Research notes on every problem, and a library of the papers behind them. Built from the open erdos repository.
Updated
For 2≤z≤y, some absolute c>0 gives
z≤p≤y∑p1≪logze−clogz+log(y/z).(1)
Proof. Partial summation of the PNT in notation gives the more
precise Mertens form
p≤t∑p1=loglogt+B+O(1/logt)(t≥10).(2)
Indeed, the prime sum equals π(t)/t+∫2tπ(u)u−2du.
Substituting π(u)=li(u)+E(u) makes the main part
loglogt plus a constant. The integral of E(u)/u2 converges,
and its tail, together with E(t)/t, is O(1/logt) by the
exponential PNT error. Bounded smaller t can be absorbed.
If y>2z, (2), with the possible endpoint term 1/z, gives
Here log(1+u)≤u and log(y/z)>log2 absorb the error.
If z≤y≤2z and z≥10, Lemma 1.6 gives
z≤p≤y∑p1≤zπ([z,y])≪zlogzy−z+e−c0logz.
On this range (y−z)/z≪log(y/z). Decrease c0 to c>0 so
e−c0logz≪e−clogz/logz.
For 2≤z<10 and y≤2z, the left side is bounded, while the
exponential term divided by logz has a positive lower bound on
that compact range. This completes every case. □
Source.Tao, published paper, published p.799, Lemma 1.7. This page uses that published version.