*Note* This software is still somewhere between alpha and beta testing - it works and we use it constantly, but it has a lot of quirks, "features", and missing bits. At least there is some documentation now... Installation instructions for the FIR filter optimization package: 1) Get one of the distribution files, either Opt-X.Y.tar.gz or Opt-X.Y.zip, from http://www.csee.umbc.edu/~dschol2/opt.html and unpack it into any convenient directory. The directory Opt should be created which is the top of the source tree. If unpacking Opt.zip under Windows it is important that the directory structure be preserved. 2) Add the Opt directory to your Matlab path. You must have Matlab 5.0 or higher, due to the heavy use of the class structure. 3) Install any optimization engine packages that will be used. Currently supported: SeDuMi, LOQO, MOSEK, SDPpack, SDPT3 and SOCP. The web page has the links. Follow the directions for each package. Make sure the main solver routines are on your Matlab path. 4) Compile mex files. Currently there are: - wherein.c in the top-level directory - pwrengine.c in @CTProcess/private - pwrengine.c in @NDProcess/private - ImpulseTime.c in @NDProcess/private - fourierEngine.c in @optArray/private If you are using Linux, we've provided a Makefile to automate this process. Simply type "make" from the top-level directory. That's pretty much it for installation. Probably the hardest part, especially on a Windows PC, is compiling the mex files for the engines. To test the implementation, go to the examples directory and try running the matlab scripts found there. test.m, test2.m: compare the same design using eigenfilter routines and the loqo QP interface testloqo.m : tests out the loqo QP interface testsocp.m,testgrid.m : tests all engines for LS and Linf designs It is very possible that somewhere in all this I use some routines that are not a standard part of Matlab, either from a toolbox or my own routines. Please let me know if you find one so I can fix it. Dan Scholnik scholnik@nrl.navy.mil 3/08/2004