# Samuel Talkington — Full Site Content

> Incoming assistant professor of electrical and computer engineering at the University of Michigan (Fall 2027), where he will lead the Eigenergy Group. Completed his Ph.D. at Georgia Tech, advised by Daniel K. Molzahn, and is spending the intervening year as a postdoctoral fellow at Harvard University. Research develops efficient algorithms for societal-scale problems in electric power networks, treating randomness as a computational resource. Supported by the NSF Graduate Research Fellowship.

## Papers

### Analyzing the Cyber Resilience of Distribution Systems with Vulnerable Inverter-Based Resources
**Authors:** Betelihem Kebede Ashebo, Anna Raymaker, Samuel Talkington, Richard Asiamah, Animesh Chhotaray, Saman Zonouz, and Daniel K. Molzahn
**Venue:** 60th Hawaii International Conference on System Sciences (HICSS) (to appear)
Analyzes how vulnerable inverter-based resources shape the cyber resilience of electric distribution systems.

### Making Every Bit Count for A-Optimal State Estimation
**Authors:** Cameron Khanpour, Daniel Turizo, and Samuel Talkington
**Venue:** 65th IEEE Conference on Decision and Control (to appear)
**arXiv:** https://arxiv.org/abs/2604.01211
Controls how bits are allocated to quantized sensor data, improving the error of linear minimum mean squared error (LMMSE) state estimators.

### Grid Trouble in Paradise: Uncovering Vulnerable Distributed Energy Resources and Their Grid-Level Risks
**Authors:** Anna Raymaker, Samuel Talkington, Zeezoo Ryu, Richard Asiamah, Emad Abukhousa, Betelihem Ashebo, Animesh Chhotaray, Daniel K. Molzahn, Frank Li, Saman Zonouz, and Raheem Beyah
**Venue:** ACM SIGSAC Conference on Computer and Communications Security (CCS) (to appear)
**arXiv:** https://arxiv.org/abs/2609.07783
Internet-scale measurement of exposed solar inverters and monitoring platforms, and a power system analysis of the risk compromised units pose to the Oahu grid.

### Admittance Matrix Concentration Inequalities for Understanding Uncertain Power Networks
**Authors:** Samuel Talkington, Cameron Khanpour, Rahul K. Gupta, Sergio A. Dorado-Rojas, Daniel Turizo, Hyeongon Park, Dmitrii M. Ostrovskii, and Daniel K. Molzahn
**Venue:** 1st PowerUp Conference
**arXiv:** https://arxiv.org/abs/2510.17798
Introduces concentration inequalities for admittance matrices to quantify uncertainty impacts on linearized power flow models.

### Zero-Sum Power Factor Games
**Authors:** Cameron Khanpour, Samuel Talkington, Mathieu Dahan, and Daniel K. Molzahn
**arXiv:** https://arxiv.org/abs/2608.20298
Solves the operator's robust choice of DER power factor settings as a minimax game, with a closed-form strategy that cancels each injection's voltage contribution.

### Interactive Optimal Power Flow Compiled to the Browser
**Authors:** Samuel Talkington
Introduces tellegen, a frontend framework that compiles exact power flow and optimal power flow solvers to WebAssembly for interactive, in-browser grid analysis.

### Proving the Limits of Quantum Power Flow
**Authors:** Cameron Khanpour and Samuel Talkington
**arXiv:** https://arxiv.org/abs/2607.19263
Proves that realistic grid structure rules out end-to-end quantum advantage for power flow, with all proofs formally verified in Lean 4.

### Error Bounds for Radial Network Topology Learning from Quantized Measurements
**Authors:** Samuel Talkington, Aditya Rangarajan, Pedro A. de Alcântara, Line Roald, Daniel K. Molzahn, and Daniel R. Fuhrmann
**Venue:** IEEE Transactions on Power Systems
**arXiv:** https://arxiv.org/abs/2508.05620
Analyzes topology learning under quantized sensor data, bounding radial grid reconstruction error versus measurement precision.

### Efficient Network Reconfiguration by Randomized Switching
**Authors:** Samuel Talkington, Dmitrii M. Ostrovskii, and Daniel K. Molzahn
**arXiv:** https://arxiv.org/abs/2510.24458
Studies randomized switching policies that improve network reconfiguration performance and reliability.

### Covert Distribution Load Tripping Attacks
**Authors:** Betelihem Kebede Ashebo, Samuel Talkington, Saman Zonouz, and Daniel K. Molzahn
**Venue:** 57th North American Power Symposium (NAPS)
Studies data-driven load tripping attacks that exploit direct load control to force reverse flows and stress distribution transformers.

### VArsity: Can Large Language Models Keep Power Engineering Students in Phase?
**Authors:** Samuel Talkington and Daniel K. Molzahn
**Venue:** 57th North American Power Symposium (NAPS)
**arXiv:** https://arxiv.org/abs/2507.20995
Reports on deploying ChatGPT-based assignments in power systems courses and how students respond to evolving model accuracy.

### Differentiating Through Power Flow Solutions for Admittance and Topology Control
**Authors:** Samuel Talkington, Daniel Turizo, Sergio A. Dorado-Rojas, Rahul K. Gupta, and Daniel K. Molzahn
**arXiv:** https://arxiv.org/abs/2510.17071
Shows how to differentiate power flow solutions with respect to admittance parameters for topology and impedance control tasks.

### Optimizing State Estimation Error with the LinDist3Flow Model
**Authors:** Jeslyn Ero, Kieran Slattery, Xianhe Qin, Samuel Talkington, and Daniel K. Molzahn
**Venue:** IEEE Opportunity Research Scholars Symposium (ORSS)
Reports undergraduate research on minimizing LinDist3Flow state estimation error using data-driven tuning strategies.

### Classifying Reactive Power Control Laws of Behind-the-Meter Solar Photovoltaic Inverters
**Authors:** Richard Asiamah, Samuel Talkington, Michael Boateng, Marta Vanin, Frederik Geth, and Daniel K. Molzahn
**Venue:** 6th IEEE Kansas Power and Energy Conference (KPEC)
Presents classifiers that infer behind-the-meter inverter reactive power control laws using only aggregated smart meter data.

### Strategic Electric Distribution Network Sensing via Spectral Bandits
**Authors:** Samuel Talkington, Rahul Gupta, Richard Asiamah, Paprapee Buason, and Daniel K. Molzahn
**Venue:** 63rd IEEE Conference on Decision and Control (CDC)
**arXiv:** https://arxiv.org/abs/2410.21270
Uses spectral bandit algorithms to schedule smart meter polling and improve distribution sensing under communication limits.

### A Data-Driven Sensor Placement Approach for Detecting Voltage Violations in Distribution Systems
**Authors:** Paprapee Buason, Sidhant Misra, Samuel Talkington, and Daniel K. Molzahn
**Venue:** Electric Power Systems Research
**arXiv:** https://arxiv.org/abs/2210.09414
Formulates a bilevel, data-driven sensor placement method that guarantees voltage violation detection in distribution grids.

### Locational Marginal Burden: Quantifying the Equity of Optimal Power Flow Solutions
**Authors:** Samuel Talkington∗, Amanda West∗, and Rabab Haider (∗ equal contribution)
**Venue:** 15th ACM International Conference on Future and Sustainable Energy Systems (e-Energy)
**arXiv:** https://arxiv.org/abs/2405.12219
Introduces locational marginal burden to link optimal power flow pricing with energy equity through differentiable optimization.

### Managing Vehicle Charging During Emergencies via Conservative Distribution System Modeling
**Authors:** Alejandro D. Owen Aquino, Samuel Talkington, and Daniel K. Molzahn
**Venue:** 2024 IEEE Texas Power and Energy Conference (TPEC)
**arXiv:** https://arxiv.org/abs/2311.16975
Demonstrates how conservative distribution system models coordinate emergency electric vehicle charging during extreme events.

### Localized Structure in Secondary Distribution System Voltage Sensitivity Matrices
**Authors:** Samuel Talkington, Santiago Grijalva, Matthew J. Reno, Joseph A. Azzolini, and Jouni Peppanen
**Venue:** Electric Power Systems Research
Analyzes secondary distribution voltage sensitivity matrices and shows how localized structure simplifies modeling and control.

### A Measurement-Based Approach to Voltage-Constrained Hosting Capacity Analysis with Controllable Reactive Power Behind-the-Meter
**Authors:** Samuel Talkington, Santiago Grijalva, Matthew J. Reno, Joseph A. Azzolini, David Pinney
**Venue:** Electric Power Systems Research
Describes a measurement-based hosting capacity analysis that relies on AMI data to estimate voltage limits without circuit models.

### Phase Retrieval via Model-Free Power Flow Jacobian Recovery
**Authors:** Samuel Talkington, Santiago Grijalva
**Venue:** 14th ACM International Conference on Future Energy Systems (e-Energy)
**arXiv:** https://arxiv.org/abs/2305.09661
Extends phase retrieval to power systems by recovering voltage angles and Jacobian structure using only magnitude measurements.

### Conditions for Estimation of Sensitivities of Voltage Magnitudes to Complex Power Injections
**Authors:** Samuel Talkington, Daniel Turizo, Santiago Grijalva, Jorge Fernandez, Daniel K. Molzahn
**Venue:** IEEE Transactions on Power Systems
**arXiv:** https://arxiv.org/abs/2212.01471
Derives uniqueness conditions for estimating voltage magnitude sensitivities to active and reactive power using limited sensor data.

### Calculating PV Hosting Capacity in Low-Voltage Secondary Networks Using Only Smart Meter Data
**Authors:** Joseph A. Azzolini, Matthew J. Reno, Jubair Yusuf, Samuel Talkington, Santiago Grijalva
**Venue:** IEEE PES Innovative Smart Grid Technologies Conference (ISGT)
Proposes a smart meter driven algorithm for PV hosting capacity in secondary networks, matching model-based accuracy at low cost.

### Improving Behind-the-Meter PV Impact Studies with Data-Driven Modeling and Analysis
**Authors:** Joseph A. Azzolini, Samuel Talkington, Matthew J. Reno, Santiago Grijalva, Logan Blakely, David Pinney
**Venue:** 49th IEEE Photovoltaics Specialists Conference (PVSC)
Data-driven modeling improves behind-the-meter PV hosting capacity studies; conference paper coauthored by Samuel Talkington.

### Sparse Time Series Sampling for Recovery of Behind-the-Meter Inverter Control Models
**Authors:** Samuel Talkington, Santiago Grijalva, and Matthew J. Reno
**Venue:** IEEE PES Innovative Smart Grid Technologies Conference (ISGT)
Presents a sparse sampling method to recover control settings of behind-the-meter inverter-based resources using limited net load data.

### Solar PV Inverter Reactive Power Disaggregation and Control Setting Estimation
**Authors:** Samuel Talkington, Santiago Grijalva, Matthew J. Reno, Joseph A. Azzolini
**Venue:** IEEE Transactions on Power Systems
Reconstructs solar PV inverter reactive power settings from aggregated AMI data using curve fitting and probabilistic disaggregation.

### Recovering Power Factor Control Settings of Solar PV Inverters from Net Load Data
**Authors:** Samuel Talkington, Santiago Grijalva, Matthew J. Reno, Joseph A. Azzolini
**Venue:** North American Power Symposium (NAPS)
Estimates fixed power factor settings of behind-the-meter PV inverters from net load AMI data using filtering and sensitivity models.

### Power Factor Estimation of Distributed Energy Resources Using Voltage Magnitude Measurements
**Authors:** Samuel Talkington, Santiago Grijalva, and Matthew J. Reno
**Venue:** Journal of Modern Power Systems and Clean Energy
Develops a data-driven method to estimate DER power factor settings from voltage magnitude measurements in distribution feeders.

### Rail Transit Regenerative Braking Energy Recovery Optimization to Provide Grid Services
**Authors:** Samuel Talkington, Santiago Grijalva
**Venue:** IEEE Power and Energy Conference at Illinois (PECI)
Optimizes regenerative braking energy recovery so rail transit systems can provide ancillary grid services.

## Blog Posts

- **[PowerIO: A Fast Parser and Converter for Power System Case Files](https://samueltalkington.com/blog/2026/powerio/)**: PowerIO: a fast Rust library that parses power system case files into one typed model and converts losslessly between formats, with Python, Julia, and C bindings.
- **[PowerUp Conference](https://samueltalkington.com/blog/2026/power-up/)**: Our new work on admittance matrix concentration was accepted at the 1st Annual PowerUp Conference.
- **[Marguerite.jl: A Differentiable Frank–Wolfe Solver](https://samueltalkington.com/blog/2026/marguerite_pkg/)**: Announcing Marguerite.jl, a minimal and differentiable Frank–Wolfe solver for constrained convex optimization in Julia.
- **[Seminar Series: Randomness as a Resource for Electric Power Systems](https://samueltalkington.com/blog/2026/faculty_seminars/)**: Seminar series on randomness as a resource for electric power systems, presented at Carnegie Mellon University, University of Michigan, and Johns Hopkins University.
- **[Interactive Visualization: Randomized Switching for Network Reconfiguration](https://samueltalkington.com/blog/2026/rand_switch_viz/)**: Demo of the randomized switching algorithm for network reconfiguration: Generate many optimized configurations for a power network.
- **[Linearizing the Power Flow Equations](https://samueltalkington.com/blog/2025/linear_acpf/)**: Tutorial on linearizing the AC power flow equations, covering DC power flow, sensitivity-based methods, and first-order Taylor approximations.
- **[Recent events: 2025 INFORMS, 2025 NAPS, and 2026 ICERM Workshop](https://samueltalkington.com/blog/2025/upcoming_talks/)**: Recap of recent conference presentations and workshops in 2025–2026.
- **[Randomness as a Resource: Scalable Algorithms for Grid Decision-Making](https://samueltalkington.com/blog/2025/pai_seminar/)**: Recap of Harvard PAI seminar on using randomness to scale decision-making algorithms for electric power grids.
- **[Generating Random Graphs and Laplacians in Julia](https://samueltalkington.com/blog/2025/julia_laplacians/)**: Step-by-step guide for building random graphs and Laplacians in Julia, with sparse matrix tricks and visualization tips.
- **[Differentiating Solutions of DC Optimal Power Flow](https://samueltalkington.com/blog/2025/diff_dcopf/)**: Explains how to differentiate DC optimal power flow solutions with respect to problem parameters.
- **[Writing DC Optimal Power Flow in Matrix Form](https://samueltalkington.com/blog/2024/matrix_dcopf/)**: Shows how to write DC optimal power flow in matrix form.
- **[Phase retrieval in power systems](https://samueltalkington.com/blog/2023/power_phase_retrieval/)**: Discusses phase retrieval in power systems and conditions for estimating complex injections using only voltage magnitudes.
- **[Checking if an admittance matrix represents a radial power system](https://samueltalkington.com/blog/2022/check_radial/)**: Walkthrough for testing whether an admittance matrix corresponds to a radial electric power network using simple counts.