSS-I/STScI image around M 81 (width of the field: $14.1\hbox {$^\prime $ }$ )