Freezer is a program which finds the best fit of a three-dimensionally 
expanding hydrodynamical source to one-particle distribution and 
two-particle correlation data from relativistic heavy ion collision 
experiments.  The parameters and model used are fully described in the 
Los Alamos preprint LA-UR-96-0782, which is also on the Los Alamos preprint
server http://xxx.lanl.gov with the designation nucl-th/9603007.

I)  The files needed to compile the programs are the following:
	freezer.f	-	program to find best fits
	plotfrz.f	-	program to plot model results
	subzero.f	-	subroutines used by both of the above
    These files must be compiled using a double precision option.  On a Sun
    SPARCstation, with f77 compiler version 4.0, this can be done using the
    following commands:

f77 -c -dalign -libmil -O4 -r8 -u -xtarget=native freezer.f
f77 -c -dalign -libmil -O4 -r8 -u -xtarget=native plotfrz.f
f77 -c -dalign -libmil -O4 -r8 -u -xtarget=native subzero.f
f77 -dalign -libmil -O4 -o freezer -r8 -u -xtarget=native freezer.o subzero.o
f77 -dalign -libmil -O4 -o plotfrz -r8 -u -xtarget=native plotfrz.o subzero.o

    On a Sun with an older fortran compiler, one could use:
f77 -c -O -r8 freezer.f
f77 -c -O -r8 plotfrz.f
f77 -c -O -r8 subzero.f
f77 -O -r8 -o freezer freezer.o subzero.o
f77 -O -r8 -o plotfrz plotfrz.o subzero.o

    On an HP workstation, one might try:
fort77 -K -w +e +ppu +autodblpad +O3 -c freezer.f
fort77 -K -w +e +ppu +autodblpad +O3 -c plotfrz.f
fort77 -K -w +e +ppu +autodblpad +O3 -c subzero.f
ld /lib/crt0.o freezer.o subzero.o /usr/lib/end.o -lcl -lc -o freezer
ld /lib/crt0.o plotfrz.o subzero.o /usr/lib/end.o -lcl -lc -o plotfrz

II) In addition to the program files, the following input files are needed
    to specify starting parameters, integration points, etc.
	freezer.in	-	input file for freezer
	plotfrz.in	-	input file for plotfrz
    Each file contains an explanation of the inputs in that file.
    The program plotfrz will calculate distributions and correlations for
    various particle momenta at a given point in parameter space.  For a
    more complete explanation, see the file plotfrz.in.

III)To run freezer, the user must also supply some of the following 
    one-particle distribution and two-particle correlation data files.  
	pim.dat		-	pi- one-particle distributions
	pip.dat		-	pi+ one-particle distributions
	kp.dat		-	K+ one-particle distributions
	km.dat		-	K- one-particle distributions
	pro.dat		-	proton one-particle distributions
	pbar.dat	-	antiproton one-particle distributions
	pim.cor		-	pi- two-particle correlations
	pip.cor		-	pi+ two-particle correlations
	kp.cor		-	K+ two-particle correlations
	km.cor		-	K- two-particle correlations
  A)  Each line of the *.dat files must contain a line of the form
	y	xmTm	P	sigma
      where each number is real, measured relative to the lab, and
	y 	= rapidity of data point
        xmTm 	= mT-m of data point (in GeV)
	P 	= Ed^3N/dp^3 (in GeV^{-2})
	sigma	= statistical error on P (in GeV^{-2}) 
      (we are assuming that in freezer.in option(2)=1 and option(3)=0)
  B)  Each line of the *.cor files must contain a line of the form
	Y	KT	q_z	q_out	q_side	C	sigma
      where each number is real, Y is relative to the lab, and
	Y	= .5*(y1+y2) is the average rapidity of the pair
	KT	= .5*(p1_out+p2_out) is the average KT of the pair (MeV)
	q_z	= (p1-p2)_z is the longitudinal momentum difference (MeV)
	q_out	= (p1-p2)_out is the out momentum difference (MeV)
	q_side 	= (p1-p2)_side is the side momentum difference (MeV)
	C	= P2(p1,p2)/(P(p1)P(p2)) is the correlation function
	sigma	= statistical error of C
      For three-dimensional binning in q, Y and KT can be taken as the
      average values of those quantities over the bin under consideration.
      In freezer.in, the value of option(4) and Ymeas determine how the 
      program should interpret q_z.  We have the cases
      when abs(option(4))
  	= 1, q_z actually means the quantity y1-y2.
      	= 2, q_z = (p1-p2)_z measured in a fixed frame defined by Ymeas.
      	= 3, q_z = (p1-p2)_z measured in the LCMS frame.
       negative values of option(4) mean to average over the relative
       sign of ql and qout.

IV) Running freezer.  Once the data files have been created and the
    appropriate options chosen in freezer.in, the Z,A, and intial 
    rapidity of each colliding nucleus should be entered, reasonable
    starting parameter values should be chosen such as at the AGS
    for example:
    120.  .5  2.     			|T, vT/c, et0
    4.  6.  6.			 	|muB/T, tau, R
    -0.5  1.5 1.			|aT, ycm, fs
    1.  1. 1. 1.			|lampi, lamK, normpi, normK 
    For complete chemical equilibrium, fs should be fixed at 1.
    For completely incoherent production, lampi and lamK should also be 1.
    For absolutely normalized data normpi and normK should be 1.
    (For unnormalized one-particle distribution data, normpi and normK
     should be allowed to vary).
    Percent systematic errors should also be specified in freezer.in.
    To do a summary run at a point in parameters space, set option(1) = 0
    The best and fastest chi^2 minimization option is option(1) = 2
    If for some reason, the above option does not work, one should
      try the slower but more durable option(1) = 3
    If this also does not find a minimum, the last resort is the
      excruciatingly slow but very durable option(1) = 1
    In general, one should start with a smaller number of integration
      points first.  Once a minimum has been found, start the program
      again with a larger number of integration points.
    The rest of the inputs are explained in the file freezer.in.
    To run the compiled program, just type freezer

V)  Output.  The main output is to the screen and to the file freezer.log.
    Everything printed to the screen is also printed to freezer.log.
    These files show the parameters and their corresponding chi^2 for each
    guess made in parameter space.  Once a minimum has been found, the
    best parameters with their 99% confidence level errors are printed.
    In addition, other quantities such as baryon density (with their errors)
    are also printed.  Finally the curvature and covariance matrices are
    printed.  In addition to the screen and freezer.log, there is some
    output to the file parameters.out which was mostly used for debugging
    and shouldn't concern most users.  

    Finally, for every *.dat (*.cor) file which is read in, a *dist.calc
    (*cor.calc) file is writen at the end.  The first four columns of the
    *dist.calc files are the same as those of the *.dat files, except that
    the sigma in the *dist.calc files is the total error (including 
    systematic).  The last two columns are the calculated value of the
    distribution at that point and the chi^2 of that point.  So each
    line will contain the six entries:
	y   xmTm   P   sigma_tot   P_calc   chi^2
    Note that chi^2 = (P-P_calc)^2/sigma_tot^2.
    The lines of the *cor.calc files are:
	Y   KT   q_z   q_out   q_side   C   sigma_tot   C_calc   chi^2
    Explicitly, the names of the possible *.calc files are
	distpim.calc	-	pi- one-particle distributions
	distpip.calc	-	pi+ one-particle distributions
	distkp.calc	-	K+ one-particle distributions
	distkm.calc	-	K- one-particle distributions
	distpro.calc	-	proton one-particle distributions
	distpbar.calc	-	antiproton one-particle distributions
	corpim.calc	-	pi- two-particle correlations
	corpip.calc	-	pi+ two-particle correlations
	corkp.calc	-	K+ two-particle correlations
	corkm.calc	-	K- two-particle correlations

    Every output file (including freezer.log) will be overwritten with
    each new run of freezer.

VI) If you have problems running the program, both Ray Nix (nix@lanl.gov)
    and Hubert Van Hecke (hubert@lanl.gov) have experience running
    freezer.  Although I may be harder to reach, you can also try me at
    my wife's email address k8chapman@aol.com.

    Good luck,

    Scott Chapman  5/21/96