10.12.2012 Views

Prime Numbers

Prime Numbers

Prime Numbers

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

60 Chapter 1 PRIMES!<br />

this recreation is [Crandall 1997a]. In that reference, what might be called<br />

preposterous physical scenarios—such as the annual probability of finding<br />

oneself accidentally quantum-tunneled bodily (and alive, all parts intact!) to<br />

planet Mars—are still not much smaller than A −A ,whereA is the Avogadro<br />

number (a mole, or about 6·10 23 ). It is difficult to describe a statistical scenario<br />

relevant to the primes that begs of yet higher exponentiation as manifest in<br />

theSkewesnumber.<br />

Incidentally, for various technical reasons, the logarithmic-integral function<br />

li0, on many modern numerical/symbolic systems, is best calculated in<br />

terms of Ei(ln x), where we refer to the standard exponential-integral function<br />

Ei(z) =<br />

z<br />

t<br />

−∞<br />

−1 e t dt,<br />

with principal value assumed for the singularity at t = 0. In addition, care<br />

must be taken to observe that some authors use the notation li for what we<br />

are calling li0, rather than the integral from 2 in our defining equation (1.3)<br />

for li . Calling our book’s function li , and the latter li0, we can summarize<br />

this computational advice as follows:<br />

li (x) =li0(x) − li0(2) = Ei(ln x) − Ei(ln 2) ≈ Ei(ln x) − 1.0451637801.<br />

1.37. In [Schoenfeld 1976] it is shown that on the Riemann hypothesis we<br />

have the strict bound (for x ≥ 2657)<br />

|π(x) − li0(x)| < 1 √<br />

x ln x,<br />

8π<br />

where li0(x) is defined in Exercise 1.36. Show via computations that none of<br />

the data in Table 1.1 violates the Riemann hypothesis!<br />

By direct computation and the fact that li (x) < li0(x) < li (x) +1.05,<br />

prove the assertion in the text that assuming the Riemann hypothesis,<br />

|π(x) − li (x)| < √ x ln x for x ≥ 2.01. (1.48)<br />

It follows from the discussion in connection to (1.25) that (1.48) is equivalent<br />

to the Riemann hypothesis. Note too that (1.48) is an elementary assertion,<br />

which to understand one needs to know only what a prime is, the natural<br />

logarithm, and integrals. Thus, (1.48) may be considered as a formulation of<br />

the Riemann hypothesis that could be presented in, say, a calculus course.<br />

1.38. With ψ(x) defined as in (1.22), it was shown in [Schoenfeld 1976] that<br />

theRiemannhypothesisimpliesthat<br />

|ψ(x) − x| < 1 √ 2<br />

x ln x for x ≥ 73.2.<br />

8π<br />

By direct computation show that on assumption of the Riemann hypothesis,<br />

|ψ(x) − x| < √ x ln 2 x for x ≥ 3.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!