Page 11 - Theory of ion dynamics and heating by magnetic pumping in FRC plasma
P. 11
072510-11 Egedal et al.
Phys. Plasmas 25, 072510 (2018)
well characterized by nc 1⁄4 0:4. Furthermore, D/B ðv; p/Þ is obtained by averaging hðD/BÞ2i over n for n > nc.
Using the substitution x 1⁄4 2n2 1, the scattering opera- tor in Eq. (16) can then be written as
8< 0 ; 0 n n c sbLmix1⁄4:D @ð1x2Þ@; n n1: (18)
We recognize Eq. (18) as the Legendre operator for which the equation LmixPl þ kPl 1⁄4 0 is solved by the Legendre functions, Pl. For the typical case where 1 x 1, the index l is an integer. However, in our case, the operator only applies in the interval 2n2c 1 x 1, and the appropriate non-integer order lj of the Legendre functions Plj must be determined by a boundary condition at xc 1⁄4 n2c 1.
To obtain the boundary condition for Pl, we note that in the following we will solve the full pumping equation in Eq. (12) through the use of the expansion:
/B@x @xc
fðv;p/;nÞ1⁄4Xfjðv;p/ÞPljð2n2 1Þ: j
(19)
@ @ @n@ @v 1⁄4@vþ@v@n;
Using that sb Lmix Plj 1⁄4 4D/B lj ðlj þ 1ÞPlj , we
now
eigenfunctions
X 1@Jz
FIG.
1Þ 1⁄4 0. The eigenvalues are kj 1⁄4 ljðlj þ 1Þ, where the non-integer values of lj are determined from Eq. (20), here for nc 1⁄4 0:4.
where
13. Eigenfunctions
of the
equation
jk jk
Lmix Plj ð2n2 1Þ þ kj Plj ð2n2
Because n is conserved for n < nc, there can be no diffusion across the x 1⁄4 xc boundary. The appropriate boundary condi- tion for the basis functions is then
Jq n Jq v 1⁄4 @ þ 1 n2 dJ0 @ :
@ @x Plj ðxÞ
@vn 2nJ0 dv @nv After multiplying Eq. (12) by sb, we find
1⁄4 0:
(20)
XX2! @fj 1@Jz @fj 1n dJ0 @Plj
x1⁄4xc
sbPlj@tv@t Plj@vþ2nJ0 dvfj@n jj
This equation is readily solved numerically for lj. For exam- ple, Table I provides the first eigenvalues and normalization factors for nc 1⁄4 0:4, such that xc 1⁄4 0:68. The correspond- ing Plj ð2n2 1Þ solutions are shown in Fig. 13. These eigen- functions form a complete set, allowing any distribution function f ðv; p/ ; nÞ to be expanded in the form given by Eq. (19). Furthermore, the functions are orthogonal on the inter- val defined through
(21)
(22)
can simplify the scattering term in the kinetic equation in Eq. (12). However, as a subtlety, given we are now using v and n as the independent variables, the @=@vjJq term in Eq. (12)
1⁄4 4D/B ljðlj þ 1Þ X Plj fj; (23) j
ð1 nc
where again Plj 1⁄4 Plj ð2n2 1Þ.
We next expand the factors in front of fj in terms of
the basis functions. Because p/ is fully conserved, we can do this separately for each p/. The kinetic equation then reads
nPlj ð2n2 1Þ Plk ð2n2 1Þ dn 1⁄4 ajdj;k; where aj is the normalization factor
ð1
aj 1⁄4 nPljð2n2 1ÞPljð2n2 1Þdn:
nc
XXT P @fjþXXA P @fj k;j lk @t k;j lk @v
þXXBk;jPlkfj 1⁄4 XXCk;jPlkfj; (24) jk jk
XTk;jPlk 1⁄4sbPlj; k
Ak;jPlk 1⁄4v@tPlj;
XBk;jPlk 1⁄41@Jz1n2dJ0@Plj ;
Xk v@t2nJ0dv@n
C P 1⁄44D lðl þ1ÞP : (25)
k
Given the orthogonality of the Plj set expressed in Eq. (21), the functions Tk;j; Ak;j, Bk;j, and Ck;j are readily isolated
must be evaluated appropriately
TABLE I. Eigenvalues and normalization factors of the Plj for nc 1⁄4 0:4.
k
j012345
lj 0 1.27 2.59 3.94 5.28 6.63 aj 0.42 0.11 0.061 0.042 0.032 0.026
k;j lk /B j j lj