An organization is seeking advanced, novel software (or tested algorithms that they can implement) to optimize the output of a highly complex, multi-input, multi-output simulation model. The time-based simulation accepts a large number of input parameters in order to predict the overall performance of a highly complex, multi-disciplinary model for a small number of output variables.
Details of the Technology Call
The simulation is focused on applying multiple sets of inputs to provide output trends over time, thus encouraging innovative and effective strategies and resource allocation levels. A traditional multi-critical decision analysis (MCDA) approach would require decision ranking and prioritization within multiple scenarios (the initial model states). It is critical to identify the most effective input parameter configurations to perturb the system toward desirable states.
The required software/algorithms would need to 1) monitor the outputs (an arbitrary subset of variables within the model) given a set of input variables over time and, 2) iteratively modify the inputs to achieve optimization of the outputs in order to maximize user-defined goals.
How we'll judge a successful solution:
Your proposed solution must clearly demonstrate its capabilities/limitations in handling highly complex dynamic systems of different types (linear, non-linear, and so forth).
Your proposed solution must:
1) Suggest potential configurations of input variables (which can change over time) to influence the system to reach desirable states.
2) Provide ranked and diversified potential configurations of the input variables (candidate solutions).
3) Provide ways to find relatively optimal candidate solutions with a small number of iterations.
There is a strong preference for a solution coded in Java (or pseudocode in the case of algorithms).
The organization is not interested in common existing solutions for multi-objective optimization. However, your solution may be based on or combine various existing multi-objective optimization approaches.
Source code, algorithms, and any programs demonstrating the solution must be disclosed, open-source, and non-proprietary.
The organization requires full rights to the technology, including the source code of any pre-compiled libraries.
POSSIBLE SOLUTION AREAS
Financial and econometric modeling
Mining and geo-data modeling
Automation and motor control software
Industrial process control
Field Of Use and Intended Application
Multiple-criteria decision making/ multiple objective decision making (MCDM/MODM) for analysis and prediction of behavior trends of various dynamic systems.
The organization would like to find existing software (or algorithms that they can implement) to optimize output performance of a multi-input/multi-output simulation that is composed of integrated, highly complex, and multi-disciplinary models. Your accepted solution will be incorporated into an existing architecture.