ChangeLog

2012/08/27  Fix bug that caused alpha to be set to beta when
	    both were fixed.  Bug found by Jeremy Neveu.
2010/06/03  Check that input SN correlation coeffs are valid
2010/05/21  Allow curvature specification in omprior, thanks to
	      Shailendra Vikas
2010/01/29  Replace WMAP5 with WMAP7 auxiliary
2009/10/23  Use empty universe relation to interpret pecvel,
             switch to double precision pretty much everywhere.
2009/09/16  Add support for w fit with fixed curvature and omega_m
2009/07/14  Update Percival BAO numbers
	    Add reading of FITS cov matrices (experimental)
2009/07/07  Made to work with 11/083 version of MKL.
	    Doc fix suggested by Stephane Blondin.
2009/03/10  Fixed lots of bugs in covMatrix woodbury if you weren't
	     using ATLAS/MKL/accelerate
2008/04/29  Fixed bug in acoustic scale calulation.  Added Percival BAO
             constraints.
2008/04/17  Fixed Woodbury Cov matrix compile errors on Mac OSX
2008/04/03  Support for Komatsu et al. (2008) parameterization of
             w(a).  Re-inclusion of Woodbury Cov matrix stuff.
	     Both covmatrix classes refactored substantially.
2008/03/12  Add beta version of WMAP5 shift parameter constraints
2008/01/31  Added simple om prior code.
2008/01/23  Fixed a bug in the BAO constraint where wa was being ignored
             when trying to fit, well, wa. (Thanks to Chris Burns for
             catching this)
2007/12/17  Added support for having all the cosmological parameters fixed
2007/12/04  Support for the OS X Accelerate framework.
2007/10/10  Minor changes.  made indivchi work again.
2007/07/20  Add support for different dispersions for different datasets.
            Added SHOWPROGBAR to show the progress bar without getting
             the full set of VERBOSE output.
            Added support for second WMAP shift parameter (l_a), although
             this requires a prior on H_0.
            Fixed a bug in om/w0/wa WMAP constraints.
            Runfitter can be given a list of parameter files now.
2007/06/18  For covariance matrix case, use BLAS to do chi^2 evaluation
             if available.  Gives a factor of ~2 speedup
2007/05/01  Some bug fixes.  get_lumdist supports multiple redshifts.
             Progress bars added for most fits.  Added width-colour,
             etc. type covariance matricies.
2007/03/13  Pre-compute luminosity distances when doing covariance
             matrix approach, trading speed for memory requirements.
             These are quite manageable for 2D, but may be a problem
             for 3D.
2007/02/15  New covariance matrix class in double precision, has 
             tie-ins for using LAPACK, which makes inversion _much_
             faster.
            Fixed some of the test programs (i.e., make check)
2007/02/09  Make wmap3 and bao constraints support 1 and 3D values,
             updated docs.
2007/02/08  Fixed bug in w_a expression, fixed bugs in contourshow,
             made the fits converters work again.
2007/02/07  Major changes:
             w_a support added
             All support for analytic marginalization over alpha/beta
              removed, since it was biasing.
             scriptM is always analytically marginalized over.
             Many organization changes.
             cosgrids is more encapsulated now.
             Errors on alpha/beta now handled properly in all cases,
              no additional cost to output them.
             Can hold more paramerters fixed in consistent fashion.
             New public interface to luminosity distances.
             Full covariance matrix support, although it hasn't been
              tested.
2006/12/11  Added WMAP 3yr shift constraint, consolidated in
             auxconstraint namespace with baryonpeak constraint.
2006/11/30  Case no longer matters in parameter file
2006/11/29  Change over to using GNU autotools for build process
2006/11/21  Fixed two bugs affecting w fits, both of which caused
             segfaults.
2006/11/14  The spline class turns out to be much slower, so removed
             it.
2006/11/13  Introduce spline and cosmos classes to store distances
             in splines.
            Doc update to point out biasing problem with fixing
             alpha, beta
            Move over to using unsigned int in many places
2006/10/16  Fixed minor output bug when parabolic peak fit fails.
2006/08/25  MWV added support for the FITS output to contourshow.pro,
             which was moved into an idl/ subdirectory.
            Fixed bug having to do with CFITSIO using fortran array
             ordering, which is irritating as hell.
2006/08/17  Use maps instead of vectors to pass around results of
             fits to cosmological and nuisance parameters.
2006/08/15  Reorganization of fit results extended to 2/3D cosmology fits.
             Fixed a bug in estimate_allnuisance that was not handling
             fixed values of one of alpha/beta correctly.
2006/08/09  Beginnings of reorganization in how fitters return results.
             So far limited to 1D cosmology fits (i.e., omega_m in a
             fixed curvature, fixed w universe)
2006/08/03  Massive changes to support analytic marginalization
             over alpha and beta.
            New structure for keeping track of parameter ranges
             and the like.
            Beginnings of new structure for sorting out which fit
             to perform.
2006/04/30  Minor change to inner loop when covariance matrix is
             present give a factor of three speedup.
2006/03/20  Added support for different power law indicies in
             baryonpeak_grid (Eisenstein used 0.98)
2006/03/13  Use variances internally for errors.  No longer ignores
             diagonal entries in COVFILE.
2006/03/02  Added support for fixed but non-zero Omega_curvature.
2006/03/01  Cov matricies now supported as long as alpha/beta are
             fixed for error propagation.
2006/02/28  Beginnings of incorporating covariance matricies.
2006/02/27  Massive directory reorg
2006/02/23  Converted some doubles to floats.
2006/01/18  Fixed bug in cosfitterexcept output routine.
2006/01/20  twopi is now actually set to 2*Pi instead of Pi.  
            Thanks to MWV for catching this.  Note that this applies
            a constant \chisq\ offset to all models, so it doesn't
            affect the cosmological contours, just the quoted chisquare.
2006/01/25  Modifications to contourshow by MWM to handle 3D grid case,
            plus some speedups.  No longer complains about under/overflow!
2006/01/26  cosgrids now support binary file input and output.
            Tests were added to cosgridtest for this functionality.
            A non-initialization error for om,od,w,alpha fits was fixed.
             Previously they wouldn't work.
2006/01/27  Parameter file parsing broken out of fitter to paramfile.
	    Added ability to convert output to fits files if cfitsio is
             available.