Time bounded frequency computations

First we obtain two new results considering the inclusion problem of polynomial time frequency classes with equal numbers of errors: 1) (m, m+d) P⊇(m+1, m+d+1) P for m<2^d; and 2) (m, m+d) P=(m+1, m+d+1) P for m⩾c(d) where c(d) is large enough. This disproves a conjecture of Kinber (1975). Next we give a transparent proof of a generalization of Kinber´s result that there exist arbitrarily complex problems admitting a polynomial time frequency computation. Several corollaries provide more insight in the structure of the hierarchy of polynomial time frequency classes