A&A 404, 1159-1163 (2003)
DOI: 10.1051/0004-6361:20030533

CVcat: An interactive database on cataclysmic variables

J. Kube^1, - B. T. Gänsicke^1,2 - F. Euchner¹ - B. Hoffmann¹

1 - Universitäts-Sternwarte Göttingen, Geismar Landstraße 11, 37083 Göttingen, Germany
2 - Department of Physics and Astronomy, University of Southampton, Hampshire, Southampton SO17 1BJ, UK

Received 29 November 2002 / Accepted 12 March 2003

Abstract
CVcat is a database that contains published data on cataclysmic variables and related objects. Unlike in the existing online sources, the users are allowed to add data to the catalogue. The concept of an "open catalogue'' approach is reviewed together with the experience from one year of public usage of CVcat. New concepts to be included in the upcoming AstroCat framework and the next CVcat implementation are presented. CVcat can be found at http://www.cvcat.org.

Key words: astronomical data bases: miscellaneous - catalogues - stars: novae, cataclysmic variables

1 Introduction

CVcat is an interactive database or "online catalogue'' that offers a number of features so far unknown to scientific catalogues. It was developed as a tool for the research community working on cataclysmic variables (CVs), a class of close interacting binaries, and as a case study for some of the concepts to be used in the development of a general catalogue software, AstroCat. CVcat can be accessed online at http://www.cvcat.org. It was first presented to the public in August 2001 during a CV conference held in Göttingen (Kube et al. 2002). Since then, the number of users of CVcat has increased to more than one hundred, the daily average of requests is around fifty (total number of delivered pages).

2 The concept

CVcat has been developed in order to overcome major conceptual shortcomings of existing CV catalogues: Ritter & Kolb (1998) include only systems with known orbital period, which limits their catalogue to $\approx$1/3 of all CVs and related objects. Downes et al. (1997) do list all known CVs, but their catalogue provides only very limited information on each individual system, i.e., the only binary parameter included is the orbital period. Our aim was to develop an online data base that combines the information of the existing catalogues and allows the users to actively contribute to the content of the data base, implementing a first version of an "open catalogue''.

CVcat differs from other CV catalogues and other astronomical databases in the concept of the data input. So far, the majority of astronomical catalogues have been compiled by relatively small editorial teams consisting of scientists knowledgeable in the fields covered by the catalogues (e.g. McCook & Sion 1999; Downes et al. 1997; Liu et al. 2001; Ritter & Kolb 1998). These catalogues typically contain more or less detailed information on a specific class of astronomical objects. Updates are published, if at all, only on a very irregular basis. The catalogues contain just one value for each listed property (e.g. distance, orbital period) of a given object. Whilethis is helpful for non-specialist users to obtain a quick overview ofthe properties of an individual object, or of the statisticalproperties of a given group of objects, the more expert user willcertainly benefit if different and possibly competing values for agiven parameter are referenced in such catalogues. This is particularly useful if the information has been obtained by different methods.

Some of the aforementioned catalogues moved from "classical'' printed publication to online web-based publication, which allows shorter update cycles (e.g. the "living edition'' of the Downes et al. 1997 CV catalogue, Downes et al. 2001), however, the overall concepts remained unchanged. In addition to these specialized catalogues for a specific object class there exists huge data bases like SIMBAD (Wenger et al. 2000), which provide very basic properties for an enormously large number of objects. However, due to the very global coverage of astronomical objects, data contained in SIMBAD are prone to be incomplete and/or inaccurate.

The concept of an "open catalogue'' implemented in CVcat permits every registered user to add data to the catalogue, which is instantly visible to all other users. The quality control is performed by an editorial team (Sect. 2.1), which may alter or remove erroneous data. For every property of an object, an arbitrary number of values can be stored (e.g. several published values for the distance). CVcat returns one of these values as the "bestavailable'' value, selected as such by the editors. However, as sucha selection process often involves some subtle subjective view of theeditor, the more expert user may decide to inspect the original sources forthe competing values, and, thereafter, decide based on his/herexperience which value is best suited for a given purpose.

	Figure 1: Classical, current, and future concepts for adding and editing catalogue data in comparison.
Open with DEXTER

A sketch of the different concepts of data input and validation is given in Fig. 1 (a) for classical catalogues and (b) for the CVcat concept. (c) introduces the concept that will be implemented in the next release of CVcat, which is adressed in Sect. 4. In the future version, newly-added data will be instantly visible as it is now (Fig. 1, (b)), but will be tagged as "unapproved'' until an editor cross-checks the data. In the current implementation (b), the user cannot see if a database entry has been approved by an editor.

2.1 Distributed editorial team

The editorial team of CVcat is recruited from the CV research community, and consists (ideally) of one expert per CV subclass. Since these editors are typically familiar with the publications on their "favourite'' objects anyway, the amount of work to cross-check newly entered data is lowered. We estimate that for a typical subclass of CVs, say, polars, the time to be spent on editorial duties in CVcat is of the order of two hours per week or less. Each editor has the privilege to remove or change erroneous data for that person's object class only. For data on all other object classes, the editor is a non-privileged user who may add new data and browse the catalogue. Note that every user may add data, in contrast to, e.g., SIMBAD, where only the editorial team can directly modify the catalogue content.

	Figure 2: Search form in CVcat: A large fraction of the search form focuses on the detailed selection of the object class.
Open with DEXTER

2.2 Database content

In contrast to the existing CV catalogues, CVcat is designed to contain a great variety of information for each object. Currently, the following object properties can be included, with the number in brackets being the number of entries on September 17, 2002 (note that because CVcat allows multiple entries for a given property of an object these numbers do not represent the number of unique objects for which a particular property is known):

High state magnitude (922), low state magnitude (794), optical spectrum exists  (665), orbital period (509), distance (221), general magnitude (209), primary mass (156), inclination (153), secondary mass (142), superhump period (128), secondary spectral type (121), mass ratio (93), Doppler tomogram exists (80), primary temperature (76), optical light curve exists (61), hydrogen column density (58), orbital ephemeris (55), eclipsing (48), spin period (36), primary radius (33), secondary radius (29), uv data exists (22), general magnetic field strength of primary star (18), field strength of primary magnetic pole (17), X-ray data exists (14), eclipse map exists (11), secondary temperature (8), colatitude of primary magnetic pole (5), field strength of secondary magnetic pole (3), spin ephemeris (3), azimuth of primary magnetic pole (2), magnitude in eclipse (2), azimuth of secondary magnetic pole (1).

Every value stored in CVcat is linked to its original publication, either in the NASA ADS using the ADS bibcode of the paper (Kurtz et al. 2000), to the astro-ph/arXiv e-print archive, or to the VSNET messages (Nogami et al. 1997). Besides references to the publications from which the data entries contained in CVcat are taken, a list of articles with general information on a given object can be stored in CVcat.

We allow inclusion of data from astro-ph, which is not necessarily identical to the datapublished in the final refereed version (or which may in some casesnever make it through the refereeing stage). Most of the astro-ph data, however, is promoted to refereed information at some point. It is a typical task of the editors to track such updates and to conduct the appropriate changes to the database, i.e. changing the source from an astro-ph to an ADS bibcode.

2.3 Searching the database

Data retrieval from CVcat works in two ways: (i) the user obtains all available data for a specific object, which can be found using its object type, its coordinates, or one of its names, (ii) the user creates a table containing selected properties for a list of objects. The latter method allows the user to create data tables suitable for easy graphics generation as well as ready-to-publish L^ATEX-tables.

	Figure 3: Search result in CVcat: Some of the data stored for SS Cygni.
Open with DEXTER

Searching in CVcat is organized as a two-step process. In the first step, the user can enter the search pattern, which can be the name or a substring of the name, a set of object classes, and the coordinate range of the object (Fig. 2). The object class is selected using a grid of logical operators ("and'', "or'', and "not''). It is possible, e.g., to look for dwarf novae, which have also been observed as novae: choose "and DN'' and "and Nova''. Another example is to look for nova-likes which do not show the SW Sex phenomenon: the corresponding selection would be "or NL'' and "not SW''.

After submitting this search request a list with all objects matching the search criteria is returned (not shown). In this list, all object names are hyperlinks to the page showing all results for the specific object (Fig. 3). Alternatively, a set of objects can be chosen from this list to generate a user-configurable list with certain properties of the objects. This list may be adjusted in a way that objects without a published value for a certain property are not included. By iterative calls of the list generator it is possible to distill a table containing e.g. all objects with known masses and periods.

An example of the results of such a list creation process is given in the following section.

2.4 Example: Orbital periods and donor masses

Using the data contained in the CVcat database, we have plotted the secondary star masses as a function of the orbital period, Fig. 4, for all CVs. A linear trend with

$$\frac{M_2}{M_{\odot}}\approx 0.11 P/{\rm h}-0.06$$ (1)

is clearly visible. This is predicted by theoretical considerations, where the reasoning is roughly this (Frank et al. 1992): using an approximation for the Roche geometry, valid for $1.3\lesssim q\lesssim 10$, and Kepler's law, one finds that the mean density of a Roche lobe filling star is a function of only the orbital period. With the knowledge of the lower main-sequence M/Rrelation (Baraffe et al. 1998; Kippenhahn & Weigert 1990), one then finds

$$\frac{M_2}{M_{\odot}}\approx 0.11 P/{\rm h}.$$ (2)

Figure 4: Secondary mass plotted against orbital period (points) and period histogram (gray bars) for all CVs with known orbital period and secondary mass. The full line shows the relation from Eq. (1), the short dashed line from Eq. (3), the long dashed line from Eq. (4). Full boxes and error bars are used for systems with estimated errors in M2, open boxes for secondary star masses with an unknown error range. The given (M2,P) data pairs represent $\approx$10% of all known CVs, the histogram contains $\approx$$39\%$ of all CVs known to CVcat.

Open with DEXTER

A more detailed analysis of secondary star masses leads to slightly different period-mass-relations (Smith & Dhillon 1998):

  

$$\frac{M_2}{M_{\odot}} (0.038\pm0.003)(P/{\rm h})^{(1.58\pm0.09)}\quad\mbox{or}$$ = (3)

$$\frac{M_2}{M_{\odot}} (0.126\pm0.011)P/{\rm h}-(0.11\pm0.04).$$ = (4)

The data currently available in CVcat are consistent with the results from Smith & Dhillon (Fig. 4). Note that some of the published secondary star masses may be derived from the orbital period using some theoretical models, hence artificially stabilizing the fit close to the theoretical predictions.

3 Usage statistics

The log file analysis of CVcat usage over a one year period shows that CVcat has a stable user community which is still slowly increasing. The typical usage of CVcat is to request all available data on a specific object, which is normally queried by its name. List generation of objects selected by their class ranks second in the usage of the database. The current growth rate of the catalogue is around 50 entries per month, mostly from the CVcat editors.

The data flow into CVcat originating from outside the CVcat core team is not yet satisfying. It is unclear why most users refrain from adding data from their own publications. Users should bear in mind that the probability of having their papers cited increases if the information from their publications can be found in the database.

   
4 Prospects

The concept of CVcat has demonstrated the benefits of a public scientific catalogue with a globally-shared expertise of its users and editors. The general structure of CVcat is also applicable to other fields of astronomy. Hence, a more general software based on the experiences with CVcat which will allow the implemenation of catalogues for arbitrary object classes, called "AstroCat'', is currently being developed. This software will also include additional features to improve the knowledge management of astronomical results and data:

• Besides the "single number data'' stored so far, the infrastructure for storing more complex data products, e.g., light curves, spectra, and finding charts will be added. This feature should help to overcome a major shortcoming in the present method of scientific publishing: most authors publish their reduced data only in the form of plots, making follow-up work on these results rather unattractive. While these data may be obtained directly from the authors, experience shows that in many cases the data have been lost due to faulty hardware or storage media, with the probability of loss increasing dramatically with time since the publication of the original work. In addition, the authors may have left astronomy, preventing access to their data. If the data is available in CVcat, the re-examination of observations would be much easier and independent of contact with the original author and the corresponding data archive. The storage of reduced and published data in a usable form at e.g. CDS, Strasbourg, is already promoted by journals like A&A.
• Information agents will allow users to be informed automatically if new data is entered into the database for their objects of interest. This is a service that keeps the user informed about new publications on a given object or class of objects without having to log in to CVcat frequently.
• An elaborated validation system will allow the users to add their comments to published data. This unique feature will add personal communication aspects to the database. Public discussion on details of the published data may arise from this.

• To allow exact quoting of a specific state of the ever-changing catalogue, a "versioning" method using a global identification number (GID) will be used: theGID is incremented with each change of the content of AstroCat (e.g. adding new data, marking data as correct, marking data as the best available etc.). Every output resulting from the use of AstroCat will includethe current GID number. An older state of the database can be exactlyreproduced by issuing a given GID (lower than the current one), andusing AstroCat with such a specific GID will return alwaysprecisely the same results, independent of the actual stateof the data base. Any statistical analysis based on data from CVcatshould therefore include the GID of the CVcat state at the time of analysis, permitting a quantitative comparison with future studies. It is not necessary to rely on a fixed or frozen state so far only available in printed catalogues.

• Data that have not yet been cross-checked by the editorial team will be tagged as such. This method combines both the unique speed of the CVcat concept and the expertise of the editorial team. As long as a new datum is visible but marked as unapproved, the user of the database can use the new entry under his own responsibility, while an approved value is authoritative.
• Hyperlinks to a large number of available web resources will be included for every object. This includes, e.g., links to other CV cataloges like SIMBAD, the ADS, arXiv.org, the different variable star observers archives, and many more.

The development of the new CVcat and the AstroCat software is done in close cooperation with the users of the current implementation. In summer 2003, the data from the current CVcat will be transferred to the new database. By the end of 2003, the AstroCat/CVcat framework will be completed. Ideas from the editors and users of other interactive catalogues, e.g. the high-z database (de Mello et al. 2002), are highly appreciated. The AstroCat project is hosted at http://astrocat.uni-goettingen.de.

5 Technical realization

CVcat is implemented as a Perl script which runs on a Linux PC. This Perl script processes the user requests (HTTP GET/POST requests) and creates HTML pages that are delivered via the Apache web server. Hence, CVcat presents itself as an interactive web page.

The data is stored in a MySQL data base to which the Perl script communicates using the standard Perl DBI/DBD interface.

For the implementation of AstroCat, an XML layer will be included between the data base and the HTML layer. This XML layer can be used to automatically include larger data sets into the database. Another possible application will be to use the AstroCat framework as an archive for reduced data from robotic telescopes that use RTML (Hessman 2001) for their observation requests, accomplished observations, and some of the metadata. A technical advantage of the usage of XML is the easy availability of many very good tools specialized in the processing of XML data and the transformation of XML documents into HTML pages. This will improve the quality and the speed of the implementation. PHP might be used in addition to Perl as the scripting language for the AstroCat programs.

6 Summary

With CVcat we have implemented an online catalogue for cataclysmic variable stars, which - for the first time - allows its users to add data instantly visible to all other users. The quality control is realized by a team of experts who share the responsibility for the different object classes. From the experiences of one year of public usage of the catalogue we have compiled a number of new concepts which will be implemented in the next generation of CVcat in the context of a more general framework for astronomical databases, AstroCat.

Acknowledgements

We would like to thank all the colleagues who volunteered to take the responsibility for one of the object classes included in CVcat: Domitilla de Martino, Don Hoard, Steve Howell, Tom Marsh, Klaus Reinsch, Matthias Schreiber, and John Thorstensen. The AstroCat/CVcat project is funded by the Deutsche Forschungsgemeinschaft (DFG) project number LIS 4 - 554 95 (1) Göttingen. The development of the feasibility study has partially been supported by the DLR under grant 50 OR99 03 6. BTG also acknowledges support from a PPARC Advanced Fellowship.

We appreciate the helpful comments from the referee, Dr. Ochsenbein, to both this paper and the CVcat database itself.

References

 
• Baraffe, I., Chabrier, G., Allard, F., & Hauschildt, P. H. 1998, A&A, 337, 403 In the text NASA ADS  
• de Mello, D., Johansson, E. P. G., & Markström, P. 2002, Ap&SS, 281, 535 In the text NASA ADS  
• Downes, R., Webbink, R. F., & Shara, M. M. 1997, PASP, 109, 345 In the text NASA ADS  
• Downes, R. A., Webbink, R. F., Shara, M. M., et al. 2001, PASP, 113, 764 NASA ADS  
• Frank, J., King, A. R., & Raine, D. J. 1992, Accretion power in astrophysics, 2nd edn., Cambridge astrophysics series (Cambridge: Cambridge University Press) In the text  
• Hessman, F. V. 2001, Astronomische Nachrichten, 322, 303 In the text NASA ADS  
• Kippenhahn, R., & Weigert, A. 1990, Stellar Structure and Evolution, A&A, Library (Berlin: Springer-Verlag), corrected 3rd printing 1994 In the text  
• Kube, J., Gänsicke, B. T., & Hoffmann, B. 2002, in The Physics of Cataclysmic Variables and Related Objects, ASP Conf. Ser., 261, 678 In the text  
• Kurtz, M. J., Eichhorn, G., Accomazzi, A., et al. 2000, A&AS, 143, 41 In the text NASA ADS  
• Liu, Q. Z., van Paradijs, J., & van den Heuvel, E. P. J. 2001, A&A, 368, 1021 In the text NASA ADS  
• McCook, G. P., & Sion, E. M. 1999, ApJS, 121, 1 In the text NASA ADS  
• Nogami, D., Kato, T., Baba, H., & Kunjaya, C. 1997, Electronic Publishing, Now and the Future, 23rd meeting of the IAU, Joint Discussion 12, 25 August 1997, Kyoto, Japan, 12 In the text  
• Ritter, H., & Kolb, U. 1998, A&AS, 129, 83 In the text NASA ADS  
• Smith, D. A., & Dhillon, V. S. 1998, MNRAS, 301, 767 In the text NASA ADS  
• Wenger, M., Ochsenbein, F., Egret, D., et al. 2000, A&AS, 143, 9 In the text NASA ADS