“…Via an interpolation argument already used in [21], this also yields the assertion of Theorem 2.3. In fact, this is the idea behind the approach we use in [24]. However, the result in [24] applies to operators satisfying generalized Gaussian estimates of any order, which means that essential technical tools such as the finite propagation speed of the wave equation for A, which one has for second-order operators, cannot be used.…”