Jaynes Centennial Symposium Agenda
8:30 a.m.-6:00 p.m.
Tuesday, October 15, 2024

8:30 a.m. Welcome Remarks
Feng Sheng Hu, PhD
Dean of the College of Arts & Sciences
Lucille P. Markey Distinguished Professor

8:40 a.m. Aaron Bobick
Dean of McKelvey School of Engineering

8:50 a.m. Kater Murch
Overview of the Day

9:00 a.m. John Clark
Personal Remembrances of Ed Jaynes

Professor Clark will provide some personal anecdotes involving Edwin Jaynes, derived from his experiences as a colleague and friend of Ed during his years at WUSTL, from his arrival from Stanford in early 60s until his death in 1998, along with Professor Clark's own views of his historical importance in the advancement of knowledge and its means of acquisition.

9:15 a.m. First Session, chaired by Shantanu Chakrabartty

9:20 a.m. Nicole Yunger- Halpern
How a paragraph by Jaynes grew into a field: Non-Abelian thermodynamics

Starting in undergraduate statistical physics, we study small systems that thermalize by exchanging quantities with large environments. The exchanged quantities—heat, particles, electric charge, etc.—are conserved globally, and the thermalization helps define time’s arrow. If quantum, the quantities are represented by Hermitian operators. We often assume implicitly that the operators commute with each other—for instance, in derivations of the thermal state’s form. Yet operators’ ability to not commute underlies quantum phenomena such as uncertainty principles. What happens if thermodynamic conserved quantities fail to commute with each other? The first known mention of any similar question appeared in Jaynes’s famous quantum 1957 paper. His one paragraph on the subject, mostly overlooked for decades, has recently blossomed into a subfield. Noncommutation of conserved thermodynamic quantities has been found to enhance average entanglement, decrease entropy-production rates, alter basic assumptions behind thermalization, and more. Non-Abelian thermodynamics illustrates how quantum information science is extending 19th-century thermodynamics, with inspiration from Jaynes. Majidy, Braasch, Lasek, Upadhyaya, Kalev, and NYH, Nat. Rev. Phys. 5, 689-698 (2023). https://www.nature.com/articles/s42254-023-00641-9

9:50 a.m. Eric Lutz
Maximum entropy principle from statistical to quantum physics

Randomness is an integral part of statistical and quantum physics. The probability distribution of a system at equilibrium is thus given by the Boltzmann-Gibbs distribution whereas the probability to obtain a given quantum measurement outcome follows Born's rule. I will discuss the application of the maximum entropy principle to both fields and highlight the difference between a heat bath and a detector.

10:20 a.m. David Wolpert
How Constraints Affect Evolution of Entropy – Strengthened Second Laws

The second law of thermodynamics gives a lower bound on the free energy needed to change one distribution (or density matrix) to another, if there are no constraints whatsoever on the precise physical process that implements that change. As an example, it has been proven that achieving that lower bound requires infinite time. This weakness makes the second law almost useless for understanding all complex systems found in biology, social science, climate modeling, etc. Here I use the modern tools of stochastic thermodynamics to show that we can (vastly) strengthen the second law if we take into account some of the constraints operating on the physical processes in such systems. In particular, I demonstrate such a strengthened second law that applies for the seemingly innocuous constraints of modularity, and / or that the physical process be periodic (as is the case in all digital systems).

10:50 a.m. Break

11:05 a.m. Second Session, chaired by Kater Murch

11:10 a.m. Shankar Mukherji (Jaynes Endowed Postdoc)
The Edwin Thompson Jaynes Postdoctoral Fellowship in Physics

11:15 a.m. Irfan Siddiqi
Harnessing the Power of the Unseen Quantum World

Quantum mechanics describes the physical world around us with exquisite precision, with no known violations of the theory. Ironically, this precision has been accompanied with philosophical uncertainty, as the theory allows an object to simultaneously exist in multiple realities, and disparate objects to have entangled destinies, that is until someone looks at them. Hence the debate: mathematical artifact or the true nature of the unseen world? The mathematical models pioneered by Jaynes and colleagues give us a lens into the quantum world, validating the most exotic predictions of the theory, and establishing the technological principles needed to wire up quantum entanglement for quantum computation.

11:45 a.m. Lan Yang
Exploring Light-Matter Interactions through Jaynes’ Lens of Inference and Physics

12:00 p.m. Jung-Tsung Shen
From the Jaynes–Cummings Model to Correlated Photon Transport: Unveiling Quantum Light–Matter Interactions

The Jaynes–Cummings model stands as a cornerstone in quantum optics, capturing the fundamental interaction between a two-level atom and a quantized mode of an optical cavity. In this talk, we will introduce the key concepts of the Jaynes–Cummings model, elucidating its profound impact on our understanding of quantum light–matter interactions. Building upon this foundation, we will delve into the realm of correlated photon transport, exploring how extensions of the model reveal novel phenomena with implications for quantum information processing and novel coherent light sources. This presentation aims to bridge foundational quantum theory with contemporary research, offering insights accessible to a diverse scientific audience.

12:15 p.m. Lunch

1:00 p.m. Third Session, chaired by Gaia Tavoni

1:05 p.m. Bill Bialek
Inference and the physics of life: From flies to flocks

1:35 p.m. Dmitry Krotov
Dense Associative Memory and its potential role in brain computation

In the spirit of Edwin T. Jaynes' long term vision of information processing and computation performed by physical and biological systems I will describe a peculiar energy-based network, called Dense Associative Memory (DenseAM), which is deeply rooted in the ideas of statistical physics and Ising-like spin models. In contrast to conventional Hopfield Networks, which were popular in the 1980s, DenseAMs have a very large memory storage capacity - possibly exponential in the size of the network. This aspect makes them appealing tools for many problems in AI and neurobiology. In this talk I will describe two theories of how DenseAMs might be built in biological “hardware”. According to the first theory, DenseAMs arise as effective theories after integrating out a large number of neuronal degrees of freedom. According to the second theory, astrocytes, a particular type of glia cells, serve as core computational units enabling large memory storage capabilities. This second theory challenges a common point of view in the neuroscience community that astrocytes play the role of passive housekeeping support structures in the brain. In contrast, it suggests that astrocytes might be actively involved in brain computation and memory storage and retrieval. This story is an illustration of how computational principles originating in statistical physics of spin systems may provide insights into novel AI architectures and brain computation.

2:05 p.m. Mark Anastasio
Approximating the Bayesian ideal observer for optimizing biomedical imaging systems

The performance of the Bayesian ideal observer (IO) acting on imaging measurements has long been advocated as a figure-of-merit (FOM) to guide the optimization of biomedical imaging systems. For computed imaging systems, the performance of the IO acting on raw imaging measurements sets an upper bound on task-performance that no image processing or reconstruction method can transcend. As such, estimation of IO performance can facilitate the design of data-acquisition protocols that maximize the amount of task-related information in the acquired data. While such IO analyses are well-known conceptually, they have generally remained infeasible to widely implement. In this talk, we describe recent advances that leverage deep learning technologies to approximate the performance of the IO for diagnostic tasks relevant to biomedical imaging applications.  Specifically, the use of deep convolutional neural networks and deep generative models for approximating the IO test statistic will be described.

2:35 p.m. Break

2:50 p.m. Fourth Session, chaired by Sadas Shankar

2:55 p.m. Keith Hengen
A Unifying Principle of Brain Function

Inspired by Edwin Thompson Jaynes' information-theoretic approach and the principle of maximum entropy, this talk presents compelling evidence that the brain operates at a critical point, optimizing its computational capacity. This framework unifies our understanding of neural dynamics under the principle of criticality, where neural systems sit delicately at the border between order and chaos, exhibiting scale-invariant dynamics, marginal stability, and optimal information processing. Our findings suggest that this critical regime serves as a universal set-point established during development, sustained by homeostatic mechanisms and sleep, with deviations linked to degraded function. This perspective offers powerful insights into neurodegenerative diseases, brain development, and evolution, laying the groundwork for future research. By framing brain function through the lens of criticality, this work extends Jaynes' contributions to statistical mechanics and provides a deeper understanding of the fundamental principles governing neural computation across species and scales.

3:10 p.m. Dave Piston
Precise sub-cellular temperature measurement through nanodiamond quantum sensing

Nanodiamonds have been used to probe local magnetic noise and temperature in numerous systems. These quantum sensors are well-suited to measurements of heat output due to the temperature sensitive of the electron spin resonance of the nitrogen-vacancy (NV) defect in the diamond lattice. We are interested in using the quantum properties of nanodiamonds as a photostable and non-cytotoxic live cell biosensor to measure sub-cellular temperature changes that arise from metabolic perturbations. Nanodiamonds loaded into the cell cytoplasm show a significant decrease in cellular temperature from macrophages that lack the heat-generating UCP-1 protein.  Since the mitochondria act as the main energy source of the cell, they have long been a target for intracellular measurements.  Towards better understanding of cellular energy flows, we have developed a robust approach to target single nanodiamonds to mitochondria for live cell intracellular temperature measurements.  We have built a chemical targeting platform demonstrate the ability to selectively bind the nanodiamonds onto a peripheral protein (TSPO) on the mitochondria membrane using a reliable conjugation protocol.  We modify the nanodiamond surface to allow conjugation to a (G4)-PAMAM dendrimer labeled with 1-(2-chlorophenyl)isoquinoline-3-carboxylic acid that binds tightly to the TSPO protein.  Using confocal microscopy imaging, we demonstrate that single nanodiamonds can be visualized in the cell with >95% colocalization to the mitochondria. The fluorescence intensity distribution is used to characterize the density of nanodiamonds and establish the presence of single nanodiamonds in the cellular environment.  By determining the position of single nanodiamonds via fluorescence and measuring their electron spin resonance frequency, we can capture local temperature changes resulting from external metabolic perturbations to the cell. With specific targeting to a TSPO protein, our conjugated nanodiamond can probe the fluctuations in mitochondrial temperature.

3:25 p.m. Peter Price
Extending Jaynes’ Bayesian View of Fickian Diffusion

Abstract In Clearing Up Mysteries - The Original Goal, Jaynes derived Fick’s Law for a dilute binary solution from Bayes’ Theorem by considering, probabilistically, the motion of the dilute solute molecules. Modifying Jaynes’ prior, changing the frame of reference, and allowing for multicomponent systems, one can follow Jaynes’ logic to arrive at several expressions for the diffusion coefficient derived from friction-based considerations. These results, however, do not generally satisfy required conditions over the full concentration range. This limitation is resolved by considering the joint motion of all components in the solution. Doing so, one arrives at new expressions for binary and multicomponent diffusion coefficients that include correlation terms to account for intra- and inter-species forces that impact diffusive motion through intermolecular associations.

3:45 p.m. Group Discussion: The Future Applications of Jaynes Research, Chaired by Nicole Moore with Bhavna Hirani

5:00 p.m. Reception, adjourn by 6:00 p.m.