specfit in mac/Intel iraf Port 2.13 beta
leighly wrote on Sep 07, 2006
I am having a problem using the STSDAS contributed program specfit on
the Mac/Intel IRAF Port 2.13 beta.
I am using the STSDAS contributed program specfit to fit spectra. I
have used this program on Suns and also on Mac PowerPcs with no
problem on thousands of spectra. However, it crashes on the mac.intel
on a 20-inch iMac in a sort of peculiar way. Generally, we fit
several times using the simplex and numrecipe algorithms because the
errors don't seem to converge right away. It fits ok several times,
and then crashes. A typical crash looks like below, when we use a iraf
script, but note that it crashes also after several fits if we do it
completely interactively.
Crash:
spfitpkg> temp2
ERROR: floating point invalid operation
"grid_par2=2, grid_blim2=0., grid_tlim2=1000., grid_step2=0., cursor= ..."
line 11: temp.cl
called as: `temp2 ()'
Script we use for one particular spectrum:
specfit ("spSpec-51901-0457-087_final_fix_z.dat",
"*", "*", 1, 1, 0., 0., 10., 1., 0.0010000000474975, 15, 100, 1, "new_comp",
debug=no, interact=yes, gridfit=no, errors_from_=no, plotdata=yes,
type_of_fit="simplex", logfiles="spSpec-51901-0457-087_logfile", database="sfdb8",
initial_fit="spSpec-51901-0457-087_newest_parameters",
final_fit="spSpec-51901-0457-087_latest_parameters", plot_file="spSpec-51901-0457-087_specfit_plot.dat",
flux_interva="", Sample_range="2200.0-3050.0", max_iteratio=100,
fit_toleranc=1.0000000000000E-5, v0=0., v1=1., key_file="specfit$",
low_mult=0.1, high_mult=10., step_mult=0.01, grid_num=1, grid_comp1=1,
grid_par1=2, grid_blim1=0., grid_tlim1=1000., grid_step1=0., grid_comp2=1,
grid_par2=2, grid_blim2=0., grid_tlim2=1000., grid_step2=0., cursor="P")
the Mac/Intel IRAF Port 2.13 beta.
I am using the STSDAS contributed program specfit to fit spectra. I
have used this program on Suns and also on Mac PowerPcs with no
problem on thousands of spectra. However, it crashes on the mac.intel
on a 20-inch iMac in a sort of peculiar way. Generally, we fit
several times using the simplex and numrecipe algorithms because the
errors don't seem to converge right away. It fits ok several times,
and then crashes. A typical crash looks like below, when we use a iraf
script, but note that it crashes also after several fits if we do it
completely interactively.
Crash:
spfitpkg> temp2
ERROR: floating point invalid operation
"grid_par2=2, grid_blim2=0., grid_tlim2=1000., grid_step2=0., cursor= ..."
line 11: temp.cl
called as: `temp2 ()'
Script we use for one particular spectrum:
specfit ("spSpec-51901-0457-087_final_fix_z.dat",
"*", "*", 1, 1, 0., 0., 10., 1., 0.0010000000474975, 15, 100, 1, "new_comp",
debug=no, interact=yes, gridfit=no, errors_from_=no, plotdata=yes,
type_of_fit="simplex", logfiles="spSpec-51901-0457-087_logfile", database="sfdb8",
initial_fit="spSpec-51901-0457-087_newest_parameters",
final_fit="spSpec-51901-0457-087_latest_parameters", plot_file="spSpec-51901-0457-087_specfit_plot.dat",
flux_interva="", Sample_range="2200.0-3050.0", max_iteratio=100,
fit_toleranc=1.0000000000000E-5, v0=0., v1=1., key_file="specfit$",
low_mult=0.1, high_mult=10., step_mult=0.01, grid_num=1, grid_comp1=1,
grid_par1=2, grid_blim1=0., grid_tlim1=1000., grid_step1=0., grid_comp2=1,
grid_par2=2, grid_blim2=0., grid_tlim2=1000., grid_step2=0., cursor="P")
Mike Fitzpatrick wrote on Sep 07, 2006
A floating point error can sometimes be a coding error and sometimes it is data related. However, there is a known issue with the V2.13 system in the FPE handling code that I suspect you might be seeing. A quick way to find out would be to install a new set of STSDAS binaries I've just
uploaded to the http://www.iraf.net/ftp/iraf/extern/ download area (see stsdas_v3.5-bin.macintel.tar.gz).
Please let me know whether these work or not. Somebody else might be able to comment on SPECFIT in general or you might also try asking the STScI helpdesk (help@stsci.edu). Good luck.
Cheers,
-Mike
uploaded to the http://www.iraf.net/ftp/iraf/extern/ download area (see stsdas_v3.5-bin.macintel.tar.gz).
Please let me know whether these work or not. Somebody else might be able to comment on SPECFIT in general or you might also try asking the STScI helpdesk (help@stsci.edu). Good luck.
Cheers,
-Mike
leighly wrote on Sep 07, 2006
Sorry to take so long to reply - I wanted to first do all the Mac
updates and reinstall everything from scratch including the new
stsdas binaries.
Unfortunately, it still doesn't work - same
"floating point invalid operation" error as before.
Thanks,
Karen Leighly
updates and reinstall everything from scratch including the new
stsdas binaries.
Unfortunately, it still doesn't work - same
"floating point invalid operation" error as before.
Thanks,
Karen Leighly
Mike Fitzpatrick wrote on Sep 07, 2006
Karen,
Then check with the STScI helpdesk (help@stsci.edu) in case this is a known issue. I'm not sure whether they're willing/able to debug a macintel problem but they know the software (I've never used it). A sample image, task parameters and a recipe for reproducing the problem will help. An FPE is sometimes a coding error and since you used the new binaries we've ruled out the one known problem with the core system so it looks like some peculiar to the task.
-Mike
Then check with the STScI helpdesk (help@stsci.edu) in case this is a known issue. I'm not sure whether they're willing/able to debug a macintel problem but they know the software (I've never used it). A sample image, task parameters and a recipe for reproducing the problem will help. An FPE is sometimes a coding error and since you used the new binaries we've ruled out the one known problem with the core system so it looks like some peculiar to the task.
-Mike
Last post on Sep 07, 2006