Adaptive Canonical Representation for Constrained Random Simulation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Constraint satisfaction problem in constrained random simulation is NP-complete, leading to performance issues in stimulus generation, as constraint solvers struggle to generate reproducible and uniformly distributed random solutions efficiently, lagging behind the speed of simulation.
Innovation Solution
Adaptive state-to-symbolic transformation in canonical representations, such as Binary Decision Diagrams (BDDs), where the system monitors parameters to dynamically add or remove state variables, optimizing the canonical representation to improve performance by intelligently managing the size and complexity of the representation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If state variables are added to the canonical representation to improve constraint solving performance, then the performance of random stimulus generation is improved, but the size of the canonical representation increases
Solution Approach 1:
The patent applies dynamics by making the canonical representation adaptive and changeable during simulation. The system dynamically adds or removes state variables from the canonical representation based on monitored parameters such as simulation progress and performance metrics. This allows the representation to evolve from a static structure to a dynamic one that optimizes performance while managing size, directly resolving the contradiction between improving constraint solver performance and controlling representation size.
2Measurement precision
If the canonical representation includes more state variables, then the accuracy of constraint satisfaction is improved, but the complexity of the representation increases
Solution Approach 1:
The patent applies parameter changes by dynamically adjusting the composition of the canonical representation based on monitored parameters. The system changes which state variables are included in the representation based on simulation progress, performance metrics, and constraint satisfaction requirements. This allows the system to maintain high accuracy by including necessary state variables while managing complexity by excluding unnecessary ones, directly addressing the contradiction between accuracy and complexity.
3Speed
If the canonical representation is optimized for performance, then the speed of stimulus generation is improved, but the uniformity of random solutions may be compromised
Solution Approach 1:
The patent applies feedback by continuously monitoring parameters during constrained-random simulation and using this information to adaptively adjust the canonical representation. The system monitors both performance metrics (for speed optimization) and constraint satisfaction quality (for uniformity assurance). This feedback loop ensures that optimizations for speed do not compromise the uniformity of random solutions, as the system can detect and correct deviations from uniform distribution by adjusting the representation accordingly.
Data Source
AI summary
Some embodiments provide a system for adaptively performing state-to-symbolic transformation in a canonical representation which is used for generating random stimulus for a constrained-random simulation. The system can construct a canonical representation for a set of constraints using the set of random variables and the subset of the state variables in the constraints. Next, the system can use the canonical representation to generate random stimulus for the constrained-random simulation, and monitor parameters associated with the constrained-random simulation. Next, the system can add state variables to or remove state variables from the canonical representation based at least on the monitored parameters. The system can then use the modified canonical representation which has a different set of state variables to generate random stimulus for the constrained-random simulation.