|
New issues from Xspec 12.3.1
Frequently Asked Questions
Mac OS X Users: When built with g77 3.4.3, PLT's file output commands (WData, WEnviron, and
WHead) are not working properly when PLT is run from XSPEC12 (by XSPEC's iplot command). They may result
in blank output files (with some of the information instead appearing in a
file named "fort.49"). If this causes a problem, we recommend installing g95 in
place of g77.
List of currently outstanding issues.
Patches:
latest update
12.3.1ap Nov 26, 2007
Requires patch installer
v2.3
To apply patches, you'll need to have downloaded and already built the source code distribution
of XSPEC. For details and more information about the patch installer, click
here.
Otherwise, just follow the instructions below.
To make sure your XSPEC installation is up to date with all listed patches:
1. If you don't already have the patch installer v2.3
tcl script, download and place it in the XSPEC src directory. This directory
should be at .../headas/Xspec/src from the source code distribution.
2. Download the latest patchfile into
the same directory. There is no need to unzip/untar the patchfile
since the installer will do that.
3. (NOTE: This assumes you
have already set your HEADAS environment variables with the headas-init script,
as you would if you were running XSPEC, and that you have not modified the structure of
your XSPEC src directories.)
At the command line, launch the script by typing
'tclsh patch_install_2.3.tcl',
and that's all. The script should detect which patches you've installed previously, copy
the new code files to the necessary directories, determine whether complete or only partial
rebuilds are necessary, and finally perform the rebuilds.
Numbers listed in red indicate that the patch modifies the interface
to one or more of XSPEC's libraries. These patches will generally cause longer rebuilds. If you
have a local model library, it is recommended (though not always necessary) that you rebuild it
with initpackage after applying red patches.
| 12.3.1a |
A very small correction to the calculated value in XSPEC's Ln(Gamma(x)) function.
This adds one more term to the series used to make the calculation.
Report added on Mar 12, 2007.
The fix is to replace headas/Xspec/src/XSUtil/Numerics/
Gamma.cxx,
and rebuild xspec.
|
| 12.3.1b |
Unable to run eqwidth on a model unless it has a fit-capable
response associated with it (ie. a genuine response or dummyrsp with
channels specified). It ought to run on a model regardless of its
response (or lack thereof). Our thanks to Ruben Farinelli for pointing this out.
Report added on Mar 12, 2007.
The fix is to replace headas/Xspec/src/XSModel/GlobalContainer/
ModelContainer.cxx,
XSUser/Handler/
xsEqwidth.cxx,
and rebuild xspec.
|
| 12.3.1c |
This patch cleans up the output of the list of model parameters when
mix models are in use. Redundant reporting of mix parameters belonging
to models for higher numbered data groups is now suppressed. Only one set of
mix parameters will be printed, at the end of the list.
Report added on Mar 12, 2007.
The fix is to replace headas/Xspec/src/XSModel/Model/
Model.cxx,
XSUser/Handler/
xsNewpar.cxx,
xsShow.cxx,
and rebuild xspec.
|
| 12.3.1d |
The compPS model can generate a segmentation fault when using
a response matrix with > 5000 energy bins. Our thanks to Hiromitsu Takahashi
for reporting this.
Report added on Mar 14, 2007.
The fix is to replace headas/Xspec/src/XSFunctions/
xscompps.f,
and rebuild xspec.
|
| 12.3.1e |
Xspec's own hardcopy command is not working. (This is not the PLT hardcopy command
accessible through iplot.) Our thanks to Jean Swank
for pointing this out. Report added on Mar 21, 2007.
The fix is to replace headas/Xspec/src/scripts/
hardcopy.tcl,
and rebuild xspec.
|
| 12.3.1f |
An improvement for the setplot area plotting option. The
effective channel area is now factored in by dividing the spectrum by
the predicted folded spectrum from a flat model.
Report added on Mar 21, 2007.
The fix is to replace headas/Xspec/src/XSModel/Data/Detector/
MultiResponse.cxx,
RealResponse.cxx,
and rebuild xspec.
|
| 12.3.1g |
An enhancement patch which adds ctrl-c breaking to the chain run command. If a
chain run is interrupted this way the chain will not be loaded, though the
incomplete chain file will still remain on disk.
Report added on Mar 30, 2007.
The fix is to replace headas/Xspec/src/XSFit/MCMC/
Chain.cxx,
.h,
ChainManager.h,
XSUser/Handler/
xsChain.cxx,
and rebuild xspec.
|
| 12.3.1h |
Improvements made to the display of spectra information. The show data command
will now always list the spectra in order of spectrum number rather than alphabetically by file name.
show file will still list the data by order of file names. Also, a couple
of missing features from xspec11 have now been implemented. show pha is now fully
functional and will display data, error, and model values for each channel. For chatter > 10, show data
will now display a map of the grouping information.
Report added on Mar 30, 2007.
The fix is to replace headas/Xspec/src/XSModel/Data/
DataSet.cxx,
.h,
EinsteinData.h,
OGIP-92aData.h,
SfData.h,
SpectralData.cxx,
.h,
XSUser/Handler/
xsShow.cxx,
Xspec/Integral/
SPI_Data.h,
and rebuild xspec.
|
| 12.3.1i |
This makes a couple of minor fixes needed to make XSPEC12 compatible
with the Intel icpc compiler.
Report added on Mar 30, 2007.
The fix is to replace headas/Xspec/src/XSModel/DataFactory/
XspecRegistry.h,
XSUser/UserInterface/
xstcl.h,
and rebuild xspec.
|
| 12.3.1j |
Bug fix for table models when using multiple data groups AND
table model parameters in the higher groups are NOT linked to the lowest group.
XSPEC12 was erroneously still using the table parameter values of the lowest group when
calculating the model for the higher data groups. Our thanks to Pierre-Olivier Petrucci
for pointing this out.
Report added on Mar 30, 2007.
The fix is to replace headas/Xspec/src/XSModel/Model/Component/
AddTableComponent.cxx,
MulTableComponent.cxx,
TableComponent.cxx,
.h,
XSModel/Model/Component/OGIPTable/
OGIPTable.cxx,
.h,
and rebuild xspec.
|
| 12.3.1k |
When running fakeit to create a spectrum that is NOT based on an
original data set, XSPEC should check the response file's F_CHAN column info
(typically the keyword TLMIN4) to set the lowest allowed channel number of
the fake spectrum. It had been just defaulting to 1 in this case. Our thanks
to Ken Ebisawa for pointing this out.
Report added on Mar 30, 2007.
The fix is to replace headas/Xspec/src/XSModel/Data/
OGIP-92aData.cxx,
XSModel/DataFactory/
OGIP-92aIO.cxx,
.h,
Xspec/Integral/
SPI_Data.cxx,
and rebuild xspec.
|
| 12.3.1l |
Bug fix for an obscure model command case: If user enters
"model < name > < ... >" where < name > refers to an already
loaded named model, and < ... > is any text other than "none", "active" or
"inactive", a segmentation fault may occur.
Report added on Apr 04, 2007.
The fix is to replace headas/Xspec/src/XSUser/Handler/
xsModel.cxx,
and rebuild xspec.
|
| 12.3.1m |
delcomp is not working properly for certain cases: Running
delcomp on a model that doesn't exist causes a segmentation fault. It is
not retaining the previously entered range for substitution when using the wildcard
'*' specifier. Also, when a hyphenated range of parameters is specified, only the
first 2 parameters in the range are being deleted. Our thanks to Suresh Sivanandam
for pointing this out.
Report added on Apr 12, 2007.
The fix is to replace headas/Xspec/src/XSUser/Handler/
xsDelcomp.cxx,
and rebuild xspec.
|
| 12.3.1n |
For type-II OGIP files, XSPEC crashes when faced with the special case of
QUALITY or GROUPING information given in a scalar column format.
Our thanks to David Band and Chris Shrader for pointing this out.
Report added on May 14, 2007.
The fix is to replace headas/Xspec/src/XSModel/DataFactory/
OGIP-92aIO.cxx,
and rebuild xspec.
|
| 12.3.1o |
On x86_64 Linux platforms, Fortran model functions which make use
of the r1mach routine for retrieving machine constants (e.g. bmc, equil, nei,
pshock, and sedov models) are exiting with invalid argument errors. Our thanks to
Terry Gaetz for pointing this out.
Report added on May 18, 2007.
The fix is to replace headas/Xspec/src/XSFunctions/
r1mach.cxx,
and rebuild xspec.
|
| 12.3.1p |
When running the compPS model with the rel_refl parameter (par 11)
greater than 0.0, a crash may occur due to a memory access error. Our thanks to
Dirk Pandel for pointing this out.
Report added on May 21, 2007.
The fix is to replace headas/Xspec/src/XSFunctions/
xscompps.f,
and rebuild xspec.
|
| 12.3.1q |
The srcut and sresc additive model components cause a
segmentation fault upon loading. Our thanks to Heather Matheson for pointing this out.
Report added on May 30, 2007.
The fix is to replace headas/Xspec/src/XSFunctions/
xsFortran.cxx,
srcut.f,
sresc.f,
and rebuild xspec.
|
| 12.3.1r |
Minor memory leak for certain cases of syntax when requesting
one-line help summaries.
Report added on Jun 15, 2007.
The fix is to replace headas/Xspec/src/XSUser/UserInterface/
unknow.cxx,
and rebuild xspec.
|
| 12.3.1s |
The tclout and help commands shouldn't activate XSPEC's autosave
facility. Writing to an autosave file greatly slows down tclout, and it
removes the ability of undo to recover the state prior to the tclout or help call.
Our thanks to Jeremy Sanders for pointing this out.
Report added on Jun 15, 2007.
The fix is to replace headas/Xspec/src/XSUser/Handler/
xsHelp.cxx,
xsTclout.cxx,
and rebuild xspec.
|
| 12.3.1t |
When displaying data file information, the label "Source File" should be
replaced with the more precise description "Spectral Data File."
Our thanks to Pat Broos for pointing this out.
Report added on Jun 22, 2007.
The fix is to replace headas/Xspec/src/XSModel/Data/
SpectralData.cxx,
src/XSUser/Handler/
xsShow.cxx,
and rebuild xspec.
|
| 12.3.1u |
A bug fix to the diagrsp command. For a response matrix which
contains a channel bin whose energy range is entirely enclosed by a single
input energy bin, diagrsp is setting the channel's diagonal response
to 0.0. This bug does not affect the similar dummyrsp command.
Our thanks to Pat Broos for pointing this out.
Report added on Jun 22, 2007.
The fix is to replace headas/Xspec/src/XSModel/Data/Detector/
UserDummyResponse.cxx,
and rebuild xspec.
|
| 12.3.1v |
Several bug fixes for the goodness command. The sim option
was still being treated as nosim. Goodness results were not
converging with increasing trials when used with chi-square fits on data with
backgrounds and non-Poisson errors. Also not properly treating the case where
background exposure time differs from spectrum exposure time. With these
fixes, goodness now follows the same algorithm as fakeit in
choosing to apply Poisson randomization to data with Poisson errors and
Gaussian randomization to data with non-Poisson errors.
Report added on Jun 27, 2007.
The fix is to replace headas/Xspec/src/XSFit/Fit/
Fit.cxx,
.h,
src/XSModel/Data/
SpectralData.cxx,
.h,
src/XSUser/Handler/
xsGoodness.cxx,
and rebuild xspec.
|
| 12.3.1w |
This adds the improve command. Essentially a wrapper for
calling the Minuit package's improve function, this will trigger
a search for additional fit-statistic local minima. (Can be used only
after fitting with a Minuit fit method, such as migrad.)
This capability existed in XSPEC v11 but was unimplemented in v12 till now.
Our thanks to Terry Gaetz for pointing this out.
Report added on Jun 27, 2007.
The fix is to replace headas/Xspec/src/XSFit/Fit/
FitMethod.cxx,
.h,
src/XSFit/FitMethod/LevMarq/
LevMarq.cxx,
src/XSFit/FitMethod/Minuit/
Minuit.cxx,
.h,
src/XSUser/
Makefile,
src/XSUser/Global/
XSGlobal.cxx,
src/XSUser/Handler/
xsImprove.cxx,
xsNot.cxx,
and rebuild xspec.
|
| 12.3.1x |
XSPEC is not allowing the loading of table models which contain
parameters whose values are stored in a scalar VALUES column
rather than a vector column.
Report added on Jun 27, 2007.
The fix is to replace headas/Xspec/src/XSModel/Model/Component/OGIPTable/
OGIPTable.cxx,
and rebuild xspec.
|
| 12.3.1y |
This makes improvements to the output display for a couple of obscure cases.
Warning messages are now given when XSPEC needs to change a 0.0 AREASCAL or
BACKSCAL value (to 1.0, but not for bad quality channels). Also, when using a
diagonal response from the diagrsp command, show data has been
displaying a "Using Auxiliary Response (ARF) File: " string even when it has no
ARF file name to report. Our thanks to Pat Broos for pointing this out.
Report added on Jun 29, 2007.
The fix is to replace headas/Xspec/src/XSModel/Data/
OGIP-92aData.cxx,
src/XSModel/Data/BackCorr/
OGIP-92aBackground.cxx,
and rebuild xspec.
|
| 12.3.1z |
A conflict can arise between XSPEC11 and XSPEC12's usage of the ~/.xspec directory.
If this directory has not yet been created and XSPEC11 is started first, it can cause
the disabling of XSPEC12's autosave and undo commands. Our thanks
to Craig Markwardt for pointing this out.
Report added on Jun 29, 2007.
The fix is to replace headas/Xspec/src/XSUser/Global/
Global.cxx,
and rebuild xspec.
|
| 12.3.1aa |
The renorm command has no effect when one or more loaded spectra have
no active models associated with them. Our thanks to Pat Broos for pointing
this out.
Report added on Jul 12, 2007.
The fix is to replace headas/Xspec/src/XSFit/Fit/
StatMethod.cxx,
and rebuild xspec.
|
| 12.3.1ab |
A bug in the handling of the active and inactive options of
the model command can cause a crash in some cases when used in tandem with the
renorm or flux commands. Our thanks to Pat Broos for pointing this out.
Report added on Jul 12, 2007.
The fix is to replace headas/Xspec/src/XSModel/GlobalContainer/
ModelContainer.cxx,
.h,
src/XSUser/Handler/
xsModel.cxx,
and rebuild xspec.
|
| 12.3.1ac |
This converts several Fortran files used during Levenberg-Marquardt
fitting to C++. The primary observable benefit is improved error handling
when using table models. Previously, out-of-bounds table parameter errors during a
fit caused XSPEC to exit.
Report added on Jul 12, 2007.
The fix is to replace headas/Xspec/src/XSFit/
Makefile,
src/XSFit/Fit/
Fit.cxx,
src/XSFit/FitMethod/LevMarq/
InvertSVD.cxx,
LevMarq.cxx,
.h,
and rebuild xspec.
|
| 12.3.1ad |
After using the recalc option of the chain command,
any newly created chain files will contain only 1 value repeated throughout
the length of the chain. Our thanks to Mark Bautz for pointing this out.
Report added on Jul 16, 2007.
The fix is to replace headas/Xspec/src/XSFit/Fit/
FitMethod.h,
and rebuild xspec.
|
| 12.3.1ae |
gain fit parameters are not able to be frozen individually during
a fit. Our thanks to Eric Miller for pointing this out. (Also with this patch,
the default parameter settings when prompting for gain
fit parameters will more closely resemble v11's behavior.)
Report added on Jul 18, 2007.
The fix is to replace headas/Xspec/src/XSFit/Fit/
Fit.cxx,
src/XSModel/Parameter/
ResponseParam.cxx,
src/XSUser/Handler/
xsGain.cxx,
and rebuild xspec.
|
| 12.3.1af |
newpar crashes with a divide-by-zero error when called on a
parameter in a non-existing model. Our thanks to Jeremy Sanders for pointing this out.
Report added on Jul 23, 2007.
The fix is to replace headas/Xspec/src/XSUser/Handler/
xsNewpar.cxx,
and rebuild xspec.
|
| 12.3.1ag |
When entering parameters from a script for a model assigned to multiple data groups,
if the skip specifier "/*" is entered for ANY parameter, it is being applied to ALL parameters
beyond the first data group. It should only be applied to the parameter for which it is
entered and those that follow.
Report added on Jul 23, 2007.
The fix is to replace headas/Xspec/src/XSModel/Model/
Model.cxx,
and rebuild xspec.
|
| 12.3.1ah |
Setting the xs_echo_script flag to 0 is not disabling the command
echo to the console when running scripts. Our thanks to Pat Broos for pointing this out.
Report added on Jul 31, 2007.
The fix is to replace headas/Xspec/src/XSUser/UserInterface/
echcom.cxx,
and rebuild xspec.
|
| 12.3.1ai |
After running fakeit, and until another data command has
been run, a segmentation fault may occur whenever models are displayed.
This affects the fit command as well as show model.
Report added on Jul 31, 2007.
The fix is to replace headas/Xspec/src/XSUser/Handler/
xsFakeit.cxx,
and rebuild xspec.
|
| 12.3.1aj |
When the flux command is run with an energy range that does
not overlap the response energy range, a segmentation fault occurs.
Our thanks to Jeremy Sanders for pointing this out.
Report added on Aug 13, 2007.
The fix is to replace headas/Xspec/src/XSFit/Fit/
Fit.cxx,
and rebuild xspec.
|
| 12.3.1ak |
This fixes dummyrsp so it can be applied to any model for a given spectrum,
and can now take advantage of XSPEC's recently added multiple sources capability
in the same way as the response command. Previously dummyrsp only worked properly
for the standard single source cases. This also fixes a Solaris specific bug,
which could cause NaN calculation errors when folding a model with a dummy response.
Report added on Aug 20, 2007.
The fix is to replace headas/Xspec/src/XSModel/Data/
SpectralData.cxx,
src/XSModel/Data/Detector/
UserDummyResponse.cxx,
src/XSModel/GlobalContainer/
DataContainer.cxx,
src/XSUser/Handler/
xsDummyrsp.cxx,
xsResponse.cxx,
and rebuild xspec.
|
| 12.3.1al |
Using delcomp to remove a table model component causes a
segmentation fault. If editmod is used on a model with a table component,
and the component is written with an abbreviated form of "atable", "mtable", etc.,
editmod erroneously considers it a modified component.
Report added on Aug 20, 2007.
The fix is to replace headas/Xspec/src/XSModel/Model/
Model.cxx,
src/XSModel/Model/Component/
Component.cxx,
and rebuild xspec.
|
| 12.3.1am |
A couple of output formatting improvements: lumin output should
be in standard "d.ddde+EE" format and not written as "d.ddd x 10^EE". The
margin command output should switch from fixed to scientific format
for a wider range of exponent values. Our thanks to Jeremy Sanders for
pointing this out.
Report added on Aug 20, 2007.
The fix is to replace headas/Xspec/src/XSFit/MCMC/
MarginGrid.cxx,
src/XSModel/Model/
Model.cxx,
and rebuild xspec.
|
| 12.3.1an |
For the tclout plink option, it is more useful to output the
link string in a format which can be used as input for other parameter
commands, such as newpar. Our thanks to Jeremy Sanders for pointing this out.
Report added on Oct 15, 2007.
The fix is to replace headas/Xspec/src/XSUser/Handler/
xsTclout.cxx,
and rebuild xspec.
|
| 12.3.1ao |
When a fakeit run is using a response with first legal channel
number (TLMIN4) = 0, and is NOT based on an already existing spectrum,
the last channel is erroneously ignored and always ends up with 0 counts.
Our thanks to David Morris for pointing this out.
Report added on Oct 31, 2007.
The fix is to replace headas/Xspec/src/XSModel/Data/
SpectralData.cxx,
and rebuild xspec.
|
| 12.3.1ap |
The gabs model is wrong and gives results dependent on the energy
bin size. Our thanks to Takao Kitaguchi for pointing this out.
Report added on Nov 26, 2007.
The fix is to replace headas/Xspec/src/XSFunctions/
xsgabs.f,
and rebuild xspec.
|
Older Versions:
|
Xspec Home Page