Final report and code for the project

Here is my final report , video and report for the project.

report: https://www.dropbox.com/s/q9srm58xlbp2qyt/GhostSPH%20Report.pdf

code: https://www.dropbox.com/s/3fffh78clw461pi/Ghost%20SPH.zip

Video: https://www.dropbox.com/s/ey1f4zbu8ls54ap/Ghost%20SPH.mp4

Problem related to the mass of each particle

When I implement the Ghost SPH, I found the mass of the particles influence the result a lot. When I dealing with 10K particles, the difference of 0.0001 in mass make the result totally different, I still can not find the perfect variables, continue working on it!

Small size Ghost SPH simulation

Due to the limitation of my mac air. I start the simulation with small size , and get following results.

Paper mistakes found

I follow the procedure  indicated by the paper to simulate Ghost SPH. For days, I can not get the satisfying results. After recheck other reference, I found two mistakes of the paper.

1. XSPH:
The equation used by the user is to calculate Viscosity velocity:

 however, it should be the term used to calculate viscosity force, not velocity.
In other word, in order to calculate the viscosity velocity, I need to divide this term by density and times dt

2. Missing of tension force
In the algorithm 3, there is no tension force term, however, it should be an important term to simulate the "real" fluid simulations, I add this force with gravity and pressure force. However, I am not sure if it follows the authors' idea.

Sampling and Relaxation finish

Finish the part to sampling and relaxation the surface and volume.

Before:

After:

neighbor search problem

When I implementing the Volume relaxation, I need to solve neighbor search problem, I use Grid search algorithm to accomplish the job. 

The neighbour search finds the neighbours of all particles in a particular grid cell.  All neighbouring particles must be contained in the current or one of the maximal 26 adjacent cells to speed up the search.

Basic SPH simulation

Before, implementing the GHOST SPH.
I decide to simulate the basic SPH simulation first. Following video is the basic SPH simulation with 8K particles.