Prove an inequality related to moments I am reading a paper and stuck with an inequality used in that paper. 
$\varepsilon^n=(\varepsilon_1^n, \varepsilon_2^n,\ldots,\varepsilon_n^n)^T$ is a vector of i.i.d. random variables with mean 0 and variance $\sigma^2$. Assume that $\varepsilon_i^n$ have finite $2k$'th moment $E(\varepsilon_i^n)^{2k}<\infty$ for an integer $k>0$. Show that for a constant $n$-dimensional vector $\alpha$, have
$$E(\alpha^T\varepsilon^n)^{2k}\leq (2k-1)!!\|\alpha\|_2^2E(\varepsilon_i^n)^{2k}.$$
The paper I am reading is "On Model Selection Consistency of Lasso" by Zhao and Yu 2006, which can be found via http://jmlr.csail.mit.edu/papers/volume7/zhao06a/zhao06a.pdf
and the inequality appears on Page 2558.
Thanks
 A: As observed in a (now deleted) previous comment, the exponent of $\|\alpha\|_2$ should be $2k$ instead of $2$ for homogeneity reasons.
If the $\varepsilon_i$ are symmetric, then this can be proven by a variant of the exponential moment generating function method used to prove Khintchine's inequality.  Indeed, if we normalise ${\bf E} \varepsilon_i^{2k}$ to be 1, then from Holder's inequality we see that ${\bf E} \varepsilon_i^j$ vanishes for odd $j$ and is bounded by $1$ for even $j$ up to $2k$.  In particular, the exponential moment generating function
$$ {\bf E} \exp( t \varepsilon_i ) = \sum_{j=0}^\infty \frac{t^j}{j!} {\bf E} \varepsilon_i^j$$
is dominated by $\cosh( t^2  )$ in the sense that the coefficients of the former power series up to $t^{2k}$ are bounded in magnitude by those of the latter.  $\cos(t^2)$ is dominated in turn by $\exp(t^2/2)$.  Since
$$ {\bf E} \exp( t \varepsilon ) = \prod_{i=1}^n {\bf E} \exp(\alpha_i t \varepsilon_i )$$
we conclude that ${\bf E} \exp( t \varepsilon )$ is dominated by $\exp( \|\alpha\|_2^2 t^2 / 2)$.  Extracting the $t^{2k}$ coefficient gives the claim.
The situation seems to be more subtle in the non-symmetric case; there does not seem to be a similarly simple argument (though one can certainly obtain a bound with $(2k-1)!!$ replaced by some weaker constant $C_k$ depending on $k$).  It might be that the authors overlooked or neglected to mention a symmetry hypothesis when using this result.
