

*** Version 1.3.2 ***

July 25, 2012

	Replaced "double precision" with "real*8" in common/hvqpdfpho.f,
	and made explicit the return typoe of ctq61pdf() in cteq61() for
	gfortran compatibility

September 30, 2010

        Fixed a compatibility problem with gfortran 4.4 (toend() produces
        and infinite loop. Replaced with "access='append'" in the open
        statement) in main/fonll.f. This seems to work with g77, g95 and
        gfortran 4.4 on Darwin

        Also, eliminated useless output to phres.tmp in 
        phresummed/photon_wps8.f

May 20, 2010
   
        Added file eps09.f in common (and to the Makefile in main)
	for providing the new EPS09 nuclear PDFs. Modified fonllpdf.f
	accordingly
	

January 29, 2010

        Eliminated writing of hdms-log.tmp, unit 66, in hdmassive/hdms.f
        to try to get rid of pointlike random crashes when multiple
        instances of the code run together in the same directory 

        Same for phmassive/phms.f

        This seems to solve most of the problems, but not all of them.

December 8, 2009

        Compactified long equation in hdresumment/struv_plus_avgo.f
        because the Intel ifort compiler does not accept more than 511
        continuation lines.

        COMPILATION WITH ifort MUST INCLUDE THE -save FLAG

November 11, 2008

        Fixed a bug in pdfpar() in common/lhapdfif.f which was
        preventing the use of LHAPDF with electrons.
        In the same routine added a check preventing repeated
        printout of the alpha_s value (happened when using photons)

July 9 and 11, 2008

        Added save statements to alfas() in common/hvqpdfpho.f,  
	alfas_p() and air() in common/fonfra_p.f, phdfpy() in 
        phmassive/phms.f, dfpy() in hdmassive/hdms.f, fonpdf() and 
        afgpdf() in common/hvqpdfpho.f  to make them compatible with 
        gfortran and g95. "save" added to all routines in hvqpdfpho.f
        containing an "if(init..." block.

        NB: "save" statemants are probably missing in other routines

        NB2: OPTIMIZATION (-O) SHOULD NOT BE USED WITH GFORTRAN OR G95
             UNLESS A -fno-automatic STATEMENT IS ALSO INCLUDED

*** Version 1.3.1 ***

Februray 1, 2008
        common/lhapdfif.f: the lambda5 value is now calculated
        from the alpha_s(mz) one. This avoids problems with sets
        which return 0 (like cteq6.5 and, more importantly, with
        MRST sets which return a slightly different lambda5, leading
        to an incorrect value for alpha_s
    

*** Version 1.3.0 ***

December 17, 2007
        Eliminated output to unit 32, hdrsout.tmp, in hdresummed/main_p.f,
        as it was essentially useless

        Patched output format (i1->i2) for isuda variable 
        and for pdf set in hdresummed/input.f

        Suppressed output to unit 15, hdres.tmp, from hdresummed/input.f 
        and hdresummed/main_p.f (can be reinstated by setting ioutput=15
        in hdresemmed/input.f)

        Shifted the wmax limit from 1 to 1-1d-16 in 
        dplus() in hdresummed/main_p.f to avoid a 1/1-w division by zero
        error in the w-1 limit

December 9, 2007
        Fixed LIBSLHA variable in main/Makefile so that it uses
        lhapdf-config from the LHAPDF installation. It must, of
        course, be correctly included in PATH beforehand

        LHAPDF does not return correctly lambda_QCD for CTEQ65.
        Value added by hand in common/lhapdfif.f

October 22, 2007
	IMPORTANT CHANGE: Added interface to LHAPDF, file common/lhapdfif.f

	modified fonllpdf.f : removed mlmpdf8 (moved to hvqpdfpho.f)
	                      jetpdflib now calls directly mlmpdf8

	renamed param to hdrsparam in hdresummed/input.f, main_p.f
	
September 3, 2007
        IMPORTANT CHANGE:
        Similarly to the massless calculation, the resummed one also suffers
        from numerical precision problems in certain limits, for 
        instance at large rapidity.
        The combination STRUV*+AVGO in the function fresc() in 
        hdresummed/main_p.f has been reimplemented in the function
        struv_plus_avgo in hdresummed/struv_plus_avgo.f by simplifying
        the proper matrix element contribution via Mathematica.
        This (together with the similar simplification described below) 
        should fix most of the numerical precision problems.
        
Aug. 30, 2007
        Fixed the wrapper to cteq6.1 in common/hvqpdfpho.f
        Bug was instead leading to the naming convention of cteq6 for
        data tables

Aug. 19, 2007
        IMPORTANT CHANGE:
	The function hqhpggm0 in hdmassive/rozlimgg.f lost precision
        in the large rapidity limit. This should now be fixed
        by a rewriting of part of the formula, through a simplification
        obtained with Mathematica.
	
	Added the option icalctype=4, to calculate massive and massless
	only (i.e. no resummed)

        Added to the function dfpy() in hdmassive/hdms.f the option
	to perform the integrals using VEGAS. Actually not used
	(hardcoded), because the precision looks good enough with the
	old (faster) option
		
          
Jul. 6, 2007
        Passed the input epsint to dovegas, so that the precision
        of the integral can be set in fonll0.f (default is 0.001).
        Also set the precision of the xes1 integral to epsint*10
        in hdresummed/main_p.f.
        This will cause small changes in numbers, but the         
        overall numerical precision should stay between 0.001 
        and 0.01

Jul. 4, 2007
        Fixed a division by zero in fonll.f, it caused crashes on
        Linux when beyond rapidity range

*** Version 1.2.1 ***

May 8, 2007
        Included a test to be used with EKS PDF's to avoid a crash
        in the x->1 limit

*** Version 1.2 ***

Sep. 6, 2005
	Paolo's trick evolved to what we have in the fragmentation paper.
	nf in Sudakov initial condition also changed to 3 for charm
	and 4 for bottom

Jul. 6, 2005
        Included Sudakov resummation for initial conditions, together
        with Paolo's trick (isuda=12 in main/fonll0.f) for handling 
        large-N values. This improves slightly the behaviour at small-pT 
        and large y (i.e. small-x, see below)	

May 20, 2005
        Added option for including shadowing corrections via the EKS
        package. To use them, put -A (atomic number) as beam type.
	For instance, Pb is -208

Apr. 27, 2005
	Changed output in fonll.f. Now the error on the resummed
	part, and an asterisk if it is larger than half the massive,
	are in output.

	Changed a little VEGAS integration in dovegas. Now 
	charm at large rapidity seems to work slightly better. The 
	probles is the convergence of the 3-dim integral for xes1
	More work on this is needed (see Sudakov above)

	If main_p.f.new is used:
	Interpreted eta = y in hdresummed/main_p.f, analogously
	to what is done with pT -> mT. This was actually done also
        before, as a consequence of using massless kinematics.
        Just rewritten the file more clearly. Old file left for comparison

Apr 8, 2005
        Version number written in file main/version.h and included
        in fonll.f and fonlltest.f


*** Version 1.12 ***

Mar. 14, 2005
        Added in fonll.f the check about the size of higher orders 
        (see BUGS)


*** Version 1.11 ***

Apr. 1, 2004
        Added in INSTALL the line about unpacking pdfdata.tgz

Feb. 18, 2004
	A few changes to make it compatible with the Intel Fortran
	compiler. Input unit changed from 5 to 55, a few lines
	shortened, a missing output in phresummed input fixed

Dec. 16, 2003
	Replaced hvqpdfpho.f. New version has the correct implementation
        of the Alekhin sets. Added datafiles in pdfdata


*** Version 1.1 ***

Oct. 28, 2003
	Small changes to dialogue in main/fonll.f and misc1/fonllgrid.f
        Updated common/hvqpdfpho.f, and added data files for hessian
		errors with CTEQ6M and MRST2001 in pdfdata
	Changed tests/tests.com to remove also links to new pdf data
	Changed maketar for dynamical main directory name
        tests/testfonll.outlog-ref renamed to tests/testfonll.outlog-v1.0
		INSTALL modified to reflect the renaming
	Added tests/testfonll.outlog-v1.1 (identical to v1.0)
	Eliminated the run directory (the user will run the code from
		her own preferred location)
	Added output unit number (itm=53) in misc2/interp.f
	Changed version number in README	
	
*** Version 1.0 ***

Jan. 2 2003
	main/fonlltest.f: print version number in .outlog file
	hdmassive/hdms.f: changed name of input log, from hvqphoms.log 
		to hdms-log.tmp;
	phmassive/phms.f: changed name of input log, from phoms.log to 
		phms-log.tmp
	README: changed to account for above changes.
	tests/tests.com:  changed to remove files with new names
	tests/testfonll.outlog-ref: changed in trivial way
	(FONLL v0.99 line appears); non-trivial changes to electr. proton
		output,	because of different interpolation (splines).
	Documentation/hvqpdfpho.txt.sh created; it generates the hvqpdfpho.txt 
		pdf numbering file for the hvqpdfpho package. hvqpdfpho.txt 
		recreated.

Dec. 23 2002
	main/fonll.f: print version number in .outlog file
	hdresummed/main_p.f: output file renamed to hdrsout.tmp (was main.out)

	misc1/ : various changes to fragmfonll, to make it easier for users
		example changed to Tevatron, as in Cacciari-Nason, prl 2002

Dec. 19 2002
	INSTALL: Sentence added on other unix platforms and CERNLIB
	common/hvqpdfpho.f newer version of cteq5 included
	main/fonll.f, fonlltest.f      prefix string up to 30 characters

Dec. 13 2002
	common/fonllpdf.f modified to use cernlib spline interpolation
		for electron pdf
	common/dcspln.f   cernlib spline added
	common/locatd.f   cernlib locatd added
	main/Makefile     dcspln.o and locatd.o included in files to link in
	
