XMM-Newton Science Analysis System


colsmooth (colsmooth-1.9.1) [xmmsas_20230412_1735-21.0.0]

Convolver.

There are several ways in which the user can specify the convolving function, governed by the parameter convolvertype. These are:

• convolvertype = `set': the task then expects the convolver to be contained in the column of a fits table described by parameters convolvertable and convolvercolumn. The data type of the column must be real32.
• convolvertype = `tophat': this is a convolver which is equal to $1/W$ within $W$ contiguous channels and 0 elsewhere. The task requires $W$ to be specified via the parameter width.
• convolvertype = `gaussian': the task reads the sigma value from the parameter of the same name. The convolver is truncated at 5 times sigma either side of the centre of the gaussian.
• convolvertype = `user': in this mode, the user can supply the convolver directly as a list of real numbers via the parameter convolver.

The user should pay careful attention to the phase of the convolver. Equation 1 (and all others that include a sum over the convolver values) implies that the number of convolver channels, $2M + 1$, is an odd number, and that the centre of the convolver is located at the centre channel. However, convolvers with an even number of channels are also permitted. In this case the lowest channel of the upper half of the channels is taken to be the centre channel. In other words, the sums in this case extend from $1-M$ to $M$ for a total number of convolver channels $2M$. To clarify with a couple of examples, calls of the form

	colsmooth convolvertype=user convolver='0 0 1 0 0'

or

	colsmooth convolvertype=user convolver='0 0 0 1 0 0'

will leave unchanged the input values in inset.ds:SPECTRUM column RATE, but

	colsmooth convolvertype=user convolver='1 0 0 0'

will left-shift all the input values by 2 channels. (Note that in this last example, the right-most two channels of the output will contain dal-type null values. Zero values would probably be better and the task may be changed in future to write zeros in this case.)

The elements of the convolver may not sum to zero. Negative and zero-valued elements should be avoided if filling of holes is desired (see section 3.3.1 below).

Note that, under default settings, no convolution is performed.