[RASMB] Global analysis requirements

John Weeks john at wavemetrics.com
Mon Mar 8 14:43:00 PST 2004


I am a programmer for WaveMetrics, Inc., makers of the scientific graphing
and numerical analysis applicaton Igor Pro. Please pardon my commercial
presence here; I am thinking about an upgrade to our global analysis
package and one of our customers suggested this list might be a good place
to get technical questions answered.

I have been looking at the documentation of SEDPHAT to get some idea of
what might be useful in a global analysis package. In particular, how real
people use multiple fitting functions in a global analysis.

1) It appears to me that the fitting functions in SEDPHAT are
pre-determined. Consequently, any linked fit parameters are known a priori;
that is, if a given fit coefficient in two different functions represent
the same thing, the software simply uses the same parameter for both. Have
I understood correctly?

2) SEDPHAT makes provision for "shared local" fit parameters. These are
parameters that have a single value for multiple data sets, but are not
global over all data sets. If I understand correctly, shared local
parameters are designated manually by the user. That is, the user selects a
fit parameter for a given data set and tells SEDPHAT to link that fit
parameter to another parameter for another data set. Did I understand that
correctly?

3) I had originally thought that I would allow the user to group data sets,
and then designate "group global" fit parameters. However, it looks to me
like that would require groups that overlap- that is, a group might be
created for data sets that share a concentration, and another group might
contain data sets that share a wavelength. These groups might contain some
common members, but not be identical. Is that correct?

Thank you in advance for any help. I appreciate your patience. I hope I
haven't mangled things too badly- my degree is in geophysics :)

Regards,
John Weeks

WaveMetrics, Inc.
Phone (503) 620-3001
Fax   (503) 620-6754
email   support at WaveMetrics.com




More information about the RASMB mailing list