Ticket #149 (closed task: fixed)

Opened 4 months ago

Last modified 2 months ago

Determine and implement correct tapering behavior

Reported by: laidler Assigned to: somebody
Priority: highest Milestone: Pysynphot support for all ETCs: 18.1
Component: pysynphot Version: Pysynphot v0.6
Severity: normal Keywords: commissioning
Cc: svn_pysyn@stsci.edu

Description

Pysynphot presently applies tapering to all Spectra and SpectralElements? that don't end in zero flux/throughput. Synphot does not.

This was originally implemented to provide a well-conditioned integrand, but it seems to have escaped to quite wide use.

This behavior seems to be causing many failures in the commissioning tests due to differences at the endpoints.

There seems to be consensus that spectra should not be tapered, but there is some disagreement over whether or not spectral elements should be.

Change History

08/08/08 17:01:42 changed by laidler

  • priority changed from normal to high.

09/12/08 10:55:38 changed by laidler

  • priority changed from high to highest.

09/15/08 14:32:23 changed by laidler

After further discussion, we have decided:

  • tapering should not be used within the pysynphot machinery. Thus it will be removed from the .resample() methods, and will no longer affect integration.
  • the .taper() method (where it already exists) will remain available to users
  • the ObsBandpass? construction should check to see that the resulting CompositeSpectralElement? produces a throughput table that is bounded by zeros, and if not, should print a warning

The responsibility for ensuring that HST obsmodes produce correctly bounded throughputs should belong upstream, at the CDBS end. We might want to build this check into the pre-delivery testing.

09/18/08 14:09:49 changed by laidler

  • status changed from new to closed.
  • resolution set to fixed.

r618 removes tapering from the .resample methods; r627 adds the check and warning to the ObsBandpass? constructor.