SUMMARY of SXT Deconvolution Workshop

(Loren W. Acton 1/21/01)
Institute for Astronomy
Honolulu, Hawaii
18-19 January 2001

Participants:

		Takeo Kosugi
		Hiro Hara
		Piet Martens
		Hugh Hudson
		David McKenzie
		Loren Acton
		Tom Metcalf
		Phil Shirts
		Nariaki Nitta
		Leon vanSpeybroeck
		Barry LaBonte
		Jing Li

Agenda:	Thursday, 1/18/01
		AM	Discussions of Point Spread Function.
		PM	Discussion of deconvolution techniques.
	Friday, 1/19/01
		Work by participants.
		Summary session at 16:00.

Results:

The workshop was considered very successful. It was agreed that cleaning of the half-resolution, full-frame, SXT Science Composite (SSC) images was a different problem than attempts to devconvolve the point spread function (PSF) from full-resolution, partial-frame, flare images. In the former case the primary objective is correcting the effects of x-ray scatter and the properties of the core of the PSF is not important. In the latter (flare) case correction for the core and near wings of the PSF is important.

Scattering Wings: McKenzie, Metcalf and Shirts performed experiments using FFT and MEM techniques to clean an SSC with a PSF comprising a delta-function core and scattering wings which fall of as the square of the distance form the scattering source. Initial efforts were encouraging. Acton verified the slope of the SXT scattering wings in sxt_psf.pro (-1.89) by normalizing the PSF to one location in a 12/26/91 starburst image and subtracting that PSF from the starburst. The correction was satisfactory. Residual noise in the corrected image centered on 0.

Flare correction: Hudson examined recent data from the flare scattering table experiment that traced the shadow of the entrance-filter "spider" to near the core of the flare. Kosugi and Hara were successful in their attempts to clean a flare image through the use of FFT with "bandwidth expansion" (rebinning to finer pixels by a factor of 2) employing a Gaussian weighted Moffat function for the core PSF. The elimination of the broad wings of the Moffat function stabilized the deconvolution.

Other: Martens investigated blind iterative deconvolution (BID) and concluded that this technique, which yields both a cleaned image and a PSF, is worth pursuing as a means of better determining the SXT PSF with partial-frame images. LaBonte demonstrated that when placing the FFI into a larger array to do an FFT it is not necessary to take into account the scattering into the extended image -- zeroes are OK. Also, when deconvolving half-resolution FFIs it is not necessary to expand the bandwidth by rebinning to mroe pixels.

Action items: