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.

1.5 Exercises 53<br />

1.17. It can happen that a polynomial, while not always producing primes,<br />

is very likely to do so over certain domains. Show by computation that a<br />

polynomial found in [Dress and Olivier 1999],<br />

f(x) =x 2 + x − 1354363,<br />

has the astounding property that for a random integer x ∈ [1, 10 4 ], the number<br />

|f(x)| is prime with probability exceeding 1/2. An amusing implication is this:<br />

If you can remember the seven-digit “phone number” 1354363, then you have<br />

a mental mnemonic for generating thousands of primes.<br />

1.18. Consider the sequence of primes 2, 3, 5, 11, 23, 47. Each but the first<br />

is one away from the double of the prior prime. Show that there cannot be<br />

an infinite sequence of primes with this property, regardless of the starting<br />

prime.<br />

1.19. As mentioned in the text, the relation<br />

1<br />

ζ(s) =<br />

∞<br />

n=1<br />

µ(n)<br />

n s<br />

is valid (the sum converges absolutely) for Re(s) > 1. Prove this. But the<br />

limit as s → 1, for which we know the remarkable PNT equivalence<br />

∞<br />

n=1<br />

µ(n)<br />

n =0,<br />

is not so easy. Two good exercises are these: First, via numerical experiments,<br />

furnish an estimate for the order of magnitude of<br />

<br />

n≤x<br />

µ(n)<br />

n<br />

as a function of x; and second, provide an at least heuristic argument as to<br />

why the sum should vanish as x →∞. For the first option, it is an interesting<br />

computational challenge to work out an efficient implementation of the µ<br />

function itself. As for the second option, you might consider the first few<br />

terms in the form<br />

1 − <br />

p≤x<br />

1 1<br />

+<br />

p pq<br />

pq≤x<br />

−···<br />

to see why the sum tends to zero for large x. It is of interest that even without<br />

recourse to the PNT, one can prove, as J. Gram did in 1884 [Ribenboim 1996],<br />

that the sum is bounded as x →∞.<br />

1.20. Show that for all x>1, we have<br />

<br />

p≤x<br />

1<br />

> ln ln x − 1,<br />

p

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

Saved successfully!

Ooh no, something went wrong!