My experience so far is that when parametrizing all interactions together of course mor memory is needed (a linear system of equations is solved for all degrees of freedom together). But if there is not enough memory, csg_fmatch crashes with an error of bad_alloc in the output. So if that not happens, not enough memory is not the problems.
If forces should be parametrized for bead distances which are not sampled at all also csg_fmatch crashes with an error of singular column in matrix or something. I am not sure where the nan numbers come from. But I would also guess they can come from insufficient sampling. So I would mabe decrease the bin size, increase the number of frames per block (if memory allows) and of course chose the min and max settings for the interaction range according to the distributions (from csg_stat) I hope this helps a bit. Best Christoph Am Freitag, 13. November 2015 03:15:43 UTC+1 schrieb nicky...@gmail.com: > > > Hello, Christoph > > In the appendix of the reference *Rühle V, Junghans C, Lukyanov A, et > al. Versatile object-oriented toolkit for coarse-graining applications[J]. > Journal of Chemical Theory and Computation, 2009, 5(12): 3211-3223.* > It requires that all interactions should be inputted by "fmatch.xml" > theoretically. > > > <https://lh3.googleusercontent.com/-ewba7Sf00LU/VkVENx2UvGI/AAAAAAAAAAc/mhClv_Xx9Hc/s1600/%25E6%258D%2595%25E8%258E%25B7.JPG> > Actually, I just put only one interaction in "fmatch.xml" as input, then > VOTCA output a normal result, whether it is correct or not. > Does it influence the result, when I input only one interaction? > Thanks a lot ! > > Nickya > > > > 在 2015年11月11日星期三 UTC+8上午5:48:06，Christoph Junghans写道： >> >> 2015-11-09 19:21 GMT-07:00 <nicky...@gmail.com>: >> > Hello, Rom, Christoph, Sikandar >> > >> > It seems that I got into the same trouble with you. >> > It was out of memory by using a large xml file. >> > Then I ran a single interaction or several interactions. >> > "Single" was much faster than "several". >> > However, some of the results were normol, some were "nan". >> > >> > ***.force >> > ... >> > 1.91 -2.89326427 3.107846992 i >> > 1.92 -1.848309842 2.648399863 i >> > 1.93 -0.5889638119 2.629085293 i >> > 1.94 0.4264542843 3.10001575 i >> > 1.95 0.7396249096 3.516476759 i >> > 1.96 0.06407534287 3.515091163 i >> > 1.97 -1.53992564 3.334205836 i >> > 1.98 -3.840262448 3.585295232 i >> > 1.99 -6.604819488 4.667911077 i >> > 2 -9.601481171 6.365280589 i >> > ... >> > >> > >> > ***.force >> > .. >> > 1.91 nan nan i >> > 1.92 nan nan i >> > 1.93 nan nan i >> > 1.94 nan nan i >> > 1.95 nan nan i >> > 1.96 nan nan i >> > 1.97 nan nan i >> > 1.98 nan nan i >> > 1.99 nan nan i >> > 2 nan nan i >> > ... >> > >> > >> > How to avoid these mistakes? >> You get NaNs if there is not enough statistics in each bin. Did you >> play with the frames_per_block option? >> The other option is to look at the rdf and pick min and max a little >> tighter, meaning so that it only covers the region where there is >> actually structure. >> >> Maybe the other Christoph can say something about that issue. >> >> Christoph >> >> >> > >> > Thanks in Advance, >> > Nickya >> > >> > -- >> > You received this message because you are subscribed to the Google >> Groups >> > "votca" group. >> > To unsubscribe from this group and stop receiving emails from it, send >> an >> > email to votca+un...@googlegroups.com. >> > To post to this group, send email to vo...@googlegroups.com. >> > Visit this group at http://groups.google.com/group/votca. >> > For more options, visit https://groups.google.com/d/optout. >> >> >> >> -- >> Christoph Junghans >> Web: http://www.compphys.de >> > -- You received this message because you are subscribed to the Google Groups "votca" group. To unsubscribe from this group and stop receiving emails from it, send an email to votca+unsubscr...@googlegroups.com. To post to this group, send email to votca@googlegroups.com. Visit this group at http://groups.google.com/group/votca. For more options, visit https://groups.google.com/d/optout.
