Mon 06 Nov 2006 10:41:30 AM CST
Mark Miller Harald Pfeiffer gave a presentation about effects of eccentricity in simulations. John Friedman gave a brief statement about how helical KV solutions might be useful. Patrick Sutton asked about how this might be used in the current numerical simulations. John thinks that this gets used by gradually turning retarded solution in these simulations. John Friedman commented that one should look for second order convergence. But asking if the solution is actually a solution, one should simply look at the accuracy with which it satisfies the Einstein equations. If you are asking if you are looking at a solution to Einstein's equations. There was a lot of discussion of Mark's suggested norm. People did not know how to say what a particular value means. Sergey mentioned that data analysts are interested in other accurate properties of spacetimes and simulations. Are you looking at something like that? Luis responded by saying that the aggregate details are being looked at. Until there are more simulations. Sergey wanted to see some more broad based information extraction. Finding errors numerically should be done with every quantity. Michele Zanolin mentioned that people compute numerically derivatives of the field with respect to the initial data. Luciano reminded people that there is consistency and convergence. If you have convergence, there is consistency. Frans said that he thinks asking what is delta h is important, but the question is what could be done to provide a delta h? Mark was thinking delta h due to each error. Pablo presented a short note about accuracy. Comparing waveforms as a form of evaluation of the differences between the waveforms. He had one plot with waveforms from PSU, Uli, NASA, UTB. Waveforms look very similar. Pablo encouraged people to keep putting their waveforms into NRwaves repository.