Reproduce “The Thermodynamic Costs of Simple Linear Regression” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Samuel H. D'Ambrosia, Sultan M. Daniels, Michael R. DeWeese, Anant Sahai
The authors apply Landauer’s principle to estimate the minimum energy cost of fitting a line from noisy data stored in floating-point registers. They compare input-state entropy with output-state entropy for exact regression and stochastic gradient descent, under idealized assumptions. In single-precision exact regression, precision contributes about 90% of calculated cost at high sample counts; these are theoretical lower bounds, not measurements.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.