Remarks on the paper of V. Guillemin and K. Okikiolu: ``Subprincipal terms in Szegő estimates'' (Q1128264): Difference between revisions

\textit{V. Guillemin} and \textit{K. Okikiolu} [ibid. 4, No. 1, 173-179 (1997; Zbl 0881.58064)] proved a two terms asymptotics of the traces , \(\text{Tr} (P_\lambda B P_\lambda)^k\), \(k=1,2,\dots\), as \(\lambda \to \infty\), where \(B\) is a pseudodifferential operator of order zero, acting in the space \(L^2(M)\) of half-densities on a smooth closed manifold \(M\), and \(P_\lambda\) is the spectral projection of a first order positive elliptic pseudodifferential operator, whose principal symbol satisfies \(a(x,\xi)=a(x,- \xi)\), while its subprincipal symbol is identically zero. Moreover, it is supposed that for any \(t>0\) the set of \(t\)-periodic points for the Hamiltonian flow associated with \(a\) is of measure zero. In the present paper, the authors show how to generalize this result to the case where the function \(\psi(t)= t^k\) is replaced by an arbitrary \(C^2\)-function, using appropriate approximation with polynomials.