SearcharxivSearch

arXiv subjects

Ethan Tam

Publications and source records attributed to Ethan Tam.

2 recordsLinked to original sources

Hydrodynamically engineered Indigenous arrows skip on water for waterfowl hunting

Across the Northern Hemisphere, Indigenous hunters developed arrows capable of skipping across the water surface to strike waterfowl. Archaeological and ethnographic records reveal remarkably similar projectile designs spanning millennia and geographically distant cultures, suggesting a convergent technological solution. Despite extensive study of water-entry dynamics, the physical principles underlying this behaviour remain poorly understood. Here we show that successful water-skipping arises from a small set of coupled geometric and dynamical parameters that define a bounded operational regime separating rebound, plunging, and overshoot. Using a combination of controlled experiments, hydrodynamic modeling, and historical reconstruction, we demonstrate that reconstructed arrow designs from independent cultures consistently fall within this predicted regime. These results demonstrate that Indigenous technologies were effectively tuned to satisfy the hydrodynamic constraints governing controlled skipping, providing evidence of convergent optimization in human-engineered systems. More broadly, our results suggest that material culture encodes physical knowledge that formal science is only beginning to articulate, and that the archaeological record and Indigenous culture may be an underexplored archive of empirical discovery.

physics.flu-dyn

Automated Neuron Labelling Enables Generative Steering and Interpretability in Protein Language Models

Protein language models (PLMs) encode rich biological information, yet their internal neuron representations are poorly understood. We introduce the first automated framework for labeling every neuron in a PLM with biologically grounded natural language descriptions. Unlike prior approaches relying on sparse autoencoders or manual annotation, our method scales to hundreds of thousands of neurons, revealing individual neurons are selectively sensitive to diverse biochemical and structural properties. We then develop a novel neuron activation-guided steering method to generate proteins with desired traits, enabling convergence to target biochemical properties like molecular weight and instability index as well as secondary and tertiary structural motifs, including alpha helices and canonical Zinc Fingers. We finally show that analysis of labeled neurons in different model sizes reveals PLM scaling laws and a structured neuron space distribution.

cs.LG