The evolvability of programmable hardware.

Area of Science:

• Computer Science
• Evolutionary Biology
• Electronic Engineering

Background:

• Biological systems utilize neutral networks, where multiple genotypes map to a single phenotype, conferring robustness and evolvability.
• The existence and properties of similar networks in technological systems, particularly electronic circuits, remain largely unexplored.
• Programmable electronic circuits computing digital logic functions are crucial for computation and evolutionary design applications.

Purpose of the Study:

• To investigate whether programmable electronic circuits exhibit properties analogous to biological neutral networks.
• To explore the potential for evolvability, robustness, and adaptability in digital logic circuit design.
• To determine if evolutionary principles can be applied to create fault-tolerant and adaptable electronic circuitry.

Main Methods:

• Exploration of a vast genotype space (10^45) of logic circuits and phenotype space (10^19) of logic functions.
• Analysis of circuit architecture and function to identify neutral networks.
• Assessment of circuit robustness, fault-tolerance, and the novelty of functions in the vicinity of neutral networks.

Main Results:

• Logic circuits computing the same function are organized into large, interconnected neutral networks spanning circuit space.
• Significant variation exists in the robustness and fault-tolerance of these circuits.
• Novel functions are readily accessible near existing neutral networks, with few architectural changes enabling interconversion between circuits computing different functions.

Conclusions:

• Electronic circuits, specifically digital logic circuits, possess neutral networks similar to those found in biological systems.
• These findings demonstrate that properties crucial for evolvability in biology are present in electronic circuitry.
• The study suggests pathways for designing fault-tolerant, adaptable, and evolvable electronic systems.