Analytical model solver framework

• US 8,117,145 B2
• Filed: 06/27/2008
• Issued: 02/14/2012
• Est. Priority Date: 06/27/2008
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Rubin, Darryl E., Mital, Vijay, Beckman, Brian C.
• Primary Examiner(s): Gaffin, Jeffrey A
• Assistant Examiner(s): BUSS, BENJAMIN J
• Application Number:
  US12/163,949
• Publication Number:
  US 20090327183A1
• Time in Patent Office: 1,327 Days
• Field of Search: 706/14, 706/45, 706/47, 703/2, 703/6, 717/104, 717/105
• US Class Current: 706/46
• CPC Class Codes: G06F 30/00 Computer-aided design [CAD]

Abstract

A solver framework for use with an analytical model. The analytical model includes multiple model parameters and includes definitions for analytical relationships between the model parameters. The solver framework coordinates the processing of multiple specialized solvers. In particular, the solver framework identifies which model parameters are input model variables and which are output model variables. The solver framework then analyzes dependencies to determine a solve order to solve for the output model variables. The solver framework then charged the specialized solvers with performing portions of the solve operation such that the specialized solvers solve for the output model variables in an order which considers the dependencies. In one embodiment, additional or replacement solvers may register with the solver framework to thereby make the specialized solver available for solving for output model variables in the future.

Claims

1. A computer program product comprising one or more physical computer-readable media have thereon computer-executable instructions that, when executed by one or more processors of the computing system, cause the computing system to generate an interpretive modeling system that is readily adaptable at runtime to any of a diverse variety of domain specific environments without having to perform any coding to adapt the interpretive modeling system for a given type of domain, the interpretive modeling system generating a composite view of visual items with which a user can interpretively interact, and wherein the interpretive modeling system comprises:

an analytics portion that defines analytical relationships between a plurality of model parameters using a plurality of analytical relations, wherein the analytical modeling component uses the plurality of analytical relations to (i) identify which of the plurality of model parameters are known and which are unknown, and then (ii) solve for any identified unknown model parameter(s), at least some of the plurality of model parameters comprising one or more input model variables and one or more output model variables;

a collection of a specialized solvers, including one or more constraints solvers, one or more equations solvers, and one or more rules solvers, that are each capable of solving for the one or more output model variables given the one or more input model variables, the analytical relations defining relationships between the plurality of the model parameters including the one or more input model variables and the one or more output model variables;

a data portion that includes a data-model binding component configured to bind portions of a plurality of data items to corresponding model parameters of the analytical modeling component;

a view portion that includes (i) a view components repository containing a plurality of heterogenic view components, each corresponding to a visual item that may be displayed, and at least some of which are parameterized, and (ii) a model-visual binding component configured to bind the plurality of model parameter values to one or more parameters of at least one of the parameterized view components contained within the view components repository, and (iii) a view composition module configured to formulate instructions for rendering a view composition that, at least in part, depends on the solved one or more output model variables and the one or more input model variables and that contains at least some of the visual items corresponding to the plurality of heterogenic view components; and

a solver framework configured to perform the following in response to accessing the one or more input model variables;

1) an act of coordinating processing by the specialized solvers in the collection to thereby cause the collection of specialized solvers to solve for the one or more output model variables, wherein coordinating said processing by the specialized solvers includes performing, in order, the following;

an act of solving for constraints using the one or more constraints solvers;

an act of solving for equations using the one or more equations solvers after the act of solving for constraints; and

an act of solving for rules using the one or more rules solvers after the act of solving for the equations; and

2) an act of outputting the one or more solved output model variables to the view composition model to control, at least in part, rendering of the view composition.

2. The computer program product as recited in claim 1, wherein the specialized solvers collectively solve only partially for the one or more output model variables.

3. The computer program product as recited in claim 1, wherein the specialized solvers collectively fully solve for the one or more output model variables.

4. The computer program product as recited in claim 1, wherein the collection of specialized solvers also include one or more simulation solvers.

5. The computer program product as recited in claim 1, wherein the act of coordinating processing comprises:

an act of identifying the one or more input model variables;

an act of identifying the one or more output model variables; and

based on the identity of the one or more input model variables, the identity of the one or more output model variables, and the analytical relations, an act of analyzing dependencies in the plurality of model parameters.

6. The computer program product as recited in claim 5, wherein the dependencies change based on the identity of the one or more input model variables.

7. The computer program product as recited in claim 5, wherein the act of coordinating processing further comprises:

an act of identifying one or more fully independent systems; and

an act of solving for the one or more fully independent systems.

8. The computer program product as recited in claim 1, wherein the act coordinating processing further comprises:

an act of repeating the acts of solving for constraints, solving for equations, and solving for rules, until the specialized solvers cause all solved values of the one or more output model variables to become consistent with the constraints, the equations, and the rules.

9. The computer program product as recited in claim 1, wherein the computer-executable instructions are further structured such that, when executed by the one or more processors of the computing system, the computing system is caused to further instantiate the following:

a solver registration component configured to register a new specialized solver with the solver framework such that the new specialized solver may be incorporated into the collection of specialized servers to thereby be managed by the solver framework.

10. The computer program product as recited in claim 1, wherein, when the collection of specialized solvers solve for the one or more output model variables, the analytical relations change.

11. The computer program product as recited in claim 1, wherein each specialized solver is configured to solve for output model variables used in a different type of analytical relation.

12. The computer program product as recited in claim 1, wherein the plurality of analytical relations include a set of one or more modifiable rules, and wherein causing the collection of specialized solvers to solve for the one or more output model variables comprises one or more of:

activating at least one rule in the set of one or more modifiable rules;

deactivating at least one rule in the set of one or more modifiable rules;

or

modifying an equation uses as part of the set of one or more modifiable rules.

13. The computer program product as recited in claim 1, wherein the identity of model parameters comprising the one or more input model variables and the identity of model parameters comprising the one or more output model variables is determined based on user input received through the view composition.

14. The computer program product as recited in claim 1, wherein the solver framework causes the collection of specialized solvers to re-solve for the one or more output model variables in response to user input received through the view composition.

15. In a computing system, a computer-implemented method for causing the computing system to generate an interpretive modeling system that is readily adaptable at runtime to any of a diverse variety of domain specific environments without having to perform any coding to adapt the interpretive modeling system for a given type of domain, the interpretive modeling system generating a composite view of visual items with which a user can interpretively interact, and wherein the method is comprised of the following acts:

instantiating within the computing system an analytics portion that defines analytical relationships between a plurality of model parameters using a plurality of analytical relations, wherein the analytical modeling component uses the plurality of analytical relations to (i) identify which of the plurality of model parameters are known and which are unknown, and then (ii) solve for any identified unknown model parameter(s), at least some of the plurality of model parameters comprising one or more input model variables and one or more output model variables;

instantiating within the computing system a collection of a specialized solvers, including one or more constraints solvers, one or more equations solvers, and one or more rules solvers, that are each capable of solving for the one or more output model variables given the one or more input model variables, the analytical relations defining relationships between the plurality of the model parameters including the one or more input model variables and the one or more output model variables;

instantiating within the computing system a data portion that includes a data-model binding component configured to bind portions of a plurality of data items to corresponding model parameters of the analytical modeling component;

instantiating within the computing system a view portion that includes (i) a view components repository containing a plurality of heterogenic view components, each corresponding to a visual item that may be displayed, and at least some of which are parameterized, and (ii) a model-visual binding component configured to bind the plurality of model parameter values to one or more parameters of at least one of the parameterized view components contained within the view components repository, and (iii) a view composition module configured to formulate instructions for rendering a view composition that, at least in part, depends on the solved one or more output model variables and the one or more input model variables and that contains at least some of the visual items corresponding to the plurality of heterogenic view components; and

instantiating within the computing system a solver framework configured to perform the following in response to accessing the one or more input model variables;

1) an act of coordinating processing by the specialized solvers in the collection to thereby cause the collection of specialized solvers to solve for the one or more output model variables, wherein coordinating said processing by the specialized solvers includes performing, in order, the following;

;

an act of solving for constraints using the one or more constraints solvers;

an act of solving for equations using the one or more equations solvers after the act of solving for constraints; and

an act of solving for rules using the one or more rules solvers after the act of solving for the equations; and

2) an act of outputting the one or more solved output model variables to the view composition model to control, at least in part, rendering of the view composition.

16. The method in accordance with claim 15, further comprising:

one or more of;

the solver framework fully solving for the one or more output model variables, or the solver framework partially solving for the one or more output model variables; and

the solver framework performing the act of outputting the one or more solved output model variables to the view composition model each time the solver framework detects that (i) a previously known model parameter value has become unknown, or that (ii) a model parameter value has changed such that one or more other values dependent on the model parameter value should be re-calculated.

17. The method in accordance with claim 16, wherein the solver framework also performs the act of outputting the one or more solved output model variables to the view composition model each time the solver framework detects that a previously known model parameter value has become unknown or has been designated as an input model variable.

18. The method in accordance with claim 15, wherein the act of coordinating processing of at least some of the of specialized solvers in the collection to thereby cause the collection of specialized solvers to solve for the one or more output model variables comprises the following:

based on the identity of the one or more input model variables, the identity of the one or more output model variables, and the analytical relations, an act of analyzing dependencies in the plurality of model parameters; and

an act of determining an order of execution of the specialized solvers based on the analyzed dependencies.