Accelerating and Parallelizing Lagrangian Simulations of Mixing‐Limited Reactive Transport

Simulations of transport in one dimension

Advection-dispersion equation is solved in numerically by using combinations of differential quadrature method (DQM) and various time integration techniques covering some explicit or implicit single and multi step methods. Two different initial boundary value problems modeling conservative and nonconservative transports of some substance represented by initial data are chosen as test problems. ...

Illumination-driven Mesh Reduction for Accelerating Light Transport Simulations

Progressive light transport simulations aspire a physically-based, consistent rendering to obtain visually appealing illumination effects, depth and realism. Thereby, the handling of large scenes is a difficult problem, as in typical scene subdivision approaches the parallel processing requires frequent synchronization due to the bouncing of light throughout the scene. In practice, however, onl...

A particle number conserving Lagrangian method for mixingdriven reactive transport

The purely Lagrangian algorithm for chemical reactions introduced by Benson and Meerschaert (2008) suffers from a low-concentration resolution problem. We alleviate the problem by redefining the probabilistic collision/reaction (birth/death) stochastic process as a mass-reduction operation. Theoretically, this corresponds to replacing an on/off particle with a large number of ‘‘subparticles’’ a...

Parallelizing OMNeT++ simulations using Xgrid

Accelerating Multiplication and Parallelizing Operations in Non-Volatile Memory

Today’s IoT applications often analyze raw data by running machine learning algorithms such as classification or neural networks in data centers [2], [3]. Sending the entire data to cloud for processing is not scalable and cannot guarantee the required realtime response [4]. Running data intensive workloads with large datasets on traditional cores results in high energy consumption and slow pro...