ckbs.omh

# -------------------------------------------------------------------
# ckbs: Constrained and Robust Kalman-Bucy Smoother Program: Copyright (C) 2006-2011
# Authors: Bradlely Bell:        bradbell at washington dot edu
#          Gianluigi Pillonetto: giapi at dei dot unipd dot it
#          Aleksandr Aravkin:    saravkin at eos dot ubc dot ca
# License: GNU General Public License Version 2
# -------------------------------------------------------------------
$begin ckbs$$
$dollar @$$
$rmark !$$
$spell
	mkdir
	Tarballs
	yyyymmdd
	yyyy mm dd
	www http https
	coin-or.org/svn/CoinBazaar
	tarball
	tgz
	xvzf
	cygwin
	ckbs
	Matlab
	xml
$$

$latex \newcommand{\R}[1]{{\rm #1}}$$
$latex \newcommand{\B}[1]{{\bf #1}}$$


$navigate%
	Across%Index%
	Prev%Previous%
	Next%Next%
	Up%Up%
	Down_up_3%_up_3%
	Down_up_2%_up_2%
	Down_up_1%_up_1%
	Down_up_0%_up_0%
	Current%Headings
%$$

$section ckbs-0.20130204.0: Constrained/Robust Kalman-Bucy Smoothers$$

$table
$cnext One web page per Section $pre $$ $cnext All as one web page
$rnext
Math in Latex  $pre $$
$cnext  $href%ckbs.htm%$$               $cnext $href%_printable.htm%$$
$rnext
Math in MathML $pre $$
$cnext $href%ckbs.xml%$$                $cnext $href%_priintable.xml%$$
$tend

$head Purpose$$
The programs in this package compute a smoothed estimate of a time series 
given process models for the state evolution, measurement
models for the observations, variance estimates, 
and the sequence of observations. 
 
$subhead Affine Constrained Smoother$$
The program $cref/constrained_affine/ckbs_affine/$$ performs 
Kalman smoothing when the process and measurement models
are affine. Affine inequality constraints may also be imposed
on the state.  

$subhead Nonlinear Constrained Smoother$$
The program $cref/constrained_nonlinear/ckbs_nonlinear/$$ performs
Kalman smoothing for general nonlinear process measurement. 
Nonlinear inequality constraints may also be imposed on the state. 

$subhead Affine L1 Robust Smoother$$
The program $cref/robust_affine/ckbs_L1_affine/$$ performs 
robust Kalman smoothing when the process and measurement 
models are affine and the measurement sequence
may contain outliers. 

$subhead Nonlinear L1 Robust Smoother$$
The program $cref/robust_nonlinear/ckbs_L1_nonlinear/$$ performs 
robust Kalman smoothing for general nonlinear process
and measurement models when the measurement sequence
may contain outliers. 

$subhead General Student's T smoother$$
The program $cref/t_general/ckbs_t_general/$$ performs 
robust Student's t Kalman smoothing for general nonlinear process
and measurement models. The measurements may contain 
very large outliers. Also, there may be sudden changes in trend data. 
The user can specify which components of process and measurement
residuals to model using Student's t. 

$head MathML$$
This documentation is in XHTML + MathML.
If you have trouble viewing it in your browser see
Viewing MathML on the web page
$pre
	$$
	$href%
		http://www.seanet.com/~bradbell/packages.htm
	%$$

$head System Requirements$$
This constrained Kalman-Bucy smoother runs under both the 
$href%
	http://www.mathworks.com/%
	Matlab!
%$$
and 
$href%
	http://www.gnu.org/software/octave/%
	Octave
%$$
While it will work under MS windows, 
these download instructions are for the Unix operating system.

$head Download$$

$subhead Release Tarballs$$
You can download a tarball release of ckbs from the directory
$pre
	$$
$href%http://www.coin-or.org/download/source/CoinBazaar%$$
$pre
$$
Theses releases are named
$codei%
	ckbs-0.%yyyymmdd%.%r%.%ext%
%$$
where $icode yyyy$$ is the year,
$icode mm$$ is the month,
$icode dd$$ is the day,
$icode r$$ is the release number,
and $icode ext$$ is $code tgz$$ or $code zip$$.
$pre

$$
If $icode ext$$ is $code tgz$$ the command
$codei%
	tar -xvzf ckbs-0.%yyyymmdd%.%r%.tgz
%$$
will create the $codei%ckbs-0.%yyyymmdd%.%r%$$ directory.
$pre

$$
If $icode ext$$ is $code zip$$ the command
$codei%
	unzip ckbs-0.%yyyymmdd%.%r%.zip
%$$
will create the $codei%ckbs-0.%yyyymmdd%.%r%$$ directory.

$subhead Using Subversion$$
If you are going to use subversion to download a version of $code ckbs$$,
start with the following commands:
$codei%
	mkdir ckbs
	cd ckbs
	web="https://projects.coin-or.org/svn/CoinBazaar/projects/ckbs"
%$$
For each release listed in the
$href%http://www.coin-or.org/download/source/CoinBazaar%download directory%$$
you can use subversion to download the release with the commands
$codei%
	svn checkout @web/releases/0.%yyyymmdd%.%r%
%$$
This will create the $codei%ckbs/0.%yyyymmdd%.%r%$$ directory.
You can also download the most recent version of ckbs (called the trunk)
using the commands
$codei%
	svn checkout @web/trunk
%$$
This will create the $codei%ckbs/trunk%$$ directory.

$head Test Install$$
Change into the directory
$codei%ckbs/0.%yyyymmdd%.%r%/example%$$ 
or
$codei%ckbs/trunk/example%$$ 
and use Matlab! or Octave to run the program $code all_ok.m$$.
It will print the test results for all the components of $code ckbs$$.
You can run some more tests by changing into the directory
$codei%ckbs/0.%yyyymmdd%.%r%/test%$$ 
or
$codei%ckbs/trunk/test%$$ 
and running the program $code all_ok.m$$
which will have a similar output.


$head Documentation$$
If you download a release, a copy of the documentation for the release
can be found be viewed in your browser by opening the file
$codei%
	ckbs/0.%yyyymmdd%.%r%/doc/ckbs.xml
%$$
If you download the latest version (the trunk),
a copy of the documentation for this version can 
be built by executing the commands
$codep
	./build_doc.sh
$$
in the $codei%ckbs/trunk%$$ directory.
You can then view the documentation in your browser by opening the file
$codei%
	ckbs/trunk/doc/ckbs.xml
%$$

$childtable%
	omh/license.omh%
	src/ckbs_t_general.m%
	src/ckbs_nonlinear.m%
	src/ckbs_L1_nonlinear.m%
	src/ckbs_affine.m%
	src/ckbs_affine_singular.m%
	src/ckbs_L1_affine.m%
	omh/utility.omh%
	example/all_ok.m%
	omh/whatsnew.omh%
	omh/wishlist.omh%
	omh/bib.omh
%$$


$end