STRESS TESTING BY AVOIDING SIMULATIONS

• US 20170004226A1
• Filed: 03/28/2016
• Published: 01/05/2017
• Est. Priority Date: 07/05/2015
• Status: Abandoned Application

• Current Assignee: SAS Institute Incorporated
• Original Assignee: SAS Institute Incorporated
• Inventors: Skoglund, Klas Jimmy, Chen, Wei
• Application Number:
  US15/082,240
• Publication Number:
  US 20170004226A1
• Time in Patent Office: Days
• US Class Current: 1/1
• CPC Class Codes: G06F 17/18 for evaluating statistical data , e.g. average values, frequency distributions, probability functions, regression analysis forecasting specially adapted for a specific administrative, business or logistic context G06Q10/04
• CPC Class Codes: G06F 30/00 Computer-aided design [CAD]
• CPC Class Codes: G06N 7/01 Probabilistic graphical models, e.g. probabilistic networks
• CPC Class Codes: H04L 67/10 in which an application is distributed across nodes in the network software deployment G06F8/60; multiprogramming arrangements G06F9/46

Abstract

Systems, methods, and computer program products are provided that perform modeling and stress testing algorithms without the need for running simulations and that provide exact or approximate solutions for predicting outcomes of states and distributions of states for components of a structure. The disclosed systems, methods, and products may employ a Markov iteration approach, such as an exact Markov iteration approach or a reduced or simplified Markov iteration approach for predicting states and distributions of states for components of a structure using an algorithm that reduces solution complexity as compared to approaches that employ simulations.

Claims

1. A stress testing system comprising:

one or more processors; and

a non-transitory computer readable storage medium including instructions that, when executed by the one or more processors, cause the one or more processors to perform operations including;

receiving a structure definition for a structure, wherein the structure includes a plurality of components, wherein the structure definition identifies characteristics of components in the structure, and wherein characteristics include a component state and a component transition history;

determining a stress scenario specification, wherein the stress scenario specification relates to time period dependent stress conditions that affect changes to characteristics;

iteratively determining transition matrices for each of a plurality of time periods and component transition histories using the stress scenario specification, wherein a transition matrix includes transition intensities, wherein a transition intensity corresponds to a likelihood that a component of the structure will change from an initial component state to a future component state within one time period, and wherein determining an individual transition matrix for a particular time period includes;

identifying allowable transitions between each component state; and

identifying transition intensities for each allowable transition using the stress scenario specification for the particular time period and the component transition histories;

determining an initial distribution of component states at an initial time, wherein determining includes using the structure definition; and

generating an output flow using the transition matrices and the initial distribution of component states, wherein the output flow provides a distribution of predicted future component states for each of the plurality of time periods.

2. The system of claim 1, wherein determining the stress scenario specification includes receiving the stress scenario specification.

3. The system of claim 1, wherein determining the stress scenario specification includes receiving a stress projection and generating the stress scenario specification using the stress projection.

4. The system of claim 3, wherein the stress projection provides macro-scale conditions for affecting the changes to characteristics of components of the structure and wherein generating the stress scenario specification includes identifying micro-scale conditions for affecting changes to characteristics of components of the structure.

5. The system of claim 1, wherein the stress scenario specification identifies predicted time period dependent stress conditions.

6. The system of claim 1, wherein a transition intensity is a transition probability.

7. The system of claim 1, wherein the transition matrices are dependent on component transition histories.

8. The system of claim 1, wherein determining an individual transition matrix includes:

generating a component state dependent transition model; and

determining transition intensities using the state dependent transition model and the stress scenario specification.

9. The system of claim 1, wherein iteratively determining individual transition matrices includes evaluating a Markov state transition model.

10. The system of claim 1, wherein determining an individual transition matrix includes generating a time dependent component state transition model using the stress scenario specification.

11. The system of claim 1, wherein a structure corresponds to a group of accounts.

12. The system of claim 1, wherein a component corresponds to an account.

13. The system of claim 1, wherein a component state identifies which of a plurality of conditions the component is associated with at a particular time.

14. The system of claim 1, wherein a component transition history identifies historical component states and transitions between states for the component.

15. The system of claim 1, wherein a component characteristic includes a value of a component.

16. The system of claim 1, wherein the output flow is used to facilitate determination of required reserves for a holder of the structure based on the definition of the structure and the stress scenario specification.

17. The system of claim 1, wherein the output flow is used to facilitate determination of predicted future values for one or more components of the structure.

18. The system of claim 1, wherein generating the output flow includes generating the output flow without requiring individual simulations of predicted future characteristics for each of the components of the structure.

19. The system of claim 1, wherein generating the output flow includes determining products of a first transition matrix corresponding to a first time period and the initial distribution of component states to generate a first distribution of characteristics for components of the structure after the first time period.

20. The system of claim 19, wherein generating the output flow includes determining products of a second transition matrix corresponding to a second time period and the first distribution of characteristics for components of the structure after the first time period to generate a second distribution of characteristics for components of the structure after the second time period.

21. The system of claim 1, wherein generating the output flow includes computing a Markov iteration for each of the plurality of time periods.

22. The system of claim 1, wherein the operations further include:

determining a time dependent growth rate, wherein generating the output flow includes using the time dependent growth rate, and wherein a time dependent growth rate provides rates at which a component characteristic increases over time.

23. The system of claim 1, wherein the operations further include:

determining a time dependent decay rate, wherein generating the output flow includes using the time dependent decay rate, and wherein a time dependent decay rate provides rates at which a component characteristic decreases over time.

24. A computer-program product tangibly embodied in a non-transitory machine-readable storage medium, including instructions configured to cause a computing device to perform operations including:

receiving, at the computing device, a structure definition for a structure, wherein the structure includes a plurality of components, wherein the structure definition identifies characteristics of components in the structure, and wherein characteristics include a component state and a component transition history;

determining a stress scenario specification, wherein the stress scenario specification relates to time period dependent stress conditions that affect changes to characteristics;

iteratively determining transition matrices for each of a plurality of time periods using the stress scenario specification and component transition histories, wherein a transition matrix includes transition intensities, wherein a transition intensity corresponds to a likelihood that a component of the structure will change from an initial component state to a future component state within one time period, and wherein determining an individual transition matrix for a particular time period includes;

identifying allowable transitions between each component state; and

identifying transition intensities for each allowable transition using the stress scenario specification for the particular time period and the component transition histories;

determining an initial distribution of component states at an initial time, wherein determining includes using the structure definition; and

generating an output flow using the transition matrices and the initial distribution of component states, wherein the output flow provides a distribution of predicted future component states for each of the plurality of time periods.

25. The computer-program product of claim 24, wherein determining the stress scenario specification includes receiving the stress scenario specification.

26. The computer-program product of claim 24, wherein determining the stress scenario specification includes receiving a stress projection and generating the stress scenario specification using the stress projection.

27. The computer-program product of claim 26, wherein the stress projection provides macro-scale conditions for affecting the changes to characteristics of components of the structure and wherein generating the stress scenario specification includes identifying micro-scale conditions for affecting changes to characteristics of components of the structure.

28. The computer-program product of claim 24, wherein the stress scenario specification identifies predicted time period dependent stress conditions.

29. The computer-program product of claim 24, wherein a transition intensity is a transition probability.

30. A computer implemented stress testing method, comprising:

receiving, at a computing device, a structure definition for a structure, wherein the structure includes a plurality of components, wherein the structure definition identifies characteristics of components in the structure, and wherein characteristics include a component state and a component transition history;

determining a stress scenario specification, wherein the stress scenario specification relates to time period dependent stress conditions that affect changes to characteristics;

iteratively determining transition matrices for each of a plurality of time periods using the stress scenario specification and component transition histories, wherein a transition matrix includes transition intensities, wherein a transition intensity corresponds to a likelihood that a component of the structure will change from an initial component state to a future component state within one time period, and wherein determining an individual transition matrix for a particular time period includes;

identifying allowable transitions between each component state; and

identifying transition intensities for each allowable transition using the stress scenario specification for the particular time period and the component transition histories;

determining an initial distribution of component states at an initial time, wherein determining includes using the structure definition; and

generating an output flow using the transition matrices and the initial distribution of component states, wherein the output flow provides a distribution of predicted future component states for each of the plurality of time periods.