 |
Hidden Markov Models and
|
This is the distribution site for the software described in Hidden Markov Models and Dynamical Systems. You can buy the book here. The software implements the algorithms described in the book. You can apply the software to your own data. The author has included the data used for examples in the book, so you can see how the code manipulates the data, creates the figures, and finally formats a copy of the book itself. You may download a pdf document that describes how the author made each of the figures in the book.
The most recent version of the software is in a file called webhmmds.tar.bz2, which you can download by clicking the link or by running wget http://dreedle.fraserphysics.com/webhmmds.tar.bz2. After retrieving the file, the following sequence of commands will build a version of the book:
>tar -xjf webhmmds.tar.bz2
>cd webhmmds
>make
Since the file is 8,861,299 bytes, and "making" the book requires almost 2GB of RAM, more than five hours of CPU time, and many supporting software packages, you may want to try some of the following alternatives:
Instead of downloading the whole package, click on the link to peruse the individual files in your browser.
Tex
algorithms.tex
appendix.tex
continuous.tex
hmmds.bib
hmmkeys.el
introduction.tex
main.tex
real.tex
software.tex
toys.tex
varients.tex
Code
Chestnut (.py files are stored as .zip and must be downloaded
to users machine)
EXT.py
Scalar.py
VARG.py
_init_.py
algorithms_0.py
algorithms_1.py
hmm (.py files are stored as .zip and must be downloaded
to users machine)
ApOb.py
ApTrain.py
DoubleClassify.py
PFsurvey.py
ScalarGaussian.py
score.py
plotscripts (.py files are stored as .zip and must be downloaded
to users machine)
EM1.gpt
EM2.gpt
Hsurvey.gpt
LDA.py
LaserLogLike.gpt
Laser_plots.py
LikeLor.gpt
LyapPlot.gpt
PFsurvey.gpt
SGO.py
STSintro.gpt
TSintro.py
ToyA.py
TrainChar.gpt
apnea_sgram.py
apnea_ts_plots.py
class1.py
gaussmix.gpt
gaussmixA.gpt
stateplot.py
python (.py files are stored as .zip and must be downloaded
to users machine)
EKF.py
Hsurvey.py
Hview.py
Laser_data.py
Ltest.c
LyapPlot.py
MakeModel.py
Sparse_hmm_lor.py
StatePic.py
TrainChar.py
VStatePic.py
cinc2000.py
em.py
lorenz.c
lorenz.py
po_speech.py
respire.py
rr2hr.py
xfigs
Markov_dhmm.fig
Markov_dhmm_net.fig
Markov_mm.fig
QR.fig
ScalarGaussian.fig
ToyCross.fig
ToyPractical.fig
dummy.fig
forward.fig
nonmn.fig
sequenceMAP.fig
sgauss1.fig
sgauss2.fig
structure.fig
viterbiB.fig
Data
Apnea
a01.Rtimes
a02.Rtimes
a03.Rtimes
a04.Rtimes
a05.Rtimes
a06.Rtimes
a07.Rtimes
a08.Rtimes
a09.Rtimes
a10.Rtimes
a11.Rtimes
a12.Rtimes
a13.Rtimes
a14.Rtimes
a15.Rtimes
a16.Rtimes
a17.Rtimes
a18.Rtimes
a19.Rtimes
a20.Rtimes
b01.Rtimes
b02.Rtimes
b03.Rtimes
b04.Rtimes
c01.Rtimes
c02.Rtimes
c03.Rtimes
c04.Rtimes
c05.Rtimes
c06.Rtimes
c07.Rtimes
c08.Rtimes
c09.Rtimes
c10.Rtimes
x01.Rtimes
x02.Rtimes
x03.Rtimes
x04.Rtimes
x05.Rtimes
x06.Rtimes
x07.Rtimes
x08.Rtimes
x09.Rtimes
x10.Rtimes
x11.Rtimes
x12.Rtimes
x13.Rtimes
x14.Rtimes
x15.Rtimes
x16.Rtimes
x17.Rtimes
x18.Rtimes
x19.Rtimes
x20.Rtimes
x21.Rtimes
x22.Rtimes
x23.Rtimes
x24.Rtimes
x25.Rtimes
x26.Rtimes
x27.Rtimes
x28.Rtimes
x29.Rtimes
x30.Rtimes
x31.Rtimes
x32.Rtimes
x33.Rtimes
x34.Rtimes
x35.Rtimes
LP5.DAT
pfsurvey.txt
Save Hview T 100.txt
Save Hview T 118.txt
Save Hview T 119.txt
a03er.seg.txt
em yt.txt
event-2-answers.txt
gutenberg-19355.txt
summary of training.txt
- make -j12
- Instead of make tells gnu-make to run up to 12 processes simultaneously. On my 8 core PC with 8 GB of RAM the -j12 option reduces the build time from more than 5 hours to one hour and one minute
- make data/po_speech
- Makes only the data for Table 1.1 which depends on very little beyond the basic HMM programs
- make figs/Statesintro.pdf
- Makes only the figure on the cover. It depends on wrapping gsl code for integrating the Lorenz system and the python matplotlib package.
Link to obsolete versions to come.