Resolving kinematic ambiguities in QCD predictions for Deeply Virtual Compton Scattering

The existing QCD predictions for the Deeply Virtual Compton Scattering (DVCS) depend on the convention used for the skewedness parameter and on the reference frame used to define helicity amplitudes. These ambiquities are formally power-suppressed but numerically significant. They are cancelled by finite-$t$ and target mass corrections that have been calculated recently to the $1/Q^2$ accuracy. It turns out that these corrections can be minimized, at least for unpolarized observables, by choosing a specific reference frame where longitudinal directions are defined by the photon momenta.