““Politehnica” University of TimisoaraPolitehnica” University of Timisoara

Course Advisor:Course Advisor:

Lucian Lucian ProdanProdan

Evolvable SystemsEvolvable Systems

Web Page:Web Page: www.acsa.upt.rowww.acsa.upt.ro

TeachingTeaching Graduate Graduate CoursesCourses

Summer Semester 2007Summer Semester 2007

What is ACSA ??What is ACSA ??

Why bother ??Why bother ??Here are some thoughts…Here are some thoughts…

Computers:Computers:• Fine exponents of the present days’ technological Fine exponents of the present days’ technological

wavewave

• Solid and trusted performance, indispensable Solid and trusted performance, indispensable support in many fieldssupport in many fields

• Moore’s law on computer performance still Moore’s law on computer performance still holding (holding (barely?barely?))

Two major problemsTwo major problems (according to Gigascale)(according to Gigascale)• Of the small: Of the small: caused by device shrinking – caused by device shrinking –

dominant thus far, industry’s main focus and dominant thus far, industry’s main focus and investmentinvestment

• Of the large: Of the large: enormity of design verification and enormity of design verification and manufacturing-test tasks – now a manufacturing-test tasks – now a limitationlimitation for for industrial progressindustrial progress

Should we care ??Should we care ??Here are some more thoughts…Here are some more thoughts…

Physical limits Physical limits of current, conventional technologies of current, conventional technologies approaching fastapproaching fast (also financial limits!) (also financial limits!)• Intel’s 90nm Prescott chip already close to the Intel’s 90nm Prescott chip already close to the thermal thermal

wallwall

• need to need to individuallyindividually place atoms inside chips place atoms inside chips throughout 2010'sthroughout 2010's

Moore’s law breakdownMoore’s law breakdown forecasted around 2018 forecasted around 2018 (source: Intel)(source: Intel)

ITRS – ITRS – two near- and longer-term challengestwo near- and longer-term challenges::• sustaining thesustaining the 17% annual increase 17% annual increase in performancein performance

• developingdeveloping beyond CMOS beyond CMOS materials and applicationsmaterials and applications

Who says this ??Who says this ??

Are there any alternatives ??Are there any alternatives ??

Emerging Emerging Technologies and Technologies and CMOSCMOS• Speed• Size• Cost• Switching

Energy

Source: ITRS reportSource: ITRS report – – 2004 update2004 update

What about this course ??What about this course ??

NewNew computing architectures: computing architectures:• Biologically-inspired computingBiologically-inspired computing

– First part of the semesterFirst part of the semester

• Molecular and DNA computingMolecular and DNA computing

• Nanoelectronics, …Nanoelectronics, …

NewNew computing paradigms: computing paradigms:• Quantum computingQuantum computing

– Second part of the semesterSecond part of the semester

• Reversible computingReversible computing

• Adiabatic computing, …Adiabatic computing, …

Bioinspired Computing: Why ??Bioinspired Computing: Why ??

TraditionTradition: engineering : engineering andand science science have developed alonghave developed along separate tracks separate tracks

• Natural scientist – a detective: seeking to analyze existing processes, to explain their operation, to model them, and to predict their future behavior

• Engineer – a builder: tries to create artificial systems (bridges, cars, electronic devices) based on a set of specifications (a description) and a set of primitives (elementary components such as bricks, beams, wires, motors, and transistors)

Bioinspired Computing: Why ?? Bioinspired Computing: Why ?? (2)(2)

Present days:Present days:

• scientists use tools created by engineers• engineers allured by certain natural

processes Living organisms – complex systems

exhibiting a range of desirable characteristics difficult to realize using traditional engineering methodologies• evolution• adaptation• fault tolerance

Living organisms ??Living organisms ??

Living systems characterized by a genetic program (the genome), that guides their development, their functioning, and their death

Considering life on Earth since its very beginning, three levels of organization distinguished:• phylogeny• ontogeny• epigenesis

PhylogenyPhylogeny

First level of organization considered for living systems

Temporal evolution of the genetic program (the genome) concerned

Replication based on genome multiplying – low error rate at individual level

Genetic mechanisms fundamentally nondeterministic –> genetic diversity -> survavibility

OntogenyOntogeny

Second level of organization considered for multicellular living systems

Temporal evolution of one individual

Successive cellular division of the zygote -> cellular differentiation

Processes essentially deterministic –> wrong genetic sequence -> notable/lethal malformations

EpigenesisEpigenesis

Third level of organization considered for living systems

Ontogenetic information limited -> another process emerge to integrate knowledge

Example: human brain 1010 neurons, 1014 connections -> too large to be encoded by the genome

Learning systems– Nervous

– Endocrine

– Immune

POE Model: How Does Affect POE Model: How Does Affect US, US, Computer EngineersComputer Engineers ?? -- ?? -- PhylogenyPhylogeny

Artificial evolution: genetic algorithms, evolution strategies, evolutionary programming, and genetic programming

Large scale programmable circuits: configure function by programming -> FPGAs - three distinct levels of configuration for an FPGA:

– logic– cell interconnection– inputs and outputs

Evolvable hardware: an evolutionary approach to digital design

POE Model: How Does Affect POE Model: How Does Affect US, US, Computer EngineersComputer Engineers ?? -- ?? -- OntogenyOntogeny

Growth, construction:– self-test and self-repair

– self-replication

– Embryonics

POE Model: How Does Affect POE Model: How Does Affect US, US, Computer EngineersComputer Engineers ?? -- ?? -- EpigenesisEpigenesis

Nervous system:– Traditionally the most investigated

– Artificial Neural Networks

Immune system:– software fault detection

– Controllers for mobile robots

Endocrine system– hormones