0.3.7

• Add field vorticity methods
• Add single to multiple methods
• Improve internal approximation accuracy
• Bug fixes

0.3.6

Fix:

• Error with NaN in some vortex filament code

Add:

• Vorticity
• (Not sure if this works) Pedritezzi relaxation

0.3.5

Fix:

• Failed OpenCL Kernel compilation
• Change F3D singularity checks - now more consistent between OCL & CPU code
• Fix OpenMP compilation error on older MSVC
• Fix CPU regularisation function incorrect formulae

Edit: cvortex_Linux_x64_release.so was originally incorrect binary.

0.3.4

Extended test coverage leads to reduction in bugs.

0.3.3

Bug fixes.

New: particle remeshing

Added:
Particle remeshing on regular grids in 2D and 3D with Gamma0, Gamma1, Gamma2, Gamma3 and M3' schemes.

0.3.1

Correct 2D impl
Binaries are actually commit c6e8e2f

Binary incompatible changes

Updates API:

• 3D vortex particle was cvtx_Particle, now cvtx_P3D
• 3D vortex filament was cvtx_StraightVortexFilament, now cvtx_F3D
• The interactions have been renamed as S2S for single to single, M2S for many to single and M2M for many to many.

2D vortex particles have also been added. Methods have therefore been added to the regularisation struct.

Performance enhancements

Small performance enhancements and bug fixes.

• Device name bug fixed
• Fixed crashing on failed OpenCL code compilation
• Overall around 20% faster on AMD Vega56 than 0.2.1

Less buggy v0.2.0

• Memory bug in vortex filament methods fixed
• Lower threshold for using CPU
• Bug in viscous vortex particle strength exchange method fixed