Cartesian Coordinates
The Fokker-Planck form of the focused transport equation is
\[\begin{split}\begin{aligned}
\frac{\partial F}{\partial t} = & \nabla\cdot\left[\nabla\cdot(\boldsymbol{\kappa}_\perp\nabla F)\right] - \nabla\cdot\left[(v\mu\boldsymbol{b} + \boldsymbol{V} + \boldsymbol{V}_d + \nabla\cdot\boldsymbol{\kappa}_\perp)F\right] \nonumber\\
& + \frac{\partial}{\partial\mu^2}\left(D_{\mu\mu}F\right) - \frac{\partial}{\partial\mu}\left[\left(\frac{d\mu}{dt}+\frac{\partial D_{\mu\mu}}{\partial\mu}\right)F\right] - \frac{\partial}{\partial p}\left(\frac{dp}{dt}F\right)
\end{aligned}\end{split}\]
where \(F=fp^2\). The corresponding SDEs are
\[\begin{split}\begin{aligned}
d\boldsymbol{X} & = (v\mu\boldsymbol{b} + \boldsymbol{V} + \boldsymbol{V}_d + \nabla\cdot\boldsymbol{\kappa}_\perp)dt + \sum_\sigma\boldsymbol{\alpha}_\sigma dW_\sigma(s) \\
dp & = \left(\frac{dp}{dt}\right)dt \\
d\mu & = \left(\frac{d\mu}{dt}+\frac{\partial D_{\mu\mu}}{\partial\mu}\right)dt + \sqrt{2D_{\mu\mu}}dW_\mu(t)
\end{aligned}\end{split}\]
where \(\sum_\sigma\alpha_\sigma^\mu\alpha_\sigma^\nu = 2\kappa_\perp^{\mu\nu}\). We can use the results from the 3D model with whole spatial diffusion tensor \(\boldsymbol{\kappa}\) but set \(\kappa_\parallel=0\). The matrix
\[\begin{split}P =
\begin{pmatrix}
0 & -b_xb_z\sqrt{2\kappa_\perp}/\sqrt{b_x^2+b_y^2} &
-b_y\sqrt{2\kappa_\perp}/\sqrt{b_x^2+b_y^2}\\
0 & -b_yb_z\sqrt{2\kappa_\perp}/\sqrt{b_x^2+b_y^2} &
b_x\sqrt{2\kappa_\perp}/\sqrt{b_x^2+b_y^2}\\
0 & \sqrt{b_x^2+b_y^2}\sqrt{2\kappa_\perp} & 0
\end{pmatrix}\end{split}\]
Since the first column is all zeros, we only need two Weiner processes to describe the spatial diffusion. For 2D problems,
\[\begin{split}P = \frac{\sqrt{2\kappa_\perp}}{\sqrt{b_x^2+b_y^2}}
\begin{pmatrix}
-b_xb_z &
-b_y\\
-b_yb_z &
b_x
\end{pmatrix}\end{split}\]
To calculate the drift velocity
\[\begin{aligned}
& \boldsymbol{V}_d=\frac{cpv}{qB}\left\{\frac{1-\mu^2}{2}\frac{\boldsymbol{B}\times\nabla B}{B^2}+\mu^2\frac{\boldsymbol{B}\times[(\boldsymbol{B}\cdot\nabla)\boldsymbol{B}]}{B^3}+\frac{1-\mu^2}{2}\frac{\boldsymbol{B}(\boldsymbol{B}\cdot\nabla\times\boldsymbol{B})}{B^3}\right\},
\end{aligned}\]
we need
\[\begin{split}\begin{aligned}
(\boldsymbol{B}\times\nabla B)_x & = B_y\partial_z B - B_z\partial_y B \\
(\boldsymbol{B}\times\nabla B)_y & = B_z\partial_x B - B_x\partial_z B \\
(\boldsymbol{B}\times\nabla B)_z & = B_x\partial_y B - B_y\partial_x B \\
\{\boldsymbol{B}\times[(\boldsymbol{B}\cdot\nabla)\boldsymbol{B}]\}_x & = B_y(\boldsymbol{B}\cdot\nabla)B_z - B_z(\boldsymbol{B}\cdot\nabla)B_y \\
\{\boldsymbol{B}\times[(\boldsymbol{B}\cdot\nabla)\boldsymbol{B}]\}_y & = B_z(\boldsymbol{B}\cdot\nabla)B_x - B_x(\boldsymbol{B}\cdot\nabla)B_z \\
\{\boldsymbol{B}\times[(\boldsymbol{B}\cdot\nabla)\boldsymbol{B}]\}_z & = B_x(\boldsymbol{B}\cdot\nabla)B_y - B_y(\boldsymbol{B}\cdot\nabla)B_x \\
\boldsymbol{B}\cdot\nabla & = B_x\partial_x + B_y\partial_y + B_z\partial_z \\
\boldsymbol{B}\cdot\nabla\times\boldsymbol{B} & = B_x (\partial_y B_z - \partial_z B_y) + \nonumber \\
& B_y (\partial_z B_x - \partial_x B_z) + B_z (\partial_x B_y - \partial_y B_x)
\end{aligned}\end{split}\]
For 2D problems, these can be simplified using
\[\begin{split}\begin{aligned}
(\boldsymbol{B}\times\nabla B)_x & = - B_z\partial_y B \\
(\boldsymbol{B}\times\nabla B)_y & = B_z\partial_x B \\
(\boldsymbol{B}\times\nabla B)_z & = B_x\partial_y B - B_y\partial_x B \\
\boldsymbol{B}\cdot\nabla & = B_x\partial_x + B_y\partial_y \\
\boldsymbol{B}\cdot\nabla\times\boldsymbol{B} & = B_x\partial_y B_z - B_y\partial_x B_z + B_z (\partial_x B_y - \partial_y B_x)
\end{aligned}\end{split}\]