35
John Doyle 道道 Jean-Lou Chameau Professor Control and Dynamical Systems, EE, & BioE tech 1 # Ca Universal laws and architectures: Theory and lessons from brains, bugs, nets, grids, planes, docs, fire, bodies, fashion, earthquakes, turbulence, music, buildings, cities, art, running, throwing, Synesthesia, spacecraft, statistical mechanics

John Doyle 道陽 Jean-Lou Chameau Professor Control and Dynamical Systems, EE, & BioE tech 1 # Ca Universal laws and architectures: Theory and lessons from

Embed Size (px)

Citation preview

Page 1: John Doyle 道陽 Jean-Lou Chameau Professor Control and Dynamical Systems, EE, & BioE tech 1 # Ca Universal laws and architectures: Theory and lessons from

John Doyle 道陽Jean-Lou Chameau Professor

Control and Dynamical Systems, EE, & BioE

tech1#Ca

Universal laws and architectures:Theory and lessons from

brains, bugs, nets, grids, planes, docs, fire, bodies, fashion,

earthquakes, turbulence, music, buildings, cities, art, running, throwing, Synesthesia, spacecraft, statistical mechanics

Page 2: John Doyle 道陽 Jean-Lou Chameau Professor Control and Dynamical Systems, EE, & BioE tech 1 # Ca Universal laws and architectures: Theory and lessons from

A minimal prologue

Page 3: John Doyle 道陽 Jean-Lou Chameau Professor Control and Dynamical Systems, EE, & BioE tech 1 # Ca Universal laws and architectures: Theory and lessons from

Diverse applications

AerospaceRoboticsBiologyFluidsPhysicsMedicineInternetSmartgridEcologyNeuroscience

Tuesday

Wednesday

Thursday

Monday

Page 4: John Doyle 道陽 Jean-Lou Chameau Professor Control and Dynamical Systems, EE, & BioE tech 1 # Ca Universal laws and architectures: Theory and lessons from

Diverse applications Diverse mathematics

AerospaceRoboticsBiologyFluidsPhysicsMedicineInternetSmartgridEcologyNeuroscience

ODE/PDEAutomataAnalysisOptimizationProb&StatsOperator ThDiff GeomAlg GeomCmp Cplxty

Details are not serious

Page 5: John Doyle 道陽 Jean-Lou Chameau Professor Control and Dynamical Systems, EE, & BioE tech 1 # Ca Universal laws and architectures: Theory and lessons from

CDS

Diverse applications(not just engineering)

Diverse mathematics(not just “applied”)

ODE/PDEAutomataAnalysisOptimizationProb&StatsOperator ThDiff GeomAlg GeomCmp Cplxty“Universal

laws and architectures”

AerospaceRoboticsBiologyFluidsPhysicsMedicineInternetSmartgridEcologyNeuroscience

Page 6: John Doyle 道陽 Jean-Lou Chameau Professor Control and Dynamical Systems, EE, & BioE tech 1 # Ca Universal laws and architectures: Theory and lessons from

CDS

Diverse applications

Domain experts have “core”

interface to diverse math and tools

AerospaceRoboticsBiologyFluidsPhysicsMedicineInternetSmartgridEcologyNeuroscience

Page 7: John Doyle 道陽 Jean-Lou Chameau Professor Control and Dynamical Systems, EE, & BioE tech 1 # Ca Universal laws and architectures: Theory and lessons from

CDS

Diverse mathematics

Math (not just applied) has “core” interface to diverse application drivers

ODE/PDEAutomataAnalysisOptimizationProb&StatsOperator ThDiff GeomAlg GeomCmp Cplxty

Page 8: John Doyle 道陽 Jean-Lou Chameau Professor Control and Dynamical Systems, EE, & BioE tech 1 # Ca Universal laws and architectures: Theory and lessons from

CDS as architecture• Architecture= “constraints that deconstrain”

– Constraint = CDS core concepts and math– Deconstraint = connections between new applications

and (also possibly new) math

• How to explain this?• Laws: Universal constraints on achievable

robust performance and efficiency• Architectures: Universal organizational

strategies to flexibly achieve what is possible• Case studies: concrete and familiar examples to

illustrate “universals”8

Page 9: John Doyle 道陽 Jean-Lou Chameau Professor Control and Dynamical Systems, EE, & BioE tech 1 # Ca Universal laws and architectures: Theory and lessons from

Core protocols

• From Architecture slides, understanding the “OS” is difficult and essential

• But term “OS” seems to confuse• Replace with “core protocols”?• Highly conserved/constrained core (knot)• Highly diverse/deconstrained edges

• What is “CDS” in this context?9

Page 10: John Doyle 道陽 Jean-Lou Chameau Professor Control and Dynamical Systems, EE, & BioE tech 1 # Ca Universal laws and architectures: Theory and lessons from

Fast

Slow

Flexible Inflexible

General Special

Apps

OS

HW

Horizontal App

Transfer

Horizontal Hardware Transfer

Page 11: John Doyle 道陽 Jean-Lou Chameau Professor Control and Dynamical Systems, EE, & BioE tech 1 # Ca Universal laws and architectures: Theory and lessons from

Fast

Slow

Flexible Inflexible

General Special

Apps

OS

HW

Tradeoffs

Page 12: John Doyle 道陽 Jean-Lou Chameau Professor Control and Dynamical Systems, EE, & BioE tech 1 # Ca Universal laws and architectures: Theory and lessons from

Fast

Slow

Flexible Inflexible

General Special

Apps

OS

HW

Tradeoffs = “laws?”

Page 13: John Doyle 道陽 Jean-Lou Chameau Professor Control and Dynamical Systems, EE, & BioE tech 1 # Ca Universal laws and architectures: Theory and lessons from

Slow

Fast

Flexible InflexibleGeneral Special

AppsOSHWDig.Lump.Distrib.

OSHWDig.Lump.Distrib.

DigitalLump.Distrib.

LumpedDistrib. Distrib.

HGT DNA repair Mutation DNA replication Transcription

Translation

Metabolism Signal

SenseMotor

Prefrontal

Fast

Learn

Reflex

Evolve

vision

VOR

Layered architectures

Page 14: John Doyle 道陽 Jean-Lou Chameau Professor Control and Dynamical Systems, EE, & BioE tech 1 # Ca Universal laws and architectures: Theory and lessons from

OS

Horizontal App

Transfer

Horizontal Hardware Transfer

Constrained

Deconstrained

Deconstrained

Architecture = Constraints that deconstrain

Page 15: John Doyle 道陽 Jean-Lou Chameau Professor Control and Dynamical Systems, EE, & BioE tech 1 # Ca Universal laws and architectures: Theory and lessons from

Horizontal Transfer

Horizontal Transfer

Constrained

Deconstrained

Deconstrained

“Hourglass”Universal architectures

Page 16: John Doyle 道陽 Jean-Lou Chameau Professor Control and Dynamical Systems, EE, & BioE tech 1 # Ca Universal laws and architectures: Theory and lessons from

Con

stra

ined

Dec

onst

rain

ed

Dec

onst

rain

ed

“Bowtie”

Horizontal versus vertical is an arbitrary convention.

Page 17: John Doyle 道陽 Jean-Lou Chameau Professor Control and Dynamical Systems, EE, & BioE tech 1 # Ca Universal laws and architectures: Theory and lessons from

CDS

Diverse applications(not just engineering)

Diverse mathematics(not just “applied”)

ODE/PDEAutomataAnalysisOptimizationProb&StatsOperator ThDiff GeomAlg GeomCmp Cplxty“Universal

laws and architectures”

AerospaceRoboticsBiologyFluidsPhysicsMedicineInternetSmartgridEcologyNeuroscience

Page 18: John Doyle 道陽 Jean-Lou Chameau Professor Control and Dynamical Systems, EE, & BioE tech 1 # Ca Universal laws and architectures: Theory and lessons from

“Pasteur’s quadrant?” (Donald Stokes)

Typically poor choice of colors and font sizes

Page 19: John Doyle 道陽 Jean-Lou Chameau Professor Control and Dynamical Systems, EE, & BioE tech 1 # Ca Universal laws and architectures: Theory and lessons from

“Pasteur’s quadrant?” (Donald Stokes)

Low

High

Edison

Bohr

Practical use

FundamentalUnderstanding

Low

High

Relative emphasis

Use inspired basic research

(Pasteur)

Page 20: John Doyle 道陽 Jean-Lou Chameau Professor Control and Dynamical Systems, EE, & BioE tech 1 # Ca Universal laws and architectures: Theory and lessons from

Low

High

Edison

PasteurBohr

Practical use

FundamentalUnderstanding

Low

High

“Pasteur’s quadrant?” Redrawn

Relative emphasis

Pasteur aimed for

both

Page 21: John Doyle 道陽 Jean-Lou Chameau Professor Control and Dynamical Systems, EE, & BioE tech 1 # Ca Universal laws and architectures: Theory and lessons from

Apps CDS Math

Practical use

FundamentalUnderstanding

Page 22: John Doyle 道陽 Jean-Lou Chameau Professor Control and Dynamical Systems, EE, & BioE tech 1 # Ca Universal laws and architectures: Theory and lessons from

Later

Deep,Universal

Apps

CDSMath

Practical impact

Specific

Now

CDS’ quadrant?

Scope

Page 23: John Doyle 道陽 Jean-Lou Chameau Professor Control and Dynamical Systems, EE, & BioE tech 1 # Ca Universal laws and architectures: Theory and lessons from

Later

Deep,Universal

Apps

CDSMath

Practical impact

Specific

Now

Fast

Slow

Flexible InflexibleGeneral Special

SW Apps

OS

HW

Related but somewhat flipped

Scope

Page 24: John Doyle 道陽 Jean-Lou Chameau Professor Control and Dynamical Systems, EE, & BioE tech 1 # Ca Universal laws and architectures: Theory and lessons from

Later

Deep,Universal

Apps

Math

Specific

Now

Circa 1975

Modern control theory

Practical impact

Scope

Page 25: John Doyle 道陽 Jean-Lou Chameau Professor Control and Dynamical Systems, EE, & BioE tech 1 # Ca Universal laws and architectures: Theory and lessons from

Later

Deep,Universal

Apps

Math

Specific

Now

Circa 1990 Complexity science

Robust control

theory

Modern control theory

Practical impact

Scope

Page 26: John Doyle 道陽 Jean-Lou Chameau Professor Control and Dynamical Systems, EE, & BioE tech 1 # Ca Universal laws and architectures: Theory and lessons from

Later

Deep,Universal

Apps

Math

Specific

Now

2014

CDS

Complexity science

Networkscience

Never

Nothing

Practical impact

Scope

Page 27: John Doyle 道陽 Jean-Lou Chameau Professor Control and Dynamical Systems, EE, & BioE tech 1 # Ca Universal laws and architectures: Theory and lessons from

Later

Deep,Universal

Specific

Now

Complexity science

Networkscience

Never

Nothing

Practical impact

Scope

Humans love this corner!

Page 28: John Doyle 道陽 Jean-Lou Chameau Professor Control and Dynamical Systems, EE, & BioE tech 1 # Ca Universal laws and architectures: Theory and lessons from

CMS

Later

Deep,Universal

Apps

Math

Practical impact

Specific

Now

Near future?

CDS

Complexity science

Networkscience

Page 29: John Doyle 道陽 Jean-Lou Chameau Professor Control and Dynamical Systems, EE, & BioE tech 1 # Ca Universal laws and architectures: Theory and lessons from

CMS

Later

Deep,Universal

Apps

Math

Practical impact

Specific

Now

The dream?

CDS

Complexity science

Networkscience

Page 30: John Doyle 道陽 Jean-Lou Chameau Professor Control and Dynamical Systems, EE, & BioE tech 1 # Ca Universal laws and architectures: Theory and lessons from

Horizontal Transfer

Horizontal Transfer

Constrained

Deconstrained

Deconstrained

Universal architectures

Universal laws

Fast

Slow

Flexible InflexibleGeneral Special

SW Apps

OS

HW

Page 31: John Doyle 道陽 Jean-Lou Chameau Professor Control and Dynamical Systems, EE, & BioE tech 1 # Ca Universal laws and architectures: Theory and lessons from

Slow

Fast

Flexible InflexibleGeneral Special

AppsOSHWDig.Lump.Distrib.

OSHWDig.Lump.Distrib.

DigitalLump.Distrib.

LumpedDistrib. Distrib.

HGT DNA repair Mutation DNA replication Transcription

Translation

Metabolism Signal

SenseMotor

Prefrontal

Fast

Learn

Reflex

Evolve

vision

VOR

Layered architectures

Page 32: John Doyle 道陽 Jean-Lou Chameau Professor Control and Dynamical Systems, EE, & BioE tech 1 # Ca Universal laws and architectures: Theory and lessons from

CDS

Diverse applications

“core” interface to

minimal math and tools

AerospaceRoboticsBiologyFluidsPhysicsMedicineInternetSmartgridEcologyNeuroscience

Morning

Universal laws and

architectures

Undergrad math

Page 33: John Doyle 道陽 Jean-Lou Chameau Professor Control and Dynamical Systems, EE, & BioE tech 1 # Ca Universal laws and architectures: Theory and lessons from

CDS

Diverse mathematics

Networks,Distributed

Cutting edge theory foundations

ODE/PDEAutomataAnalysisOptimizationProb&StatsOperator ThDiff GeomAlg GeomCmp Cplxty

Afternoon

Page 34: John Doyle 道陽 Jean-Lou Chameau Professor Control and Dynamical Systems, EE, & BioE tech 1 # Ca Universal laws and architectures: Theory and lessons from

CMS

Later

Deep,Universal

Apps

Math

Practical impact

Specific

Now

The dream?

CDS

Complexity science

Networkscience

Page 35: John Doyle 道陽 Jean-Lou Chameau Professor Control and Dynamical Systems, EE, & BioE tech 1 # Ca Universal laws and architectures: Theory and lessons from

Caveats

• Many ideas are classic “robust control”, but much of the organization (the architecture) is fairly new

• Motivating case studies also new

• Many rough edges• Nobody is working on it quite like this• We want to change that