Assistant Professor of Physics Natalia Connolly has been awarded computing time on the TeraGrid, an open scientific super-computing infrastructure funded by the National Science Foundation. This award is for her work with ultra-high luminous infra-red galaxies (ULIRGs) and will allow her to continue her research begun last summer with Will Eagan '11 and University of Pennsylvania Postdoctoral Researcher Brian Connolly.
The three researchers’ goal is to determine whether distant (or high-redshift) ULIRGs go through the same evolutionary steps as low-redshift ones. In order to understand the evolution of ULIRGs, they developed a novel statistical technique that allows them to perform pair-wise spectral comparisons of ULIRGs. While highly promising, the method is quite computationally intensive. The use of the TeraGrid, which currently includes more than a petaflop of computing capability and more than 30 petabytes of online and archival data storage, will faciliate their work.
A petaflop (super) computer is be capable of performing 1 quadrillion computations per second. A modern desktop is typically capable of performing in the range of a few billion operations.
A single spectral comparison takes about half an hour on a modern desktop. Connolly and her fellow researchers have a sample of about 150 spectra, including both low- and high- redshift ULIRGs, requiring approximately 22,000 comparisons or about 11,000 hours, or 1.3 years! If they were to use a single desktop it would take them well over a year to complete the calculation.
Connolly and her partners are trying to figure out how ULIRGs evolve. ULIRGs are among the most luminous objects known to astronomers; but it's not at all clear how they live out their lives or how much time they spend at each phase in their lives. In their earlier work, the three researchers proposed what they hope is a fundamental evolutionary plane for low-redshift ULIRGs; now they are extending their reach to high-redshift ones as well.
Clarifying the evolutionary scenario for ULIRGs would be a great step toward understanding galaxy evolution. It would also generate a plethora of future studies, from observing specific objects whose properties are predicted by our evolutionary scenario to understanding objects that are thought to be related to ULIRGs but may not be.
The three researchers’ goal is to determine whether distant (or high-redshift) ULIRGs go through the same evolutionary steps as low-redshift ones. In order to understand the evolution of ULIRGs, they developed a novel statistical technique that allows them to perform pair-wise spectral comparisons of ULIRGs. While highly promising, the method is quite computationally intensive. The use of the TeraGrid, which currently includes more than a petaflop of computing capability and more than 30 petabytes of online and archival data storage, will faciliate their work.
A petaflop (super) computer is be capable of performing 1 quadrillion computations per second. A modern desktop is typically capable of performing in the range of a few billion operations.
A single spectral comparison takes about half an hour on a modern desktop. Connolly and her fellow researchers have a sample of about 150 spectra, including both low- and high- redshift ULIRGs, requiring approximately 22,000 comparisons or about 11,000 hours, or 1.3 years! If they were to use a single desktop it would take them well over a year to complete the calculation.
Connolly and her partners are trying to figure out how ULIRGs evolve. ULIRGs are among the most luminous objects known to astronomers; but it's not at all clear how they live out their lives or how much time they spend at each phase in their lives. In their earlier work, the three researchers proposed what they hope is a fundamental evolutionary plane for low-redshift ULIRGs; now they are extending their reach to high-redshift ones as well.
Clarifying the evolutionary scenario for ULIRGs would be a great step toward understanding galaxy evolution. It would also generate a plethora of future studies, from observing specific objects whose properties are predicted by our evolutionary scenario to understanding objects that are thought to be related to ULIRGs but may not be.
