More fun in 2015, Part 41

Find L = log ϕ ( n ) d n ϕ ( d ) L=\log_{\phi(n)}\prod_{d|n}\phi(d) for n = 2014 × 2015 n=2014\times 2015 .

Bonus Problem : Find this logarithm for any square-free n n .


Inspired by Chinmay Sangawadekar .


The answer is 32.

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

Otto Bretscher
Dec 26, 2015

We can group the divisors of a square-free integer n > 1 n>1 into (unordered) pairs d d and n / d n/d , with ϕ ( d ) ϕ ( n / d ) = ϕ ( n ) \phi(d)\phi(n/d)=\phi(n) since ϕ \phi is multiplicative. If τ ( n ) \tau(n) is the number of divisors of n n , then there are τ ( n ) / 2 \tau(n)/2 such pairs, so that d n ϕ ( d ) = ϕ ( n ) τ ( n ) / 2 \prod_{d|n}\phi(d)=\phi(n)^{\tau(n)/2} and L = τ ( n ) / 2 L=\tau(n)/2 . Note that τ ( n ) = 2 m \tau(n)=2^m if n n has m m prime factors.

In the case of n = 2014 2015 n=2014*2015 we have m = 6 m=6 prime factors so that L = 2 5 = 32 L=2^5=\boxed{32} .

Moderator note:

Great observation of how to split out the terms.

If n n has m m distinct prime factors, then we are looking at choosing 1 to m of them to multiply to get a factor of n. If we choose k of them, then k m \frac{k}{m} of them will multiply out till we get n n .

Thus we are evaluating k = 1 m k m ( m k ) = 2 m 1 \displaystyle\sum_{k=1}^{m} \frac{k}{m}\binom{m}{k}=2^{m-1}

Trevor Arashiro - 5 years, 5 months ago

Yes! I had done exactly the same!

A Former Brilliant Member - 5 years, 5 months ago

But why to take only sq. free no.? @Otto Bretscher

A Former Brilliant Member - 5 years, 5 months ago

Log in to reply

The problem does not look like a lot of fun when n n isn't square-free ;)

Otto Bretscher - 5 years, 5 months ago

What if we take totient of 1/d or d/phi(d)? @Otto Bretscher

A Former Brilliant Member - 5 years, 5 months ago

Log in to reply

ϕ ( 1 d ) \phi(\frac{1}{d}) does not make sense... ϕ \phi is defined for integers.

Otto Bretscher - 5 years, 5 months ago
Jake Lai
Dec 26, 2015

This solution will focus on n squarefree. Since n is squarefree, we may let n = i = 1 ω ( n ) p i \displaystyle n = \prod_{i=1}^{\omega(n)} p_i , where ω ( n ) \omega(n) is the number of distinct prime factors of n.

"Zoning in" on a single factor p 1 p_1 , we can ask how many divisors of n have p 1 p_1 as a factor. If a divisor has m factors, one of which is p 1 p_1 , then we can choose m-1 other factors in ( ω ( n ) 1 m 1 ) \dbinom{\omega(n)-1}{m-1} ways. It is then easy to conclude that, in total, there are ( ω ( n ) 1 1 1 ) + ( ω ( n ) 1 2 1 ) + + ( ω ( n ) 1 ω ( n ) 1 ) = 2 ω ( n ) 1 \dbinom{\omega(n)-1}{1-1} + \dbinom{\omega(n)-1}{2-1} + \ldots + \dbinom{\omega(n)-1}{\omega(n)-1} = 2^{\omega(n)-1} divisors of n with p 1 p_1 ; an identical argument follows for all other p i p_i .

Using the fact that ϕ \phi is multiplicative, we may simply separate the factors of each divisor. Following this, we may observe that

d n ϕ ( d ) = i = 1 ω ( n ) ϕ ( p i ) 2 ω ( n ) 1 = [ i = 1 ω ( n ) ϕ ( p i ) ] 2 ω ( n ) 1 = ϕ ( n ) 2 ω ( n ) 1 . \begin{aligned} \prod_{d|n} \phi(d) &= \prod_{i=1}^{\omega(n)} \phi(p_i)^{2^{\omega(n)-1}} \\ &= \left[ \prod_{i=1}^{\omega(n)} \phi(p_i) \right]^{2^{\omega(n)-1}} \\ &= \phi(n)^{2^{\omega(n)-1}}. \end{aligned}

This implies, then, that (for ϕ ( n ) 1 \phi(n) \neq 1 ),

log ϕ ( n ) ( d n ϕ ( d ) ) = log ϕ ( n ) ϕ ( n ) 2 ω ( n ) 1 = 2 ω ( n ) 1 . \begin{aligned} \log_{\phi(n)} \left( \prod_{d|n} \phi(d) \right) &= \log_{\phi(n)} \phi(n)^{2^{\omega(n)-1}} \\ &= 2^{\omega(n)-1}. \end{aligned}

Here, ω ( 2014 × 2015 ) = 6 \omega(2014 \times 2015) = 6 so 2 ω ( n ) 1 = 32 2^{\omega(n)-1} = \boxed{32} .

Note that this solution is not by any means rigorous, but more motivational of the thought process/method of solving than anything.

Jake Lai - 5 years, 5 months ago

Log in to reply

It does not take much to make your argument rigorous. If n = p 1 p 2 p K n \,=\, p_1p_2 \ldots p_K is a product of K K distinct primes, then the divisors of n n are numbers of the form d A = j A p j , A { 1 , 2 , , K } . d_A \; = \; \prod_{j \in A} p_j \qquad , \qquad \qquad A \subseteq \{1,2,\ldots,K\} \;. and hence d n ϕ ( d ) = A { 1 , 2 , , K } ϕ ( d A ) = A { 1 , 2 , , K } j A ( p j 1 ) = j = 1 K ( p j 1 ) N ( j ) \prod_{d \vert n} \phi(d) \; = \; \prod_{A \subseteq \{1,2,\ldots,K\}} \phi(d_A) \; = \; \prod_{A \subseteq \{1,2,\ldots,K\}} \prod_{j \in A} (p_j-1) \; = \; \prod_{j=1}^K (p_j - 1)^{N(j)} where N ( j ) = { A { 1 , 2 , , K } j A } = 2 K 1 N(j) \; = \; \big\vert \{ A \subseteq\{1,2,\ldots,K\} \; \big\vert \; j \in A \} \big\vert \; = \; 2^{K-1} for each 1 j K 1 \le j \le K . Thus d n ϕ ( d ) = j = 1 K ( p j 1 ) 2 K 1 = ϕ ( n ) 2 K 1 \prod_{d \vert n} \phi(d) \; = \; \prod_{j=1}^K (p_j-1)^{2^{K-1}} \; = \; \phi(n)^{2^{K-1}} and hence log ϕ ( n ) d n ϕ ( d ) = 2 K 1 . \log_{\phi(n)} \prod_{d \vert n} \phi(d) \; = \; 2^{K-1} \;.

Mark Hennings - 5 years, 5 months ago

i did it the exact same way!did you try to generalize for all n?

Aareyan Manzoor - 5 years, 5 months ago

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...