Papers where the original authors have engaged with our plain-language explanation.
For every paper on this page, at least one of the original authors has seen our plain-language explanation and engaged with it — either confirming it reads accurately or requesting corrections that we then applied. An endorsement does not mean the authors formally approve every sentence, but it does mean the explanation has passed the eyes of the people who wrote the paper.
1336 papers reviewed by authors · 321–330 / 1336
This study presents the first quantitative bottom-up model estimating that building one kilogram of human skeletal muscle requires a total metabolizable energy intake of approximately 14,570 kJ (3,481 kcal), accounting for stored tissue energy, synthesis costs, physiological deposition efficiency, maintenance during accretion, and diet-induced thermogenesis.
This paper proposes a conceptual framework for episodic Markov decision process planning as Bayesian inference over policies, where the policy serves as the latent variable and expected return defines the posterior, offering a distinct alternative to trajectory-centric or entropy-regularized approaches while demonstrating its efficacy through variational sequential Monte Carlo methods across various environments.
The paper introduces GEMSS, a variational method that utilizes a single mixture model with built-in repulsion to efficiently discover multiple distinct, sparse, and statistically plausible solutions for high-dimensional classification and regression problems, outperforming existing baselines in both synthetic benchmarks and real-world applications.
This paper proposes that the divisor structure of the exponent introduces a detectable arithmetic bias in the distribution of Mersenne primes, leading to a refined Wagstaff heuristic where exponents with higher normalized divisor complexity are statistically more likely to yield prime Mersenne numbers.
This paper forecasts that the next-generation CHORD radio telescope will revolutionize the census of neutral hydrogen galaxies by detecting roughly ten times more sources than current catalogs, extending the survey to significantly lower-mass galaxies () and uncovering thousands of massive high-redshift giants to better understand the evolution of galactic gas reservoirs.
The paper introduces Syntriod, a robust template-based algorithm that provides reliable initial orbital parameter estimates for spectroscopic binaries across diverse and sparse observational sampling regimes, outperforming classical period-search techniques and serving as an effective pre-solver for modern orbit-fitting pipelines.
This paper investigates high-energy photon production in a quark-gluon plasma by deriving analytical expressions for both leading-logarithmic and beyond-leading-logarithmic contributions to the photon production rate using two-loop photon self-energies in thermal QCD, confirming consistency with kinetic theory calculations.
This paper introduces a new simulation method to accurately model cosmological collider signatures, presenting the first measurements of oscillating halo bias in simulations and demonstrating that its mass-dependent amplitude and phase provide a robust, unique probe of primordial high-energy physics that is distinct from observational systematics.
This paper presents a forensic reproducibility audit of a chest-radiograph vision-language model benchmark that uncovered critical discrepancies in data rendering, the tracing of preserved import and call paths to support an inference regarding prompt binding, and statistical analysis—including the clarification that while the fixed-cohort analysis produced 27/45 unadjusted and 20/45 Holm-adjusted comparisons, the 28/45 result was reproduced in a separate pairwise available-case analysis where each comparison used its own denominator—leading to the withdrawal of the original performance and automated report-label discordance claims and the proposal of machine-verifiable controls to ensure future artifact integrity.
This paper empirically demonstrates that contemporary commercial large language models possess significant and inconsistent vulnerabilities in their safety guardrails, allowing them to be easily manipulated into generating believable, customized medical notes that are visually indistinguishable from authentic documents.