CDS 270 Course details (Spring 2017)
CDS 270 Course Details
There are videos, slides, and reading in the subfolder New_CDS270_VideosSlidesPapers in this dropbox folder.
I'm adding material various subfolders of CDS270, labeled to suggest their content. I'll continue to add subfolders with mixes of videos, slides, and reading. I'll try to make videos of the highest priority material.
Current topics are:
1.Overview These give a quick overview of the questions motivating this class, primarily using the human nervous system, but mentioning bacteria and the Internet. These will also be the first videos to watch for the class. There is also a folder with some related papers.
2.NeuroIntroTheory A quick introduction to the theory that connects speed/accuracy tradeoffs in spiking neurons with speed/accuracy tradeoffs at the system level.
3.BalanceWaterbed Summary of stick balancing tradeoffs including unstable poles and zeros. The "waterbed" effect and oscillations. Chaotic dynamics and sensitivity to initial conditions.
3.x1.ExtrasVision has extra slides on vision and visual illusions.
3.x2.TuringArchMiscDetails has more extra old slides on architecture details, with an emphasis on connecting with Turing's work.
4.MedPhysio&HeartRateVar Videos, slides, and a PNAS paper on explaining cryptic patterns in heart rate variability using robust performance tradeoffs and optimal control theory.
5.BioGlycoOscillate Videos, slides, and a Science paper on explaining the mechanism underlying glycolytic oscillations at the cellular level. Also involves robust efficiency, robust control theory, and the waterbed effect. Lots of extra old slides on architecture in biology, emphasizing bacteria. New and more accessible theory explaining the ubiquity of cryptic oscillations in controlled systems.
6.DynamicsNorms A minimal introduction to dynamics, input-output responses, (finite) impulse response, and norms on signals and systems.
7.Turbulence Robust control applied to study the origin and nature of turbulence in the high shear flows that dominate much of technology and nature. Particular focus on large coherent structures, their connection with blunting of the turbulent profile, and the resulting impact on drag. Emphasizes collaborations with Dennice Gayme and Beverley McKeon.
8.SysLevelSyn System Level Synthesis is our new framework for doing distributed and localized control.
9.Javad Javad Lavaie gave guest lectures.
I'll continue to add to these, but at any time there will be a mix of materials, and not very well organized.
You might want to download the videos, as dropbox will run them in preview mode, which limits their length.
Aims and Motivation
Aims: Complex tech, bio, neuro, med, eco, and socio-econ networks have both strikingly universal shared architectural features and constraining "laws" but with extremely different domain specific details. This course will use familiar case studies to motivate a new mathematical framework for understanding these similarities and differences, emphasizing layering, dynamics, optimization, nonlinearity, learning, communications, and control, sparsity and structure, and tradeoffs between robustness, efficiency, and evolvability. The aim will be to establish a common core of concepts accessible to anyone at Caltech, with optional additional domain and math details more aimed at experts. Lectures will primarily be by video with class time devoted to discussions. Lots of live demos and games involving audience participation.
Motivating case studies and projects will be drawn from a variety of areas. Students with domain expertise can help form teams to explore using new theory in these areas. There will be videos and reading on these topics, but students are urged to suggest others.
Neuroscience: sensorimotor control, spiking neurons, and limits to performance; planning and reflex; modularity, plasticity, learning, and evolution; vision and the vestibular ocular reflex (VOR); balance and bipedalism; automation and (un)conscious; fMRI and EEG analysis; throwing and hitting a 100 mph fastball;
Internet: Software defined networking (SDN), Internet of things (IOT), Application driven networking (ADN), content centric, operating systems, Network function virtualization (NFV), routing, congestion control, end-to-end and local, protocol stacks, aggregation
Microbial cell: metabolism; glycolytic oscillations; stress response; signal transduction; gene regulation; evolution, horizontal gene transfer; immune mechanisms; pathogenesis; microbiomes; phage physiology;
Power: Future smartgrid; optimization, control; cascading outages; challenge of renewables; role of storage and controllable loads
Ecosystems: wildfire ecology; Mediterranean climates; fire in the earth system
Medicine and physiology: cardiovascular physiology, heart rate variability, (an)aerobic metabolism, fatigue, training; sepsis and shock; electrocardiographic diagnosis of acute myocardial infarction; development, wound healing, and cancer
Physics: multiscale and complexity; high shear flows, turbulence, coherent structures, blunting, and drag; statistical mechanics, fluctuation-dissipation, back action and measurement, uncertainty principles, quantum measurement; phase transitions, criticality, optimization;
Social sciences: behavorial economics, casinos, the hot hand fallacy (fallacy), (non)stationary stochastic process, and evolutionary psychology; risk and uncertainty; chimps, bonobos, baboons, and gorillas; orcas and elephants; ants and bees; violence; patriarchy and matriarchy; fear, anger, hatred; empathy, compassion; language; bipedalism and weapons; agriculture and owners, soldiers, and slaves; Egypt, Sparta, Athens, Rome, China; Adam Smith; American Civil War ; technology, industrialization, and automation; World wars and depressions; wealth inequality; the 60s (1860s and 1960s) and "rights"; financialization; tech and social networks; current politics and media; Putin, Trump, and Brexit; Article on animal social justice
Transitions in evolution, architecture, and complexity: e.g. life, metabolism, ribosomes, oxygen, mitochondria, brains, flight, warm blood, maternal care, bipedalism, weapons, language, agriculture, states, money, industry, technology, IOT, ...
Universal laws and architectures: across all domains there are hard limits or "laws" on robustness vs efficiency, speed vs accuracy, with efficiency and robustness conservation laws that limit what is possible; universal architectures allow components to achieve what laws allow, and mix diverse components to create systems that achieve what no component can alone; universals including layering, hourglasses and bowties with thin waists and knots; planning and reflex layers; apps, OS, and hardware; horizontal swapping of apps, genes, memes; these are all largely unknown in science and poorly taught in engineering, with little or no theory, until recently; LEGOS as a model system; Darwin, Turing, Bode, Wiener, Shannon, Von Neumann, the death of cybernetics;
Infectious hijacking: layered architectures are vulnerable to infectious hijacking; mechanisms that swap apps, genes, and memes facilitate hijacking by viral, pathogenic versions; toxoplasmosis and zombie parasites; phage kill half the ocean's bacteria daily; network and automation technologies with increasingly dangerous, pervasive cybersecurity vulnerabilities; human's ultimate weakness may be strongly held beliefs (memes) that are contagious, false, dangerous, and unhealthy; parasites vs predators;
New “sciences”: complexity science; network science; edge of chaos; power laws; self-organized criticality; scale-free networks; small worlds; zombie science; hoaxiness; emergulence
Theory foundations: We will aim for a more integrated and accessible treatment of control theory, dynamical systems, optimization, probability and statistics, machine learning, communications theory, and computational complexity. Focus on "deep" fragilities. Previous familiarity or study is helpful but not essential as the questions arising in the case studies can be largely appreciated with minimal math, though the answers go through layers of increasing theory complexity. Students who do have backgrounds in these areas will be particularly useful in teams with domain experts.
Recommended lite reading (listening)
These are popular books for a general audience available in audio format.
Nick Lane, The Vital Question: Energy, Evolution, and the Origins of Complex Life Audible
Nick Lane, Life Ascending: The Ten Great Inventions of Evolution Audible
Robert M. Sapolsky, Behave: The Biology of Humans at Our Best and Worst Audible
Geoffrey G. Parker, Marshall W. Van Alstyne, Sangeet Paul Choudary, Platform Revolution: How Networked Markets Are Transforming the Economy - and How to Make Them Work for You Audible
Ed Yong, I Contain Multitudes: The Microbes Within Us and a Grander View of Life Audible
Michael Lewis, The Undoing Project: A Friendship That Changed Our Minds Audible
Kathleen McAuliffe, This Is Your Brain on Parasites: How Tiny Creatures Manipulate Our Behavior and Shape Society Audible
Noam Chomsky, Requiem for the American Dream: The Principles of Concentrated Wealth and Power Audible
Robert J. Gordon, The Rise and Fall of American Growth: The U.S. Standard of Living Since the Civil War Audible
Elizabeth Warren, This Fight Is Our Fight: The Battle to Save America's Middle Class Audible