About
New astronomical surveys such as Pan-STARRS and LSST are under development and will collect petabytes of data. These surveys will repeatedly capture large areas of sky to great depth, and will detect vast numbers of moving, variably bright, and transient objects, such as supernovae, transiting in planetary systems, and lensing of starlight by moving objects.
The data products of these surveys are series of observations taken over time.
Such time series observations are common to many disciplines: prices of shares, exchange rates, price indices, time development of biological populations, seismic observations at times of earthquakes and aftershocks, and weather observations.
Across all of these disciplines, a common set of problems in the analysis of time series concerns issues of feature extraction, similarity, and the search for anomalies. Bringing together scientists from a diverse set of fields, all interested in the temporal changes in their data, fosters vigorous collaborations that create interesting algorithms and experiment with new ideas.
Our methodology is to use the massive datasets from the field of astronomy to bootstrap the center. We address issues in database technology, data integration, statistical analysis and data mining. Currently, multidisciplinary teams of astronomers, computer scientists, and statisticians are working on such problems and gaining know-how and expertise that advance the state of art in multiple subject domains.
In the long term, we envisage the world's largest center for time series where issues of analysis, distribution and storage can be addressed by teams of experts.
Goals
The Time Series Center has the following immediate goals:
- We are making a home for astronomy databases such as TAOS, MACHO, Pan-STARRS, and ESSENCE that is:
- widely accessible,
- includes innovative access features that support scientific inquiries,
- and scalable to even larger datasets such as LSST, or beyond.
- Developing new database and data management tools that serve as enabling technologies for such an effort.
- Developing techniques for mining and analyzing times series to help in categorization and analysis of the data.
These goals are being realized by building tools for automatic classification, categorization, integration and disbursement of astronomical light curves. Many challenges need to be overcome. In particular, size impacts database design and architecture; standardization impacts outside access to data, both incoming and outgoing; online queries and subscriptions impact the development of distributed, real-time analysis pipelines; the design of algorithms impacts database architecture; and novel algorithms for cataloging, filtering and discovering patterns impacts classification of data and the real-time search for interesting events in the data.
Last Updated at: 2007-10-05 09:58 AM.