This paper defines a simple component of the CMB 4point function which they claim vanishes for lensing but provides a signature of kinetic SZ, hence allowing one to probe the latter independently.
My first comment is that they are using a series expansion in the deflection angle. As they mention this is an approximation, and in fact is not really very good (e.g. see Fig 8 of astroph/0601594). However doesn't the lensed < T_lens(x)^3 T_lens(y) > vanish more generally (for uncorrelated lensing potential) because T_lens is linear in T, we assume T is Gaussian and isotropic, and <T(x)^2> is unchanged by lensing?
My second comment is that the SZ signal will be correlated with the lensing signal to some extent (see e.g. astroph/0208325), which presumably complicates the picture.
[astroph/0610007] Separating the Weak Lensing and Kinetic SZ Effects from CMB Temperature Maps
Authors:  Mario A. Riquelme, David N. Spergel 
Abstract:  A new generation of CMB experiments will soon make sensitive high resolution maps of the microwave sky. At angular scales less than $\sim$10 arcminutes, most CMB anisotropies are generated at z $< 1000$, rather than at the surface of last scattering. Therefore, these maps potentially contain an enormous amount of information about the evolution of structure. Whereas spectral information can distinguish the thermal SunyaevZeldovich (tSZ) effect from other anisotropies, the spectral form of anisotropies generated by the gravitational lensing and the kinetic SunyaevZeldovich (kSZ) effects are identical. While spectrally identical, the statistical properties of these effects are different. We introduce a new realspace statistic, $<\theta (\hat{n})^3 \theta (\hat{m})>_c$, and show that it is identically zero for weakly lensed primary anisotropies and, therefore, allows a direct measurement of the kSZ effect. Measuring this statistic can offer a new tool for studing the reionization epoch. Models with the same optical depth, but different reionization histories, can differ by more than a factor of 3 in the amplitude of the kSZgenerated nonGaussian signal. 
[PDF] [PS] [BibTex] [Bookmark] 

 Posts: 1618
 Joined: September 23 2004
 Affiliation: University of Sussex
 Contact:

 Posts: 3
 Joined: January 18 2005
 Affiliation: Princeton University
 Contact:
[astroph/0610007] Separating the Weak Lensing and Kinetic S
Yes, I also get that it vanishes.However doesn't the lensed < Tlens(x)3 Tlens(y) > vanish more generally (for uncorrelated lensing potential)