Universally Programmable Intelligent Matter

1
Universally Programmable Intelligent Matter

• A Systematic Approach to Nanotechnology
• Bruce MacLennan
• University of Tennessee, Knoxville
• Supported by NSF NER grant
• and UTK Center for Information Technology Research

2
Introduction Definitions

• Intelligent matter individual molecules or
  molecular clusters function as agents
• Universally programmable intelligent matter made
  from a small set of building blocks that are
  computationally universal

3
Goals

• Avoid case-by-case nano-engineering of materials
• Design one universal material that can be
  programmed for many applications
• Requires a small, fixed set of molecular building
  blocks ( reactions), which can be arranged for
  varied purposes
• Suggestive evidence for such sets

4
K-substitution
((K X) Y) ? X
5
K-Substitutionas a Molecular Process
6
S-Substitution (with Copying)
(((S X) Y) Z) ? ((X Z) (Y Z))
7
S-Substitution (with Sharing)
(((S X) Y) Z) ? ((X Z) (Y Z))
8
Computational Properties

• Universality S and K can compute anything that
  is computable
• Church-Rosser Property substitutions can be done
  in any order without affecting result

9
SK Computation

• Good model of nondeterministic parallel
  computing
• Has been studied as model of massively parallel
  computer architecture
• Functional computer programs can be compiled into
  SK networks

10
Example Computing a Ring

• Ring (X,Y) R
• where rec R Aux (X,Y,R)
• Aux (X, nil, R) R
• Aux (xX, yY, R)
• (x,y) Aux (X,Y,R)

11
Example Computing a Tube

• Tube (nil, X, Y)
• Ring (X, Y)
• Tube (aN, X, Y)
• Ring X, Tube(N,X,Y)

12
Extensions

• Sensor operations respond to environmental
  conditions
• Effector operations have physical effects on
  environment
• Execution of these imperative operations must
  be controlled

13
Some Static Applications

• Complex physical structures chains, tubes,
  spheres, fibers, networks, quasi-crystalline
  structures
• Membranes with pores or channels
• Very dense analog neural networks
• Sensor effector organs for microrobots
• Conventional computation

14
Some Dynamic Applications

• Membrane with controllable channels
• Free-floating clusters controlling fluid
  properties
• Semiautonomous agents to recognize and bind
  molecules
• Sensory transducers, such as artificial retinas
  cochleas
• Effectors, such as cilia artificial muscle
  fibers
• Self-repair

15
Developing an Application

• Write debug program
• Compile into SK network
• Simulate on computer
• Flatten into DNA sequence
• Replicate DNA
• Construct molecular network from DNA
• Supply reactants for computation
• Optionally, replace by permanent groups

16
Issues

• Appropriate model of computation
• Replication/sharing problem
• Appropriate choice of combinators
• Blocking computation
• Nontraditional effects on computation
• Dealing with substitution error
• Geometrical issues
• Supply of reactants
• Identifying/synthesizing appropriate reactions

17
Current Activities

• Developing mathematical model
• Theoretical analysis
• Developing simulation tool
• Programming sample applications