I've attached a PDF that describes (in somewhat excruciating detail)
exactly how we're writing binary trajectories using the NetCDF format.
Dave, Tom C., Bob and I are (I think) reasonably happy with the format
as described here, and have it implemented in pmemd, sander and ptraj.
Before I send this off to John Stone and the VMD folks, though, I want
to give amber-developers a chance to comment on it and point out
anything crucial that I/we may have missed.
Note that one of the points of this format is that it has
forward-compatible extensibility. So if, for instance, sander starts
writing additional data into the file, an older version of ptraj will
still be able to read that file (of course it won't do anything with the
additional data). My point being: if there are additional things you'd
like to see in the trajectory file that aren't in this document, it
doesn't necessarily need to be fixed now, because it can be painlessly
added down the road. On the other hand, if you see a problem or
desperately need a change to what's already described in this document,
then *now* is the time for changes, before we have problems with
backwards compatibility or programs outside the AMBER suite.
Thanks!
John
Received on Wed Apr 05 2006 - 23:49:43 PDT
