Bug fix by Jason Dossett for existing (SDSS) BAO likelihood in 07(08)/11
version of CosmoMC:

Cause: If not using a likelihood that calls get_cls in calclike, inithermo is
never run and z_drag is never calculated.  This causes the BAO likelihood to 
blow up:

Fix:

1.) In camb/cmbmain.f90:  added InitVars to the list of public routines.

2.) In sources/calclike.f90: added the following lines to the declarations in
	the function GetLogLikePost:
			use cambmain, only: initvars
    		use Camb, only: CAMBParams_Set 
    		type(CAMBParams)  P
    		
    		
    changed the following lines:
    
    	else
           if (Use_SN) GetLogLikePost = GetLogLikePost + SN_LnLike(CMB)
           if (Use_BAO) GetLogLikePost = GetLogLikePost + BAO_LnLike(CMB)
           if (Use_HST) GetLogLikePost = GetLogLikePost + HST_LnLike(CMB)           
        end if
        
    to:
    	
    	else
           if (Use_SN) GetLogLikePost = GetLogLikePost + SN_LnLike(CMB)
           if (Use_BAO)then
           		!JD had to change below so new BAO will work without CMB
           		!previous way z_drag was not calculated without calling get_cls.
           		if (RecomputeTransfers(CMB, Info%LastParams))  then
     				call CMBToCAMB(CMB, P)
         			call CAMBParams_Set(P)
           			call InitVars
           			Info%Theory%BAO_loglike = Bao_lnLike(CMB)
           			Info%LastParams = CMB
           		end if
           		GetLogLikePost = GetLogLikePost + Info%Theory%BAO_loglike
           end if
           if (Use_HST) GetLogLikePost = GetLogLikePost + HST_LnLike(CMB)           
        end if


With these changes in place the existing BAO likelihood will work even when
not using CMB or another dataset that requires a call to get_cls.


################################################################################
New BAO likelihood code to use multiple BAO data sets including the new 
WiggleZ Dark energy survey BAO measurements from Blake et al arXiv:1108.2635, and
the SDSS DR7 bao measurements of Percival et al. arXiv:0907.1660 with corrections 
to rs from Hamann et al., arXiv:1003.3999.
By Jason Dossett
################################################################################

Use:

Each .dataset file contains settings and filenames for a bao dataset, these
inclue:

		name = ;
		
		type_bao = 
		This is given an integer value and specifies the type of bao measurement
		the data set uses.  Currently the only options allowed are:
			1-for rs/D_V measurments   
		and 
			2-for A(z) (acoustic parameter) measurements
		More options can be added later and modifications to the likelihood code
		would be necessary;
		
		num_bao = 
		The number of data points in the data set
		
		bao_measurements_file = 
		The file that contains the bao measurement currently in the form:
			z	measurement
		If the file is in the your data directory start the file name with 
		%DATASETDIR% per the usual convention in CosmoMC;
		
		bao_invcov_file = 
		the inverse covariance matrix file for your bao data, named in same
		way as bao_measurements file

To use select bao data sets use the following format in the params.ini file
as is usual in CosmoMC:
#filenames for BAO data sets
bao_numdatasets = 2
bao_dataset1 = data/wigglez_bao.dataset
bao_dataset2 = data/sdss_bao.dataset



Changes to cosmoMC to implement new BAO likelihood:

1.) Added bug fix above

2.) New sources/bao.f90 has structure for reading multiple datasets based upon
	mpk.f90.It takes snippits of code from the existing bao.f90 and adds new 
	lines as well.

3.) Added the following files to data/:
		#These files are for the existing BAO data
		sdss_bao_invcov.txt
		sdss_bao.txt
		sdss_bao.dataset
		
		#These files are for the new WiggleZ data
		wigglez_BAO_invcov.txt
		wigglez_BAO.txt
		wigglez_bao.dataset
		
4.) In driver.f90:
		Added the following lines to the declarations:
			integer numbaosets
			character(LEN=Ini_max_string_len) bao_filename(100)
			
		Below the lines:
			nummpksets = Ini_Read_Int('mpk_numdatasets',0)
			if (Use_mpk) then
			 do i= 1, nummpksets
			  mpk_filename(i) = ReadIniFileName(DefIni,numcat('mpk_dataset',i)) 
			  call ReadMpkDataset(mpk_filename(i))
			 end do
			 if (Feedback>1) write(*,*) 'read mpk datasets'
			end if
			
		the following lines were added:
		
			!JD copied mpk structure above for new bao likelihood file.
			numbaosets = Ini_Read_Int('bao_numdatasets',0)
			if (Use_BAO) then
			 do i= 1, numbaosets
			  bao_filename(i) = ReadIniFileName(DefIni,numcat('bao_dataset',i)) 
			  call ReadBaoDataset(bao_filename(i))
			 end do
			 if (Feedback>1) write(*,*) 'read bao datasets'
			end if
			
		Removed the lines:
			if(Use_BAO .and. use_dr7lrg) &
         	call MpiStop('DR7 LRG and BAO are based on the same dataset. You cannot use both.')
        
        This line is no longer needed because bao.f90 ignores the sdss bao data if
        if the dr7lrg mpk is used. 
        
5.) In the Makefile, because the new bao.f90 uses mpk.f90 we reordered the
	OBJFILES:
		Formerly:
			bao.o lrggettheory.o mpk.o supernovae.o
		Now: 
			lrggettheory.o mpk.o bao.o supernovae.o
			
	New build rules as well:
		Formerly: 
			bbn.o: settings.o
			bao.o: cmbtypes.o
			mpk.o: cmbtypes.o lrggettheory.o
			
		Now:
			bbn.o: settings.o
			mpk.o: cmbtypes.o lrggettheory.o
			bao.o: cmbtypes.o mpk.o
			
6.)Added the following lines to params.ini
	#filenames for BAO data sets
	bao_numdatasets = 2
	bao_dataset1 = data/wigglez_bao.dataset
	bao_dataset2 = data/sdss_bao.dataset