Ongoing Developments
Implementation of an AW node in Brazil
We are discussing with people involved in the development of Astro WISE in order to install the first AW node in South America. The main points of AW were presented before and from that it is easy to see the importance of having an environment capable to doing image processing at different levels in a modern way. Brazil has now not only to its own observing facility (SOAR) but also access to Gemini and CFTH. AW will serve as a national facility and will be openly available. Given the importance of its usage by the whole community we plan on organizing a national committee that can examine the global needs and suggest further development to the scientific committee of INCT-A. The realistic schedule for having AW fully operational at INPE is late 2009.
Development of Decision Tree and fof algorithm
We are investing a considerable amount of time in establishing better and more efficient ways of providing accurate star-galaxy separation in different data sets. Several Decision Tree algorithms (DTA) have been developed in the recent years and most of them exhaustively tested in other data outside astronomy. Here, we are conducting several experiments where using different sets of photometric attributes we reach higher and higher discrimination between stars and galaxies. A key ingredient in this project is to be able to provide high performance at the faint magnitude level and that is why we are testing several algorithms on the deep data available from SDSS called stripe-82. This data will be of paramount importance to test several procedures in the search for clusters of galaxies in the DES (Dark Energy Survey) starting from star-galaxy separation. Preliminary results based on the plain SDSS data show that DTA work better than parametric methods traditionally used in the past. This is an ongoing development although we expect to have a VO service with this within six months.
One of the main problems in using the friend-of-friends (FoF) algorithm is the time it takes to process large numbers of objects, scaling with N2log N. It is necessary to weaken this dependence on the total number of objects and thus be able to treat the hundreds of millions of particles found in current large cosmological simulations. First experiments on reducing the dependence on N have shown that after a domain decomposition (subdividing the data in redshift shells) combined with a post-processing step we have already reduced the scaling to NlogN2 , a considerable improvement. Our parallel version has fully reproduced previous results (Caretta et al. 2008; A&A,487,445) for computing the potential gravitational energy spectrum for galaxies and clusters of galaxies at many redshifts. A VO service will be made available allowing the user to run the FoF algorithm over the most important cosmological simulations available to date and those inputed by the user.
Developing an engineered version of 2DPHOT
2DPHOT was initially developed as a suite of FORTRAN programs that carry out wide-field image analysis. It is been used by several research groups as a standalone package. Now, we are planning to bring 2DPHOT to the AW environment, which requires extra investment in re-designing 2DPHOT as an engineered environment. This is mandatory if we want to process Terabytes of data over a reasonable timescale. This will be tested in processing the entire SDSS collection of images in four bands, which will establish a data processing benchmark. We hope to deliver 2DPHOT as a VO service for processing limited amount of data until the end of 2009.
Processing of a 60 Square Degree SDSS area
The most important photometric data available for extragalactic research is the SDSS, providing images in four bands over ˜9000 square degrees. These data are restricted to theoptical part of the electromagnetic spectrum and it should be matched against data obtained in other wavelengths, like X-ray, Infrared, Ultraviolet, etc. This is what this project is intending to do by selecting non-contiguous areas over the northern sky covered by SDSS and with available deep catalogs in other wavelength regimes. We are already processing the SDSS images and soon will be matching the output catalogs (stars and galaxies) against the available deeper catalogs. This data set will be crucial in panchromatic studies of different classes of objects and also an important resource for data mining exploitation.
Creation of a Dynamical dB for the SPIDER Project
The most important photometric data available for extragalactic research is the one created by SDSS. Starting from the main SDSS database several sub sets can be created. Westarted a project to study the global properties of elliptical galaxies in the nearby universe - SPIDER (Spheroidal’s Panchromatic Investigation in Different Environmental Regimes). The first part of this development was to use 2DPHOT to measure the photometric parameters in eight different passbands (grizYJHK) spanning from optical to near infrared. Also, we analyzed the spectra of these galaxies provided by the SDSS database resulting in measurements of age, metallicity, and alfa-enhancement for the stellar populations dominating these systems. The sample has ~40100 early type systems and the database been created will harbor not only all the photometric and spectroscopic info but also images, spectra and other quantities characterizing the environment where these galaxies reside. This is an example of a small size dB but very specialized and it should be an important laboratory for developing statistical and visualization tools in the framework of BRAVO@INPE. This will be the first database made available by this project to the community.
Conducting the First Brazilian Grid Processing Experiment in Astronomy
We recently carried out a census of the current hardware/software resources situation in Brazilian Astronomical Community participating of the INCT-A initiative in order to scale our future needs. As a result, we are setting up the Brazilian Network for Processing Grid in Astronomy (BNPGA), which will attend, in a limited way, the demands of our community in terms of image processing. This will help us to trace the roadmap of what is really needed from now on. We see this experiment as an essential part of the whole BRAVO project since it touches all the fundamental aspects of the VO idea. Using the expertise of the members of our team who developed the OurGrid middleware, we plan on building up the BNPGA.
We are discussing with people involved in the development of Astro WISE in order to install the first AW node in South America. The main points of AW were presented before and from that it is easy to see the importance of having an environment capable to doing image processing at different levels in a modern way. Brazil has now not only to its own observing facility (SOAR) but also access to Gemini and CFTH. AW will serve as a national facility and will be openly available. Given the importance of its usage by the whole community we plan on organizing a national committee that can examine the global needs and suggest further development to the scientific committee of INCT-A. The realistic schedule for having AW fully operational at INPE is late 2009.
Development of Decision Tree and fof algorithm
We are investing a considerable amount of time in establishing better and more efficient ways of providing accurate star-galaxy separation in different data sets. Several Decision Tree algorithms (DTA) have been developed in the recent years and most of them exhaustively tested in other data outside astronomy. Here, we are conducting several experiments where using different sets of photometric attributes we reach higher and higher discrimination between stars and galaxies. A key ingredient in this project is to be able to provide high performance at the faint magnitude level and that is why we are testing several algorithms on the deep data available from SDSS called stripe-82. This data will be of paramount importance to test several procedures in the search for clusters of galaxies in the DES (Dark Energy Survey) starting from star-galaxy separation. Preliminary results based on the plain SDSS data show that DTA work better than parametric methods traditionally used in the past. This is an ongoing development although we expect to have a VO service with this within six months.
One of the main problems in using the friend-of-friends (FoF) algorithm is the time it takes to process large numbers of objects, scaling with N2log N. It is necessary to weaken this dependence on the total number of objects and thus be able to treat the hundreds of millions of particles found in current large cosmological simulations. First experiments on reducing the dependence on N have shown that after a domain decomposition (subdividing the data in redshift shells) combined with a post-processing step we have already reduced the scaling to NlogN2 , a considerable improvement. Our parallel version has fully reproduced previous results (Caretta et al. 2008; A&A,487,445) for computing the potential gravitational energy spectrum for galaxies and clusters of galaxies at many redshifts. A VO service will be made available allowing the user to run the FoF algorithm over the most important cosmological simulations available to date and those inputed by the user.
Developing an engineered version of 2DPHOT
2DPHOT was initially developed as a suite of FORTRAN programs that carry out wide-field image analysis. It is been used by several research groups as a standalone package. Now, we are planning to bring 2DPHOT to the AW environment, which requires extra investment in re-designing 2DPHOT as an engineered environment. This is mandatory if we want to process Terabytes of data over a reasonable timescale. This will be tested in processing the entire SDSS collection of images in four bands, which will establish a data processing benchmark. We hope to deliver 2DPHOT as a VO service for processing limited amount of data until the end of 2009.
Processing of a 60 Square Degree SDSS area
The most important photometric data available for extragalactic research is the SDSS, providing images in four bands over ˜9000 square degrees. These data are restricted to theoptical part of the electromagnetic spectrum and it should be matched against data obtained in other wavelengths, like X-ray, Infrared, Ultraviolet, etc. This is what this project is intending to do by selecting non-contiguous areas over the northern sky covered by SDSS and with available deep catalogs in other wavelength regimes. We are already processing the SDSS images and soon will be matching the output catalogs (stars and galaxies) against the available deeper catalogs. This data set will be crucial in panchromatic studies of different classes of objects and also an important resource for data mining exploitation.
Creation of a Dynamical dB for the SPIDER Project
The most important photometric data available for extragalactic research is the one created by SDSS. Starting from the main SDSS database several sub sets can be created. Westarted a project to study the global properties of elliptical galaxies in the nearby universe - SPIDER (Spheroidal’s Panchromatic Investigation in Different Environmental Regimes). The first part of this development was to use 2DPHOT to measure the photometric parameters in eight different passbands (grizYJHK) spanning from optical to near infrared. Also, we analyzed the spectra of these galaxies provided by the SDSS database resulting in measurements of age, metallicity, and alfa-enhancement for the stellar populations dominating these systems. The sample has ~40100 early type systems and the database been created will harbor not only all the photometric and spectroscopic info but also images, spectra and other quantities characterizing the environment where these galaxies reside. This is an example of a small size dB but very specialized and it should be an important laboratory for developing statistical and visualization tools in the framework of BRAVO@INPE. This will be the first database made available by this project to the community.
Conducting the First Brazilian Grid Processing Experiment in Astronomy
We recently carried out a census of the current hardware/software resources situation in Brazilian Astronomical Community participating of the INCT-A initiative in order to scale our future needs. As a result, we are setting up the Brazilian Network for Processing Grid in Astronomy (BNPGA), which will attend, in a limited way, the demands of our community in terms of image processing. This will help us to trace the roadmap of what is really needed from now on. We see this experiment as an essential part of the whole BRAVO project since it touches all the fundamental aspects of the VO idea. Using the expertise of the members of our team who developed the OurGrid middleware, we plan on building up the BNPGA.
