Welcome to FSRI Research! We’re super excited to introduce you to lots of amazing stuff at Caltech!
List of Research Projects
Click on a project’s title to see more details!
Volcanic history of the Isla Tortuga, Gulf of California, Mexico
Description: Isla Tortuga corresponds to a young volcano in the Guaymas Basin, Gulf of California and it is part of the of the volcanism produced by the extensional tectonics (the plates separate from each other) in the Gulf of California. During the 70s (e.g. Batiza, 1978) some research was conducted in this Island to understand its origin. Hoever, the age of the volcano is still unknown. Last summer we conducted an expedition to Isla Tortuga to collect samples for geochronological and geochemical analyses. In addition to our samples, we are going to recover some samples from the 70’s expeditions to conduct more analysis. For this project, the students would help to process the samples from the 70s expedition. The rock processing involves: crushing, ball-milling, and making beads for the geochemical analysis.
Required Knowledge: Enthusiasm about rocks and geology! I would recommend that the students learn the following book, however is not mandatory: Tarbuck, E. J., Lutgens, F. K., Tasa, D., & Tasa, D. (2005). Earth: an introduction to physical geology (p. 744). Upper Saddle River: Pearson/Prentice Hall
AT THE CORE OF URANUS
Description: Our project focuses on the fascinating H2-H2O system, which is crucial for understanding the complex structures and phases within Uranus. Participants will dive into the literature to collect data on the phase stability of hydrogen and water under extreme conditions. They will then use the machine learning code provided to construct phase diagrams that contribute to our understanding of the internal dynamics of Uranus.
Required Knowledge: No programming experience is required (some Python experience would be helpful), just a keen interest in the cosmos and a willingness to learn!
Imaging chicken eggs at early embryonic stage via laser speckle imaging
Description: The student will assist us in our efforts in building a compact an optical device to image chicken eggs at early embryonic stage via laser speckle imaging. The aim of our project is to sex chicken egg at early stage, to reduce/suppress chick culling. The student will learn programming tools (Matlab), some optical techniques and biological methodology.
Required Knowledge: None
Graham's pebbling conjecture for novel products of graphs
Description: Given a graph, we place pebbles on some of the vertices. We can trade 2 pebbles on one vertex for 1 pebble on an adjacent vertex. A vertex is reachable in a configuration of pebbles if we can get at least 1 pebble to that vertex. The pebbling number of a graph is the smallest number of pebbles so that every vertex is reachable given any configuration with that number of pebbles. Graham’s conjecture famously claims that the pebbling number of a Cartesian product of graphs is at most the product of the pebbling numbers of the separate graphs. We will study Graham’s conjecture for different graph product constructions.
Required Knowledge: There are no pre-requisites and no required knowledge. We will start the project by reading about the problem and trying early examples of pebbling.
Robotic Mapping for Earth Science Studies
Description: Efficient and accurate mapping of surface fractures in geologic fault zones is critical for earthquake research and crucial in facilitating rapid response and rescue efforts. However, existing methods, including human-led field surveys and satellite imaging, present some significant limitations. Field surveys can be time-consuming, risky, and often impractical. Satellite techniques like InSAR or pixel tracking can suffer from delayed data acquisition (e.g. 1-2 weeks) and inadequate resolution for detecting fine fractures on the scale of 10-30 meters. To address these challenges, we plan to develop a heterogeneous multi-UAV system, designed to significantly improve the speed, safety, and resolution of fracture mapping in fault damage zones. Our goal is to map fault zones up to 5 km in length with an unprecedented ground sampling resolution of 5 mm, offering critical insights into fault fracturing processes and enhancing rapid response capabilities. We will test our system on recent fractures in the Southern San Andreas Fault system near the Salton Sea.
Required Knowledge: Python Programming, Linear Algebra
Eddy Dynamics in the Ross Sea
Description: Eddies are swirling ocean currents with a scale of approximately 100 km which carry heat and other properties with them as they traverse the ocean. These currents, accounting for over 80% of the ocean’s kinetic energy, play a crucial role in regulating Earth’s climate. However, in polar regions, observing eddies poses challenges. This project aims to utilize new observations and high-resolution models to investigate how eddies contribute to transporting heat from the open ocean to the Ross Sea, a body of water bordering Antarctica.
Required Knowledge: The main pre-requisite is an interest in physics, climate or both! Familiarity with Python or another scientific programming language, particularly for data visualization, is desirable. Basic calculus would be useful but not essential. The goals can be flexible depending on the students’ backgrounds.
Probing evolution of neural representations during skill learning
Description: We aim to study how humans learn complex cognitive/ motor skills. WE wish to trace this process through computational hypotheses and neuroimaging. For FSRI students, we have a few human behavioral tasks that they can help us design and launch.
Required Knowledge: It would be great to have some programming experience in either python / HTML & Javascript.
A Distributionally Robust Approach to Adaptive Control
Description: Uncertainty mitigation is a fundamental challenge in safety-critical control and decision making. Traditional approaches to uncertainty mitigation consider two extreme settings: 1) optimizing average performance under stochastic uncertainty arising from entirely random and irregular disturbances, 2) optimizing the worst-case performance under adversarial uncertainty arising from structured and patterned disturbances. The dichotomy of stochastic vs adversarial uncertainty has led to many successes (e.g., moon landing, safe air travel), these approaches do not capture complex and dynamic structure of the real-world systems. This necessitates the need to adopt new approaches to uncertainty mitigation to better adapt to realistic dynamic systems. In my research, I develop the theoretical foundations for a novel approach, distributionally robust control. This approach allows us to tackle uncertainties with varying degree of structured randomness, thereby bridging the gap between stochastic and adversarial realms.
Required Knowledge: Calculus (derivatives, integrals, minimum/maximum of functions), Basic linear algebra (matrix arithmetics, inner/dot product, etc), Calculus-based probability theory (probability densities, expectation, conditional probability, Gaussian distribution, etc.)
Predict fast sodium-ion conducting electrolytes for solid state sodium ion batteries.
Description: Fast sodium-ion conducting electrolytes are materials that can transport sodium ions at high rates and low resistance, enabling the development of solid-state sodium-ion batteries. These batteries have advantages over conventional lithium-ion batteries in terms of cost, safety, and sustainability. There is a ceramic material with a layered structure that allows sodium ions to move along the planes. There are also family of compounds called NASICON with a general formula of Na1+xZr2SixP3−xO12, where x can vary from 0 to 3. They have a three-dimensional framework structure with channels for sodium ions. These are some of the promising candidates for fast sodium-ion conducting electrolytes. However, there are still many challenges and opportunities for improving their performance, stability, and compatibility with electrodes and interfaces.
Required Knowledge: Knowledge of basic mathematics and chemistry are pre-requisites. Knowledge of python is added advantage.
Heterologous production of drug candidate molecules in a model plant (Nicotiana benthamiana)
Description: Students will work with me to apply synthetic biology principles to build plasmid constructs to reconstruct a novel metabolic pathway that produces a valuable plant metabolite with medicinal applications. In a current project in our lab, we are learning the biosynthetic pathways that novel plants use to produce medicinal alkaloids using genomic, transcriptomic and metabolomic data. Once we have learnt these pathways, we aim to produce these valuable compounds using plants as our biofactories. This research seeks to mediate the challenges with meeting the needs for medicinal compounds that are extracted directly from plants and the challenges with chemical synthesis of these complex chemical stuctures.
Required Knowledge: The minimum prerequisite for this time of work would be general molecular biology principles governing DNA, RNA and proteins in prokaryotic and eukaryotic organisms. The rest of the knowledge is easily teachable in a way that makes it understandable for an incoming undergraduate student.
From standard neural networks to recent advances
Description: Machine learning algorithms have made remarkable advancements in recent years. This project will introduce the foundational concepts of neural networks, including their structure and optimization. Students will gain hands-on experience by developing their own neural networks code, both from scratch and using libraries, to address regression and classification. Further, we will explore recent advancements in machine learning. We will tailor the specific topics to the interests of the students, such as diffusion-based generative models, and learning with neural operators.
Required Knowledge: Basic knowledge in Linear Algebra. Basic skills in Python.
Mathemaics of Enzyme Kinetics
Description: Integrate and display the simulation of differential equations describing enzyme kinetics. Develop simpler algebraic equations to characterize the work and improve the present state of the art. Compare the work with recent Covid-19 data results.
Required Knowledge: Calculus
Understanding the role of supporting electrolytes in Cu-catalyzed electrochemical CO2 reduction in acetonitrile
Description: Electrochemical reduction of CO2 (CO2RR) to value-added products provides a promising pathway for the production of fuels and commodity chemicals. Understanding the effect of the electrolyte and optimizing the electrolyte composition are important in tuning the selectivity of Cu-catalyzed CO2RR. We have found that the activity of H2O can affect the rate of product formation and the mechanistic pathway toward C1 and C2 products in acetonitrile using TBABF4 supporting electrolytes. However, the effect of the identity of the supporting electrolyte on the selectivity of CO2RR is still not well understood. We will be studying CO2RR using different supporting electrolytes to understand the effect of anion on CO2RR product selectivity.
Required Knowledge: Fundamental high school level chemistry knowledge.
Directed Evolution of Enzymes toward New-to-Nature Chemistries
Description: Natural evolution has provided a wide range of enzymes that assemble the molecules of life with a chemical prowess unmatched by human chemists. In the Arnold lab, we mimic Nature’s intricate process of evolution using protein engineering methods, such as Directed Evolution. Our goal is to engineer enzymes to perform chemical transformations that are unknown to the biological world, but that have applications focused on modern human needs (e.g. pharmaceuticals, agriculture, bioremediation, etc.). During this research stint, you will perform 1-2 rounds of directed evolution on a new-to-nature chemical transformation using concepts and techniques in molecular biology, organic chemistry, analytical chemistry, and data visualization.
Required Knowledge: Some knowledge of molecular biology and organic chemistry would be useful; but is not required.
Monitor the health of Radio array using Raspberry Pi and LED
Description: We will use the 3d printing facility on campus to make a mode of the Radi array. Connect LED strips to each antenna. Take health data of the array from a database and then code the raspberry pi to light up the LEDs based on if the antenna in the array is functioning or not
Required Knowledge: Algorithm building. Any coding is a plus but not essential. If they have worked with any DIY or at-home simple electronic projects using LEDs - that would be helpful too.
Genetic Circuit Development for Non-Invasive Cellular Control with Ultrasound
Description: In this research project, you will collaborate with a fellow FSRI student to develop and possibly test innovative genetic circuits. These circuits are designed to make mammalian cells express a specific gene in response to ultrasound stimulation. This capability allows for remote control of gene expression in deep tissue. Throughout the project, you will master key molecular cloning techniques, which are fundamental to the field of synthetic biology.
Required Knowledge: Basic knowledge of molecular biology, especially the central dogma and its molecular components.
Simulation and Experimental Verification of Inflatable Lunar Landing Pad
Description: PILLARS is the Plume-deployed Inflatable for Landing and Launching Abrasive Regolith Shielding, a finalist of the NASA BIG Idea Challenge for 2024. Work closely with a large group of undergraduates to develop simulations of rarified plume flow on the moon and fluid-membrane dynamic interactions, or work on prototyping autonomous deployment and our hot-fire rocket test bed. You will have the opportunity to join the team at the Big Idea Forum at NASA’s Langley Research Center in November.
Required Knowledge: For students interested in Computation Fluid Dynamics, competency in c++ is required. Other projects have no required skills.
Computational Identification of Virus Entry Rules for Cell-Type Specific Infection
Description: There are an estimated 10^31 individual viruses on Earth, which amounts to 10 million for every star in the known universe. More than 300,000 virus species are estimated to cause human infectious disease, however, less than 300 have been successfully detected in humans. Of the known viruses infecting humans, many have been implicated in complex diseases such as heart disease, cancer, and neurodegeneration. Excitingly, the application of large-scale transcriptomics analyses and machine learning-based structural biology tools make possible, for the first time, a systematic and expansive exploration of the virome at single-cell resolution, and thus hold the promise of new insights into the mechanisms governing viral entry into cells. This project will combine a novel method for accurate, efficient detection of viral infections from sequencing data with AlphaFold2-based structural modeling to identify protein-protein interactions governing viral entry into cells and tissues.
Required Knowledge: Basic computer science and coding (Linux, Python/R helpful); basic theoretical knowledge of molecular and cell biology (central dogma, protein structure and function)
Characterizing solar storms for a Parker Solar Probe catalog
Description: Solar energetic particle (SEP) events are the fastest moving component of solar storms which pose a hazard to astronauts and spacecraft alike. Despite 24-hour monitoring with a wide variety of instruments such as particle detectors, magnetic field antenna, and telescopes across the electromagnetic spectrum, the conditions on the Sun which lead to SEP events remain elusive. We would like FSRI students to parameterize and catalog solar storms, over the last 5 years, via in-situ & remote data from several spacecraft. This resulting catalog will become a training dataset for a machine learning project which aims to predict which types of solar storms produce SEP events.
Required Knowledge: Understanding of basic astronomy and coordinate systems. Ideal candidates should have a sharp attention to detail to annotate solar observations. Observational tools are based on IDL and Python code, but coding is not necessary.
Optical Cavity Construction and Characterization for LIGO Upgrade
Description: We are working on an upgrade to the LIGO readout cavity that will improve the quality of gravitational wave signals. Construction of our Output Mode Cleaner cavities is happening now in our labs, and we want you to help! As we build the cavities, we measure their performance, which presents an interesting opportunity for participants to conduct experimental physics research and to develop data analysis skills.
Required Knowledge: Engineers, physicists, and anyone with an interest in optical and electrical measurement techniques will be a great fit. No pre-requisite skills or competencies.
Analysis and testing of mixing efficiency of rocket injector elements
Description: When designing an injector for use in a rocket engine, one of the most important things to categorize is the ability of the element to properly mix the two propellants. However, this is hard to categorize due to the fact that in actual operation high pressure combustion creates an adverse environment for most standard measurement techniques. A technique used in literature to estimate the mixing without having exposure to any dangerous pressures, temperatures, or chemicals is running the element at low pressure with two liquids that do not dissolve into one another (i.e. mineral oil and water). The project is testing various elements the club has developed with this technique to characterize the mixing efficiency
Required Knowledge: Hands-on experience with prototyping techniques such as 3D printing would be helpful but is not strictly necessary. Furthermore, experience with introductory physics, especially anything to do with fluid mechanics would allow a faster start on the project but again is not necessary