How big can multispec files get?
Frank Valdes wrote on Mar 09, 1999
Q: I'm trying to make a multi-spec file of a catalogue of galaxy spectra. I have 558 objects drawn from 8 fields. Whan I combine 7 of them using scopy (a total of 487 objects) everthing works OK. However, when I try 8 it crashes citing `too many WCS systems` and exhibiting problems with the resulting header. The resulting spectrum is produced but has zero spectra from spectrum 545 onwards. I've tried increasing len_minuserarea to 1000000 to squeeze everything it but to no avail. Nothing is wrong in terms of disk space either. Nor is it anything worng with the last file as I've played with the order. A: There are definite limits to the number of spectra in the multispec format. The one you are encountering is a limit in how many "attributes" can be in an image. For multispec format the "attributes" are the key=value things in the WAT keywords. This mostly means the specN values. The current limit is 512 attributes which means about 500 specN values (to allow for a few other attributes such as the units, etc). This limit applies to the more complex multispec format where the coordinate systems can be independent. If all your catalog spectra have the same linear coordinate systems (same CRVAL1, etc) or can be put into that form through resampling with DISPCOR then the "equispec" format would be used. This has a simple WCS but needs many APNUM/APID type keywords. This would run into the indexing limit (provided the header size was set big enough) of 999.
Peter R Newman wrote on Mar 10, 1999
Frank Valdes wrote: > > Q: I'm trying to make a multi-spec file of a catalogue of galaxy spectra. > I have 558 objects drawn from 8 fields. Whan I combine 7 of them using > scopy (a total of 487 objects) everthing works OK. However, when I try > 8 it crashes citing `too many WCS systems` and exhibiting problems with > the resulting header. I ran into a similar problem producing multispec files. I found that scopy would crash after about 140 spectra were copied, but it I reloaded the noao.imred.specred package cleanly before each scopy, I could get the 320 spectra copied that I was trying for (and the file looked good). All spectra had identical dispersion in this case. Sorry, I don't have the error message -- "some kind of resource limit" say my notes made at the time. Pete Newman -- http://sa1.star.uclan.ac.uk/~prn
Last post on Mar 10, 1999