Tracker Component Library Release 1.0, August 2015

These are the release notes for the first release of the Tracker Component Library. The Tracker Component Library is a collection of Matlab routines (some requiring compilation) for simulating and tracking targets in various scenarios. Due to the complexity of the target tracking problem, a great many routines can find use in other areas including combinatorics, astronomy, and statistics.

To use the full library, add the library to your active path in Matlab and run the CompileCLibraries function. This assumes that a C/ C++ compiler been set up in Matlab for the mex command. This does not require the purchase of any Matlab toolboxes. For those who do not have such a compiler that will work with Matlab, free options are the Windows SDK from Microsoft and XCode for Mac OS X. Note that the command line tools must be installed under Mac OS X for Matlab to be able to use the compiler.

Due to GitHub's limit that files not exceed 100MB, the ephemeris data file for the astronomical code has been omitted. To use any of the files requiring planetary ephemerides, such as solarBodyVec.m, please download the file de430.bsp from http://naif.jpl.nasa.gov/pub/naif/generic_kernels/spk/planets/ and place it in the folder "Astronomical Code/data".

The subroutines in the folders are part of the Distribution A release of the tracker component library. The algorithms have been broken into a number of different categories. Each routine is commented, though when dealing with compilable code without a native Matlab implementation, the comments will not show up using the "help" command in Matlab.

Some routines in the Tracker Component Library are available in compiled as well as native Matlab versions. When both versions are available, comments on the usage of the file are generally more detailed in the Matlab implementation file. If the compilation fails, then make sure that the compiler has been properly setup for Matlab. Additionally, when using older versions of Matlab under Mac OS X, Matlab has numerous problems as described below. Compilation has been tested under Mac OS X 10.10 and Windows 7. Compilation should also work under various versions of Linux and Unix.

The third party code folder consists of open source code with permissive licenses (e.g. not GPL) that are used as subroutines in functions in the Tracker Component Library.

COMPILING WITH MATLAB 2013b UNDER MAC OS X:

No special work should be necessary to compile the Tracker Component Library under Matlab 2015a under Mac OS X or Windows. One the other hand, Matlab 2013b contains a number of bugs in its default compile script that must be corrected in order to successfully compile the code. Most of these have been fixed in Matlab 2014a. Additionally, under both versions of Matlab, one must make sure that the command line versions of the developer tools have been installed.

To compile the code under Mac OS X, make sure that XCode is installed along with the command-line developer tools. XCode can be obtained for free through the App Store. The command line developer tools are not installed by default. To install them, open a terminal window (In /Applications/Utilities/Terminal ) type the command "cc" on the command line and hit return. If the command line developer tools are installed, you will get the message clang: error: no input files If the developer tools are not installed, then a window should open up saying The "cc" command requires the command line developer tools. Would you like to install the tools now? In that case, just click "Install". If the command line tools are not installed, one will generally get an error like xcrun: error: unable to find utility "clang++", not a developer tool or in PATH when trying to compile in Matlab.

Whereas the default compile script for Matlab2014a appears to work properly, the default compile script for Matlab2013b does not work properly under Mac OS X, because Matlab tries to link to an older library that might not be installed. Thus, if one is running Mac OS X 10.8 or later, one will get the error
xcodebuild: error: SDK "macosx10.7" cannot be located when trying to compile. The fix for this is to open the terminal and enter the command open ~/.matlab/R2013b/mexopts.sh This should open the compile script file that is used by Matlab in XCode. The reason that the terminal is used to open the file is because Matlab hides the file inside of an invisible folder in the user's home directory. In the file, change the deployment target and the SDK to the version of Mac OS X being used. For example, if one is using Mac OS X 10.9, then the following lines should be present in various parts of the file (instead of using versions less than 10.9) CC='xcrun -sdk macosx10.9 clang' MACOSX_DEPLOYMENT_TARGET='10.9' CXX='xcrun -sdk macosx10.9 clang++'

Next, Matlab 2013b does not properly define all of the types in its own headers. Thus, the '-U__STDC_UTF_16__' flag is used in all of the mex commands in CompileCLibraries to prevent errors.