$ \alpha $ contours on a magnetic surface

Figure 19 compares a small number of $ \phi $ contours and $ \alpha $ contours on a magnetic surface.

Figure 19: Comparison between a series of $ \phi $ contours (left) and a series of $ \alpha $ contours (right) on a magnetic surface. Here the values of $ \phi $ are $ \phi _j = \alpha _j = (j - 1) 2 \pi / 4 / (10 - 1)$ with $ j = 1, 2, \ldots , 10$, i.e., only $ 1 / 4$ of the full torus. The values of $ \alpha $ contours are $ \alpha _j = (j - 1) 2 \pi / 4 / (10 - 1)$ with $ j = 1, 2, \ldots , 10$, i.e., only $ 1 / 4$ of the full range $ 2 \pi $. Every $ \phi $ and $ \alpha $ contours start from the lower-field-side midplane and go one full poloidal loop. Magnetic field from EAST discharge #59954@3.03s.
\resizebox{8cm}{!}{\includegraphics{/home/yj/project_new/fig_lorentz/fig3e/p.eps}} \resizebox{8cm}{!}{\includegraphics{/home/yj/project_new/fig_lorentz/fig3d/p.eps}}

As is shown in the left panel of Fig. 19, with $ \phi $ fixed, an $ \theta $ curve reaches its starting point when $ \theta $ changes from zero to $ 2 \pi $. However, as shown in the right panel of Fig. 19, with $ \alpha $ fixed, an $ \theta $ curve (i.e. a magnetic field line) does not reach its starting point when $ \theta $ changes from zero to $ 2 \pi $. There is a toroidal shift, $ 2 \pi q$, between the starting point and ending point. Therefore there is generally no periodic condition along $ \theta $ since $ q$ is not always an integer. A mixed periodic condition involves both $ \theta $ and $ \alpha $ is given in (290).

In field-line-following coordinates $ (\psi , \theta , \alpha )$, a toroidal harmonic of a physical perturbation can be written as

$\displaystyle \delta A (\psi, \theta, \alpha) = \delta A_0 (\psi) \cos (m' \theta + n \alpha + \alpha_0)$ (292)

where $ n$ is the toroidal mode number, $ m'$, which may not be an integer, is introduced to describle the variation along a field line. The periodic condition given by Eq. (290) requires that

$\displaystyle \cos (m' \theta + n \alpha + \alpha_0) = \cos [m' (\theta + 2 \pi) + n (\alpha - 2 \pi q) + \alpha_0],$ (293)

To satisfy the above condition, we can choose

$\displaystyle m' 2 \pi - n 2 \pi q = N 2 \pi,$ (294)

where $ N$ is an arbitrary integer, i.e.,

$\displaystyle m' = N + n q.$ (295)

We are interested in perturbation with a slow variation along the field line direction (i.e., along $ \partial \mathbf{r}/ \partial \theta$) and thus we want the value of $ m'$ to be small. One of the possible small values given by expression (295) is

$\displaystyle m' = n q - \ensuremath{\operatorname{NINT}} \left( n \frac{q_{\max} + q_{\min}}{2} \right),$ (296)

where N $ \ensuremath{\operatorname{INT}}$ is a function that return the nearest integer of its argument, $ q_{\max}$ and $ q_{\min}$ is the maximal and minimal value of the safety factor in the radial region in which we are interested. Note that $ m'$ depends on the radial coordinate $ \psi $ through $ q (\psi)$. Also note that $ m'$ here is differnt from the poloidal mode number $ m$ in $ (\psi , \theta , \phi )$ coordinate system. It is ready to show that the perturbation given by Eq. (292) with $ m' \sim 1$ and $ n \gg 1$ has large poloidal mode number $ m$ when expressed in $ (\psi , \theta , \phi )$ coordinates. [Proof: Expression (292) can be written as

$\displaystyle \delta A = \delta A_0 (\psi) \cos [m' \theta + n (\phi - \overline{\delta} (\psi, \theta)) + \alpha_0]$ (297)

If $ \theta $ is the staight-field-line poloidal angle in $ (\psi , \theta , \phi )$ coordinate system, then $ \overline{\delta} (\psi, \theta) = q \theta$ and the above eqaution is written as

$\displaystyle \delta A = \delta A_0 (\psi) \cos [(m' - n q) \theta + n \phi + \alpha_0],
$

which indicates the poloidal mode number $ m$ in $ (\psi , \theta , \phi )$ coordinates is given by $ m = m' - n q$. For the case with $ m' \sim 1$ and $ n \gg 1$, $ m$ is much larger than one.]

[In the past, I choose $ m' (\psi) = n q - \ensuremath{\operatorname{NINT}} (n q)$. However, $ m'
(\psi)$ in this case is not a continuous function of $ \psi $ and thus is not physical.]

[In passing, let us introduce the binormal wavenumber, which is frequently used in presenting turbulence simulation results. Consider the toroidal phase in expression (292), i.e.,

$\displaystyle \ensuremath{\operatorname{phase}} = n \alpha$ (298)

Define the binormal direction $ \mathbf{s}$ by

$\displaystyle \mathbf{s}= \frac{\mathbf{B} \times \nabla \Psi}{\vert \mathbf{B} \times \nabla
\Psi \vert}, $

which a unit vector lying on a magnetic surface and perpendicular to $ \mathbf {B}$. The binormal wavenumber is defined by

$\displaystyle k_{b n} =\mathbf{s} \cdot \nabla \ensuremath{\operatorname{phase}},$ (299)

which can be written as
$\displaystyle k_{b n}$ $\displaystyle =$ $\displaystyle \frac{\mathbf{B} \times \nabla \Psi}{\vert \mathbf{B} \times
\nabla \Psi \vert} \cdot \nabla (n \alpha)$  
  $\displaystyle =$ $\displaystyle n \frac{\nabla \Psi \times \nabla \alpha}{\vert \mathbf{B} \times \nabla
\Psi \vert} \cdot \mathbf{B}$  
  $\displaystyle =$ $\displaystyle n \frac{B^2}{\vert \mathbf{B} \times \nabla \Psi \vert}$  
  $\displaystyle =$ $\displaystyle n \frac{B}{\vert \nabla \Psi \vert},$ (300)

Using $ B_p = \vert \nabla \Psi \vert / R$, the above equation is written

$\displaystyle k_{b n} = n \frac{B}{R B_p}$ (301)

which indicates the binormal wavenumber generally depends on the poloidal angle. For large aspect-ratio tokamak, we have $ B_{\phi} \approx B$, $ q
\approx B_{\phi} r / (B_p R)$. Then Eq. (301) is written

$\displaystyle k_{b n} \approx \frac{n q}{r},$ (302)

which indicates the binormal wavenumber are approximately indepenent of the poloidal angle. For modes with field aligned structure (i.e., $ k_{\parallel} \ll k_{\perp}$), we have $ m \approx n q$, where $ m$ is the poloidal mode numer along $ \theta $ with $ \phi $ hold fixed. In this case, the above equation is written $ k_{b n} \approx m / r$, which is the usual poloidal wave number. Due to this relation, the binormal wavenumber $ k_{b n}$ is often denoted by $ k_{\theta}$ in papers on tokamak turbulence. In some papers the binormal wavenumber is denoted by $ k_y$.]

Since $ \alpha $ contours are magnetic field lines, they span out the 3D shape of the magnetic surface when there are enough numbe of $ \alpha $ contours on a magnetic surface, as is shown by the left-panel of Fig. 20.

Figure 20: Comparison between a series of $ \phi $ contours(left) and a series of $ \alpha $ contours (left) on a magnetic surface. The $ \alpha $ contours correspond to magnetic field lines. Here the $ \alpha $ values of adjacent $ \alpha $ contours differ by $ d \alpha = 2 \pi / 20$ and each $ \alpha $ contour goes one full poloidal loop. Magnetic field from EAST discharge #59954@3.03s.
\resizebox{8cm}{!}{\includegraphics{/home/yj/project_new/fig_lorentz/fig3c/p.eps}} \resizebox{8cm}{!}{\includegraphics{/home/yj/project_new/fig_lorentz/fig3b/p.eps}}

yj 2018-03-09