Stochastic computational model parameter synthesis system

• US 9,558,300 B2
• Filed: 11/09/2012
• Issued: 01/31/2017
• Est. Priority Date: 11/11/2011
• Status: Active Grant

• Current Assignee: Carnegie Mellon University
• Original Assignee: Carnegie Mellon University, University of Central Florida Research Foundation Inc. (State University System of Florida)
• Inventors: Jha, Sumit K., Langmead, Christopher J.
• Primary Examiner(s): Rivas, Omar Fernandez
• Assistant Examiner(s): Khan, Iftekhar
• Application Number:
  US13/673,575
• Publication Number:
  US 20130124164A1
• Time in Patent Office: 1,544 Days
• Field of Search: 703/2
• US Class Current: 1/1
• CPC Class Codes: G06F 2111/08 Probabilistic or stochastic CAD
• CPC Class Codes: G06F 30/00 Computer-aided design [CAD]

Abstract

A stochastic computational model parameter synthesis system comprising at least one processor capable of executing processor executable code, and a non-transitory computer memory operably coupled with the at least one processor and storing processor executable code, which when executed by processor, causes processor to synthesize at least one parameter of a stochastic computational model to satisfy one or more behavioral specifications of properties observed in a modeled system. The processor generates and searches randomized projections of a first parameter space having n dimensions into one or more second abstract parameter space having d dimensions, where d is less than n, and outputs a signal indicative of a synthesized parameter value to the user.

Claims

1. A stochastic computational model parameter synthesis system comprising at least one processor capable of executing processor executable code, and a non-transitory computer memory operably coupled with the at least one processor and storing processor executable code, which when executed by the at least one processor, causes the at least one processor to create at least one synthesized parameter value of a stochastic computational model to satisfy one or more behavioral specifications of properties observed in a modeled system by:

receiving a first input signal from a user over a network interface, the first input signal indicative of the stochastic computational model to be carried out by the at least one processor, the stochastic computational model having n parameters in a first parameter space, the first parameter space being n-dimensional;

receiving a second input signal indicative of at least one behavioral specification corresponding to at least one property observed in the modeled system;

receiving a third input signal from the user over a network interface, the third input signal indicating a minimum probability with which the stochastic computational model must satisfy the at least one behavioral specification;

generating one or more second parameter space having d parameters from the first parameter space having n parameters where d is less than n, the one or more second parameter space being d-dimensional;

generating one or more randomized projections of the first parameter space into the one or more second parameter space, the randomized projections probabilistically associated with n+1 linearly independent anchor points in the first parameter space;

searching the one or more second parameter space by exhaustively searching discretized points corresponding to at least one parameter value in the one or more second parameter space to identify the discretized point with a highest probability of satisfying the at least one behavioral specification;

wherein the one or more randomized projections preserve a distance between the n+1 linearly independent anchor points in the first parameter space and the discretized point corresponding to the at least one parameter value with the highest probability of satisfying the at least one behavioral specification in the one or more second abstract parameter space;

verifying that the probability of the discretized point with the highest probability of satisfying the at least one behavioral specification is at least equal to or greater than the minimum probability input by the user;

creating the at least one synthesized parameter value in the first parameter space from the verified discretized point in the second parameter space to satisfy the one or more behavioral specifications observed in the modeled system.

2. The system of claim 1, wherein distances of discretized points from projections of n+1 anchor points are used to identify candidate parameter values in the first parameter space.

3. The system of claim 1, wherein randomized metric embeddings matrices are used to construct the one or more second parameter space.

4. The system of claim 1, wherein the randomized projections are generated by two or more parallel computer processors.

5. The system of claim 1, wherein the modeled system is a physical system.

6. The system of claim 5, wherein the modeled system is a biological system.

7. The system of claim 6, wherein the biological system is described using an agent-based model.

8. The system of claim 7, wherein the at least one synthesized parameter value is a kinetic parameter of the biological system.

9. The system of claim 1, wherein the modeled system is a non-physical system.

10. The system of claim 9, wherein the modeled system is a financial trading system.

11. The system of claim 10, wherein the at least one synthesized parameter value is a risk of at least one asset in an investment portfolio.

12. The system of claim 1, wherein the stochastic computational model is selected from a continuous time Markov chain, a discrete time Markov chain, a stochastic differential equation, and a jump diffusion process.

13. The system of claim 1, wherein the property specification is an adapted finitely monitorable specification.