Next: Development of Radio Astronomical Data Reduction Software NEWSTAR
Up: Software Applications
Previous: CCD Charge Shuffling Improvements for ICE
Table of Contents - Subject Index - Author Index - Search - PS reprint - PDF reprint

He, X. H., Wise, M., & Glotfelty, K. J. 2001, in ASP Conf. Ser., Vol. 238, Astronomical Data Analysis Software and Systems X, eds. F. R. Harnden, Jr., F. A. Primini, & H. E. Payne (San Francisco: ASP), 518

Enhancements of MKRMF

X. Helen He
Harvard-Smithsonian Center for Astrophysics, 60 Garden Street, MA 02138

Michael Wise
MIT Center for Space Research, Cambridge, MA 02139

Kenny J. Glotfelty
Harvard-Smithsonian Center for Astrophysics, 60 Garden Street, MA 02138


MKRMF, a data analysis tool of the Chandra X-ray Science Center (CXC), has evolved to more effectively create response matrix files (RMF). It provides new and enhanced features: uniform binning syntax, all-inclusive FITS embedded function (FEF) file extraction, and non-linear EBOUNDS calculation. This paper describes the algorithm, application interfaces and highlights of the future development.

1. Introduction

MKRMF generates a response matrix for an arbitrary redistribution function over a 2-dimensional grid plane. Since last reported by He (1999) it has evolved to include:

The tool's parameter file (mkrmf.par) has been accordingly updated, as listed in Table 1. The table highlights the most important parameters.

\tablecaption{Input Paramet...
verbose &0 & verbosity level (0 = no display)\nl

2. Uniform Binning Syntax and Rebinned Output

In conjunction with CXC Data Model (DM) filtering, MKRMF adopts the DM grid binning syntax for the ``axis1'' and ``axis2'' parameters. The binning syntax is consistent with other CIAO 1 tools. As previously defined, the binning syntax is grouped into two categories: command line and ASCII/FITS table file. The command line input allows the user to specify discrete grids in lower bound, upper bound, and binning type, separated by ":". The grid type can be specified as linear or logarithmic binning for either the binning step or the total number of bins. The file input allows the user to tabulate an arbitrary grid. The ASCII file format contains two columns: lower and upper bounds. Two columns extracted from the FITS file define the bounds by following DM filtering syntax. Table 2 summarizes the binning syntax.

\scriptsize\tablecaption{Binni... FITS format&\verb*+[cols ENERG_LO, ENERG_HI)])+\nl

MKRMF always computes the matrix (and EBOUNDS) at full resolution at the specified <min>:<max> range. The matrix can be binned by specifying bin type and step size. MKRMF will use this information to scale down the output to the requested bin size.

3. New FEF File

A FEF FITS binary table is a CXC format convention to allow specification of an n-dimensional image in the form of an analytic function of n variables in a FITS binary table HDU. This file format has been developed to be extremely generic and to allow very cost effective reuse (Rots 2001).

MKRMF takes the analytic redistribution functions which are conventionally expressed by columns of independent function variables. These function variable columns are stored in the FEF format and used by MKRMF to create standard response matrices. Prior to the release of CIAO 2.0, a FEF file existed for each individual spatial region for which (different) redistribution functions were defined. Because of this, those who wished to perform analyses over large spatial regions or over multiple chips potentially had to keep track of a large number of files. In CIAO 2.0, these files have been merged into a single file, an all-inclusive and spatial varying FEF file. In addition to columns containing redistribution function parameter values, the new FEF file contains new columns CCD_ ID, CHIPX, and CHIPY, which provide the location and bounds for each spatial region. Another new column, REGNUM, contains an integer value that identifies each spatial region uniquely. The new FEF format also encapsulates the gain relationship of the CCD for energy in a specific region. This relationship is defined by a new column, CHANNEL (or PHA), mapped to ENERGY in that region. By introducing the ( ENERGY, CHANNEL) pair columns, the generic FEF extractor can effectively result in data block consisting of one region for each energy when a DM filter is applied to the FEF file. Therefore, the generic FEF file is effectively reusable and backward compatible.

The MKRMF user must now use DM syntax to retrieve information for one spatial region, as shown in this example:

    mkrmf infile=fef.fits[ccd_id=7, chipx=(1:100),chipy=(1:32)]

When mkrmf executes the program with an input FEF file named "fef.fits", it extracts the data containing ( ENERGY, CHANNEL) gain and redistribution function variables internally sorted for CCD_ ID of 7 with chip pixel range of 1 to 100 and 1 to 32 along X- and Y-Axis, respectively. Figure 1 illustrates two MATRIX components generated on a FEF file with the filter above for a combination of ten embedded Gaussian functions on linear binned PHA-energy grids.

Figure 1: RMF MATRIX Output.

4. Non-linear EBOUNDS Interpolation

In previous releases, the EBOUNDS array was calculated through a linear analytic expression of two constant parameters (keywords): SCALE and OFFSETS. MKRMF now employs a new algorithm for the gain calculation from data tabulated in the new FEF file. The original linear interpolation still exists for backwards compatibility.

The CHANNEL and ENERGY pairs in the new FEF file encapsulate the piecewise linear gain information within a given ( CHIPX, CHIPY) region for each specific energy value. Our new scheme to calculate the EBOUNDS array is to linearly interpolate the energy for a given PHA value from the pairs of channel and corresponding energies that bound the PHA value. The EBOUNDS energy derived in this way represents the variation of the CCD gain in that region, or an effectively non-linear and spatial varying relationship.

5. Future Developments

Future focus will be on multi-region RMF calculations. The current tool provides RMF output limited to one region characterized by the same response function within the given ( ENERGY, CHANNEL) range. However, those parameters are also functions of spatial variables. This limitation will be removed by taking a weighting factor into consideration for each region, so a multi-region RMF output can be achieved by applying the weighting to redistribution functions of the interesting regions. As such, MKRMF data I/O is expected also to be updated in the future. The user interface to MKRMF will also be evaluated to make it easier for users.


We are grateful for many fruitful discussions with various CXC members. This project is supported by the Chandra X-ray Science Center as part of NASA contract NAS8-39073.


He, H. Wise, M., & Ljungberg, M. 1999, in ASP Conf. Ser., Vol. 216, Astronomical Data Analysis Software and Systems IX, ed. N. Manset, C. Veillet, & D. Crabtree (San Francisco: ASP), 636

Rots, A. H. 2001, this volume, 479


... CIAO1

© Copyright 2001 Astronomical Society of the Pacific, 390 Ashton Avenue, San Francisco, California 94112, USA
Next: Development of Radio Astronomical Data Reduction Software NEWSTAR
Up: Software Applications
Previous: CCD Charge Shuffling Improvements for ICE
Table of Contents - Subject Index - Author Index - Search - PS reprint - PDF reprint