Description of the program: av-d2

This program takes one of the output files of the d2, c1, or c2naive programs as its input and smoothes the content of the file by arithmetically averaging over a given interval. Furthermore, it is possible to specify a range of embedding dimensions to be printed. In the case of d2 output, the application of av-d2 makes most sense on the local slopes file (extension .d2).

Suppose the input file contained values ri and di, then the output will be in two columns

                  __ a
            1    \
   r    ,  ----   |       d
    i      2a+1  /__ j=-a  i-j

where a is the argument to the -a option. We recommend to experiment with the parameter a for optimal results.


av-d2 input-file [Options]
input-file has to be one of the files created by d2, c1 or c2naive.

Possible options are:

Option Description Default
-m# minimal dimension to print 1
-M# maximal dimension to print whole file
-a# smooth over an interval of 2#+1 points 1
-E show the length scales in rescaled units [0:1] units of the data
-o# output file name; without -o default is used
-o without a datafile means 'filename'.av
-V# verbosity level
  0: only panic messages
  1: add input/output messages
-h show these options none

Description of the Output:

The output is the same as the one from the d2 and c2naive programs. The only difference is that the output is smoothed.
Other post-processing for correlation sum data is available: c2d computes local slopes from correlation sum data (extension .c2) by fitting straight lines. This is essentially redundant with the present program. Takens' estimator can be obtained using c2t while c2g provides Gaussian kernel correlation integrals. Both act on the correlation sum data (extension .c2).
View the C-sources.
Table of Contents * TISEAN home