Put and . For any positive integer shift , translating the interval changes its sum by at most . Averaging the shifts gives the bilinear shift averaging for a logarithmic phase identity
Since , the alternating Taylor expansion of the logarithm has remainder at most . Thus
For , we have , hence the error is at most . The exponential function on an imaginary argument changes by at most the change in that argument. Therefore
Use and . Taking absolute values proves
The boundary error comes from integer shifts, so this argument also covers intervals shorter than a shift. Here range over positive integers; no zero term is needed.

Articles by others on the same topic (0)

There are currently no matching articles.