The thermal distribution function is expressed in its general relativistic form
| (2) |
where Θ = Te∕mec2 is the ratio of the local electron temperature Te to the electron rest mass energy mec2 while the relativistic Lorentz correction factor is defined as γ2 = p2 + 1. Here, ξ is the cosine of the pitch angle. By symmetry, the Maxwellian distribution function is independant of ξ. By definition, fM is normalized to unity in the interval p ∈, where pmin is a given upper cut-off limit for the Maxwellian. The parameter αM is then given by the integral
| (3) |
Recalling that γdγ = pdp,
| (4) |
where
| (5) |
By integrating by parts
and in the limit pmin → +∞,For Θ ≪ 1, using the large argument asymptotic development
| (9) |
the usual expression keeping the first in the expansion
is well recovered and
| (12) |
In the LUKE code, the momentum p here expressed in relativistic units mec is normalized to the thermal reference value pth† = mevth†,
| (13) |
and consequently
| (14) |
since thermal electrons are only weakly relativistic. The well known relativistic Lorentz correction factor γ is then simply given by the relation
| (15) |
and in the non-relativistic limit, i.e. the condition p2βth†2 ≪ 1 or γ ≈ 1 holds.
Since in relativistic units, the total energy is linked to the relativistic momentum by the expression,
| (16) |
it is straightforward to express the kinetic energy Ec as a function of p in units of electron rest mass energy mec2
| (17) |
Finally, concerning the normalization of the electron velocity v, one has
| (21) |
or
| (22) |
with
| (23) |
Consequently, the non-relativistic expression of fM is only valid when γ ≃ 1, or in an equivalent form
| (24) |
Since p may be as large as 30 in numerical calculations, in order to correctly describe momentum dynamics of the fastest electrons, it results that
| (25) |
or
| (26) |
since βth† = and Θ† = Te†∕mec2. Therefore, in a thermonuclear plasma, one must always consider the relativistic form of the Maxwellian distribution function. In addition, considering the asymptotic limit of K2 for large arguments (??), the condition is
| (27) |
or
which is always satisfied, since Te† never exceeds a few ten keV in tokamak plasmas. This means that the approximate formulation of f is fully valid.
In normalized units,
| (30) |
since pth† = βth†, with
| (31) |
and the modified normalisation coefficient becomes
| (32) |
The Maxwellian distribution function be expressed in an alternative form, useful for calculating interpolation between full and half-grids,
| (33) |
One can then easily cross-check that
| (34) |
is well recovered.