26-Mar-2007
Dear RHESSI software users, We've changed the default values for some control parameters in the RHESSI and OSPEX objects. The purpose of these changes is to make the defaults the parameter values that work best in most common situations. Of course you should continue to change these and other parameters for particular cases. The names of the parameters that were changed and a short explanation are listed below. In the RHESSI spectrum object: 1. sp_energy_binning The default energy binning code has been changed from 7 to 14. 14 has 77 bins ranging from 3 to 250 keV (1 keV wide bins at low energies and wider bins at higher energies) and is good for analyzing hard x-ray flares. You may also want to consider binning code 22 which is similar, but has 101 bins ranging from 3 to 300 keV with 1/3 keV wide bins from 3 to 15 keV. 2. sp_time_interval The default time bin width is now 4 seconds. This is more suitable for the analysis of temporal structure. Previously it was 60 seconds. 3. simplify The full SRM will now be written in the FITS output file when you write the spectrum and SRM files. Previously it was the diagonal SRM. In some cases calculation of the full SRM will take longer (if you have a large number of energy bins), but for spectral analysis you almost always want to use the full SRM. In the RHESSI image object: 1. natural_weighting, uniform_weighting The default is now natural weighting (all collimators have equal weight). Previously it was uniform weighting (weight each collimator by the inverse of its FWHM). Natural weighting is better for extended sources and allows the finer grids to be utilized without adding to the noise; uniform weighting provides greater sensitivity for compact sources. This is used only in the back-projection and clean algorithms. 2. use_flux_var use_flux_var is now enabled by default. When enabled, the software attempts to construct a demodulated profile of the true incident flux variation in the energy band, unmarred by grid modulation. This can be critical for algorithms that try to minimize the fitting statistic. 3. use_rate use_rate is now enabled by default. This enables the rate-based back projection. This is particularly critical to set when deadtimes become greater than 50%. For smaller deadtimes, it has little effect. For dirty (back-projection) maps made with use_rate set to 0, the true source positions may have the lowest (negative) values in the field. Setting use_rate to 1 corrects that and the source positions will again have the highest values and appear brightest in the dirty map images. For the CLEAN algorithm to work, the source positions must have the highest values. 4. image_algorithm The MEM VIS algorithm has been disabled. It didn't work. The MEM Sato algorithm is no longer an option from the hessi GUI. You can continue to use it from the command line, but we don't have confidence in the results. Two new image algorithms that use visibilities will be available soon. 5. clean_niter The number of iterations in the CLEAN algorithm is now 100. Previously it was 33. 6. clean_frac The gain in the CLEAN algorithm is now 0.05. Previously it was 0.10. In each iteration of CLEAN, the clean_frac fraction of the peak value is multiplied by the point spread function and subtracted. We want to encourage users to explore using parameter values other than the defaults. In particular, users should vary their choice of the range of collimators used (det_index_mask parameter), pixel size (pixel_size parameter), and energy range (im_energy_binning parameter). And we want to remind users to exclude detector 2 when using energies below 20 keV, and to use smaller pixel size with the finer grids. In the OSPEX object: 1. spex_uncert The systematic uncertainty is now 0.02. Previously it was 0.05. Setting this parameter too high results in an artificially low value for chi-squared. You may want to set it 0.0. Please let us know if you have any questions about this. Kim