A more accurate form of the guiding center drift

Equation. (61) can also be written as

$\displaystyle \left( 1 + \frac{v_{\parallel}}{\Omega} \ensuremath{\boldsymbol{b...
... t} + \frac{v_{\parallel}^2}{\Omega} \nabla \times \ensuremath{\boldsymbol{b}},$ (67)

which can be arranged in the following form

$\displaystyle \left( 1 + \frac{v_{\parallel}}{\Omega} \ensuremath{\boldsymbol{b...
...{\boldsymbol{b}} \times \frac{\partial \ensuremath{\boldsymbol{b}}}{\partial t}$ (68)

As discussed above, the last term $ v_{\parallel} / \Omega \ensuremath{\boldsymbol{b}} \times
\partial \ensuremath{\boldsymbol{b}} / \partial t$ is usually ignored. Thus, the above equation is written as

$\displaystyle \left( 1 + \frac{v_{\parallel}}{\Omega} \ensuremath{\boldsymbol{b...
...mbol{b}} \times \left( - Z e \ensuremath{\boldsymbol{E}} + \mu \nabla B \right)$ (69)

Define

$\displaystyle B^{\star} = B \left( 1 + \frac{v_{\parallel}}{\Omega} \ensuremath{\boldsymbol{b}} \cdot \nabla \times \ensuremath{\boldsymbol{b}} \right),$ (70)

which is related to $ \ensuremath{\boldsymbol{B}}^{\star}$ defined in Eq. (17) through $ B^{\star} = \ensuremath{\boldsymbol{b}} \cdot \ensuremath{\boldsymbol{B}}^{\star}$, then Eq. (69) is written as

$\displaystyle \dot{\ensuremath{\boldsymbol{X}}} = \frac{v_{\parallel}}{B^{\star...
...+ \frac{1}{m \Omega B^{\star}} \ensuremath{\boldsymbol{B}} \times \mu \nabla B,$ (71)

Define

$\displaystyle \ensuremath{\boldsymbol{v}}_{\parallel}^{\star} \equiv \frac{v_{\...
... \frac{v_{\parallel}}{\Omega} \nabla \times \ensuremath{\boldsymbol{b}} \right)$ (72)

Eq. (71) is written as

$\displaystyle \dot{\ensuremath{\boldsymbol{X}}} = \ensuremath{\boldsymbol{v}}_{...
... + \frac{1}{m \Omega B^{\star}} \ensuremath{\boldsymbol{B}} \times \mu \nabla B$ (73)

which agrees with Eqs. (8)-(14) in Todo's paper[4]. Note that in this form of the guiding center drift, the curvature drift is included in $ \ensuremath{\boldsymbol{v}}_{\parallel}^{\star}$ (the second term in Eq. (72)). Compared with Eq. (66), Eq. (73) is more accurate because it does not use the approximation that $ \ensuremath{\boldsymbol{b}} \cdot \nabla \times \ensuremath{\boldsymbol{b}} \approx 0$. The numerical results form my numerical code indicate that Eq. (73) can conserve the toroidal angular momentum more accurately than Eq. (66).

YouJun Hu 2014-05-19