CYCLIZE change log
-Jon Lapham
==============================================================================

-------------------------------------
v1.4
7 July, 2000
 * Added the btilt, broll, btwist harmonic force constant parameters to the
   CHAINPARMS structure (see src/cyclize_lib.h).
 * Added the params_calc_b() function to calculate the harmonic force 
   constants from the temperature and flexibilities.
 * Added 'float Temp', a global to cyclize_lib.c, to store the temperature of
   the simulation.

-------------------------------------
v1.3
7 July, 2000
 * Added the btilt, broll, btwist harmonic force constant parameters to the
   CHAINPARMS structure (see src/cyclize_lib.h).
 * Added the params_calc_b() function to calculate the harmonic force 
   constants from the temperature and flexibilities.
 * Added 'float Temp', a global to cyclize_lib.c, to store the temperature of
   the simulation.
 * Rewrote the toplevel Makefiles, seperating the gcc, pgcc, sgi makefiles.
 * Rewrote the src/Makefile to link using the Fortran compiler.
 * Replaced the routine gauss_box_muller with a free version
 * Moved the ranu() function from random.c to math_routines.c, this required
   deleting random.c and random.h, and removing references to both files in 
   various places.

-------------------------------------
v1.2
5 July, 2000
 * Data fitting can be accomplished using EITHER the numerical recipies 
   routines, or the odrpack routines (Weighted Orthogonal Distrance 
   Regression, the default).  The odrpack routines can be redistributed for
   free, whereas the NR routines require purchase of a license.
 * Use all chains in the radial distribution function that have an end-to-end
   distance less than the radial cutoff.  This costs nothing in CPU time, and
   gives much better statistics in a poorly sampled region of the radial
   distribution function.  Thus, you can still specify a 'nrad_stats' value
   less than the total number of whole chains, to speed up the calculation.

-------------------------------------
v1.1
27 June, 2000
 * Use the CYCLIZE_HOME env. var. in show_results.pl to find the 
   cyclize_params file.  This is much better than expecting the file to 
   always reside in /usr/local/lib/cyclize ...
 * Fixed the embarrasing Y2K bug in generate_chains.c and analyze_chains.c
   ... yes, I forgot to add 1900 to the year.  Only affected the *.header
   files... doesn't change any results.
 * Changed all the perl #! lines to read "#! /usr/bin/perl" rather than
   "#! /usr/local/bin/perl" since perl is more likely to be installed under
   /usr/bin these days.

-------------------------------------
v1.0
12 February, 2000
 * Converted cyclize_lib.pl to use 2 environment variables:
   CYCLIZE_HOME (mandatory) and CYCLIZE_NICE (optional) to set
   the installation root directory and 'niceness' level of the
   calculations, respectively.
 * Updates to the Makefile, uses more standardized 'sh' commands.
   Should work with most decent UNIces.

-------------------------------------
v1.0b
23 April, 1999
 * Updates to the Makefile, is run from the primary directory now.

-------------------------------------
v0.4
14 November, 1998
 * Wrote the "Cyclize" perl subroutine, removing the need to
   make the "system" calls.  Ahh, this is good.  
 * Cleaned up the code a bit...  stupid unused vars, etc... stopped
   a "-Wall" compile from complaining too much...  made sure the
   SGI "cc" compiler can handle the code, in addition to 'gcc"...
 * Many updates to the manual.
 * New analysis files, "distfreq_halfchain.xm" and 
   "dist_freq_wholechain.xm" are created.  They contain informaiton on
   the distribution frequency of half and whole chain end-to-end
   distances, see the manual for a better description :)
 * The global distribution arrays (Rdist, Adist, Tdist) were 
   converted to "double" from "unsigned long int" to protect 
   from overflow errors.
 * The numerical recipies "machar" routine was added to 
   "math_routines.c" so we can report on any potential problems
   with machine precision, etc.  This hasn't been implimented, but
   the routine exists!
 * Fixed the cpu_time calculation, it was "forgetting" 1/10th of
   the CPU time!
 * The analyze_chains routine takes nrad_stats in whole_chains,
   instead of half_chains
 * The &find_positions subroutine has been added to work with 
   &read_sequence for people who prefer to read the actual chain
   sequence in from a text file.  &find_positions looks for and
   returns the base pair positions where a particular sequence is
   found.  See the "program" section of the manual for a better
   description.
 * Ttracts are recognized by fill_Atract_params.  The parameters
   used for the Ttracts are automatically generated from the
   parameters used in the Atracts.

-------------------------------------
v0.3
3 November, 1998
 * Rise per helix axis "dz" was added to the parameter hash
   input.
 * $nkeepers variable was added to allow for limiting the
   number of cyclized chains that are stored in the 
   "cyclized_chains.dat" file.  This keeps this file from 
   becoming huge.
 * A final "pseudo" basepair is calculated at the end of the
   chain.  This makes more intuitive sense (?) in that the
   end-to-end distance should TRULY be zero and the Phi angle
   between the first BP and this pseudo BP should be zero.

-------------------------------------
v0.2
13 September, 1998
 * The addition of the ability to specify "independent 
   calculations" was added to give more reproducible 
   results by reducing the "importance" of rare cyclizing 
   half-chains.
 * This manual was written.

-------------------------------------
v0.1
2 July, 1998
 * The original Fortran routines were converted to C (actually, I just 
   translated Levine's original paper into C) and a simple PERL front 
   end was written to allow for simple construction of new chain parameters.
