Carlos Gabriel & Aitor Ibarra (XMM-Newton SOC)
This version (4.0.4) has been integrated in January 2005. The main aim of this release is to make SciSim more accessible to non-experts. At the same time it should make it possible to perform the sequence of activities: Simulation > Data-Conversion-into-ODF > Data-Analysis-with-SAS in a simple and standardised way, and should allow calibration experts to re-assess the reliability of the different elements of XMM-Newton simulations.
SciSim pages at the XMM-Newton SOC are here.
The work on SciSim has been complemented with an upgrade of the SAS task odffix, which makes possible the ingestion of SciSim generated ODFs into the SAS. Also, a release of all calibration files to be used in the reduction, through the SAS, of SciSim generated data, is coordinated with this release.
In addition to the natural documentation upgrade related to the changes with respect to the former SciSim version, new documentation for guiding the user especially through the analysis of SciSim data has been generated for each of the instruments, and can be found in the following links ( PN, MOS, RGS) .
While systematic studies by the XMM-Newton SOC calibration teams should follow this release in order to revise strengths, accuracies and short-comings of SciSim, a first attempt to summarize this is given in this technical note (see section "Status of SciSim by V4.0.4"), including results from limited evaluation.
SciSim 4.0.4 is released as binary for following Linux and Solaris platforms:
Due to evolution of the system characteristics and calibration through the mission, SciSim V3.0 was presenting serious deficiencies and had to be refurbished. Especially necessary for an assessment of the calibration reliability of simulations of the different observing modes (still to be performed by the XMM-Newton calibration teams) was the closure of the loop between simulation and data analysis. This has been done complementing the SciSim release with the release of 90 specific SciSim calibration files included in the SciSim CCF tarfile , together with the corresponding Calibration Index file CCF_scisim.cif. Furthermore, two SAS tasks (odffix and epicproc) had to be updated to make possible a proper SAS data reduction of simulated data. They build a patch to SAS 6.1.0. which can be found in any of the ftp directory repositories containing the different versions of SAS 6.1.0 ( ftp://xmm.vilspa.esa.es/pub/sas/6.1.0/ ), as xmmsas_20041122_1834-platform-patch4SciSim.tgz.
Testing of the SciSim functionality has shown satisfactory results for all SciSim implemented modes of all X-ray cameras, for which output conversion into ODF and subsequent data reduction through SAS is possible. This is not yet the case for OM simulated data due to limitations in the ODF generation still present. The upgrade of the main GUI interface including the chip geometries and the different FOVs of EPIC-MOS and EPIC-PN makes possible now a fast assesment of the position of sources with respect to the cameras and especially their inter-chip gaps. The astrometry is now correctly linked to the SAS, as shown in the PN example included below, where several sources are simulated and correctly found at their expected positions after SAS data reduction.
The main SciSim interface shows a large number of sources distributed in and out the FOV falling onto the EPIC PN chips.
After ODF ingestion and SAS data reduction the sources are found exactly on the expected positions.
While the ray tracing elements of SciSim including the whole optics, material descriptions, etc, seem to be rather complete and working properly, the main limitations of SciSim so far are given by the poor representation of the response functions of the cameras as well as of the implemented CTI model. The SciSim user must be aware of the limitations in this area, which make difficult if not impossible the assessment of effects related to CCD properties.
The best results can be obtained simulating EPIC full frame mode observations. As an example, a simulation of a source with the Raymond-Smith model with a temperature of 0.86 kT (keV), and an abundance of 0.27 in solar units, has been performed with all instruments. SAS data reduction, including spectral extraction of the non-piled-up source-signal parts, leads to the spectrum shown in the following figure:
Fitting with Xspec yields a temperature of (0.858 +/- 0.002) keV and an abundance of (0.256 +/- 0.009) in solar units, which represent an excellent result. EPIC-MOS simulations in full frame give similar results, as well as RGS spectroscopy data simulations. However, the spectra derived from windowing or timing modes EPIC simulations are shifted in energy scale, showing problems related to the gain corrections.
Experience within the next months should show, in a more quantitative way, limits and shortcomings of this SciSim version, as well as building the basis for improvements in future ones.