absorb.md

Tanishq Mathew Abraham, Ph.D.

Chronological feed of everything captured from Tanishq Mathew Abraham, Ph.D..

Fair Division of Goods with Independence Structures and Epistemic Envy-Freeness

This paper studies the fair division of indivisible goods under an independence structure, motivated by network connections. It establishes the existence of epistemic EF1 partitions for agents with arbitrary additive valuations, and provides conditions for envy-freeness up to one good when valuations are uniform. The work connects to matroid theory and offers constructive algorithms for fair allocation.

Convex Analysis at Infinity: A Generalized Differentiation Framework

This paper introduces a generalized differentiation theory at infinity specifically for convex sets and functions. It extends fundamental convex analysis concepts—tangent cones, normal cones, and subdifferentials—into the context of infinity, maintaining convexity and linking with recession analysis and epigraphical geometry. The resulting framework offers simpler, more broadly applicable results than general nonsmooth analysis, providing calculus rules and geometric characterizations for optimization problems with unbounded feasible sets.

Precision-Weighted Bias: A Better Metric for Including Adaptive Designs in Meta-Analyses

This paper introduces precision-weighted bias, a new metric for statistical performance that accounts for an estimator's bias and its precision. It argues that the aggregate bias in a meta-analysis is better represented by a precision-weighted average of individual study biases, rather than a simple average. Simulations indicate that adaptive designs, while potentially biased in unweighted terms, often exhibit negligible precision-weighted bias, suggesting their inclusion in meta-analyses may not significantly inflate overall bias and that precision-weighted bias is a more appropriate measure for such decisions.

Quantifying Order Unity Couplings and Their Ratios in Quantum Field Theories

This paper introduces a measure, the 'spread,' to quantify the adherence of quantum field theories to the expectation that dimensionless couplings are order unity. It demonstrates that even with assumptions of independent, identically distributed couplings, the ratios of these couplings can exhibit significant deviations. The analysis reveals that the distribution of such ratios develops fat power-law tails, the extent of which increases with the number of independent couplings, even under conditions of exponentially suppressed tails for individual couplings.