arXiv Science⌕ Search

arXiv subjects

Ankit Agarwal

Publications and source records attributed to Ankit Agarwal.

5 recordsLinked to original sources

Taking Band-Limited Maximum Likelihood to High Dimensions: Linear Memory, the Parametric Rate, and a Self-Tuning Cut-off

Band-limited maximum likelihood (BLML) estimates a density as the square of a band-limited function. The estimate is non-negative, and when the square root of the true density is band-limited its error falls at the parametric rate $n^{-1}$. Its algorithm bins the data, which is fast only in one or two dimensions: from three on every point has its own bin, the kernel system is dense, and memory grows as $n^2$, 40 GB at $10^5$ points. Whether the rate survives in higher dimension, where kernel estimators slow down, had not been tested. We study dense data, densities that fill their space, not data near a low-dimensional manifold. We show that the binned equations are the stationarity condition of a strictly convex objective and minimise it by block coordinate descent without forming the kernel matrix: memory is linear in $n$, a sweep costs $O(n^2 d)$ and a fit takes about ten, and one A100 GPU fits $10^6$ points in eight dimensions in about ten minutes. On band-limited truths the error then decays as $n^{-1}$ in every dimension we tested, up to eight, while the slopes of kernel estimators flatten from -0.67 to -0.28, and it is 2 to 18 times lower than a neural spline flow's. A streaming fit learns the cut-off, BLML's only tuning parameter, from analytic gradients of the held-out likelihood, for about the cost of one more fit. On real data BLML is the strongest nonparametric estimator we tested: it leads kernel and nearest-neighbour estimators on electricity prices by 0.27 to 0.47 nats per hour, and a tuned kernel estimator on the full POWER and GAS benchmarks by 0.39 and 1.58 nats per point, placing it between early and recent normalizing flows. It loses on densities with a kink.

stat.ML↗

Physics for the environment and sustainable development

A reliable understanding of the Earth system is essential for the life quality of modern society. Natural hazards are the cause of most life and resource losses. The ability to define the conditions for a sustainable development of humankind, to keep the Earth system within the boundaries of habitable states, or to predict critical transitions and events in the dynamics of the Earth system are crucial to mitigate and adapt to Earth system related events and changes (e.g., volcanic eruptions, earthquakes, climate change) and to avert the disastrous consequences of natural hazards. In this chapter, we discuss key concepts from nonlinear physics and show that they enable us to treat challenging problems of Earth sciences which cannot be solved by classic methods. In particular, the concepts of multi-scaling, recurrence, synchronization, and complex networks have become crucial in the very last decades for a substantially more profound understanding of the dynamics of earthquakes, landslides, or (palaeo-)climate. They can even provide a significantly improved prediction of several high-impact extreme events. Additionally, crucial open challenges in the realm of methodological nature and applications to Earth sciences are given.

nlin.CD↗

Loading-unloading contact law for micro-crystalline cellulose particles under large deformations

A semi-empirical mechanistic contact law for micro-crystalline cellulose (Avicel PH-200) particles is proposed and characterized experimentally using force-displacement curves obtained from diametrical compression of single particles. The concepts of a shape factor and a master contact law are introduced first for elastic ellipsoidal particles, and subsequently generalized to plastic irregular particles. The proposed loading-unloading contact law is a function of three characteristic diameters (lengths of the principal axes of an approximated ellipsoid), a geometric parameter associated with the loading condition, three plastic and one elastic material properties. The force-displacement curves obtained using a micro-compression tester exhibit an apparent strain-hardening at distinctly different strain values, which is captured by the shape factor function and its geometric parameter. The three plastic material properties are log-normal distributions estimated from the loading experimental curves, while the elastic property is estimated from the unloading experimental curves. The study shows a very good agreement between predictions of the calibrated loading-unloading contact law and the experimental values.

cond-mat.mtrl-sci↗

Contact radius and curvature corrections to the nonlocal contact formulation accounting for multi-particle interactions in elastic confined granular systems

We present contact radius and curvature corrections to the nonlocal contact formulation that take into account multi-particle interactions in elastic confined granular systems. The nonlocal contact formulation removes the classical assumption of independent contacts by taking into account the interplay of deformations due to multiple contact forces acting on a single particle. The contact radius correction considers the components of these deformations that contribute to the inter-particle contact area. The curvature correction improves the description of the contacting surface profiles by including higher order terms in their Taylor series expansions. To validate the corrected formulation, we restrict attention to rubber spheres under different loading conditions, in the absence of gravitational forces, adhesion or friction. Specifically, we show that the predictions of contact force and radius are in remarkable agreement with finite-element simulations and experimental observations up to levels of deformation at which contact impingement occurs, which was not possible with the original elastic nonlocal contact formulation. Convergence of the curvature corrected formulation is observed at a four-term correction.

cond-mat.soft↗

Partial wavelet coherence analysis for understanding the standalone relationship between Indian Precipitation and Teleconnection patterns

Hydro-meteorological variables, like precipitation, streamflow are significantly influenced by various climatic factors and large-scale atmospheric circulation patterns. Efficient water resources management requires an understanding of the effects of climate indices on the accurate predictability of precipitation. This study aims at understanding the standalone teleconnection between precipitation across India and the four climate indices, namely, Niño 3.4, PDO, SOI, and IOD using partial wavelet analysis. The analysis considers the cross correlation between the climate indices while estimating the relationship with precipitation. Previous studies have overlooked the interdependence between these climate indices while analysing their effect on precipitation. The results of the study reveal that precipitation is only affected by Niño 3.4 and IOD and a non-stationary relationship exists between precipitation and these two climate indices. Further, partial wavelet analysis revealed that SOI and PDO do not significantly affect precipitation, but seems the other way because of their interdependence on Niño 3.4. It was observed that partial wavelet analysis strongly revealed the standalone relationship of climatic factors with precipitation after eliminating other potential factors. Keywords: Indian Precipitation, wavelet coherency, partial wavelet coherence, teleconnections patterns.

physics.ao-ph↗