changelog

Log of changes to the RH code

--	Jun 14 2001: chemequil.c --> ChemicalEquilibrium
	
	Found bug that caused population numbers of molecules
	that include the active atom to be zero when starting from
	initial solutions other than OLD_POPULATIONS and OLD_POPS_AND_J.
	Put in appropriate conditional for the initialization of fn0.

--      Jun 20 2001: kurucz.c --> readKuruczLines

	Changed the way Kurucz linelists are read. The keyword
	KURUCZ_DATA in keyword.input now specifies a file that lists
	a series of file names, one line each, the each contain
	a list of Kurucz format line transitions. This allows a
	list of separate files to be read instead of the one specified
	previously under KURUCZ_DATA.
	
--      Jun 26 2001: initial_xdr.c --> initSolution

	Found bug in initialization of active molecule. The rate
	matrix mutex lock variable molecule->Gamma_lock was not
	initialized at all, which meant the program would hang in
	the case of an active molecule with NTHREADS set to a value
	of 2 or more.

--      Aug 15 2001: rhsphere:formal.c

	Fixed a typo in the evaluation of angle-averaged mean intensity J.
	This would cause problems for larger values of Ncore, so that
	J != B in the deep photosphere.

--      Aug 15 2001: rhsphere:feautrier.c

	Routine was not threadsafe. It still contained some static
	variables that got overwritten when running multiple threads.
	Replaced them with purely temporary variables which now get
	malloced and freed at the beginning and end, respectively.

--      Aug 15 2001: rhsphere:geometry.c, rhsphere:geometry.h

	Tidied up the code a bit. Renamed Ray->Nz and Ray->z to
	Ray->Ns and Ray->s to reflect more properly the fact that these
	are pathlength variables rather than vertical distance variables.
	
--      Oct  5 2001: rhsc2d:geometry.c

	Solved bug in geometry.c. The interpolation Stencils where not
	properly filled when ray was exactly vertical. Added a new
	function setVertStencil for this particular case. There are now
	three cases for mux > 0.0, mux = 0.0, and mux < 0.0.

--	Oct 17 2001: solvene.c --> Solve_ne

	New routine for calculating LTE electron density in atmosphere
	given T and Hydrogen density. Set SOLVE_NE = TRUE in keyword.input
	to activate. Routine is called from routine Background

--	Oct 18 2001: ltepops.c --> SetLTEQuantities

	New routine needed to calculate LTE populations and active
	atom collisional rates after the electron density is calculated.
	These quantities can no longer be calculated from readAtom
	when the electron density is unknown. Atomic models
	have to be read before the electron density is calculated.
	Specifically, the hydrogen model is needed to calculate partition
	function values.
	
--      Oct 18 2001: background.c --> Background.c

	Added additional keyword equilibria_only. When this is set
	the routine exits after calculating electron density,
	LTE populations, collisional rates for the active atom,
	and chemical equilibrium. Neede when called from solveray.

--	Oct 18 2001: backgropac_xdr.c --> backgrOpac

	Added calls to Solve_ne and SetLTEQuantities. See previous
	three items.
	
--	Nov  6 2001: chemequil.c --> ChemicalEquilibrium.c

	Modified routine so that it is no longer necessary to include
	detailed atomic models for all the constituents that can appear
	in molecules. It is now required that Kurucz partition functions
	are read in always (KURUCZ_PF_DATA is now KEYWORD_REQUIRED,
	accordingly). For bound-free background opacities it is still
	required to specify detailed model atoms in background.input
	
--	Nov 29 2001: zeeman.c --> determinate.c

	Slightly improved the atomic label dissection algorithm so
	that it now recognizes label that contain elements like e5DE
	in Fe model atoms.

--	Dec 13 2001: voigt.c --> VoigtHumlicek.c

	Older versions (Oct - Nov 2000) had a problem in the check
	for the four different domains. Instead of fabs(v) abs(v) was
	used, leading to erroneous choices for the domain. In later
	versions this was (fortuitously) implemented correctly.

--	Dec 13 2001: humlicek.c, voigt.c

	Cleaned up the expressions for complex arithmatic a bit.
	Changed the constants a[] and b[] in domain IV of VoigtHumlicek.c
	to the proper accuracy, consistent with the F90 version in
	humlicek_.f90.

--	Dec 19 2001: collision.c --> CollisionRate.c

	Fixed error that was result of SetLTEQuantities when proton
	density was required in calculation of collisional excitation.
	Routine was still referring to atmos.nH, which is already
	freed in distribute_nH. Now we make the proper reference through
	atmos.H->n.

--	Jun 12 2002: initscatter.c, rhf1d.c, rhsc2d.c, rhsc3d.c, rhsphere.c

	Fixed logical error that would prevent calculation of all four
	Stokes profiles in the case that STOKES_MODE = FIELD_FREE and
	N_MAX_ITER = 0. This setting could be used for calculating LTE
	profiles with polarization (set STARTING_SOLUTION = LTE_POPULATIONS
	of course). Now background scattering is only performed if
	N_MAX_ITER > 0.

--	Jul 09 2002: initspectrum.c --> initSpectrum.c

	Wavelength table is now read in XDR format to improve cross-platform
	operability.

--	Jul 15 2002: zeeman.c --> ZeemanStrength.c

	Added check on absolute value of \delta J. Previously strength
	s could be uninitialized if none of the three values of the case
	statement on dJ was satisfied. Such lines are now rejected
	for polarization (see below).

--	Jul 15 2002: readatom.c --> readAtom.c

	Reject polarization calculations for lines that have \delta J
	of larger than one. See previous entry.

--	Jul 31 2002: writecoll_xdr.c

	New routine to write collisional rates to file. Rates are
	written when COLLISION_OUTPUT is set to anything other than
	"none" in keyword.input. Also adapted readatom.pro to read
	this additional output.

--	Jul 31 2002: writedamp_xdr.c

	New routine to write broadening velocity and damping parameters
	for each line to file. Rates are written when DAMPING_OUTPUT is
	set to anything other than "none" in keyword.input. Also adapted
	readatom.pro to read this additional output.

--	Oct 01 2002: rhf1d/solveray.c

	The ray angle cosines mux and muy where improperly initialized.
	This lead to problems for full Stokes solutions as the angle
	with the magnetic field is undetermined.

--	Nov 18 2002: {rhf1d,rhsc2d,rhsc3d}/solveray.

	Improved the section reading the wavelength indices for the
	wavelengths for which chi and S have to be stored. Indices can
	now be arbitrarily broken over more than one line in ray.input.

--      Feb 05 2003: readBackground.c, writeBackground.c --> readj.c
	             MolecularOpacity.c --> opacity.c
	             Background.c --> background.c

	Added option for polarized transfer in background molecular
	lines. Streamlined storage of background opacities. These
	can now deal with emissivity of all four Stokes parameters.

--	Feb 12 2003: molzeeman.c

	New routines to compute Zeeman splitting paterns for molecular
	transitions. Includes routines to calculate strength and
	shift of Zeeman multiplets from quantum numbers, and a routine
	for the calculation of effective Lande factors. Both are capable
	of dealing with Hund's cases (a) and (b). Enhanced routine
	ReadMolecularLines.c for the Kurucz case to read additional line
	parameters.
	
--	Mar 31 2003: addtoGamma.c --> fillgamma.c

	Approximate operator now accounts for all four Stokes parameters
	in the case of polarized transfer in an atomic line.

--	Apr 14 2003: kurucz.c --> rlk_opacity.c

	Added support for polarized background lines. Also changes
	were made to routines in readj.c to store and retrieve background
	opacity values.

--	Nov 13 2003: backgrOpac.c --> backropac_xdr.c

	Fixed bug in check for existence of He atomic model that
	would lead to a bus error if He was not present in the
	emplyed background input file (by default background.input
	but user defineable with the -b command line option).

--	Jan 14 2004: readKuruczLines --> kurucz.c

	Fixed bug that caused problems with unitialized Q, U, and V
	profiles for lines that have no broadening specified (Grad =
	GStark = GvdWaals = 0.0). These lines now have polarizable = 
	FALSE.

--	Jan 14 2004: background.c, stokesopac.c, kurucz.c, opacity.c

	Added magneto_optical effects for LTE background lines. This
	can be set as for the active lines before by setting
	MAGNETO_OPTICAL = TRUE (the default) in keyword.input.

--	Jan 15 2004: rhsc2d/anglequad.c, rhsc3d/anglequad.c, readvalue.c

	Added support for new angle sets in 2- and 3-D code. You can
	now specify ANGLESET = SET_GL_NiXNa, where Ni and Na are the
	number of grid points in inclination and azimuth, respectively.
	Quadrature in inclination is Gauss-Legendre (hence the name of
	the set), in azimuth it is equidistant in angle. Number of rays
	is 2 x Ni x Na in 2-D, and 4 x Ni x Na in 3-D.
	
--	Jan 21 2004: rhf1d/formal.c, rhsc2d/formal.c

	Added support for background polarization due to scattering
	by Thomson and rayleigh processes. Set the keyword
	BACKGROUND_POLARIZATION = TRUE in keyword.input. The second
	radiation tensor J^2_0 is written to output file J20.out in
	this case.

--	Feb 25 2004: rhsc3d/formal.c, rhsc3d/geometry.c, rhsc3d/geometry.h
	  rhsc3d/shortchar.c rhsc3d/solvelong.c

	Added support for polarized radiation, including through
	background scattering. Also revised formal solution to be
	really three-dimensional, with long characteristics in cases
	where no nearest cell crossing in a horizontal plane occurs.
	Interpolation in the horizontal plane can be either linear
	or by cubic convolution. It is linear in vertical planes.

--	May 12 2004: profile.c, initial_xdr.c, readj.c, opacity.c,
	  {rhf1d,rhsc2d,rhsc3d,rhsphere}/formal.c

	Implemented an LIMIT_MEMORY option that writes profile data
	and mean intensity J to file instead of keeping it in memory.
	This strongly limits the amount of memory in multi-dimensional
	cases, at the cost of I/O overhead of course.

--	Sep 22 2004: writemolec_xdr.c, background.c, chemequil.c:

	Implemented NON_ICE (for Non- Instantaneous Chemical Equilibrium)
	option for keyword.input. It allows to by-pass the chemical
	equilibrium calculation by reading the file molecules.out
	with pre-calculated molecular concentrations. See IDL routine
	readmolecules.pro for details on how this file can be read
	and written in IDL. Allocation of atmos.nHmin is now in
	readatom.c instead of in chemequil.c. Enable by setting
	NON_ICE = TRUE (default is FALSE). KEYWORD_OPTIONAL

--	Jan 18 2005: readinput.c, multiatmos.c:

	Enabled option for irradiation from the TOP of the atmosphere in 1-D.
	Set keyword ATMOS_ITOP to the name of the file with irradiation
	values. This is a binary file in XDR with values for each wavelength
	and each angle (wavelength is the fastest index).

--	Jan 27 2005: rhsc2d/shortchar.c:

	Included proper boundary conditions at lower boundary when
	THERMALIZED is chosen.
	
--	Jan 27 2005: rhsc3d/shortchar.c:

	Included proper boundary conditions at lower boundary when
	THERMALIZED is chosen.
	
--	Jan 28 2005: rhsphere: readatmos.c, faeutrier.c geometry.h

	Added optional irradiation at outermost layer in the spherical
	version. Set keyword ATMOS_ITOP to the name of the file with
	irradiation values. This is a binary file in XDR with values
	for each wavelength. Infalling radiation is isotropic.
	
--      Apr 22 2005: readmolecule.c

	Added new format for molecular line list: KURUCZ_NEW
	This format allows for better specification of the
	vibrational levels.

--	Aug 11 2005: solvene.c

	Fixed bug in getKuruczpf where the log of Uk should
	be taken for stage == 1. Thanks to Sven Wedemeyer-B\"ohm
	for pointing this out.

--	Dec 23 2005: rhsc3d/solveray.c

	Fixed bug. Call to Bproject() was missing to fill the
	arrays with the projection angles of the field along
	the line of sight.
	
--	Feb 24 2006: rhsc2d/chartflow.c

	Fixed bug. When LIMIT_MEMORY is set J should be read from
	file wavelength by wavelength.
	
--      Feb 14 2007: writeinput_xdr.c

	Write Endianness of calculating host to input.out.
	The IDL routines on the analyzing host can now figure
	out what to do with the J.dat file, swapping bytes or not.

--	Oct 16 2007: zeeman.c

	Fixed bug in adjustStokesMode.c. Cannot free pointer
	line->phi in case input.limit_memory is set to TRUE.

--	Oct 24 2007: broad.c

	Fixed bug in Damping.c. When line->cStark is positive
	AND negative the Stark broadening should be computed.
	For negative values the broadening is fabs(line->cStark) *
	atmos.ne, in MULTI style (but note we have [ne] = m^-3,
	See also tools/convertatom.c)

--	Nov  5 2007: barklem.c, broad.c, readatom.c

	Added option for vd Waals broadening in Barklem fashion.
	Option BARKLEM in .atom file sets this option.
	
--	Nov  8 2007: barklem.c, kurucz.c

	Fixed bug. SWAP_VALUE erroneously used int for the values
	to be swapped. Now defined SWAPDOUBLE macro and put it in
	rh.h. Bug erroneously ed to integer values for neff1,2 when
	swapped.
	
--	Dec  7 2007: profile.c, sortlambda.c
	rhf1d/rhf1d.c, rhf1d/solveray.c,
	rhsc2d/rhsc2d.c, rhsc2d/solveray.c,
	rhsc3d/rhsc3d.c, rhsc3d/solveray.c,
	rhspehere/rhsphere.c

	Implemented routine getProfiles in profile.c to manage the
	calculation of line absorption profiles, only after all
	hydrogen populations are determined.

--	Jan  4 2008: rhf1d/hydrostat.c

	New routine to solve for hydrostatic equilibrium in 1-D
	plane geometry. Small changes in other routines to accommodate
	new routine. H2 is included in the calculation.

--	Aug 29 2008: rhsc2d/solvelong.c, rhsc3d/solvelong.c

	Fixed bug. In case I_uw < 0.0, the just computed value
	of I_uw would be used in the evaluation of linear interpolation,
	rendering the result meaningless. Thanks to Rene Holzreuter
	for finding this problem.

--	Dec 22 2008: readj.c, initial_xdr.c

	Fixed bug that prevented convergence in multi-threaded mode
	when angle-dependent PRD was used. Routines readImu and writeImu
	still employed fseek() and fread()/fwrite(), which is not thread-safe.
	Now pread/pwrite are used. This requires a file descriptor, rather than
	a file pointer.
	
--      Dec 23 2008: fixed various print formatting discrepancies
	
--      Apr 22 2009: Major revision to version 2.

	Multiple active atoms, and molecles, as well as wavetable.
	Input file background.input is replaced by two files:
	atoms.input and molecules.input
	
--      Oct 13 2009: nemetals.c, rhf1d/hydrostat.c

	Fixed bugs in implementation for hydrostatic equilibrium
	In nemetals.c f_n was not zeroed for every element, and in
	hydrostat.c atmos.H->ntotal was wrongly used as the total
	neutral hydrogen density (i.e., without subtracting np).

--	Jan 12 2010: profile.c

	Fixed a bug in the PRD case. When Profile() is called from
	adjustStokes() the PRD profile should be kept and not
	reinitialized. Added check for line->rho_prd and only
	initialize when it == NULL
	
--	Jan 14 2010: rhf1d/formal.c, rhsc2d/formal.c, rhsc3d/formal.c

	Fixed bug that resulted in segmentation fault when running
	with BAKGROUND_POLARIZATION and LIMIT_MEMORY both set to TRUE.
	In this case no memory for spectrum.J20 is allocated and J20
	is written to file per wavelength.

--	Mar  3 2010: ohchbf.c, background.c, readmolecule.c

	Added routines to add bound-free opacity from the OH and CH
	molecules in the UV. Pointers to the OH and CH molecular structure
	are stored in the atmos structure. This is done in
	readMolecularModels