The stochastic optimization option facilitates optimization over an array of random or user-specified scenarios without first arraying the model. The total payoff is the sum of the payoffs of the individual runs in the sensitivity ensemble.
This feature is simple to use. Choose the Stochastic option in the optimization control dialog, or add the keyword :STOCHASTIC to an existing optimization control file (.voc), and provide a savelist (.lst) and sensitivity control file (.vsc) in the simulation control dialog. When the optimizer encounters the :STOCHASTIC keyword, it will automatically use the sample specified in the sensitivity controls for every point in the optimization.
Example:
:OPTIMIZER=Powell
:SENSITIVITY=Off
:MULTIPLE_START=Off
<…snip…>
:VECTOR_POINTS=25
:STOCHASTIC
-2.048<=x[i]=0<=2.048
The aggregate payoff seen by the optimizer is the sum of the payoffs of the individual ensemble simulations. If one simulation in a sensitivity sample fails due to a floating point error, the entire ensemble is considered to fail.
If the sensitivity setup uses the File option, the input sample is cached for speed. If the sample size exceeds storage capacity, it will be downsampled such that each row has equal probability of remaining in the sample.
Examples
See the examples in OptSensi .
