With the new Cluster the research is 10 times faster

Share
condividi

Ca’ Foscari researchers have a new working companion. They call it Cluster (System for the scientific calculations at Ca’ Foscari). It is so fast at calculations that it saves hours, days, even weeks of the scholar’s time with mathematic models.

The Cluster works at the Scientific Campus. It is a mix of powerful computers, a hybrid between a common PC and a supercomputer. It is a precious infrastructure for computerized research in all fields, whether Chemistry or Statistics, Linguistics or Finance.

Technically speaking, the system is formed of 7 ‘calculation knots’ for a total of 240 cores and 2.25 TB of RAM (Random Access Memory). It was created thanks to fundings of the University, of the Department of Environmental Sciences, Informatics and Statistics and the Department of Economics, and with the collaboration from scholars from the Department of Molecular Sciences and Nanosystems. Recently the European Centre for Living Technology has also contributed to funding the system.

After a first trial phase in 2015,  the Cluster entered into a full regime: today it is used by around eighty scholars, doctors, researchers and professors. It has contributed to the production of numerous scientific publications in various fields as well as for students’ theses (click here for the list of publications).

Ca’ Foscari researchers  have explained how it has become a necessary infrastructure for them. Andrea Pietropolli Charmet, researcher from the Department of Molecular Sciences and Nanosystems, is one of the frequent users of the calculation system: “Thanks to the Cluster I get hold of results in adequate time, but we’re still talking about weeks of calculations, so I limit the access to the supercomputer in Cineca’s SCAI Centre to certain cases, those that are more complicated in terms of calculation and resources". The interuniversity consortium Cineca manages 4 of the 6 Italian supercomputers, on the list of the top 500 most powerful supercomputers in the world.

Pietropolli Charmet is a physical chemist that specializes in spectroscopy: “With infrared spectroscope I primarily study compounds of atmospherical interest, for which I determine the potential impact on the phenomenon of global warming, and of astrophysical interest. With the Cluster I can calculate a series of parameters that support and integrate with the experimental investigation, allowing me to obtain a more complete and accurate description of the systems that I am studying."

On the Economics Campus there is a group of econometric scholars that use the calculation system intensely. “In terms of the research that we are faced with, there are ties with computerized econometrics” explains Roberto Casarin, a professor from the Department of Economics. “We use the Cluster both for the analysis of properties of new mathematical models and new methods of statistic analysis, and for the application of these models and methods. The Cluster is used to analyze the information contained in large databases and elaborate indications on the state of the economic system and the financial market."


According to the researchers at San Giobbe, the durations of calculations have reduced by nearly 10 times. Casarin demonstrates an emblematic case, showing the graphics of a project studying the spread of bankrupcy in the financial system. It shows the connections (the coloured lines in the image above) between the principal European financial institutions (the coloured knots) that have a fundamental role in the transmission of the financial crisis. “Extracting the connections at a daily frequency would take around a year and a half to calculate on a normal computer. With the Cluster it takes 30 days to obtain it“ he affirms.

The new challenges of research and the rapid obsolescence of hardware demand a continued investment in the infrastructures of calculation. “Our Cluster will become obsolete in no more than 3 years” explains Flavio Sartoretto, professor at the Department of Environmental Sciences, Informatics and Statistics and member of the committee that manages the system. “The progress of cloud computing will lead to making a choice between obtaining new hardware or investing in staff able to configure and manage calculation services provided by big cloud players, like Amazon. Personally I agree with going in the second direction, taking advantage of the calculation services that will be made available to us in cloud by the public consortium GARR".