Asynchronous Archipelago Model for Distributed Evolutionary Computation

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Distributed evolutionary computation techniques face challenges such as convergence to local optima and inefficiency in handling large populations, requiring synchronized processing capabilities and stable island connections, which limits their robustness and scalability.

Innovation Solution

The implementation of an archipelago model with an archipelago manager computing device and island computing devices connected via a network, allowing for asynchronous and modular distribution of candidate solutions through an archipelago management queue, enabling robustness to processing power differences and device failures, and facilitating global optimization without specific island connection structures.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Power

If distributed evolutionary computation techniques are implemented with synchronized processing, then processing capability is improved, but robustness to device failures and processing power differences deteriorates

Engineering Contradiction:

Improveprocessing capabilityVSAvoidrobustness to device failures

Core Design Contradiction:

PowerVSReliability

Solution Approach 1:

The system segments the evolutionary computation population into multiple islands distributed across different computing devices. Each island independently processes a subset of the population, eliminating the need for synchronized processing while maintaining distributed processing power. This segmentation allows devices with different processing capabilities to operate autonomously without requiring coordination.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

An archipelago manager acts as an intermediary between island computing devices, managing candidate solution exchanges without requiring direct peer-to-peer communication or synchronization. The manager coordinates the migration of candidate solutions between islands, enabling robust distributed processing where devices can operate independently while still contributing to the overall evolutionary computation process.

Inventive Principle:

Principle #24Intermediary (Mediator)

2Manufacturing precision

If stable island connections are maintained, then convergence accuracy is improved, but scalability to large populations deteriorates

Engineering Contradiction:

Improveconvergence accuracyVSAvoidscalability to large populations

Core Design Contradiction:

Manufacturing precisionVSProductivity

Solution Approach 1:

The system transitions from direct peer-to-peer island connections to a hierarchical structure where the archipelago manager provides a central coordination layer. This dimensional change in the communication architecture allows the system to scale to large populations by distributing the coordination burden, enabling more islands to be managed without requiring each island to maintain stable direct connections with all other islands.

Inventive Principle:

Principle #17Another dimension (Dimensionality change)

3Manufacturing precision

If specific island connection structures are used, then convergence control is improved, but system complexity and robustness deteriorates

Engineering Contradiction:

Improveconvergence controlVSAvoidconnection structure complexity

Core Design Contradiction:

Manufacturing precisionVSDevice complexity

Solution Approach 1:

The archipelago manager serves as a mediator that simplifies the connection structure by providing a central coordination point. Instead of requiring complex peer-to-peer connection management between islands, the manager handles all coordination, migration, and convergence control, reducing system complexity while maintaining or improving convergence control through centralized management.

Inventive Principle:

Principle #24Intermediary (Mediator)

4Productivity

If synchronous processing is implemented, then processing efficiency is improved, but adaptability to different processing powers deteriorates

Engineering Contradiction:

Improveprocessing efficiencyVSAvoidadaptability to processing power differences

Core Design Contradiction:

ProductivityVSAdaptability or versatility

Solution Approach 1:

The system implements dynamic asynchronous processing where each island computing device operates independently at its own pace based on its processing power. The archipelago manager coordinates candidate solution migration without requiring synchronization, allowing devices with different processing capabilities to contribute efficiently to the overall evolutionary computation process. This dynamic approach maintains high processing efficiency while being highly adaptable to heterogeneous computing environments.

Inventive Principle:

Principle #15Dynamics

Data Source

PatentUS20230252306A1Asynchronous architecture for evolutionary computation techniques

Publication Date: 2023.08.10 X DEVELOPMENT LLC

AI summary

In some embodiments, an archipelago model is provided for distributed execution of evolutionary computing techniques. In some embodiments, an archipelago manager computing device manages a centralized archipelago management queue, and provides population subsets to a plurality of island computing devices. The archipelago manager computing device receives candidate solutions from the island computing devices, stores the candidate solutions in the archipelago management queue, and transmits candidate solutions from the archipelago management queue to the island computing devices in order to exchange candidate solutions between the island computing devices. The use of an archipelago management queue allows transfer of candidate solutions between different island computing devices, is robust to failure of any given island computing device, and does not require homogeneity within the plurality of island computing devices.