jianwu21 / deep-contact Goto Github PK

Grid fill pattern for 3 points:

With values of grid datastructure (not the above fill patern):

SPH poly6 2D Kernel plot . Does this look right ?

The issue is that the sum of W coefficients is for this case : 1.1641047283126023

• remove unnecessary code from PyBox2d
• every point in a contact becomes a contact in XML
• generate world data (same radius, friction=0.5, random position/sampled from normal distribution)
• import XML file into a simulation
• set up build of pyBox2d since we are hacking on it
• add the modifications in here
• early stopping for velocityiterations
• set up http://www.erda.dk/
• makefile

• transfer to grid
• transfer from grid back to graph
• choose a kernel similar to the ones used in Smooth Particle Hydrodynamics (poly6 for example)
• extract error measure from pybox2d solver (for example using L_inf norm)