Improved non-perturbative renormalization without $c_{NGI}$

Abstract

Recently, a method for O(a) improvement of composite operators has been proposed which uses the large momentum behavior of fixed gauge quark and gluon correlation functions (G. Martinelli et al., hep-lat/0106003). A practical problem with this method is that a particular improvement coefficient, $c_{NGI}$, which has a gauge non-covariant form, is difficult to determine. Here I work out the size of the errors made in improvement coefficients and physical quantities if one does not include the $c_{NGI}$ term.