E-science meets computational science and information technology

Over the last decade or so, there has been much discussion of and progress in computational science. There was the HPCC (High Performance Computing and Communications) Initiative and the NSF Grand Challenge programs which largely focused on this. However recently there has been a subtle change – for example NSF initiated the ITR (Information Technology Research) program based on the recommendations of PITAC – the President's Information Technology Advisory Committee. We did not have the CSR (Computational Science Research), CSITR or PCSAC. One will find classic computational goals as one part of the Information Technology agenda but they are just one part and perhaps a part that is getting smaller. One sees the same trend in research with work on grids and distributed computing overshadowing that on classic parallel computing. Now NSF has a new cyberinfrastructure report ((http://www.cise.nsf.gov/b_ribbon/) continuing the same emphasis on distributed systems. What does this mean for computational science and its associated technology and research? Once we spoke of the three approaches to science: experiment, theory and the emerging computational view. Does Information Technology and cyberinfrastructure enable a fourth paradigm? Such a proliferation of scientific methods does not seem too reasonable. However the National Virtual Observatory (http://www.us-vo.org/) eloquently describes the new approach to astronomy. Just a few years ago, astronomy involved individual groups designing new instruments, developing particular observing strategies and managing the data taking on some instrument either in the space or on the ground – this data was lovingly analyzed to discover and publish new scientific insights. This " stove-pipe " approach to observational (experimental) science is typical in many fields. It ensures that those who build the equipment can ensure their data is properly interpreted with the usually difficult instrumental corrections properly applied. Note that it is usual to compare " reasonably corrected experimental data " with hypotheses (models) to which other instrumental effects are applied. Often it is not possible to remove all instrumental effects from a dataset so that it can be compared with a pristine theory (developed by a pristine theorist who needs no significant understanding of the instrument). Nevertheless we imagine a new astronomy where our investigator has the best analysis and visualization capabilities connected to the global internet. This analysis will integrate the data from multiple instruments spanning multiple wavelengths and multiple regions of the sky. This vision stresses the high performance networks, data access, management and processing enabled …