Computations in Science Seminars
Apr
22
Wed 12:15
Joseph Vallino, Marine Biological Laboratory
e-mail:
Host: Wendy Zhang ()
Organizer: Kim Weirich ()
Living systems defined in the context of maximum entropy production and information: A computational approach

The maximum entropy production (MEP) principle holds that non equilibrium systems with sufficient degrees of freedom will likely be found in a dynamic state that maximizes entropy production or, analogously, maximizes potential energy destruction rate. The theory does not distinguish between abiotic or biotic systems; however, I will show that systems that can coordinate function over time and/or space can potentially dissipate more free energy than purely Markovian processes (such as fire or a rock rolling down a hill) that only maximize instantaneous entropy production. Biological systems have the ability to store useful information acquired via evolution and culled by natural selection in genomic sequences that allow them to execute temporal strategies and coordinate function over space. For example, circadian rhythms allow phototrophs to “predict” that sun light will return and can orchestrate metabolic machinery appropriately before sunrise, which not only gives them a competitive advantage, but also increases the total entropy production rate compared to systems that lack such anticipatory control. Similarly, coordination over space, such a quorum sensing in microbial biofilms, can increase acquisition of spatially distributed resources and free energy and thereby enhance entropy production. In this talk a computational modeling framework will be presented to describe microbial biogeochemistry based on the MEP conjecture constrained by information and resource availability. Results from model simulations will be compared to laboratory experiments to demonstrate the approach.

Apr
29
Wed 12:15
Michael Rubenstein, Harvard
e-mail:
Host: Leo Kadanoff ()
Organizer: Ivo Peters ()
Taming the Swarm: Control and Design of Multi-Robot Systems.

Advances in technology have begun to allow for the production of large groups, or swarms, of robots; however, there exists a large gap between their current capabilities and those of swarms found in nature or envisioned for future robot swarms. These deficiencies are the result of two factors, difficulties in algorithmic control of these swarms, and limitations in hardware capabilities of the individuals. Creating a hardware system for large robotic swarms is an open challenge; cost and manufacturability pressure hardware designs to be simple with minimal capabilities, while algorithm design favors more capable hardware. The robot design must balance these factors to create a simple robot that is, at the same time, capable of performing the desired behaviors. To investigate these challenges, I created the Kilobot robot swarm, a swarm of 1024 (“kilo”) robots. In this talk, I will discuss the many challenges associated with creating a robot swarm at this scale and the implications this has for creating even larger, more capable swarms in the future.

Controlling these swarms is also a challenge, as the properties desired from these systems, e.g. shape, locomotion, are generally a global property; however, we can only control local interactions between individuals. Furthermore, the mapping between controllable local behaviors and desired global results is not well understood. Their control is further complicated by the very nature of these systems which are composed of decentralized, distributed, asynchronous, error-prone individuals with often limited capabilities. I will discuss two examples of algorithms recently implemented on the Kilobot swarm, self-assembly of user-defined 2D shapes, and the collective transport of objects. Both of these examples provide guarantees of correctness and performance bounds of the swarm, and provide examples of reliable global-to-local control over a robot swarm. I will describe unexpected challenges faced while trying to control the Kilobot swarm, and how these challenges will influence the design of future swarm algorithms.

May
6
Wed 12:15
Tim Sanchez, Harvard
e-mail:
Host: Leo Kadanoff ()
Organizer: Shiladitya Banerjee ()
Self-organization in Cytoskeletal Materials

Inspired by biological functions such as ciliary beating and cytoplasmic streaming, we developed a highly tunable and robust model system from cytoskeletal components that self-organizes to produce a broad range of far-from-equilibrium materials with remarkable emergent properties. Using only simple components -- microtubules, kinesin motor clusters, and a depletion agent that bundles MTs -- we reconstituted analogues to several essential biological functions, including cilia-like beating, metachronal waves in bundle arrays, and internally generated flows in active cytoskeletal gels. Beyond these biomimetic behaviors, we have also used the same components to engineer novel active materials which have no biological analogues: active streaming 2D nematics, self-propelled emulsion droplets, and self-deforming vesicles. Since these initial observations, theoreticians have recapitulated many of these experimental results with physical models of cytoskeletal mixtures. This underscores the value of model systems such as ours for better understanding the fundamental principles that drive self-organized processes. This could one day lead to the systematic engineering of far-from-equilibrium materials with highly sought-after collective and biomimetic properties.

In my graduate work, I systematically varied energy levels (ATP) and characterized the response in our system’s collective dynamics. I will also discuss my current research, investigating the possible effects that varying cellular energy levels may have on the self-organized properties of the mitotic spindle. The spindle is also composed of microtubules in a liquid crystalline phase and motor proteins, and is essential to life because it mediates chromosome segregation. In vivo, energy levels are determined by the mitochondria and the cell's metabolism. We are able to quantitatively characterize the metabolic state of cells using Fluorescence Lifetime Imaginge Microscopy (FLIM), and are now investigating whether metabolic activity affects spindle function and chromosome segregation.

May
13
Wed 12:15
Luis Bettencourt, Santa Fe Institute
e-mail:
Host: Daniel Holz ()
Organizer: Sayantan Majumdar ()
The Mathematics of Cities

Human cognitive and social systems are perhaps the final frontier for mathematical scientific theory. While well-known methods of statistical physics and scientific computation are useful as entry points to a fast growing body of data, critical formal innovations are also necessary that describe these systems in their own terms.

Cities, in particular, provide a rich, novel and increasingly empirically available set of problems where open-ended adaptation at different scales builds large-scale socioeconomic networks in interaction with infrastructural systems embedded in space and time.

In this talk, I will describe the emerging mathematics of cities. The crucial starting element deals with the quantification of the general properties of urban areas, which become apparent through scaling analysis and associated statistics. Based on a set of regularities that I will demonstrate empirically, I then build a mean-field theory that derives the scaling of many socioeconomic, infrastructural and physical properties of cities and reveals the basic trade-offs involved in these systems.

I will then demonstrate how the detailed fabric of cities can be understood through a process of spatial selection and show how the complexity of explanations at the local level (groups, neighborhoods) can be quantified in units of information relative to more coarse-grained descriptions, in a way analogous to renormalization group transformations in statistical physics.

I will end with some general (speculative) thoughts on the convergence between methods of statistical physics, the mathematics of selection and basic aspects of human social behavior and cognition that may provide a path to a more integrated quantitative understanding of complex adaptive systems.

May
20
Wed 12:15
Andrew Ferguson, University of Illinois at Urbana-Champaign
e-mail:
Host: Leo Kadanoff ()
Machine learning of viral fitness landscapes and protein folding funnels

“Big computing” – petascale systems and the multicore paradigm – has enabled rapid, large-scale biomolecular simulation and property prediction. Similarly, “big biology” – high-throughput sequencing and the “-omics” revolution – has heralded voluminous bioinformatics databases. These large data sets present exciting opportunities to advance scientific understanding, but their size presents new challenges, and demands new paradigms, for their analysis. In the first part of this talk, I will discuss the translation of clinical sequence databases into viral fitness landscapes based on spin glass models from statistical physics. In an application to hepatitis C virus, we identified particular viral vulnerabilities and rationally designed T-cell vaccines to hit the virus where is hurts. In the second part of this talk, I will describe an approach integrating ideas from dynamical systems theory and nonlinear machine learning to infer multidimensional biomolecular folding funnels from univariate experimental measurements.

May
27
Wed 12:15
Matthew Pinson, University of Chicago
e-mail:
Host: Leo Kadanoff ()
Signal Transmission through Disordered Hypostatic Materials

A bag of sand is a slightly hypostatic system: the number of constraints is just a little smaller than the number of degrees of freedom. As a result, several linearly independent modes of motion are available at zero energy cost. The question naturally arises: can we use these modes to transmit information from one side of the system to the other? In this talk, I will explain why we cannot. Even though each mode considered on its own spans a large portion of the system, combining the modes yields only a few independent long range modes, and many localised modes. Thus the effective number of free modes seen by any small portion of the system is much smaller than we would have guessed based on Maxwell counting. This provides an unexpected limitation on the perturbations that can be applied, and even most of those that are accessible are not transmitted.

Jun
3
Wed 12:15
Vishal Soni, University of Chicago
e-mail:
Host: William Irvine () *
Jun
8
Mon 12:15
Stephan Herminghaus, Max Planck Institute
e-mail:
Host: Tom Witten
Jun
17
Wed 12:15
Marilena Loverde, University of Chicago
e-mail:
Host: Leo Kadanoff ()
Jun
24
Wed 12:15
OPEN *
Jul
1
Wed 12:15
OPEN *
Jul
8
Wed 12:15
OPEN
Jul
15
Wed 12:15
Bradford Benson, University of Chicago
e-mail:
Host: Leo Kadanoff ()
Jul
22
Wed 12:15
OPEN
Jul
29
Wed 12:15
Shiladitya Banerjee, University of Chicago
e-mail:
Host: Leo Kadanoff ()
Aug
5
Wed 12:15
Radha Ramachandran, Eve Tulbert, University of Chicago, Freedom Games
e-mail:
Aug
12
Wed 12:15
Stephane Perrard, University of Chicago
e-mail:
Host: William Irvine ()
Aug
19
Wed 12:15
Sayantan Majumdar, University of Chicago
e-mail:
Host: Leo Kadanoff ()
Aug
26
Wed 12:15
Kim Weirich, University of Chicago
e-mail:
Host: Leo Kadanoff ()
Oct
7
Wed 12:15
Zvonimir Dogic, Brandeis University
e-mail:
Host: William Irvine ()
Oct
14
Wed 12:15
Michael Oppenheimer, Princeton
e-mail:
Host: Leo Kadanoff ()
Oct
21
Wed 12:15
Kathleen Stebe, University of Pennsylvania
e-mail:
Host: Leo Kadanoff ()
Oct
28
Wed 12:15
Tom Lubensky, University of Pennsylvania
e-mail:
Host: Leo Kadanoff ()
Nov
4
Wed 12:15
OPEN
Nov
11
Wed 12:15
David Schuster, University of Chicago
e-mail:
Host: Leo Kadanoff ()
Nov
18
Wed 12:15
OPEN
Dec
2
Wed 12:15
OPEN
Dec
9
Wed 12:15
Charles Kane, University of Pennsylvania
e-mail:
Host: Leo Kadanoff ()
Dec
16
Wed 12:15
OPEN