Question: What is the difference between CGM and CGM-RS?
Question: Can the setup steps in MVO be saved, reused, and automated?
Question: Can I add my own optimization algorithm to MVO?
Question: How must my data be formated for MVO import?
What is the difference between CGM and CGM-RS?
Concurrent Gradient Method and Concurrent Gradient Method: Response Surface are approaches for finding improved or optimal designs. They employ different methods for calculating gradients.
GGM uses a finite difference method to calculate a gradient. CGM-RS, uses a response surface. There is higher degree of assumption with CGM-RS compared to CGM. Because of this, CGM will be more accurate than CGM-RS. But, CGM is more time consuming. Generally CGM-RS should be used when the dimensionality of the problem is at or above 10 to 15 variables.
Using CGM-RS is able to estimate the most significant projections of gradients. Accomplished in 5 evaluations. The apporximation occurs when the building of the gradient doesn't include all variables, only the most significant ones.
Can the setup steps in MVO be saved, reused, and automated?
In the Controls, the user specifies variabled limits, variable initial conditions, objective functions to be optimized, what algorithm to use, and paraters of tuning the algorith. Changes to everything in the Controls is saved when a solution is saved. To reuse a setup, save it as a solution.
Can I add my own optimization algorithm to MVO?
Located at \Multistat\MTVOptimizer 4.00\DemoOptimizations, there is sample C++ code for your use in creating your own algorithms to be incorporated to MVO. The process is edit, compile, register, then it will be included in the drop down list of available optimization algorithms.
How must my data be formated for MVO import?
Multistat Visual Optimizer gives you the following data import options: