Measurement, representation, and mechanism
Making complex systems tractable, interpretable, and predictive.
I develop measurement strategies, compact representations, and simulation methods that connect complex biological and physical data to mechanism, prediction, and efficient computation. I lead the Biophysical and Biomedical Measurement Group at the National Institute of Standards and Technology (NIST).
Research program
Three connected research pillars
Biological Measurement & Data Quality
Quantitative cytometry, single-molecule sensing, fluorescence/FRET, calibration, uncertainty, and the propagation of measurement quality into biological inference.
Biomolecular Representations & Simulation
Compressed, physically meaningful descriptions of biomolecular ensembles for more efficient simulation, model inference, comparison with experiment, and prediction.
Algorithms for Many-Body Physical Systems
Computational strategies for classical and quantum many-body systems: finite-size scaling, physically informed representations, open-system methods, and tensor networks.
A measurement-to-model approach
The quality of a prediction begins with the quality of the measurement.
My work starts from how data are generated: the instrument response, sample preparation, calibration, standards, noise, bias, and the physical meaning of a measured signal. From there, I develop representations and mechanistic models that preserve what matters for inference while discarding irrelevant complexity.
This creates a continuous path from measurement to representation to model to prediction—a path that is useful for both scientific discovery and trustworthy AI.
Current work
Where the program is moving now
Single-cell measurement and inference
Current quantitative-cytometry work examines how measurement quality and analytical choices alter inferred cell-state distributions and how the remaining uncertainty should follow the inference.
Interpretable molecular states
Secondary-structure distances and clustering compress atomistic DNA and RNA ensembles into physically meaningful states. Ongoing work extends this representation-first approach toward simulation efficiency, model inference, and prediction.
Entanglement at the Fermi edge
New work identifies the natural entanglement structure that makes tensor-network simulations of transport efficient and explains the logarithmic complexity governing the relevant representation.
The unifying idea
Better representations, better predictions.
Across biological measurements, biomolecular ensembles, molecular simulation, and quantum transport, the central challenge is the same: identify the variables and structures that retain the information needed for mechanism and prediction without carrying the full complexity of the underlying system.