SearcharxivSearch

arXiv subjects

Aryav Das

Publications and source records attributed to Aryav Das.

2 recordsLinked to original sources

Catalytic formation of H_2 on carbonaceous dust grains - implications for interstellar observations

We use kinetic Monte Carlo (KMC) simulations to study molecular hydrogen formation on carbonaceous dust grain surfaces, validated against recent laboratory measurements of H$_2$ formation on coronene films at temperatures from 10 to 250 K. The model uses a three-dimensional amorphous carbon lattice with heterogeneous physisorption ($45 \pm 5$ meV) and chemisorption ($1.75 \pm 0.25$ eV) sites, and tracks both Langmuir--Hinshelwood (LH) and Eley--Rideal (ER) formation channels within a stochastic Gillespie event-driven framework. The model reproduces the measured efficiency curve within the experimental uncertainties, including the isothermal (constant surface temperature) measurements at 100 - 250 K. The simulations correctly describe the phase boundary between the LH and ER driven processes as functions of grain temperature and the observed crossover. Under interstellar medium conditions, 10 - 250 K and n = 10 - 10$^4$ cm$^3$, the model predicts three distinct regimes for the formation efficiency $\epsilon$, the fraction of impinging H atoms released as H$_2$. At 10 K diffusion is slow and $\epsilon \approx 0.06$. Between 20 K and 80 K, LH dominates and $\epsilon \approx 0.28$. Above 150 K, an ER plateau at $\epsilon = 0.19$ is sustained by chemisorption-trapped H atoms. The LH-to-ER crossover occurs between 100 and 120 K. At 100 K we observe a 16\% density-dependent stochastic enhancement, which rate-equation models cannot capture. At T$_{dust}$ = 60 K, n = 10$^3$ cm$^3$ we find the ratio of H$_2$ formation to free-fall time $t_{{\rm H}_2}/t_{\rm ff} \approx 0.93$, so dust-catalysed H$_2$ chemistry can keep pace with gravitational collapse in high-redshift star-forming environments.

astro-ph.GA

Placing Degree Scales After LayerNorm

Graph neural networks (GNNs) are widely used to learn node-selection policies on graphs, and most stack graph attention (GAT) blocks with LayerNorm. On degree-sensitive tasks, LayerNorm tends to remove the degree signal these models need to rank nodes. Much recent work addresses this by redesigning normalizers or aggregators, which changes what these components compute but does not ask where, relative to LayerNorm, a degree scale should be applied. In this paper, we show that the answer follows from a single algebraic fact about LayerNorm. When a positive per-node scale is applied before LayerNorm, LayerNorm divides it out, and it never reaches the model's output. Applied after LayerNorm, the same scale comes through and reaches the score head as magnitude. From this placement rule we derive PostDeg, a parameter-free inverse-degree scale that we add as the single change to a fixed GAT backbone. PostDeg multiplies each node's normalized representation by an inverse function of its degree, and we compare it against controls in the same position. At every evaluation size, PostDeg improves over the LayerNorm backbone on influence maximization, network dismantling, and maximum independent set, and these controls show where the improvement comes from. The same scale before LayerNorm stays at the backbone, as the absorption identity predicts, and a constant scale after LayerNorm stays there too on all but the most heavy-tailed graphs, so the effect needs both the position after LayerNorm and a degree-dependent scale. The exact form of the scale matters much less, so we recommend PostDeg, which needs no tuning.

cs.LG