PSS Reference

  This section contains a brief outline of PSS's processing method followed by a number of topics which describe the full functionality of the program.

At the core of PSS are a set of routines which evaluate a statistic comparing model and data. These all provide the same facility to the rest of the PSS application, and are applied in different ways at different stages of processing or in different PSS modes. The facility is the evaluation of the significance surface at a point in image coordinates. The mechanism is as follows (See Figure 1 in Appendix D),

• The input image pixel containing the point is located.
• This pixel defines the centre of the psf box. The size of the box is controlled by the PSFPIX parameter is expert mode, otherwise is chosen to include 68% of the psf energy.
• The area of overlap of the psf box and the image grid is found, taking due account of the edges of the latter. Note that any user defined image slice does not affect the overlap of the psf box outside the slice.
• The psf values in the psf box are found. If the point lies on the exact centre of pixel , and the parameter PSFCON is true, then the psf values are simply retrieved from a storage array. If the former is true, but not the latter, then the psf is evaluated using a call to the psf system. The last alternative, PSFCON true and lying in the pixel but not at its centre, causes a resampling of the psf by a vector where is the image coordinate of the centre of pixel .
• The statistic is evaluated as a summation over the area of overlap of the image and the psf box. The exact algorithms used are specified in Appendix B. Image data points with bad quality are ignored in these summations. Because the significance is a summation over the surrounding area, it is possible to produce a significance even if the pixel has bad quality. If all the pixels in the overlap region have bad quality, the the significance is set to zero.
• If only a single grid position is called for, the routine sets the optimum flux and any other relevant information in a storage area.

The SEARCH mode algorithm uses a statistic routine to calculate the significance map over the entire image region selected by the user. PSS then puts down a small grid around each candidate source with the grid chosen in such a way that the optimum source position lies on a grid vertex. The grid used is just large enough to perform centroiding to derive the next guess at the optimum source location.

PARAM mode invokes a statistic routine to evaluate significance at each test position lying on the image. Only a single grid position is used.

Both UPLIM and UPMAP modes use special versions of the statistic routines which perform the additional task of finding a flux upper limit after the optimum flux has been found.

• Global Properties
• Input data
• Background Model
• Point Spread Function
• Fitting Statistics
• Background Rescaling
• Oversampling
• Source Confusion
• Source Extension
• Specifying Confidence Levels
• Source Parameter Errors
• Restarting PSS
• Source Subtraction
• Diagnostic Mode
• Limitations