Computer Graphics and Surface Tension Modelling in Smoothed Particle Hydrodynamics
Amal S Sebastian
Year of Study: 5th
Contact details: amalssebastian@gmail.com
Project type: BTP.
Motivation to pursue project: Videos like these:
https://www.youtube.com/watch?v=tl4mx0TtaAc
https://www.youtube.com/watch?v=elZieJNBYqk
https://www.youtube.com/watch?v=tCsuEDkxU30&t=301s
Summary of Project: Over the years, Smoothed Particle Hydrodynamics has gained sig- nificant popularity in the graphics community for its ability to produce stunning visual results. This project is an effort to explore and realize certain algorithms in Smoothed Particle Hydrodynamics, for the purposes of computer graphics. We also look into the simulating a dam break problem with graphics schemes taste of what can be achieved with the discussed algorithms. Multiphase flow is a typical phenomenon which widely exists in nature, engineering applications and produces stunning visual anima- tions. To capture a lot of interesting visual effects in graphics, it requires us to model surface tension in our system. This project is an also effort to implement and explore certain surface tension models in Smoothed Particle Hydrodynamics. This report provides the reader with a brief introduction to Smoothed Particle Hydrodynamics a meshfree Lagrangian particle method and then introduces the reader to the graphics algorithms which are widely used in SPH. We then introduce few surface tension models commonly used in SPH literature. We also look into the simulating certain benchmark test cases com- monly found in SPH literature to give better clarity about the dis- cussed models.
Link to the report: https://drive.google.com/file/d/1m0rgDXo5DLKHHfSSN4_nBS0A1Lx28X7V/view?usp=sharing
References can be found at the end of the report
Experience and Time commitment (1: very low, 5: very high):
| Frequency of meetings |  |
| Involvement of the professor | |
| How strict were the deadlines | |
| Learning outcome | |
| How likely are you to recommend working under this professor | |
Parallel Fast Multipole Methods
Harshvardhan Baldwa
Year of Study: 5th
Contact details: harshbaldwa@gmail.com
Project type: BTP.
Motivation to pursue project: Learnt about Fast Multipole Methods in AE625 - Particle Methods for Fluid Flow Simulation. I was very intrigued with the possibility to solve the N body problem in linear time.
Summary of Project: N-body simulation is encountered while solving various problems in the fields ranging from celestial mechanics to molecular dynamics. If the interactions between these N particles are computed directly, this results in a computational cost of O(N2). For large N, these simu- lations become impractical. So, to reduce this computational cost, treecodes were developed. These methods use a tree data structure, which reduces the computational cost to O(N log N ) or O(N) and hence the name. These methods can also be parallelised, thus reducing the com- putational time even more. In this project, treecodes by [Barnes and Hut, 1986], FMM by [Greengard and Rokhlin, 1987] and AFMM developed by [Carrier et al., 1988] are studied and implemented in serial as well as parallel for two-dimensional N-body problems. For 3D, a hybrid of methods developed by [Anderson, 1992] and [Makino, 1999] is implemented, and the results from these algorithms are presented. All the steps in these methods are fully parallel which makes it possible to execute on multi-core CPUs as well as on GPUs. The reduction in time is very significant, and high accuracy can be achieved by choosing the right parameters.
Link to the Report: https://drive.google.com/file/d/15M360Cyhd_mBgwcfBX8PrAU6cwKZ-4Ih/view?usp=drivesdk
Experience and Time commitment (1: very low, 5: very high):
| Frequency of meetings | |
| Involvement of the professor | |
| How strict were the deadlines | |
| Learning outcome | |
| How likely are you to recommend working under this professor | |