Wednesday, May 31, 2006

xselect and filter col PI=...

Fixed the case that the user does a filter column on the spectrum column (PHA or PI). Before this was not working because I automatically add a filter between the minimum and maximum allowed values. The two sets of filters are ORed in extractor so the filter column has no effect. Removed the min to max filter if a filter column on the spectrum is present. There is a potential conflict if the user does both a filter pha and a filter column on the PI/PHA. In this case the filter pha is ignored.

It would be to good to support exclude ranges in extractor so that the user could specify eg filter column PI=!1600:1700 to get all channels except 1600-1700.

Keywords: HEAsoft, xselect

more on xselect and filter time UT

Fixed further problems in filter time UT. Wrote a new wrap-up routine str2sec which converts a string of comma-delimited time specifications into an array of double precision SCC times. This replaces large chunks of code in xsl_time_cursor and xsl_enter_filter and eliminates most of the problems. Also fixed xsl_choose so that if the user loads an obscat and does a choose instead of read'ing files then the mjdrefi/f and timesys internal globals are set.

Keywords: xselect, HEAsoft

Tuesday, May 30, 2006

xselect and XMM response generation

Modified the XMM EPIC response generation perl script so that it uses the badpixel data. This required a new perl script (and code in xselect to call it) which copies the appropriate BADPIX## and EXPOSU## extensions from the event file to the spectrum file. At present this requires that there only be one event file from which the spectrum was extracted. If there are multiple event files then the user is advised to do an "extract event" then save the spectrum.

Moved XSL_RSPSAV from xsel_mission.f to xsel_utils.f since it has no mission-dependent code. Note that I slightly changed the behaviour of XSL_GRPSAV in the event that something goes wrong in getting the grouping.

Keywords: HEAsoft, xselect, XMM

Wednesday, May 24, 2006

Gary Hinshaw: WMAP 3 year results

EUD colloquium by Gary Hinshaw, data lead on the WMAP mission. He summarized the science justification and history then launched into the the 3 year results. The gain calibration has been improved leading to slightly different results on the temperature power spectrum - particularly at the low l end - and of course the S/N is improved. The big new result is the polarization map which provides a much better measure of the optical depth to free electron scattering - this is significantly less than the best-fit from the 1-year results implying a later reionization epoch (400,000 years after inflation rather than 100,000 years). Since the optical depth is highly correlated with the power-spectrum index this result also enables a more precise measurement of the index which is now inconsistent with Harrison-Zel'dovich n=1 at 2 sigma. The best fit is around n=0.96 consistent with two of the most popular inflation models. In answer to my question Gary also noted that the octopole is now more consistent with LCDM. The quadrapole best fit is still low however examination of the complete likelihood function shows a tail to larger values such that there is a 1/10 probability of getting the observed result. Since precision is limited by cosmic variance (the quadrapole is the variance of 5 independent numbers) we can't do any better.

Laurie Leshin: First results from Stardust

Goddard Science Colloquium by Laurie Leshin (GSFC) on the preliminary (and as yet unpublished) results from Stardust. She explained why comets are interesting - they were formed far enough out in the proto-Solar nebula that they should be unprocessed material which are deflected into the inner system and hence available for study. Sample return offers the benefit that all the heavy equipment can be kept on the ground and instrument development continues after the mission is completed. Stardust successfully collected cometary dust in its aerogel and preliminary examination is underway. The most unexpected result so far is the discovery of a refractory grain which should not be present in the comet if it is unprocessed.
She finished up by describing SCIM, a Mars Scout proposal, for a sample return mission which dips into the Martian atmosphere during the dust storm season to collect particles then returns them to Earth.

Phil Kronberg: Intergalactic CRs and mag. fields

EUD seminar by Phil Kronberg (LANL & U.Toronto) on some of his recent work on intergalactic magnetic fields observed via Faraday rotation. He is using Arecibo & DRAO interferometer to do a low surface brightness, wide-angle survey for intergalactic emission. The first field of the Coma cluster and surroundings shows emission from the same extended region seen in the X-rays.

Monday, May 22, 2006

string justification in the PS driver

Fiddled around with a hack to GRLEN to deal with the difference in character widths of for PS and PGPLOT symbols. The width ratio varies between different characters so had to give separate numbers for different categories. I only included the most common cases (a-z, A-Z, ".", "(", ")", " ", "-") but that should be close enough for most use.

The proper way to do string justification would be encode it in the PS file (using stringwidth and some tricky manipulation) however to do this the FJUST variable would have to be passed down to the GRTEXT routine and I didn't want to change the signatures of standard PGPLOT calls.

Keywords: HEAsoft, PGPLOT

Friday, May 19, 2006

more on the PS driver

Improved handling of text for the PS driver. Now pass the text bounding box values in the RBUF array for the IFUNC=31 option. These are calculated in GRTEXT where the only special case required for PS drivers is not to call GRLIN0 to actually plot the lines. This is much simpler than the previous method of setting the color to the special value of -1.

Also moved the PS driver-specific code related to the PS_VERBOSE_TEXT environment variable (which is obsolete) from GRTEXT into PSDRIV and reformatted and renamed the subroutines in PSDRIV to conform better to PGPLOT standards.

Thursday, May 18, 2006

xspec 12 start-up script

Eliminated the xspec v12 start-up script so that we just need the executable. The XSPECROOT environment variable is not used anywhere. SPECTRAL is used by routines in Global.cxx and XSGlobal.cxx where it can be replaced by HEADAS. POW_LIBRARY is set in headas-init. TCLRL_LIBRARY is used in xspec.cxx but the information it contains is available in tclreadline.h so just include that. The PGPLOT variables can all be removed once we merge the versions of pgplot under the Xspec and tcltk trees. The difference is Ben's PS driver. This uses PGPLOT_DIR for the directory containing - at the moment this is in the bin directory but should be moved to lib. PGPLOT_FONT is the font file which is an ASCII file in the Xspec tree and a binary unformatted file in the tcltk tree.

Fixed Ben's PS driver so that it is no longer necessary to set PGPLOT_PS_BBOX to MAX to avoid the chance of text being lost outside the bounding box. The trick is to have GRTEXT perform all the operations used for other devices which divide text up into line segments but not actually draw any of the lines. This ensures that the bounding box is set correctly. The
text is then written to the PostScript output. Also changed the write text option from IFUNC=30 to 31 because 30 is used for scrolling a rectangle in the X-window driver. The functions in src/grpssy.f are properly part of the PS driver so appended them to the end of psdriv.f.

Note that writing text the way we do at the moment does not conform to the pgplot architecture since it requires an explicit test for a PS device type in the routine grtext, which ought to be device-independent. The correct way to handle this is to define a new driver attribute to specify whether text needs to be split into line segments or can be written directly by the driver. grtext would then query this attribute rather than check for a specific device type.

PGPLOT_PS_BBOX should be set to MAX for Ben's PS driver otherwise text may end up lost off the edge of the bounding box.

Note that Ben's PS driver pretty clearly subverts the way that PGPLOT works. In standard drivers the text is written by drawing lines. Since PS handles text internally Ben set up a device-dependent diversion in GRTEXT to pass the text to the driver (using IFUNC=30). One consequence of this is that the bounding box variables (BBXMIN etc) are not updated for the position of the text hence the need to set PGPLOT_PS_BBOX. This ought to be fixable.

Wednesday, May 17, 2006

Suzaku XIS bad channels

Modified xselect.mdb to set the QUALITY on XIS channels 1-81 and 3290-4095 so they are ignored when the xspec command "ignore bad" is used.

Keywords: HEAsoft, Suzaku, xselect

mkfbin and hkbin

Modified mkfbin/hkbin so that if an event lightcurve has been extracted then the mkf/hk parameters will be binned to match the event lightcurve and the RATE will be added to the mkfbin/hkbin output file. plot mkf/hk then enables the user to plot correlations between housekeeping parameters and the event rate. Much of the code for this change has been moved out of Fortran and into a perl program (xsl_mkf_or_hk_bin).

This required a change to extractor to save the threshold used to choose whether to write out a lightcurve bin (ie minimum value of FRACEXP). This is extractor v4.65.

Added filter mkf and filter hk options as aliases for select mkf and select hk. Added save mkfbin and save hkbin options.

Set the xselect version number to 2.4 for the next HEAsoft release.

Keywords: HEAsoft, xselect, extractor

Tuesday, May 09, 2006

filter time ut in xselect

Scott Porter pointed out that the filter time ut command didn't work in xselect. This was due to errors in parsing the input and in converting UT to MJD. The manual doesn't state the format required for time - it is the standard used in FITS files ie yyyy-mm-ddThh-mm-ss.sss.

Keywords : xselect, HEAsoft

Thursday, May 04, 2006

IoA models in next v12 release

Added Andy & Roderick's models kdblur, kdblur2, rdblur, and laor2 to v12 as standard models for the next release.

Keywords: HEAsoft, xspec

Wednesday, May 03, 2006


Sutton & Wandelt present a Bayesian approach to optimal image reconstruction in radio interferometry. They use Gibbs sampling an incorporate an Occam factor.

Seljak et al. combine Ly alpha forest power spectrum results with 3-year WMAP, SN, and galaxy clustering data to constrain cosmological parameters finding the simple 6 parameter model is preferred with no running of the spectral index.

Belanger et al. claim a detection of a 22.2 min quasi-periodic modulation in the X-ray flux during the 2004 Aug 31 Sag A* flare. If this transient event was at the marginally stable orbit this implies an angular momentum parameter of 0.22.

Zucker et al. find a new Milky Way dwarf satellite galaxy at a distance of ~220 kpc using the SDSS data release 5. This discovery suggests other such objects are to be found. The same group report another dwarf galaxy.

Eisenstein et al. argue that the baryon acoustic peak can be reconstructed to correct for non-linear degradation providing significant improvements in determining the distance scale from galaxy surveys.

Bryans et al. present collisional ionization equilibrium calculations using state-of-the-art calculations from DR and RR rates.

Lawrence et al. describe the UKIRT Infrared Deep Sky Survey, a 7-year survey begun in May 05.

Begelman et al. suggest that SS433 is a ULX observed from the side with the radiative output collimated by a massive outflow from the disk at a distance of ~10% of the disk edge.

Sanders & Fabian construct a model for cluster X-ray emission including resonance scattering for several thousand lines. They find that including resonance scattering does not account for the observed dip in the central abundance however allowing for absorbing material within the inner 40 kpc of NGC4696 significantly improves the quality of the fits to the Centaurus cluster.

Forman et al. present the first results from the Chandra 500 ksec observation of M87.

Parkinson et al. apply Bayesian model selection theory to the 3-year WMAP data using the CosmoNest code and find weak evidence for a perturbation slope != 1.

Reiprich compares the 3-year WMAP results with those from the HIFLUGCS galaxy cluster sample and concludes that the new results are in much better agreement than the 1-year WMAP results.

Lieu: Soft X-ray cluster excess and SZ effect

EUD seminar by Richard Lieu from U. Alabama in Huntsville. The first half of the talk was a summary of soft X-ray observations of clusters. There are currently no confirmed observations of redshifted OVII so no definite evidence for WHIM in emission. The soft excess in the outer parts of clusters may still be thermal due to low S/N however that in the brighter regions must be non-thermal - possibly due to relativistic electrons accelerated in shocks or perhaps more exotic sources such as neutralino decay. The second and newer work was looking at WMAP data to measure the SZ effect from a sample of 30 bright, nearby clusters. The results are mixed - some clusters are clearly detected at the expected level (Coma), some are ambiguous, and some appear not to be detected at the level predicted by the X-ray data. Summing all 30 clusters and comparing with sets of randomly selected fields finds a net signal consistent with background (driven by the ~1 degree first peak in the power-spectrum). This is difficult to understand but implies some fraction of clusters do not show an SZ signal of the strength expected. Gary Hinshaw of the WMAP team confirmed that they also detect an SZ effect from Coma and that the noise level in Richard's plots is correct.

Keywords: Clusters, WMAP, SZ