This app provides an interactive environment to solve deterministic optimization problems with multiple local optima, problems with little structural
properties, and black-box optimization problems, using the Gradient-based adaptive stochastic search (GASS) method (you can find more details in publication or an overview in slides).
This method iteratively finds high quality solutions by sampling from a parameterized probabilistic model and updating the parameter
of the probabilistic model by direct gradient search. A list of algorithms provided for different types of problems is shown below.
Installation Instruction
This software can be installed as an application in Matlab or as a standalone application on computer. The instruction below is based on Matlab 2012b.
Application in Matlab
1. Download the installation file here: GASS.mlappinstall
2(a). Double click on the installation file, and it will open Matlab and guide you through to install the app.
2(b). Or, instead of step 2(a), you can open Matlab --> click on the tab "APPS" --> click on "Install App", which will guide you through to install the app.
3. The app "GASS" should be installed immediately and appear under "MY APPS" category in Matlab.
Standalone Application
1. Download the installation file here: GASS_pkg.exe.
2. Double click on the installation file to extract files. It will also lead to an installation wizard.
3. To run standalone applications written by MATLAB, users have to install MCR (MATLAB Compiler Runtime) first.
Following the instructions in the installation wizard or in readme.txt to install the correct version of MCR. The installation of MCR may take some time.
4. After that users can use GASS as a standalone application by double clicking on .
User Manual
Main Dialog
Once opened, this is the main dialog. It allows you to write their own objective function in MATLAB and input the function name in "Set Objective".
Then users can choose to max or min this objective function, set problem type as continuous or discrete, choose the constraint type, and select a method.
Parameter Dialog
By clicking on the button "Parameters", you open the dialog window below to set the parameters that are necessary
for solving the objective function. Valid parameter values have to be entered here; otherwise, errors will be reported.
By clicking on the button "Advance Settings", you open the dialog window below for complete parameter setting.
You can use the default advanced settings, but sometimes you can get better results by fine tuning the algorithm parameters.
To understand more about the advanced parameter settings, please refer to the paper GASS.pdf.
Result Dialog
This dialog provides information such as the average number of iterations, optimal objective function value found,
and outputs optimal solution in GASS_solution.mat file under the current directory.
A plot shows the average function value found as the algorithm runs.
Example
After writing a user-defined objective function, input the function name, choose a solving method, set parameters, and click on "Solve" to solve this objective function.
.
Parameter Dialog
Example
