GitLab

Maser

See merge request !9

This adds a commented-out debug print statement to the
routine that calls esc_CRDwing_1side.  The report
extends over the range of interesting optical depths.  The
range extends from close to the lower
limit of the maser optical depths we can handle due to
floating point limits to a high enough value well into
the positive tau case.

Note that this printout occurs for whatever damping constant
the routine is called the first time.  The print code
exits after doing one loop.  It does not test a range
of damping constant and would not illustrate the problem
being fixed on this branch..

An assert for positive escape probability has been
added to esc_CRDwing_1side.  That assert would have made
it much easier to trace down the original thrown assert.

Add tfile to list of ignored files. 

Improve the usefulness of error print when maser optical depth
becomes too small by giving the smallest optical depth considered.

Correct algebra on line 197 to preserve intent
of original source.

A sim that used all levels of the new Fe II and Fe III
datasets threw an assert at
 Failed: escgrd_v > 0.
 It happened in the file ../rt_escprob.cpp at line number 330

It was due to a long-standing bug/oversight.  The complete
redistribution with wings escape probability did not protect
against masers and a negative escape probability resulted.
We now use the appropriate routine when the optical depth is
negative.  The sim follows.

This oversight has been present for a long time.  It is
surprising that it has only now bitten.  This is likely
because we now have atomic models that are much larger
than in the past.

---

````
set save prefix "assert"
TURBULENCE= 0.600000 LOG 3.000000
species "Fe+" dataset="Smyth19"
species "Fe+" levels=all
species "Fe+2" levels=all
stop column density 23
stop temperature exceeds 1e7 K
print lines sort wavelength
print lines column
print lines faint 1
iterate
print last
hden 14
table SED "AGN_Jin12_Eddr_mid.sed"
phi(H) 23
normalize to "Inci" 1215 scale factor = 1215
````

Refactor dynamical iteration test into method

See merge request !8

Gargi Shaw docs CR background and PAH interactions

See merge request !7

A plethora of syntaxes exists across the code for testing whether an
iteration is to be considered an initial relaxation iteration in dynamical
models.  For consistency, it is better for the task to be encapsulated in
a function, and to call that instead.

The only remaining instance of the check that is not replaced by the
method call occurs in source/conv_itercheck.cpp, which applies to
advective models.  The intent there is to guarantee that one iteration is
done after the initial ones with dynamical sources, and without declaring
convergence prematurely.

https://groups.google.com/g/cloudy-dev/c/3jd_sNqfCtE

This updates two parts of Hazy 1 and adds a citation to
Shaw & Ferland (2021arXiv210103732S).  The discussion
reviews how the PAH abundance affects the cosmic ray secondary
ionization rate and gives the recommended Galactic background
for the case where PAHs are present.  Our recommended default
CR background rate remains the Indriolo+ value but we give
the Cloudy command to change this rate to our recommended value
when PAHs are present.

Docs

See merge request !6

Silence Oracle compiler warning

See merge request !5

Gargi Shaw investigated why we have warnings about duplicate
reactions in the development version.  These still occur in
release versions but have been silenced.  Her document
was converted to markdown and is posted on the GitLab wiki
and is included here to make it easy to find.

Update monitors of two sims to new timestep logic

See merge request !4

Incorporate changes to:

- Revert adaptive timestep logic

- Add 'save dt' command

- Prevent FPEs in cooling sims with external radiation fields

- Add columns for advective terms to save cool/heat

- Force consistency of equilibrium solves in non-eq sims (bugfix)

- Ensure consistent policy for equilibrium in non-eq runs

The notation in the description of the U and xi forms
of the ionization parameter was slightly inconsistent,
which caused problems for one user.  These are updated.
There is no change in content.

The TexShop app on macOS cares a number of helper files
to cross-reference the source and compiled text.  These
have been added to .gitignore

Code for deciding whether to include non-equilibrium terms in population
solves was repeated across the codebase, allowing for inconsistencies to
creep in.

Establish and use method to ensure consistent treatment.

https://groups.google.com/g/cloudy-dev/c/kFDQe9yq0Ts

The command 'set dynamics population equilibrium' forced the solves for
the populations of neutrals and ions to their steady-state setup, but it
did not do so for molecules.  Add checks for the command's status in the
molecular network matrix setup to keep equilibrium treatment in
non-equilibrium runs consistent across the different branches of the code.

Reported-by: M. Chatzikos
https://groups.google.com/g/cloudy-dev/c/kFDQe9yq0Ts

additional entries for .gitignore

See merge request !3

## The documentation

By default, we build both documentation and executables in folders
below master.  The code's documentation is in the
docs/latex and we generate PDF copies of the three parts
of Hazy and the QuickStart.  The default .gitignore does
ignore the latex helper files so only the final PDF needs be added.

## The executable

We build in the sys_* directories below source.  The default
.gitignore does ignore the compiler-generated files so only
a few implementation-dependent configuration files needed to be added.

## The test suite

The tsuite directory has several folders, a large number of models,
and an even larger number of files created during the test suite.
My workflow involves regularly testing with gcc and llvm and on
occasion with the Intel compiler.  All three versions of the
test suite should be available.  I created three symbolic link
folders, gtsuite, ltsuite, and itsuite, to hold the test suite
for the gcc, llvm, and icc compilers.  These three directories
are ignored.

## The documentation

By default, we build both documentation and executables in folders
below master.  The code's documentation is in the
docs/latex and we generate PDF copies of the three parts
of Hazy and the QuickStart.  The default .gitignore does
ignore the latex helper files so only the final PDF needs be added.

## The executable

We build in the sys_* directories below source.  The default
.gitignore does ignore the compiler-generated files so only
a few implementation-dependent configuration files needed to be added.

## The test suite

The tsuite directory has several folders, a large number of models,
and an even larger number of files created during the test suite.
My workflow involves regularly testing with gcc and llvm and on
occasion with the Intel compiler.  All three versions of the
test suite should be available.  I created three symbolic link
folders, gtsuite, ltsuite, and itsuite, to hold the test suite
for the gcc, llvm, and icc compilers.  These three directories
are ignored.

Superfluous spaces in the Makefile if-statement caused the script
gitversion.sh to not be executed, except from a sys_* directory.
Remove.

Adjust printed Cloudy version to Git

See merge request !1

Clean out deprecated versioning variables

See merge request !2

Switching to a new workflow has made superfluous variables that tell if
the code is in beta-version, or on a release branch.  Delete them.

As part of the switch to git, a new workflow is to be established.
The previous practice of releasing every few years would better be
replaced by a scheme where we release every few months.  We seem to be
gravitating toward a model where feature branches are merged onto
master, which is merged onto a release branch for periodical releases.

The Cloudy version reported in the .out file treats the release branch
as any other, which would be opaque to users.  The version number stored
as a tag to the appropriate git commit is now employed as the Cloudy
version in user-facing runs.

Reported-by: M. Dehghanian
https://groups.google.com/g/cloudy-dev/c/N_tKJTriDDw

The output of the save cooling/heating commands treats advective heating
and cooling terms ambiguously.  It excludes them from the total when
reporting fractional contributions, but reports them as contributing
agents.  As a result, advective terms with fractions >>100% are commonly
encountered in cooling simulations.

Add columns to report these contributions, but exclude them from the
list of cooling/heating agents.

https://groups.google.com/g/cloudy-dev/c/69NLm1Ru_cs

Including external radiation fields (e.g., the CMB) in cooling sims
may lead to FPEs.  The cumulative radiation field intensity involves
multiplication by a factor that may be near the maximum value a float
can hold.  This is not a problem with diffuse radiation.  However, when
the CMB is included, the product exceeds that limit and an FPE occurs.

The adopted bugfix hides in a new class the details of how the
integrated flux is accumulated, while preserving how the flux vectors
are accessed.  The latter is mainly useful for the instantaneous flux.

As an added feature, the time-weighted mean flux is also computed.

Reported-by: M. Chatzikos
https://groups.google.com/g/cloudy-dev/c/GSWoDN_hxqc