Anshuman Pandey | Searcharxiv

arXiv2026

Hypersonic Shock-Wave/Boundary-Layer Interaction on a Three-Dimensional Expansion-Compression Geometry

This experimental work explores the flow field around a three-dimensional expansion-compression geometry on a slender cone at Mach 5 and 8 using high-frequency pressure sensors, high-framerate schlieren, temperature-sensitive paint, shear-stress measurements and oil-flow visualizations. The $7^\circ$ cone geometry has a hyperbolic slice acting as an expansion corner which is then followed by a $30^\circ$ finite-span compression ramp. The freestream Reynolds number was varied so that the boundary layer approaching the expansion corner was either laminar, transitional or turbulent. At laminar or early transitional conditions, the separation shock locks onto the expansion corner and the separation region encompasses most of the slice, with the separation shear layer flapping at a preferred frequency. As Reynolds number is increased, the separation shock moves downstream onto the slice, the separation bubble shrinks, and the shear layer flapping frequency increases while its amplitude drops. In all cases, large-scale low-frequency breathing motions are observed. The strong relaminarization across the expansion corner at Mach 8 prevents the shock/boundary-layer interaction from reaching truly turbulent conditions and fundamentally changes its behavior on this non-canonical geometry.

physics.flu-dyn↗

arXiv2021

Tom: Leveraging trend of the observed gradients for faster convergence

The success of deep learning can be attributed to various factors such as increase in computational power, large datasets, deep convolutional neural networks, optimizers etc. Particularly, the choice of optimizer affects the generalization, convergence rate, and training stability. Stochastic Gradient Descent (SGD) is a first order iterative optimizer that updates the gradient uniformly for all parameters. This uniform update may not be suitable across the entire training phase. A rudimentary solution for this is to employ a fine-tuned learning rate scheduler which decreases learning rate as a function of iteration. To eliminate the dependency of learning rate schedulers, adaptive gradient optimizers such as AdaGrad, AdaDelta, RMSProp, Adam employ a parameter-wise scaling term for learning rate which is a function of the gradient itself. We propose Tom (Trend over Momentum) optimizer, which is a novel variant of Adam that takes into account of the trend which is observed for the gradients in the loss landscape traversed by the neural network. In the proposed Tom optimizer, an additional smoothing equation is introduced to address the trend observed during the process of optimization. The smoothing parameter introduced for the trend requires no tuning and can be used with default values. Experimental results for classification datasets such as CIFAR-10, CIFAR-100 and CINIC-10 image datasets show that Tom outperforms Adagrad, Adadelta, RMSProp and Adam in terms of both accuracy and has a faster convergence. The source code is publicly made available at https://github.com/AnirudhMaiya/Tom

cs.LG↗