These are small changes that will be incorporated in the next revision of the code. The version of the code is followed by hot fixes to that version.  These corrections will be included in later versions of the code. 

In the following, the corrections that should be made are indicated by line numbers within specific files.

These corrections where not present when the output from the downloaded test suite was created.  The test suite should be run before these are applied. 

fixes to version 06.02c

An assert can be thrown at line 176 or 177 of cont_ffun.c.  Remove these two asserts, which read as follows:

and replace with the following block

The power law command could result in a floating point exception when the vary option was included.  This was caused by incrementing the number of continua before the command was fully parsed.  To correct, edit the file parse_powerlawcontinuum.c and move the block of code between lines 92 and 99, which reads as follows:

to after the block that ends at line 153.  After the change the code will read as follows (this includes some code before and after the change):

Many thanks to Pavel Abolmasov for reporting this crash.  This change is only needed if you want to vary the options on the power law command.  06 Nov 10.

fixes to version 06.02b

The table stars command could cause crashes on certain types of Macs.  Edit file stars_rebinatmosphere.c and change the block starting at line 89 so that it reads as follows:

This need only be done if the table stars commands do not work on your Mac.  If does not affect the calculations other than preventing the crash on some Macs.

Many thanks to Dick Henry for uncovering the problem and to Peter van Hoof for fixing it.  2006 Aug 12.

The code could incorrectly detect level populations insanity and end with an assert.  Some important quantities are calculated with two independent methods, their results compared, and insanity declared if the results differ.  This is a basic concept in safe programming (discussed in this link).  The predicted densities of trace stages of ionization could differ by a fraction of a percent when the ionization fraction fell below one part in 10¹⁵ due to numerical imprecision.  The code detected this and threw an assert around line 138 of radius_increment.c.  The fix is to not do this test for ions which have extremely small fractional abundances. 

Edit the file radius_increment.c and change line 99 and add new lines so that they read as follows

Change line 123 and add two new lines of code:

Change line 142 and add new lines so that they read as

Change line 148 and add two lines so that it reads as

Finally change line 160 so that it reads as

This fix need only be made if the code throws an assert near line 138 of radius_increment.c.  The correction will not affect results.   Thanks to Joseph Meiring for reporting this problem.  2006 July 23

The mean H⁰ gas temperature that is weighted by the 21 cm opacity should be weighted over radius but was weighted with respect to volume.  This is given with the label T(<nH/Tspin> in the final printout.  To fix this edit file prt_final.c and change line 1515 from

to read as follows:

Thanks to Eric Pellegrini & Jack Baldwin for catching this problem.  2006 July 21.

fixes to version 06.02a

The UV FeII optical depths predicted by the large FeII model are not correct when gas becomes very cold.  The large model trims down the number of levels it computes when gas become very cold and the populations of highly excited levels becomes very small.  When this happened the code did not continue to update optical depths of lines between lower levels and these high levels that are not computed.  As a result the optical depth printed for the transition is too small.  The fix, which will be part of 06.02b, is to continue to update optical depths for such lines. 

2006 Jun 25

Thermal instabilities, followed by a thrown assert, could occur in a cosmic-ray dominated gas.  There are two fixes needed.  Edit file helike_recom.c and add the following after line 2260

The comes just before the statement that reads "if( !helike.lgNoRecombInterp )".

Second, edit file heat_sum.c and change the code at lines 445 to 450 from the old code

to the following:

Many thanks to Jennifer West for reporting this bug.  2006 June 07.

The punch FeII lines command only outputs emission from the lower levels in an optically thick cloud.  This is because the number of FeII levels is adjusted as the calculation progresses.  The number of levels considered at the end of the calculation may be small if the gas becomes cold.  To fix this edit atom_feii.c and change the variable FeII.nFeIILevel to FeII.nFeIILevelAlloc at lines 1900 and 1902.  After the change the lines will read as follows:

2006 May 20

The interpolated continuum was truncated at a certain cell number when very high resolution continua with very large numbers of points were used.  To fix change two files.  First edit edit parse_interp.c and change rfield.nupper to NCELL at line 260 and line 263.  Before the fix the lines will read a

if( npairs + 2 < rfield.nupper )
{
    /* zero out remainder of array */
    for( i=npairs + 1; i < rfield.nupper; i++ )
    {
        rfield.tNuRyd[rfield.nspec][i] = 0.;
    }
}
 

after the change this will read as

if( npairs + 2 < NCELL )
{
    /* zero out remainder of array */
    /*>>chng 06 may 17, same as above */
    /*for( i=npairs + 1; i < rfield.nupper; i++ )*/
    for( i=npairs + 1; i < NCELL; i++ )
    {
        rfield.tNuRyd[rfield.nspec][i] = 0.;
    }
}

The second change is in file cont_ffun.c.  Change rfield.nupper to NCELL-1 at line 160.  Before the fix the lines will read as

/* loop starts at second array element, [1], since want to
 * find next continuum energy greater than desired point */
/*for( i=1; i < rfield.nupper; i++ )*/
i = 1;
while( i< rfield.nupper && rfield.tNuRyd[rfield.ipspec][i]>0. )
{
    if( xnu < rfield.tNuRyd[rfield.ipspec][i] )
    {

after the change these lines will read as

Many thanks to Christophe Morisset for discovering this problem and to Swarna K Ghosh for helping clarify this fix.  2006 May 17

The abundances starburst command did not work with the vary option.  If the starburst abundances are to be varied during an optimization run then it will be necessary to change the format used for the abundances scale facror.  Edit routine abund_starburst.c and change line 317 from the following:

strcpy( optimize.chVarFmt[optimize.nparm], "ABUNDANCES STARBURST %9.5F LOG" );

to read as follows:

strcpy( optimize.chVarFmt[optimize.nparm], "ABUNDANCES STARBURST %f LOG" );

That is, change %9.5F to %f.  06 May 14

The FeII spectrum produced with the punch FeII continuum command is shifted by about 600 km/s.  It was initially only intended to make rough plots of the UV continuum and imprecise physical constants were used.  Additionally, vacuum rather than air wavelengths are used for all wavelengths.  The first problem can be fixed by editing atom_feii.c and changing the code around line 1500 to read as follows:

The next bug fix roll up with include this fix and also use air wavelengths for wl > 2000A   This is a very small effect and so not done here as a hot fix.  Many thanks to Yumihiko Tsuzuki for reporting this bug.  2006 April 10.

fixes to version 06.02

The atom H2 dissociation collisions command turned off all collisions within the large H₂ molecule.  Edit routine parse_atomh2.c and add an "else" before the second if within the collisions branch at line 142.  The changed code will read as follows:

2006 March 1

Approximations to Nussbaumer & Storey DR rates could diverge for very low temperatures.  The approximations in their published papers were not mean to be used outside certain temperature limits.  These rates are still the default but are being replaced by the Badnell values.  One test on the range of validity of the NS rates was left out, and the rates could strongly diverge for very low temperatures.  Edit routine ion_recom.c near line 214 and add a test on the value of tefac.  After the change the code will read as follows (code before and after the change is included for reference):

2006 Feb 27

The wavelength used to identify the continuum bands (given in bands_continuum.dat and entered into the emission line printout) gave the wavelength in Angstroms when it should have been in microns.  Edit routine cont_createpointers.c and add the following code to line1809 (code before and after the change is also shown)

2006 Feb 27

FeII may fail with insanity under some conditions when more than one iteration is performed.  To fix this edit routine atom_feii.c and add two lines, defining and setting the variable nFeII_old, and another line testing its value.  After these changes the code between lines 1203 and 1222 will read as follows (snippet begins and ends with neighboring lines of code to place in context):

2006 Feb 27

The code will generate a bogus warning when the density is set by an external function, such as fabden or dlaw.  It will complain that the density is changing by too much, even when it is constant.  The warning is benign and can simply be ignored.  To fix the code, edit routine pressure_change.c and change the code near line 572 to read as follows:

Many thanks to Valentina Luridiana for finding this problem.  2006 Feb 23.

fixes to version C05_07

The array index for the Rauch 2002 stellar atmospheres is incorrect when the default is used.  To fix, edit stars_rauch02.c and change line 802 so that it sets ipTeff to 14 rather than 23.  After the correction the line will read as follows:

Many thanks to Marcin Hajduk for finding this problem.  2006 Feb 08.

The continuum array could have insane energies when the continuum intensity changes dramatically at certain energies.  When this occurs the code will stop and declare insanity.  To fix, edit cont_setintensity.c and add an if test following line 156.  After the change the code (including two lines before and one line after the change) will read

Many thanks to Sergio Fajardo-Acosta for reporting this problem.  06 Feb 03.

The optimize temperatures command only recognized the keyword VOLUME and not volume.  Change the variable chCard to chCap on line 598 of parse_optimze.c.  After the change the line and the one before it will read

The recombination contribution to [N II] l5755 was incorrect.  Equation 1 from Liu et al. MNRAS 312, 585 was written in terms of the temperature scaled to 1e4 K.  The code works in terms of the real temperature so the rate was too large by a factor of 1e4^0.3 or 15.8489.  To fix this edit routine prt_lines_lv1_li_ve.c and change the code near line 365 to read as follows:

The code can crash with an fpe for very density models.  To fix, edit routine helike_level.c, and change line 783, variable iso.numLevels[ipHE_LIKE][nelem] to helike.numLevels[nelem].  After the change the code around this line will read:

Many thanks to Kirk Korista for reporting this problem.  05 Nov 18.

The grain emission continuum predicted by the punch grain continuum command is incorrect on second and later iterations.  The storage arrays need to be reset to zero.  Correct this by adding the following loop after line 314 of grains.c - after the correction the code will appear as follows:

if( gv.lgDustOn && gv.lgGrainPhysicsOn )
{
    long i;
    for( i=0; i < rfield.nupper; i++ )
    {
    /* >>chng 05 sep 08, zero on every iteration as per John Everett comment, PvH */
    gv.SilicateEmission[i] = 0.;
    gv.GraphiteEmission[i] = 0.;
}

for( nd=0; nd < gv.nBin; nd++ )
{
/* >>chng 97 jul 5, save and reset this
* save grain potential */

This only affects the punch grain continuum command.  Many thanks to John Everett for discovering this problem.  05 Sep 08

An assert could be thrown for cases where the He⁺⁺ abundance became very small.  Edit routine radius_increment.c and change ERR_CHK at line 112 from 1.001 to 1.01.  After the change the line will read as follows:

Many thanks to Yoshiki Matsuoka and Carmen Sanchez Contrera for reporting this problem.  05 Sep 04.

The set spectrum command incorrectly raised its argument to a power.  This results in an FPE for reasonable resolving power.  If you wish to use this command then file routine parse_set.c and change line 647 from the following:

so that it reads as follows - simply remove the "pow(10.," and the trailing ")". 

/* >>chng 05 aug 25, following had pow(number), so crashed for high resolv power
 * bug caught by Yan Changshou */
i = 5;
punch.cp_resolving_power[punch.cp_npun] = (float)FFmtRead(chCard,&i,INPUT_LINE_LENGTH,&lgEOL);
if( lgEOL )

Bugs come in groups of three.  The same mistake occurs at line 630 and 631 of the file.  Do the same edit - change the existing lines, which read as follows;

punch.cp_range_min[punch.cp_npun] = (float)pow(10.,FFmtRead(chCard,&i,INPUT_LINE_LENGTH,&lgEOL));
punch.cp_range_max[punch.cp_npun] = (float)pow(10.,FFmtRead(chCard,&i,INPUT_LINE_LENGTH,&lgEOL));

after the change it will read as follows:

punch.cp_range_min[punch.cp_npun] = (float)FFmtRead(chCard,&i,INPUT_LINE_LENGTH,&lgEOL);
punch.cp_range_max[punch.cp_npun] = (float)FFmtRead(chCard,&i,INPUT_LINE_LENGTH,&lgEOL);

Note that this command accepts only the energy in Rydbergs.

This only affect the set spectrum commands.  Many thanks to Yan Changshou for reporting this problem.  05 Aug 25

An incorrect array index is used to zero a continuum array.  Edit parse_interp.c lines 402-404, which originally read as

/* zero out remainder of array */
for( i=npairs; i < rfield.nupper; i++ )
{
    rfield.tFluxLog[rfield.ipspec][i] = 0.;
    rfield.tslop[rfield.ipspec][i] = 0.;
    rfield.tNuRyd[rfield.ipspec][i] = 0.;;
}

changing the occurrences of ipspec to nspec so that the block reads as follows

/* zero out remainder of array */
for( i=npairs; i < rfield.nupper; i++ )
{
    rfield.tFluxLog[rfield.nspec][i] = 0.;
    rfield.tslop[rfield.nspec][i] = 0.;
    rfield.tNuRyd[rfield.nspec][i] = 0.;
}

This has no effect on the physics but caused a memory access exception when the code is used as a routine.  Many thanks for Ralf Quast for finding and fixing the problem.  05 Jul 27

 An assert may be thrown in helike_cs on some system due to numerical imprecision.  Edit helike_cs.c and change line 1647 from

ASSERT( aveRadius >= Bohr_rad );

to read as follows (this includes a few lines of code around it to place it in context)

ASSERT( aveRadius < 1.e-4 );
ASSERT( aveRadius > 3.9/LIMELM * Bohr_rad );
global_an = aveRadius;

This has no effect on results but will cause thrown asserts on some systems.  05 Jul 14

Last changed 02/03/07.
Return to the Cloudy Home Page.
Copyright 1978-2006 Gary J. Ferland