Purdue researchers using SubmitR on DiaGrid to get results
January 17, 2013
Agricultural economics Professor Michael Delgado uses computationally demanding statistical models to study questions like the effectiveness of voluntary pollution abatement programs and the impact of education in economies of developing nations.
“All of my statistical research is done using R,” says Delgado, referring to the open source R software environment widely used by statisticians and economists, “so having a lot of computational power behind R is crucial for me.”
Lately, Delgado has been getting the kind of computing power his research requires using SubmitR, a Web-based R tool available to Purdue faculty and their students through an accessible graphical interface and Purdue’s DiaGrid hub.
The SubmitR tool on DiaGrid can make thousands of processors available at once for R computing jobs, a particular advantage for tasks like parameter sweeps, which can take advantage of a large processor pool.
ITaP will hold an event to demonstrate SubmitR from 2-3 p.m. Wednesday, Feb. 20, in Stewart Center, Room 311. Registration isn’t required but those who plan to attend are asked to register so adequate refreshments can be ordered. To register, go to firstname.lastname@example.org.
The new tool is on Purdue’s DiaGrid hub web portal at http://diagrid.hubzero.org. DiaGrid and SubmitR are available at no cost to Purdue users.
In addition to statisticians and economists, engineers, scientists and other researchers employ R to explore and mine data sets statistically and to create publication-quality graphics from results.
Delgado says the processing power and long run times available with SubmitR and DiaGrid have been a good fit for several large Monte Carlo simulations on which he and colleagues have been working.
“SubmitR is particularly suited for running large simulations that require thousands of independent computations, at least for my purposes,” Delgado says.
The DiaGrid hub R tool is being developed by ITaP research computing staff members Rob Campbell and Kevin Chen, who made an initial version available fall semester 2012.
In addition to jobs like parameter sweeps that can take advantage of numerous processors, R on DiaGrid can be good for large, long-running jobs, like Delgado’s Monte Carlo simulations, because the DiaGrid pool features an extensive time limit, Chen says. Meanwhile, even advanced R users may benefit from SubmitR’s ease of use.
The DiaGrid hub is built on Purdue’s HUBzero platform, http://hubzero.org, which is designed to make high-level computational research software and access to high-performance and cloud computing about as easy as using a Web browser.