Asynchronous Liquids: Regional Time Stepping for Faster SPH and PCISPH

09/30/2020
by   Prashant Goswami, et al.
0

This paper presents novel and efficient strategies to spatially adapt the amount of computational effort applied based on the local dynamics of a free surface flow, for both classic weakly compressible SPH (WCSPH) and predictive-corrective incompressible SPH (PCISPH). Using a convenient and readily parallelizable block-based approach, different regions of the fluid are assigned differing time steps and solved at different rates to minimize computational cost. Our approach for WCSPH scheme extends an asynchronous SPH technique from compressible flow of astrophysical phenomena to the incompressible free surface setting, and further accelerates it by entirely decoupling the time steps of widely spaced particles. Similarly, our approach to PCISPH adjusts the the number of iterations of density correction applied to different regions, and asynchronously updates the neighborhood regions used to perform these corrections; this sharply reduces the computational cost of slowly deforming regions while preserving the standard density invariant. We demonstrate our approaches on a number of highly dynamic scenarios, demonstrating that they can typically double the speed of a simulation compared to standard methods while achieving visually consistent results.

READ FULL TEXT

page 4

page 9

page 11

page 12

page 18

page 19

page 23

research
05/04/2021

A fluid simulation system based on the MPS method

Fluid flow simulation is a highly active area with applications in a wid...
research
08/21/2022

Tensor-based flow reconstruction from optimally located sensor measurements

Reconstructing high-resolution flow fields from sparse measurements is a...
research
10/20/2021

Asynchronous parareal time discretization for partial differential equations

Asynchronous iterations are more and more investigated for both scaling ...
research
11/30/2017

Constraint Bubbles: Adding Efficient Zero-Density Bubbles to Incompressible Free Surface Flow

Liquid simulations for computer animation often avoid simulating the air...
research
02/06/2023

Rethinking Gauss-Newton for learning over-parameterized models

Compared to gradient descent, Gauss-Newton's method (GN) and variants ar...
research
11/24/2020

SegBlocks: Block-Based Dynamic Resolution Networks for Real-Time Segmentation

SegBlocks reduces the computational cost of existing neural networks, by...
research
06/09/2023

Novel Simulation-Inspired Roller Spreading Strategies for Fine and Highly Cohesive Metal Powders

When fine powders are to be used in powder bed metal additive manufactur...

Please sign up or login with your details

Forgot password? Click here to reset