S = ∑ g cd ( k − 1 , 2 0 1 5 )
If the sum above is taken over all integers k with 1 ≤ k ≤ 2 0 1 5 and g cd ( k , 2 0 1 5 ) = 1 , find S .
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.
Very nice! You are proving Menon's identity in the case when N is square-free, which makes the proof more transparent (compare with the general proof ) . Your sum in the last two lines is simply ( 9 − 1 ) ( 2 5 − 1 ) ( 6 1 − 1 ) = p ∣ N ∏ ( Σ ( p ) − 1 ) = p ∣ N ∏ ( 2 p − 2 ) = 2 m p ∣ N ∏ ( p − 1 ) = τ ( N ) ϕ ( N ) as in Menon's identity, where m is the number of prime factors of the square free integer N .
Stated succinctly, the sum we seek is the Dirichlet convolution of the Mobius function with the gcd-sum function, ∑ d ∣ N μ ( N / d ) Σ ( d ) .
By Menon's elegant identity, the answer is τ ( 2 0 1 5 ) ϕ ( 2 0 1 5 ) , where τ denotes the number of positive integer factors. Now 2 0 1 5 = 5 ∗ 1 3 ∗ 3 1 so that τ ( 2 0 1 5 ) ϕ ( 2 0 1 5 ) = 2 3 ∗ 1 4 4 0 = 1 1 5 2 0 .
Menon Identity would be a nice way.
But is there any way to solve it using elementary way? I tried it but failed.
Log in to reply
Now that school is out, I had a bit of time to think about this. For a number like 2015, with just 3 (distinct) prime factor, it's not hard to find this sum by inspection. All we need is some careful counting and the Euclidean algorithm (EA) that allows us to write 1 as a linear combination of any two co-prime integers.
Let me try to work this out in an "elementary way".
It might be slightly easier to find C = ∑ g cd ( k − 1 , 2 0 1 5 ) where we sum over all k with g cd ( k , 2 0 1 5 ) = 1 . Then we have S = 1 3 7 2 5 − C where 13725 is the gcd-sum of 2015.
Let p 1 = 5 , p 2 = 1 3 , p 3 = 3 1 be the prime factors of 2 0 1 5 . All integers we consider will be between 1 and 2015, inclusive.
How often does the summand p 1 p 2 appear in C ? By the EA, there is exactly one multiple j p 3 such that j p 3 − 1 is divisible by p 1 p 2 . Thus the summand p 1 p 2 appears exactly once in C ; likewise for p 1 p 3 and p 2 p 3 .
How often does the summand p 1 appear in C ? There are p 2 − 1 multiples j p 3 such that g cd ( j p 3 − 1 , N ) = p 1 , and, likewise, there are p 3 − 1 multiples j p 2 such that g cd ( j p 2 − 1 , N ) = p 1 . With one double count for the multiples of p 2 p 3 , the summand p 1 appears p 2 + p 3 − 3 times in C . Likewise for p 2 and p 3 , mutatis mutandis.
The remaining summands of S are all 1; we count 483 of them.
Now we are ready for the final tally: C = 3 ( 5 ∗ 1 3 + 5 ∗ 3 1 + 1 3 ∗ 3 1 ) − 3 ( 5 + 1 3 + 3 1 ) + 4 8 3 ∗ 1 = 2 2 0 5 and S = 1 3 7 2 5 − 2 2 0 5 = 1 1 5 2 0
Problem Loading...
Note Loading...
Set Loading...
Set N be a square-free number (and we will take N = 2 0 1 5 = 5 ⋅ 1 3 ⋅ 3 1 ), and d be a divisor of N .
Lemma: ⟨ gcd ( k − 1 , N ) ∣ k is a multiple of d ⟩ = ⟨ gcd ( ℓ , N / d ) ∣ 0 ≤ ℓ < N / d ⟩ . (This is an equality of "bags", i.e. the lists of numbers are identical except for the order.)
Proof : Apply the Chinese Remainder Theorem to the system of equations { x ≡ − 1 mod d , x ≡ ℓ mod N / d . to be guaranteed a unique solution 0 ≤ x = k − 1 < N for every value of ℓ . (It is essential that N is square-free, so that d and N / d are coprime.) This defines a bijection between the values of k and ℓ . Moreover, g cd ( k − 1 , N ) = g cd ( x , N ) = g cd ( x , N / d ) = g cd ( ℓ , N / d ) . □
Now define Σ ( n ) : = ∑ ℓ = 0 n − 1 gcd ( ℓ , n ) for square-free integers n .
Lemma: Σ is a multiplicative function and for prime numbers p , Σ ( p ) = 2 p − 1 . (I leave the proof to you.)
Now to solve the problem we observe that the set of k values over which we sum is { multiples of 1 } − { multiples of 5 } − { multiples of 13 } − { multiples of 31 } + { multiples of 65 } + { multiples of 155 } + { multiples of 403 } − { multiples of 2015 } (E.g. after subtracting multiples of 5 and multiples of 13 we have subtracted all multiples of 65 twice, so we need to add them back in once, and so on. This is a common counting strategy.)
We obtain the desired sum by adding and subtraction various sums along the same lines; and note that k is multiple of d ∑ gcd ( k − 1 , 2 0 1 5 ) = ℓ = 0 ∑ 2 0 1 5 / d − 1 gcd ( ℓ , 2 0 1 5 / d ) = Σ ( 2 0 1 5 / d ) . Thus we get as result Σ ( 2 0 1 5 ) − Σ ( 4 0 3 ) − Σ ( 1 5 5 ) − Σ ( 6 5 ) + Σ ( 3 1 ) + Σ ( 1 3 ) + Σ ( 5 ) − Σ ( 1 ) = 9 ⋅ 2 5 ⋅ 6 1 − 2 5 ⋅ 6 1 − 9 ⋅ 6 1 − 9 ⋅ 2 5 + 6 1 + 2 5 + 9 − 1 = 1 1 5 2 0 .