Researchers at four Ontario universities will be able to spend more time fact-finding and less time writing computer programs thanks to an agreement that will give members of the High Performance Computing Virtual Library (HPCVL) access to the library
of the Numerical Algorithms Group (NAG).
The HPCVL is a consortium of four Eastern Ontario universities, including Carleton University, Queen’s University, the Royal Military College of Canada and the University of Ottawa. It is dedicated to providing researchers with access to high-performance computing resources.
The partnership with Downers Grove, Il.-based NAG is designed to benefit HPCVL member researchers doing work in fields like bioinformatics, computer-aided molecular design, econometrics, computational fluid dynamics, drug discovery, photonics and other disciplines. It will give access to its math, statistical and data mining libraries, compilers and tools with more than 2,500 component routines.
“”It’s designed to provide our researchers with the tools and routines that are already optimized for use on their systems,”” says HPCVL executive director Ken Edgecombe. “”It will help to optimize a lot of the math routines that people need to do effective work.””
The resources available to HPCVL members will include the Fortran Library, C-Library, f95 Library, SMP and Parallel Libraries for parallel execution.
Edgecombe says the agreement with NAG is a unique one, and will save its member researchers a lot of money by giving them access to NAG’s routines both on their own desktops and HPCVL’s mainframes without having to buy the software themselves.
“”It will allow the researchers to simply call up those routines and insert them into their code, as opposed to having to write the routines, some of which are very difficult,”” says Edgecombe. “”Essentially, it’s a way not only to speed up their calculations and therefore their research, but it’s also a way to save time putting their own software packages together.””
NAG president Rob Meyer says most researchers aren’t computer scientists, and having access to NAG’s library of routines will help them focus on their own disciplines.
“”We bring a set of software resources to enable these researchers to use this software on a project by project basis as they need it, rather than everybody having to go out to try and assemble the hardware and software resources on a piecemail basis,”” says Meyer. “”Our software is the tools that allow them to build better models, faster models, to do their research work.””
One of the researchers hoping to benefit from NAG’s library is Doug Mewhort, a psychology professor at Queen’s University doing research into human memory. Mewhort uses supercomputers to build a model of how he thinks human memory works, testing things like how we remember things we just learned, or how we recognize an object. He says his computational model matches human data very well.
“”With the models we’re able to describe the control structure by which you can retrieve things, we’re able to describe the mechanism of retrival that you use,”” says Mewhort. “”NAG is the premier supplier of high quality algorithisms for doing standard things, and they will allow us to speed up quite a bit.””
By using the standardized routines in the NAG library, Mewhort says it will also be easier for him to share his research and his models with his colleagues at other institutions.
“”I think we also gain portability,”” says Mewhort. “”I can give someone else my code with the assurance that if they also run a NAG library on another kind of machine it will work the same way.””