Open uguuuuuu opened 1 year ago
Is it because you guys follow an orientation such that for a point on a curve in a plane, the tangent, curve normal, and plane normal form a right-handed coordinate system?
Boundary normal direction of $x_0^D$ is same as geometry-term $G^{+}$. In eq[40], $\Delta G = G^{-} - G^{+}$
When your normal direction is right, $G^{+}$ equals 0 and $G^{}$ equals G computed normally between $x_0^D$ and $x_0^S$; and when your normal direction is left, $G^{+}$ equals G computed normally between $x_0^D$ and $x_0^S$ and $G^{-}$ equals 0.
There are two possible directions opposite to each other, and the choice affects the sign of $\Delta G$ in the multi-directional form of the boundary integral.
$x{0}^{D}$ and $x{0}^{S}$ are reversed in the code. But following the convention of the picture, I believe
e
is the normal of the orange plane,bss.edge
the edge direction,_dir
$\omega^{B}$,proj
the tangent of the discontinuity curve at $x_{0}^{D}$,_its2.n
the surface normal at $x_{0}^{D}$,n
the normal of the discontinuity curve $\Delta\mathcal{M}(\pi)$,value0
$\Delta G$. It seems that it reduces to how you choose the normal directione
of the orange plane. Why did you choose to orient the orange plane bycross(bss.edge, _dir)
but notcross(_dir, bss.edge)
? And why the tangent vectorproj
of the discontinuity curve goes in the directioncross(e, _its2.n)
but not the opposite? Finally, why the curve normaln
to point in the direction ofcross(_its2.n, proj)
but not the opposite?