> Dear Colleagues, > > As an author the 2009 ZEUS paper on isolated photon emission > (arXiv:0909.4223) I was pleased to see your recent paper applying the kT > factorisation approach to our data and noted your success in describing > the low-Q^2 and low-x bins. > > This appears to be a major success for kT factorisation as compared to > the DGLAP approach. May I ask two questions: > > 1) Is this the clearest experimental evidence so far in favour of > the BFKL approach? First of all, we would prefer to use the term of "k_t-factorization" rather than BFKL. In fact, the k_t-factorization is a kind of more general term than BFKL, and, within the k_t-factorization, there are several different schemes based on different parton evolution equations (one of them is BFKL, and another one is CCFM). In fact, in our present calculations we have been using the KMR parametrization of gluon densities. In this approach, the BFKL kernel is only used for the last splitting (and not for all splittings in the gluon evolution). Prompt photons show one of clear enough experimental signatures of the k_t-factorization, among other ones (such as J/psi and Upsilon meson spin alignement, jet-jet azimuthal correlations, etc.) > 2) Why is this a good place to test BFKL v. DGLAP The reason is that, BFKL is focused on collecting the so called "large logarithms" of the type [alpha_s * log(1/x)]^n up to infinetely large n, while DGLAP collects [alpha_s * log(Q^2)]^n. So, The k_t-factorization shows the best applicability in the region of small x and small Q^2; and any process in that region is a good place. Prompt photons are especially useful as they directly probe the parton densities and are not affected by the subsequent fragmentation and hadronization. The latter are excluded because of photon isolation cuts. In conclusion, we would like to add that many of the k_t-factorization effects are seen with the NLO or NNLO matrix elements in the collinear (DGLAP) approach. In many cases we see that including the higher order contributions in the DGLAP calculations makes the DGLAP predictions closer to the k_t-factorization. The NLO DGLAP looks like the k_t-factorization, but is not identical to it. > > The ZEUS paper has recently been accepted by Phys Lett B. This version of > the paper includes a new high-Q^2 bin. I would be very interested to see > how your predictions compare to the data. see > http://arxiv.org/PS_cache/arxiv/pdf/0909/0909.4223v2.pdf > Yes, of course we will consider doing calculations for that. With our best regards, Sergey Baranov, Artem Lipatov, Nikolai Zotov