Hi, everybody. I recently do some reserch on MLPG using in 3D problems, and I have modified the program in G.R. Liu's book.
But what makes me crazy is that I must always run the program many times to idenetifiy the proper quardrature sub-domain size to get the right result.
So I confused here, then I did a comparision of the various sub-domain size to examinate the effect of them in G.R. Liu's MLPG2D program. This time, sub-domain size seems have liittle effect on the result. I tried 0.1*nodalspace situation, and results has a about 20% error.
So in here, I want to ask you the experts in MLPG and Meshless to give me some tips. Is this existed in the root of MLPG, or maybe just some bugs in my code? Give me some directions, Thank U!