2008-07-03  Brian Gough  <bjg@hp2.network-theory.co.uk>

	* Makefile.am (INCLUDES): use top_srcdir instead of top_builddir

2007-10-02  Brian Gough  <bjg@network-theory.co.uk>

	* variance_source.c: added functions for sum of squares and
	weighted sum of squares

2006-01-07  Brian Gough  <bjg@network-theory.co.uk>

	* test_float_source.c: additional tests for NaNs

	* minmax_source.c: handle NaNs correctly, if any element is NaN
	then the max and min are NaN

Sat Jul 15 12:23:09 2000  Brian Gough  <bjg@network-theory.co.uk>

	* test_nist.c: added url/reference to the original NIST datasets

Thu May  4 15:11:11 2000  Brian Gough  <bjg@network-theory.co.uk>

	* covariance.c: added covariance function

Wed Mar  8 15:29:19 2000  Brian Gough  <bjg@network-theory.co.uk>

	* Function Name Reorganization: The function suffix "_with_mean"
 	has been changed to "_m", the function suffix "_with_mean_and_sd"
 	has been changed to "_m_sd". The variance function est_variance
 	has been changed to simply _variance, (similarly for est_sd, which
 	is now _sd), while the function _variance is now
 	_variance_with_fixed_mean (and _sd is _sd_with_fixed_mean).

Wed Mar  1 11:09:21 2000  Brian Gough  <bjg@network-theory.co.uk>

	* wabsdev.c wabsdev_source.c wkurtosis.c wkurtosis_source.c
 	wmean.c wmean_source.c wskew.c wskew_source.c wvariance.c
 	wvariance_source.c:  added support for weighted statistics

	* removed source.h (no longer used)

Tue Feb 29 10:35:23 2000  Brian Gough  <bjg@network-theory.co.uk>

	* changed arguments of pvariance and ttest to have consistent form
 	(DATA,STRIDE,SIZE)

Mon Feb 28 20:29:08 2000  Brian Gough  <bjg@network-theory.co.uk>

	* removed sort function since this is now available in the sort
 	directory.

	* converted all functions to take a stride argument, in
 	preparation for supporting statistics on vectors

1998-11-06    <bjg@ancho.lanl.gov>

	* test.c: added prototype for memcpy using #include <string.h>

Wed Aug 12 12:34:24 1998  Brian Gough  <bjg@vvv.lanl.gov>

	* quantiles.c: Renamed percentile to quantile throughout. This is
 	the correct name for the quantity, mathematically speaking.

Wed Aug  5 12:34:58 1998  Brian Gough  <bjg@vvv.lanl.gov>

	* mkheaders.pl: script to generate all the headers from
 	gsl_statistics_int.h

	* test.h: removed test.h, not needed

	* test.c: added tests for all types

	* converted functions to use 'long double' for internal calculations,
	for extra range.

	* converted all the functions to use templates_on.h and
 	templates_off.h

Mon Jun  1 23:47:23 1998  Brian Gough  <bjg@vvv.lanl.gov>

	* lag1_source.c: added a new function to compute the
 	lag1-autocorrelation

	* test_nist.c: added some numerical accuracy tests from NIST

Fri Apr 10 15:11:51 1998  Brian Gough  <bjg@vvv.lanl.gov>

	* renamed pooled_variance.c, etc to p_variance to avoid linker
 	complaints about long filenames on some platforms

Wed Apr  8 18:11:48 1998  Brian Gough  <bjg@vvv.lanl.gov>

	* Added a new recurrence algorithm for the mean, variance, skew
 	and kurtosis. It uses a running value which is less likely to
 	overflow than a global sum.

Mon Mar 30 22:18:59 1998  Brian Gough  <bjg@vvv.lanl.gov>

	* completely reorganised the source system to use macros as
	a primitive form of templates. 

Sun Mar 29 16:25:24 1998  Brian Gough  <bjg@vvv.lanl.gov>

	* make everything const where possible (to reduce the chance of errors)

Sun Mar 22 23:34:05 1998  Brian Gough  <bjg@vvv.lanl.gov>

	* skew.c: compute the skewness of a dataset

	* percentiles.c: find a given percentile from a sorted dataset

	* minmax.c: allow the user to find the indices of the max/min
 	data points, in addition to the max/min values

	* median.c: compute the median of a sorted dataset

	* kurtosis.c: compute the kurtosis

	* absdev.c: compute absolute deviations

	* split statistics functions into double versions (gsl_stats) and
 	integer versions (gsl_stats_int). There is a script
 	(convert_double_to_int.pl) to make the integer versions from the
 	double versions.