Noah Olsman

Ph.D. Candidate in Control and Dynamical Systems

California Institute of Technology


My name is Noah Olsman, and I am currently a Ph.D. candidate at the California Insitute of Technology in the Department of Computing and Mathematical Sciences, where I am getting my degree in Control and Dynamical Systems coadvised by professors Lea Goentoro and John Doyle. My research broadly focuses on Systems Biology, where we use the mathematical tools and concepts from engineering, such as sensitivity, robustness, and efficiency, to analyze biological processes.

I was born and raised in Los Angeles where I attended the University of Southern California, majoring in Electrical engineering and minoring in Mathematics. While at USC my interests largely revolved around robotics. I was a member (and for a while team lead) of the USC Competition Robotics team for four years working on an autonomous submarine for the annual RoboSub competition. After graduating in 2012, I decided to try something completely new and spent about a year working at Yale University in the lab of Thierry Emonet where I developed some new mathematical models of bacterial chemotaxis. In the Fall of 2013 I began my Ph.D. at Caltech, where I am working on whatever interesting problems come my way. You check out the work I've done so far on my Google Scholar page.

Outside of academia, I am the Chief Analytics Officer of the 501(c)3 non-profit Seed Consulting Group, which provides pro bono consulting services to environmental groups in California. My initial work with Seed primarly focused on water policy, where I've served as a consultant to both the Natural Resource Defense Council and Heal the Bay. Now I am focusing on developing Seed Insights, a public-facing medium for discussing the environmental issues that our clients and consultants work on. We're just getting this started, so stay tuned!

My biggest hobby is playing music. I have been playing the slide guitar for years now and recently started playing the mandolin. I try to expose myself to as much different music as possible, right now I am focusing on learning classical (mostly Bach) and bluegrass music. My favorite genre of music will always be the blues, primarily from the early 20th century. Musicians like Son House and Blind Willie Johnson are my biggest influences, and I spend a lot of time just trying to figure out how they made the sounds they did. My two favorite recordings in the world are Blind Willie Johnson's Dark Was the Night, Cold Was the Ground and Chris Thile's mandolin arrangement of the Prelude from Bach's Violin Partita No. 3 in E Major.

Research Projects

Logarithmic Sensing in Biological Signaling Systems

Sensory systems in biology are faced with two goals: they must be able to sensitively respond to changes in stimuli, while also being able to sense a broad range of possible intensities. For example, our eyes can detect changes of just a few photons in light intensity, yet we can also see in a dim room just as well as we can in broad daylight. One way that signaling systems can achieve both of these goal simultaneously is with a phenomenon known as Fold-Change Detection, where the internal response of the cell is a function not of the absolute change in intensity, but the ratio of stimulus to background. This means that a cell can respond to a change from 1 to 3 just as well as it can respond to a change from 100 to 300. I am currently using mathematical models of protein dynamics to understand the molecular mechanisms that facilitate Fold-Change Detection in real biological systems. This work was primarily with my adviser Lea Goentoro, and was presented at the 2015 Winter q-bio Meeting. It has since been published in the Proceedings of the National Academy of Sciences, with open access under the title Allosteric Proteins as Logarithmic Sensors.

Modular Structure in Biological Networks

For decades, biologists have sought to understand how the vast array of genes, proteins, and cellular structures that exist come together to form living the world. In recent years, it has become possible to extensively catalog not only the individual components in various biological systems, but also the networks of interactions between them. We are now at a point in time where it is possible to analyze the structure of biological networks consisting of dozens, or even hundreds, of interacting components. To this end, I am working in a collaborative project across biology, computer science, and mathematics to try to understand precisely how network structure is related to function. Our first investigation of this was in a newly identified fat-storage network of about 100 genes in yeast, where we studied how a gene's location in the network is related to its importance, measured by how much fat storage in mutants changed relative to the wild type. We are now broadening this analysis to much larger genetic, proteomic and neuronal networks. This work has involved many collaborators, and is primarily through the Zinn Lab at Caltech.

Efficient Bayesian Social Learning

This project focuses on understanding social learning, the process through which a network of agents can make observations of each other and try to make inferences about the world. For example, imagine a large group of people each individually trying to solve a puzzle. Each person might be able to observe the progress of their neighbors in order to solve the puzzle faster than any individual could on their own. This is an example of a consensus problem, where many agents try to come to a common conclusion by sharing information and observing each other. I worked with Professors Omer Tamuz and Elchanan Mossel to study a particular model of social learning to try to understand how efficiently agents can come to a consensus given a particular network structure. With this work we hope to shed light on what particular stuctural properties of a network determine its capacity to transmit information. This paper was accepted to the 2016 Allerton Conference, you can read the ArXiv preprint here.

If you would like to contact me, please send an email to