Annotation of OpenXM_contrib/gnuplot/docs/old/README.pro, Revision 1.1
1.1 ! maekawa 1: The modifications to gnuplot 3.2 in this directory add the
! 2: capability of plotting probability distributions.
! 3:
! 4: This mod adds the following extra standard gnuplot functions:
! 5:
! 6: ibeta - incomplete beta function
! 7: igamma - incomplete gamma function
! 8: lgamma - natural log of gamma
! 9: gamma - gamma function
! 10: erf - error function (= trivial variation of infamous bell curve)
! 11: erfc - 1.0 - error function (more accurate than 1.0 - erf yourself)
! 12: rand - pseudo random number generator (in compliance with Knuth)
! 13:
! 14: using these things as building blocks, a gnuplot command file 'stat.inc'
! 15: is provided with definitions of almost all statistical distributions.
! 16:
! 17: Author: Jos van der Woude, jvdwoude@hut.nl
! 18:
! 19: ==========================================================
! 20: Directions to install gnuplot 3.3:
! 21: (diffs are relative to gnuplot 3.0, patchlevel 2.0)
! 22:
! 23: If you have a BSD machine, add the following to
! 24: the makefile.unx:
! 25:
! 26: OPTIONS = -DERF -DGAMMA
! 27:
! 28: and specfun.c to the SOURCES and specfun.o to the OBJECTS.
! 29:
! 30: For MSDOS machines (and non-BSD machines) do the following:
! 31:
! 32: ASSUMPTIONS:
! 33: The diffs provided work for msdos based pc's, using turbo C, version 2.0,
! 34: and Borland C++, version 2.0 or 3.0.
! 35: If you use a different setup, you might have to make (some) changes.
! 36:
! 37: GOAL:
! 38: To extend the standard function library of gnuplot 3.2 with the following
! 39: functions:
! 40:
! 41: ibeta - incomplete beta function
! 42: igamma - incomplete gamma function
! 43: erf() - error function (= trivial variation of normal distribution function)
! 44: erfc() - 1.0 - erf() (more accurate than computing 1.0 - erf() yourself)
! 45: gamma() - gamma function (for entire domain!)
! 46: lgamma() - natural logarithm of gamma function
! 47: rand() - a statistical acceptable version of standard C rand() function
! 48:
! 49: These additions to the repertoire of standard functions turn gnuplot into
! 50: an excellent tool for plotting statistical density and distribution functions.
! 51:
! 52: DETAILED INSTRUCTIONS:
! 53: 0. Split the bottom part of this file into the appropriate files:
! 54: readme.p3, patch3.dif, stat.inc, nomo95.dem, random.dem, prob.dem,
! 55: prob2.dem, and specfun.c using an editor or a PD PC shar utility.
! 56:
! 57: 1. Use the patch program to generate new versions of:
! 58: gnuplot.doc, command.c, internal.c, misc.c, parse.c, plot.c, plot.h,
! 59: standard.c, util.c and version.c
! 60:
! 61: 2. Make sure that file specfun.c is in the gnuplot directory.
! 62:
! 63: 3. Use the make program to create new version of gnuplot.
! 64:
! 65: 4. Test the new functions with commmand files prob.dem, prob2.dem,
! 66: random.dem and nomo95.dem.
! 67:
! 68: KNOWN BUG:
! 69: On MSDOS machines this modification can result in a DGROUP segment overflow
! 70: error at link time. The reason is that gnuplot uses the resources of the
! 71: large memory model right up to the limit.
! 72: There are two things you can do about this:
! 73:
! 74: 1. Compile gnuplot using the huge memory model.
! 75: This was tested using Borland C++, version 2.0: Works fine, even with
! 76: -DREADLINE enabled.
! 77:
! 78: 2. If you dont want to compile under the huge memory model you can try to
! 79: disable unused terminal drivers in term.h. See instructions at top
! 80: of file term.h.
! 81:
! 82:
! 83: Happy plotting,
! 84:
! 85: Jos van der Woude
! 86: Amsterdam, The Netherlands
! 87: jvdwoude@hut.nl
FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>