Description of the program: ghkss

This program performs a noise reduction as proposed in Grassberger et al. In principal, it performs a orthogonal projection onto a q-dimensional manifold using a special (tricky) metric. In case the -2 parameter is set, an euclidean metric is used. This is done in Cawley et al. as well as in Sauer and is sometimes useful for flow systems.

Usage:

ghkss [Options]

Everything not being a valid option will be interpreted as a potential datafile name. Given no datafile at all, means read stdin. Also - means stdin

Possible options are:

Option Description Default

-l# number of points to use whole file

-x# number of lines to be ignored 0

-c# column to be read 1,...,dimension of the vectors
-m#,# number of components, embedding dimension 1,5

-d# delay for the embedding 1

-q# dimension of the manifold to project to 2

-k# minimal number of neighbours 30

-r# minimal size of the neighbourhood (interval of data)/1000

-i# number of iterations 1

-2 use euclidean metric instead of the tricky one tricky metric

-o# output file name without file name: 'datafile'.opt.n, where n is the iteration
(or stdin.opt.n if data was read from stdin)
without -o the last iteration is also written to stdout

-V# verbosity level
  0: only panic messages
  1: add input/output messages
  2: add average correction and trend
  4: add how many points were corrected for which epsilon 7

-h show these options none

Option	Description	Default
-l#	number of points to use	whole file
-x#	number of lines to be ignored	0
-c#	column to be read	1,...,dimension of the vectors
-m#,#	number of components, embedding dimension	1,5
-d#	delay for the embedding	1
-q#	dimension of the manifold to project to	2
-k#	minimal number of neighbours	30
-r#	minimal size of the neighbourhood	(interval of data)/1000
-i#	number of iterations	1
-2	use euclidean metric instead of the tricky one	tricky metric
-o#	output file name	without file name: 'datafile'.opt.n, where n is the iteration (or stdin.opt.n if data was read from stdin) without -o the last iteration is also written to stdout
-V#	verbosity level 0: only panic messages 1: add input/output messages 2: add average correction and trend 4: add how many points were corrected for which epsilon	7
-h	show these options	none

Description of the Output:

Each file produced contains the filtered time series as one column. The standard error device shows some statistics, namely for each iteration (i) the number of vectors corrected up to the actual value of the neighborhood size, (ii) the average shift and (iii) the average correction. (iv) The next line shows for how many points the correction was unreasonably large and the last line shows (v) the file, to which the corrected data was written.

View the C-sources.