-
Notifications
You must be signed in to change notification settings - Fork 1
Fast Wavelet Transform
License
bindatype/fwt
Folders and files
| Name | Name | Last commit message | Last commit date | |
|---|---|---|---|---|
Repository files navigation
Change Log:
1.) Removed ifwt.c from project. Was never really part of it anyway.
2.) Replaced pipein.c with get_data.c which is much more flexible and
allows for arbitrary M by N ascii data files to be imported. Note
that currently only 1, 2, or 3 column input is supported.
3.) Speaking of that, 1, 2, or 3 column input is now supported. See
below in the DESCRIPTION section for more information.
4.) Added a gaussian routine for bootstraping (Box-Muller).
5.) Added a poisson routine for bootstraping (Used Knuth's algorithm from
The Art of Computer Programming, Vol2).
6.) Added wavlet_variance and heatmap routines to the project.
################## PURPOSE #############################################
################## MAN PAGE #############################################
NAME
fwt - decomposes an array of numbers into wavelet coefficients.
SYNOPSIS
fwt [OPTION]... < [file]...
DESCRIPTION
This software expects input in the form of ascii numbers with one,
two, or three columns. If one column then that column must be the
series to to transform. If two columns then the first column is
the index (such as time) and the second is the series (signal) to
be transformed. If three columns are supplied then the first
column is the index, the second is the series (signal) to be
transformed, and the third column is the uncertainty. If only one
or two columns are given then the uncertainty is assumed to be
the square root of the signal. columns must be of length 2^m where
m is a positive integer. The maximum size of m depends on the
machine running the code. It is the user's responsibility to
guarantee the length of the data. In any case the signal is assumed
to be uniformly sampled.
Input is supplied in the form of pipes or redirection. This can be
accomplished via:
cat INPUTFILE | fwt
or
fwt < INPUTFILE
Different wavelet filters can be chosen via command line options.
cat INPUTFILE | fwt -d2
or
cat INPUTFILE | fwt -c12 --offset 3
If no wavelet filter is specified the default, -d2 (aka the Haar
basis), is assumed.
A positive offset to the signal may be specified by the --offset
option or -o For example, if there were an array 1,2,3...7,8 and
an offset of one were given then the signal would be processed as
8,1,2...6,7. This is useful when adjusting the appearance of a
scalogram. Any statistical measure which sums over position
values in an octave will be insensitive to offset.
The output of the code is also two column ascii data. The first column
is index information and the second column holds the transformed signal.
OPTIONS
-d NUM, --daub=NUM Daubechies family of wavelet filters. NUM=2 gives
the Haar filter. NUM determines wavelet support and can be 2, 4, 6.
-c NUM, --coif=NUM Coiflet family of wavelet filters. NUM determines
wavelet support and can be 6, 12, or 18.
-i, --inverse Does inverse of fwt.
-o NUM, --order=NUM Specifies offset of the signal. If offset is not
specified a default value of zero is assumed.
-g, --gaussian Bootstrap based on a gaussian probability distribution. The mean
is zero and the width (sigma) is taken from the third column.
If a third column isn't given then the width is assumed
to be the square root of the number of counts.
This option is incompatible with poisson bootstrap.
-p, --poisson Bootstrap based on a poisson probability distribution
(Knuth's The Art of Computer Programming, Vol. 2.)
The parameter lamda is taken from the third column.
If a third column isn't given then the width
is assumed to be the square root of the number of counts.
This option is incompatible with gaussian bootstrap.
AUTHOR
Written by Glen Alan MacLachlan.
REPORTING BUGS
Report bugs to <bindatype@gmail.com>
COPYRIGHT
Copyright © 2010 Glen Alan MacLachlan.
License GPLv3+: GNU GPL version 3
or later <http://gnu.org/licenses/gpl.html> This is free software: you
are free to change and redistribute it.
There is NO WARRANTY, to the extent permitted by law.
REQUIREMENTS
gcc
make
COMPILING
To make the project type:
make
To clean the project, i.e., remove the objects and executable:
make clean
About
Fast Wavelet Transform
Resources
License
Stars
Watchers
Forks
Releases
No releases published
Packages 0
No packages published