Good Estimate

If any two integers m , n m ,n are chosen from the interval [ 1 , ) [ 1, \infty) , find the probability that they are relatively coprime to 3 decimal places.


The answer is 0.607927101.

This section requires Javascript.
You are seeing this because something didn't load right. We suggest you, (a) try refreshing the page, (b) enabling javascript if it is disabled on your browser and, finally, (c) loading the non-javascript version of this page . We're sorry about the hassle.

2 solutions

\rightarrow We may consider the sum 2 n x ϕ ( n ) 1 2 \sum_{n\leq x}\phi (n)-1 , which ϕ ( n ) \phi(n) is the totient function. )

\rightarrow When m n m \leq n , there is n x ϕ ( n ) \sum_{n\leq x}\phi (n) number of pairs of ( m , n ) (m,n) .

\rightarrow The case when n m n \leq m is the same, since we have counted the pair ( 1 , 1 ) (1,1) twice, we minus one from the total, and we get the stated formula.

\odot We now prove that n x ϕ ( n ) = ( 3 / π 2 ) x 2 + O ( x l o g x ) \sum_{n\leq x}\phi (n) = (3/ \pi ^2 ) x^2 + O( x logx ) .

As d n ϕ ( d ) = n \sum_{d \mid n}\phi (d) = n , by Möbius inversion formula, ϕ ( n ) = d n μ ( d ) ( n / d ) \phi(n) = \sum_{d \mid n}\mu (d) (n/d) .

To get a good estimation of n x ϕ ( n ) \sum_{n\leq x}\phi (n) , we proceed as following.

n x ϕ ( n ) \sum_{n\leq x}\phi (n)

= n x d n μ ( d ) ( n / d ) = \sum_{n\leq x} \sum_{d \mid n} \mu (d) (n/d)

Define k = n / d k=n/d

= d , k , a n d , d k x μ ( d ) ( k ) = \sum_{{d,k},and, {dk\leq x}} \mu (d) (k)

= d x μ ( d ) k x / d ( k ) = \sum_{d\leq x} \mu (d) \sum_{k \leq x/d} (k)

Since k x / d ( k ) \sum_{k \leq x/d} (k) is sum of integers less than or equal to x x ,

the sum is ( x / d ) ( x / d + 1 ) / 2 ( \left \lfloor x/d \right \rfloor ) ( \left \lfloor x/d \right \rfloor +1 ) /2

\rightarrow n x ϕ ( n ) \sum_{n\leq x}\phi (n) = d x μ ( d ) ( x / d ) ( x / d + 1 ) / 2 = \sum_{d\leq x} \mu (d) ( \left \lfloor x/d \right \rfloor ) ( \left \lfloor x/d \right \rfloor +1 ) /2

Since x / d \left \lfloor x/d \right \rfloor = x / d + O ( 1 ) = x/d + O(1) ,

So ( x / d ) ( x / d + 1 ) ( \left \lfloor x/d \right \rfloor ) ( \left \lfloor x/d \right \rfloor +1 ) = ( x / d + O ( 1 ) ) ( x / d + O ( 1 ) ) = x 2 / d 2 + O ( x / d ) = ( x/d + O(1) ) ( x/d + O(1) ) = x^2/d^2 + O(x/d)

so d x μ ( d ) ( x / d ) ( x / d + 1 ) / 2 \sum_{d\leq x} \mu (d) ( \left \lfloor x/d \right \rfloor ) ( \left \lfloor x/d \right \rfloor +1 ) /2

= 1 / 2 ( x 2 ) d x μ ( d ) / d 2 + O ( d x ( x / d ) μ ( d ) ) = 1/2 (x^2) \sum_{d\leq x} \mu (d) /d^2 + O ( \sum_{d\leq x} (x/d) \mu (d) )

= 1 / 2 ( x 2 ) d x μ ( d ) / d 2 + O ( d x ( x / d ) μ ( d ) ) = 1/2 (x^2) \sum_{d\leq x} \mu (d) /d^2 + O ( \sum_{d\leq x} (x/d) \left | \mu (d) \right | )

= 1 / 2 ( x 2 ) d x μ ( d ) / d 2 + O ( d x ( x / d ) ) = 1/2 (x^2) \sum_{d\leq x} \mu (d) /d^2 + O ( \sum_{d\leq x} (x/d) )

= 1 / 2 ( x 2 ) d x μ ( d ) / d 2 + O ( x d x ( 1 / d ) ) = 1/2 (x^2) \sum_{d\leq x} \mu (d) /d^2 + O ( x \sum_{d\leq x} (1/d) )

= 1 / 2 ( x 2 ) d x μ ( d ) / d 2 + O ( x 1 x ( 1 / d ) d d ) ) = 1/2 (x^2) \sum_{d\leq x} \mu (d) /d^2 + O ( x \int_{1}^{x} ( 1/d) {\mathrm{d} d}) )

= 1 / 2 ( x 2 ) d x μ ( d ) / d 2 + O ( x l o g x ) = 1/2 (x^2) \sum_{d\leq x} \mu (d) /d^2 + O ( x log x)

So as x x \rightarrow \infty ,

lim x ( 2 n x ϕ ( n ) 1 ) / ( x 2 ) ) \lim_{x\rightarrow \infty}( 2 \sum_{n\leq x}\phi (n)-1 ) / (x^2))

= lim x [ x 2 ( d x μ ( d ) / d 2 + O ( x l o g x ) ] / x 2 = \lim_{x\rightarrow \infty}[x^2 ( \sum_{d\leq x} \mu (d) /d^2 + O ( x log x)]/x^2

= lim x ( d x μ ( d ) / d 2 ) + O ( x l o g x ) ] / x 2 = \lim_{x\rightarrow \infty} ( \sum_{d\leq x} \mu (d) /d^2 ) + O ( x log x)]/x^2

As d x μ ( d ) / d 2 = 1 / ζ ( 2 ) = 6 / ( π ) 2 \sum_{d\leq x} \mu (d) /d^2 = 1/ \zeta(2) = 6/( \pi )^2

so lim x ( 2 n x ϕ ( n ) 1 / ( x 2 ) ) = 6 / ( π ) 2 \lim_{x\rightarrow \infty}( 2 \sum_{n\leq x}\phi (n)-1 / (x^2)) = 6/( \pi )^2

where ζ ( s ) \zeta (s) is the Riemann zeta function, and so the solution is 6 / ( π ) 2 \mathfrak{6/( \pi )^2 } .

Brian Moehring
Jan 30, 2017

Let A p A_p denote the event that at least one of the numbers is not divisible by a given prime p p . Then we can see that P [ A p ] = 1 1 p 2 \mathbb{P}[A_p] = 1 - \frac{1}{p^2} . The event that the two nonzero integers are relatively prime is just p A p \bigcap_p A_p , and the events are independent, so we have P [ p A p ] = p P [ A p ] = p ( 1 1 p 2 ) = 1 ζ ( 2 ) = 6 π 2 . \mathbb{P}\left[\bigcap_p A_p\right] = \prod_p \mathbb{P}[A_p] = \prod_p \left(1 - \frac{1}{p^2}\right) = \frac{1}{\zeta(2)} = \frac{6}{\pi^2}.

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...