Page 1
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, docs, planes, fire, fashion,
art, turbulence, music, buildings, cities, earthquakes, bodies, running, throwing,
Synesthesia, spacecraft, statistical mechanics and
zombies
Page 2
• Brains• Nets• Grids (cyberphys)• Bugs (microbes, ants)• Medical physiology
• Lots of aerospace• Wildfire ecology• Earthquakes• Physics:
– turbulence, – stat mech (QM?)
• “Toy”: – Lego– clothing, fashion
• Buildings, cities• Synesthesia
Fundamentals!
Case Study
Page 3
• Neuroscience+ People care+ Live demos
• Internet (& Cyber-Phys)+ Understand the details- Flawed designs- Everything you’ve read is wrong (in science)*
• Cell biology (bacteria)+ Perfection Some people care
* this comment is for scientists
Page 4
• Neuroscience+ People care+Live demos!
1. experiments2. data3. theory4. universals
Page 5
control feedback
RNA
mRNA
RNApTranscription
Other Control
Gene
AminoAcids
Proteins
Ribosomes
Other Control
Translation
Metabolism Products
Signal transductionATP
DNANew gene
SensoryMotor
Prefrontal
Striatum
Reflex
LearningSoftware
Hardware
DigitalAnalog
Horizontal Gene
Transfer
Horizontal App
Transfer
Horizontal Meme
Transfer
Page 6
Slow
Flexible
Fast
Inflexible
Ideal
FragileArchitecture?
Page 7
Slow
Flexible
Fast
Inflexible
Ideal
Fragile
Impossible (law)
Architecture
Architecture (constraints that
deconstrain)
Page 8
Slow
Flexible
Fast
Inflexible
Impossible (law)
Architecture
Architecture (constraints that
deconstrain)
General Special
Universal laws and architectures (Turing)
ideal
Page 9
Robust vision with motion• Object motion• Self motion
Vision
Motion
Page 11
Fast
Slow
Flexible Inflexible
Vision
Explain this amazing system.
LayeringFeedback
Page 12
Experiment• Motion/vision control without blurring• Which is easier and faster?
Robust vision with• Hand motion• Head motion
Page 13
Fast
Slow
VOR
vision
Why?• Mechanism• Tradeoff
Page 14
Fast
Slow
Flexible Inflexible
VOR
vision
Vestibular Ocular Reflex (VOR)
Mechanism
Tradeoff
Page 15
Slow
Flexible
vision
eye vision
Actslow
delay
Fast
Inflexible
Page 16
Slow
Flexible
vision
eye vision
Actslow
delay
Fast
Inflexible
VOR
fast
Page 17
Slow
Flexible
eye
Act
Fast
Inflexible
VOR
fast
Vestibular Ocular Reflex (VOR)
Page 18
Slow
Flexible
eye
Act
Fast
Inflexible
VOR
fast
It works in the dark or with your eyes closed, but
you can’t tell.
Page 19
Slow
Flexible
vision
eye vision
Actslow
delay
Fast
Inflexible
VOR
fast
Page 20
vision
eye vision
Actslow
delay
VOR
Slow
Flexible
Fast
Inflexible
Illusion
Highly evolved (hidden)
architecture
LayeringFeedback
Page 21
eye vision
Act
VOR
Layering
AutomaticUnconscious
PartiallyConscious
Page 22
eye vision
Actslow
delay
VOR
fastLayeringFeedback
Page 23
vision
eye vision
Actslow
delay
VOR
Slow
Flexible
Fast
Inflexible
Illusion
fastArchitecture
Page 24
Slow
Flexible
Fast
Inflexible
Impossible (law)
Architecture
Architecture (constraints that
deconstrain)
General Special
Universal laws and architectures (Turing)
ideal
Page 25
Fast
Slow
Flexible Inflexible
Apps
OS
HW
AppsOS
HardwareDigital
LumpedDistributed
Tech implications/extensions
Page 26
Layered Architecture
AppsOS
HardwareDigital
LumpedDistributed
Any layer needs all
lower layers.
Page 27
Layered architectures 101
Apps
OS
HW
Operating SystemAppsOS
HardwareDigital
LumpedDistributed
Page 28
OS
DiverseDeconstrained(Hardware)
Deconstrained(Applications)
DiverseHorizontal App
Transfer
Horizontal HW
Transfer
Layered architectures
Apps
OS
HW
Page 29
OS
Layered architectures
Minimal diversity
and change
Apps
OS
HW
Page 30
DiverseDeconstrained
DeconstrainedDiverse
Horizontal Transfer
Horizontal Transfer
diversity and change
diversity and change
Page 31
DiverseDeconstrained(Hardware)
Deconstrained(Applications)
DiverseHorizontal App
Transfer
Horizontal HW
Transfer
Layered architectures
Apps
OS
HW
Maximal diversity
and change
Page 32
DiverseDeconstrained
(Hardware)
Deconstrained(Applications)
Diverse
Layered architectures
ConstrainedBut hidden
Apps
OS
HW
Core Protocols
Page 33
Fast
Slow
Flexible Inflexible
General Special
Apps
OS
HW
Page 34
Fast
Slow
Flexible Inflexible
General Special
Apps
OS
HW
Page 35
FastCostly
SlowCheap
Flexible Inflexible
General Special
Layered Architecture
AppsOSHW
DigitalLumpedDistrib.
OSHW
DigitalLumpedDistrib.
DigitalLumpedDistrib.
LumpedDistrib. Distrib.
Any layer needs all lower layers.
Page 36
DigitalLumped Lumped
.
AppsOS
HardwareDigital
Lumped
Act
Decide
Sense
Page 37
Act
Sense
AppsOS
HardwareDigital
Lumped
DecideSlow
Inflexible
Fragile
Slow
Inflexible
Fragile
Expensive
Page 38
Efficiency/instability/layers/feedback
• Money/finance/lobbyists/etc• Society/agriculture/weapons/etc• Bipedalism• Maternal care• Warm blood• Flight• Mitochondria• Translation (ribosomes)• Glycolysis (2011 Science)
• All create new efficiencies but also instabilities• Requires new active/layered/complex/active control
Slow
Inflexible
Fragile
Expensive
Page 39
accessibleaccountableaccurateadaptableadministrableaffordableauditableautonomyavailablecredibleprocess
capablecompatiblecomposable configurablecorrectnesscustomizabledebugabledegradabledeterminabledemonstrable
dependabledeployablediscoverable distributabledurableeffectiveefficientevolvableextensiblefail transparentfastfault-tolerantfidelityflexibleinspectableinstallableIntegrityinterchangeableinteroperable learnablemaintainable
manageablemobilemodifiablemodularnomadicoperableorthogonalityportableprecisionpredictableproducibleprovablerecoverablerelevantreliablerepeatablereproducibleresilientresponsivereusable robust
safety scalableseamlessself-sustainableserviceablesupportablesecurablesimplicitystablestandards
compliantsurvivablesustainabletailorabletestabletimelytraceableubiquitousunderstandableupgradableusable
Requirements on systems and architectures
Page 40
accessibleaccountableaccurateadaptableadministrableaffordableauditableautonomyavailablecompatiblecomposable configurablecorrectnesscustomizabledebugabledegradabledeterminabledemonstrable
dependabledeployablediscoverable distributabledurableeffective
evolvableextensiblefail transparentfastfault-tolerantfidelityflexibleinspectableinstallableIntegrityinterchangeableinteroperable learnablemaintainable
manageablemobilemodifiablemodularnomadicoperableorthogonalityportableprecisionpredictableproducibleprovablerecoverablerelevantreliablerepeatablereproducibleresilientresponsivereusable
safety scalableseamlessself-sustainableserviceablesupportablesecurablesimplestablestandardssurvivable
tailorabletestabletimelytraceableubiquitousunderstandableupgradableusable
efficient
robust
sustainable
Sustainable robust + efficient
Page 41
accessibleaccountableaccurateadaptableadministrableaffordableauditableautonomyavailablecompatiblecomposable configurablecorrectnesscustomizabledebugabledegradabledeterminabledemonstrable
dependabledeployablediscoverable distributabledurableeffective
evolvableextensiblefail
transparentfastfault-tolerantfidelityflexibleinspectableinstallableIntegrityinterchangeabl
einteroperable learnablemaintainable
manageablemobilemodifiablemodularnomadicoperableorthogonalit
yportableprecisionpredictableproducibleprovablerecoverablerelevantreliablerepeatablereproducibleresilientresponsivereusable
safety scalableseamlessself-sustainableserviceablesupportablesecurablesimplestablestandardssurvivable
tailorabletestabletimelytraceableubiquitousunderstandableupgradableusable
efficient
robust
sustainable
Simple dichotomous
tradeoff pairs
PCA Principal Concept Analysis
wasteful
fragile
efficientrobust
Page 42
wasteful
fragile
efficientrobust
Laws
Architectures
LayeringFeedback
Page 43
Robust
FragileUniversal
laws
wasteful
efficient
Page 44
Slow
Flexible
Fast
Inflexible
Robust
FragileUniversal
laws
wasteful
efficient
Page 45
Chandra, Buzi, and Doyle
UG biochem, math, control theory
Most important paper so far.
Page 46
k
z pz p
10-1 100 101100
101
too fragile
complex
No tradeoff
expensive
fragile
2 20
1 ln
ln
zS j dz
z pz p
Law #1 : ChemistryLaw #2 : Autocatalysis
Law #3:
Page 47
I recently found this paper, a rare example of exploring an explicit tradeoff between robustness and efficiency.
This seems like an important paper but it is rarely cited.
Page 48
1m
Bacteria
Phage
Page 49
Survive
Phage lifecycle
Page 50
Survive
Phage lifecycle
Infect
Page 51
Multiply
Survive
Phage lifecycle
Infect
Page 52
Multiply
Survive
Phage lifecycle
InfectLyse
Page 53
Multiply
Survive
Phage lifecycle
Tradeoff?
Page 54
Costly/slow
fragile
Efficient/fast
robust
Survive
Grow/Multiply
ideal
Ideal for phage, bad for bacteria
Page 55
fragile
robust
Survive
SlowCostly
FastEfficientCheap
Multiply
thinsmall
thickbig
Capsid sizeGenome size
What the data says
Page 56
fragile
robust
Survive
SlowCostly
FastEfficientCheap
Multiply
thinsmall
thickbig
Good architecture?
Evolut
ion
Laws?
Impossible?
Not viable?
Page 57
thinsmall
thickbig
Capsid sizeGenome size
Shared architecture
Page 58
thinsmall
thickbig
Capsid sizeGenome sizeWhy fragile?
Instability?
Page 59
Capsid sizeGenome sizeWhy fragile?
Instability?
Page 60
Capsid sizeGenome sizeWhy fragile?
Instability?
Defect
Page 61
Capsid sizeGenome sizeWhy fragile?
Instability?
Defect
Page 62
thinsmall
thickbig
Capsid sizeGenome size
Active control
Passive control
Page 63
fragile
robust
Survive
SlowCostly
FastEfficientCheap
Multiply
thinsmall
thickbig
Capsid sizeGenome size
What the data says
Page 64
Multiply
Survive
Phage lifecycle
InfectLyse
costly
fragile
cheap
robust
Survive
Multiply
thinsmall
architectures
thickbig
Evolut
ionLaws
Universals • Laws, constraints, tradeoffs– Robust/fragile– Efficient/wasteful– Fast/slow– Flexible/inflexible
• Architecture• Hijacking, parasitism,
predation
Page 65
control feedback
RNA
mRNA
RNApTranscription
Other Control
Gene
AminoAcids
Proteins
Ribosomes
Other Control
Translation
Metabolism Products
Signal transductionATP
DNANew gene
Horizontal Gene
TransferBacteria
Phage
Page 66
control feedback
RNA
mRNA
RNApTranscription
Other Control
Gene
AminoAcids
Proteins
Ribosomes
Other Control
Translation
Metabolism Products
Signal transductionATP
DNANew gene
Horizontal Gene
Transfer
Sequence ~100 E Coli (not chosen randomly)• ~ 4K genes per cell• ~20K different genes in total (pangenome)• ~ 1K universally shared genes • ~ 300 essential (minimal) genes
Clinical issue: antibiotic resistance?
Page 67
control feedback
RNA
mRNA
RNApTranscription
Other Control
Gene
AminoAcids
Proteins
Ribosomes
Other Control
Translation
Metabolism Products
Signal transductionATP
DNANew gene
SensoryMotor
Prefrontal
Striatum
Reflex
LearningSoftware
Hardware
DigitalAnalog
Horizontal Gene
Transfer
Horizontal App
Transfer
Horizontal Meme
Transfer
What can go wrong?
Page 68
Horizontal Bad Gene Transfer
Horizontal Bad App Transfer
Horizontal Bad Meme
Transfer
Parasites &
Hijacking
Fragility?
Exploiting layered
architecture
Virus
Virus
Meme
Page 69
easyhard
Law #1 : Mechanics ChemistryLaw #2 : Gravity Autocatalysis
Page 70
hard harder
Why?vision
Act
delay
Law #3 : Light
Page 71
Universal laws and architectures:brains, bugs, networks, physiology,
grids, medicine, wildfire, turbulence,literature, fashion, dance, earthquakes,
art, music, Lego, buildings, citiesvision
Act
delay
Control theory+ NeuroscienceBalancing
an inverted pendulum
Mechanics+Gravity +Light +
Page 72
Crashes can be
made rare with active
control.
Page 73
easy
hard
Gravity is stabilizing
Gravity is destabilizing
harder
More unstable
Law #1 : MechanicsLaw #2 : Gravity
Page 74
hard harder hardest!
Easy to prove using simple models.
What is sensed matters.
Why?
Why?!?
Page 75
Efficiency/instability/layers/feedback
• Money/finance/lobbyists/etc• Society/agriculture/weapons/etc• Bipedalism• Maternal care• Warm blood• Flight• Mitochondria• Translation (ribosomes)• Glycolysis (2011 Science)
• All create new efficiencies but also instabilities• Requires new active/layered/complex/active control
Slow
Inflexible
Fragile
Expensive
Page 76
Efficiency/instability/layers/feedback
• Money/finance/lobbyists/etc• Society/agriculture/weapons/etc• Bipedalism• Maternal care• Warm blood• Flight• Mitochondria• Translation (ribosomes)• Glycolysis (2011 Science)
• All create new efficiencies but also instabilities• Requires new active/layered/complex/active control
easy hard
Page 77
Universal laws
vision
Act
delay
+ Neuroscience
Balancing an inverted pendulum
Mechanics+Gravity +Light +
2 20
1 ln
ln
pT j dp
z ppz p
Some math details, which can be skipped
Page 78
easy
Law #1 : MechanicsLaw #2 : Gravity
2cos sin
cos sin 0sin
M m x ml u
x l gy x l
0
M m x ml u
x l gy x l
linearize
Mechanics+Gravity
Page 79
easyhard
Law #1 : MechanicsLaw #2 : Gravity
2cos sin
cos sin 0sin
M m x ml u
x l gy x l
0
M m x ml u
x l gy x l
linearize
Page 80
hard harder
Easy to prove using simple models.
Why?vision
Act
delay
Law #3 : Light 0
y x
M m x ml u
x l gl
Page 81
eye vision
slow
Act
delay
Control
l
1pl
noiseerror
ET jN
Page 82
Universal laws and architectures:brains, bugs, networks, physiology,
grids, medicine, wildfire, turbulence,literature, fashion, dance, earthquakes,
art, music, Lego, buildings, citiesvision
Act
delay + Neuroscience
Balancing an inverted pendulum
Mechanics+Gravity +Light +
2 20
1 ln
ln
pT j dp
z ppz p
Page 83
1
2
4
8
.1 1
Length l (meters)
.05 .5.2
1pl
Fragility
p .3s
2 20
1 2ln pT j d pp
Law #4 :
.1s
Page 84
1
2
4
8
.1 1
Length l (meters)
.05 .5.2
1pl
Fragility
p .3s
2 20
1 2ln pT j d pp
Law #4 :
.1s
Shorter
Fragile
Page 85
1
2
4
8
.1 1
Length l (meters)
.05 .5.2
1pl
Fragility
p .3s
2 20
1 2ln pT j d pp
Law #4 :
.1s Slow
er
Page 86
hard harder hardest!
Easy to prove using simple models.
What is sensed matters.
Why?
Page 87
hard harder hardest!
What is sensed matters.
Unstable poles Unstable zeros
0l l0l l
Page 88
hardest!
0l l
2 20
l1 n n2l z ppT j dp p
pz
0l l
Page 89
1
2
4
8
.1 1.05 .5.2Length (meters)
Fragility
ln z pp
p
z
.1s
2 20
l1 n n2l z ppT j dp p
pz
.3s
0 1l l
0l l
Page 90
Sense(Measure Lengthl0, m)
Length l to CoM, m
0.2 0.4 0.6 0.8 1
0.4
0.6
0.8
1
1.2
1
1.5
2
2.5
3
3.5easy
hardhard
robust(easy)
fragile(hard)
Page 91
Sense(Measure Lengthl0, m)
Length l to CoM, m
0.2 0.4 0.6 0.8 1
0.4
0.6
0.8
1
1.2
1
1.5
2
2.5
3
3.5
hardhardrobust(easy)
hard!
0l l
easy
Page 92
vision
Act
delay
2 20
1 ln
ln
pT j dp
z ppz p
Holds for all controllers.
Like Turing, a “law” about intrinsic problem difficulty.
Page 93
Efficiency/instability/layers/feedback
• Money/finance/lobbyists/etc• Society/agriculture/weapons/etc• Bipedalism• Maternal care• Warm blood• Flight• Mitochondria• Translation (ribosomes)• Glycolysis (2011 Science)
• All create new efficiencies but also instabilities• Requires new active/layered/complex/active control
Slow
Inflexible
Fragile
Expensive
Page 94
1
2
4
8
.1 1.05 .5.2
0l l
0 1l l
Length (meters)
Fragility
lnz pp
p
z
.1s
k
z pz p
10-1 100 101100
101
too fragile
complex
No tradeoff
expensive
fragile
eye vision
Actslow
VOR
fast
Act
delay
Balancing an inverted pendulum
Control
costly
fragile
cheap
robust
Survive
Multiply
thinsmall
thickbig
Laws
Page 95
1
2
4
8
.1 1.05 .5.2
0l l
0 1l l
Length (meters)
Fragility
lnz pp
p
z
.1s
k
z pz p
10-1 100 101100
101
too fragile
complex
No tradeoff
expensive
fragile
costly
fragile
cheap
robust
Survive
Multiply
thinsmall
thickbig
Laws
Slow
Inflexible
Fragile
Expensive
Page 96
Slow
Flexible
vision
eye vision
Act slowdelay
Fast
Inflexible
VOR
fastSlow
Inflexible
Fragile
Expensive
Slow
Flexible
Fast
Inflexible
Impossible (law)
General Special
Page 97
1
2
4
8
.1 1.05 .5.2
0l l
0 1l l
Length (meters)
Fragility
ln z pp
p
z
.1s
k
z pz p
10-1 100 101100
101
too fragile
complex
No tradeoff
expensive
fragile
costly
fragile
cheap
robust
Survive
Multiply
thinsmall
thickbig
Laws
Slow
Flexible
vision
eye vision
Act slowdelay
Fast
Inflexible
VOR
fast
Slow
Flexible
Fast
Inflexible
Impossible (law)
Universal laws
Page 98
What (some) reviewers say• “…to establish universality … is simply wrong. It
cannot be done…• … a mathematical scheme without any real
connections to biological or medical… • …universality is well justified in physics… for
biological and physiological systems …a dream …never be realized, due to the vast diversity in such systems.
• …does not seem to understand or appreciate the vast diversity of biological and physiological systems…
• …a high degree of abstraction, which …make[s] the model useless …
Page 99
Control, OR
CommsCompute
Physics
Shannon
Bode
Turing
Gödel
EinsteinHeisenberg
Carnot
Boltzmann
Theory?Deep, but fragmented, incoherent, incomplete
Nash
Von Neumann
Page 100
Control, OR
Compute
Bode
Turing
Delay and risk are most important
• Worst-case (“risk”)• Time complexity (delay)
• Worst-case (“risk”)• Delay severely degrades
robust performance
Computation for control• Off-line design• On-line implementation• Learning and adaptation
Page 101
Control, OR
CommunicateCompute
Physics
Shannon
Bode
Turing
Einstein
Heisenberg
Carnot
Boltzmann
Delay and risk are
most important
Delay and risk are
least important
Page 102
Communicate
Physics
Shannon
Einstein
Heisenberg
Carnot
Boltzmann
Dominates “high
impactscience” literature
• Average case (risk neutral)• Random ensembles• Asymptotic (infinite delay)
• “Layering” by averaging
• Space complexity
Page 103
Control, OR
CommunicateCompute
Physics
Shannon
Bode
Turing
Delay and risk are
most important
Delay and risk are
least important
New progress!
Page 104
Physics of Fluids (2011)
wU
z x
y
uz x
yFlow
upflowhigh-speed
region
downflowlow speed
streak
Blunted turbulent velocity profile
Laminar
Turbulent
wU3D coupling
Coherent structures and turbulent drag
Page 105
wasteful
fragile Laminar
Turbulent
efficient
robust
Laminar
Turbulent
wU
?
Control?
Fundamentals!