Engines of Adaptation: Self-Programming via Emergent Thermodynamic Machines
Abstract
Research Problem: Remarkably robust and intelligent systems operate across the tremendous time and length scales of the natural world. What is common in the behaviors spanning these immense scales is what one can call adaptive emergent intrinsic computation. These systemsÕ functioning relies on a kind of self-programmingÑspontaneously-generated ex- tended patterns of control and self-monitoring. Our goal is to experimentally demonstrate and theoretically predict this kind of emergent adaptive computation. Technical approach: We take a particularly novel approach that harnesses recent advances in nonlinear physics, nanoscale thermodynamics, genetic engineering, and high-performance computing. Recent remarkable progress in these arenas and our teamÕs deep expertise with them suggests that our attack on the challenge of adaptive emergent intrinsic computation is likely to succeed, where others have fallen short. To guide and ground theoretical developments we will use four unique experimental platforms, two natural and two engineered, coupled with large-scale simulations to quantitatively bridge between theory and experiment. Anticipated Outcomes: In light of recent foundational progress we anticipate the following results: (i) information principles of pattern formation in large-scale complex systems, (ii) design, prediction, and monitoring of a novel class of spatial, networked information engines; (iii) thermodynamics of adaptation that lays out the trade-offs between cognitive processing and energetics, and (iv) identifying and using the natural programmability of emergent patterns. Impact on DoD Capabilities: By implementing networks of information engines connected in various topologies we will explore the principles driving the emergence of system organizations that support computation and adaptation. Success will introduce radically new and functional platforms that exploit intelligent thermodynamically-embedded information processing in a way that can be programmed to generate emergent functioning in distributed complex systems. Such capabilities will impact many DoD applications from endowing materials with computationally advanced sense-response capabilities and directing the flow of energy and entropy to and from storage devices to porting such programmable patterns to information processing on radically smaller spatial and faster temporal scales and a first-principles understanding of adaptive behavior in autonomous systems.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Jan 04, 2021
- Source ID
- W911NF2110048
Entities
People
- James P. Crutchfield
Organizations
- Army Contracting Command
- United States Army
- University of California, Davis