2024 Secondary P9
Tahsin Khan
November 2024
1 Solution
The answer is all m such that gcd(m, k!) = 1 (i.e. m is co-prime to k!).
Let f (n, m) = (n + m)(n + 2m)...(n + km) for a fixed k.
Proof that m doesn’t work if gcd(m, k!) ̸= 1:
Let p be a prime divisor of gcd(m, k!). For n = 1, we have n ≡ 1 (mod p) and
m ≡ 0 (mod p).
So, n + sm ≡ 1 (mod p) for all integers s.
So, p ∤ n + sm =⇒ p ∤ f (1, m).
But, p | k!. This implies k! ∤ f (1, m)
Proof that m works if gcd(m, k!) = 1:
We will prove vp (f (n, m)) ≥ vp (k!) for all n and for all prime p.
We know that among any consecutive p integers, one of them is divisible p.
Call an arithmetic progression of positive integers with common difference d
”t, p-good sequence”, if all of its elements are divisible by pt and vp (d) = t.
Lemma 1: In a t, p-good sequence a1 , a2 , ...ap with p terms, there exists at least
one i such that pt+1 | ai .
Proof: Let the common difference be d.
For the sake of contradiction, assume that vp (ai ) < t + 1 for all i.
But, vp (ai ) ≥ t for all i as well.
So, vp (ai ) = t for all i.
Let bi = apti .
So, p ∤ bi =⇒ bi ̸≡ 0 (mod p) for all i.
Using Pigeon-hole Principle, there exist i, j such that bi ≡ bj (mod p)
So, bi − bj ≡ 0 (mod p)
So, p | pdt (i − j)
But, 0 < i, j ≤ p =⇒ i − j < p =⇒ p ∤ i − j
So, p | pdt =⇒ pt+1 | d =⇒ vp (d) ≥ t + 1. Contradiction.
Lemma 2: In a t, p-good sequence a1 , a2 , a3 , ...aq with q terms, ⌊ pq ⌋ of the
terms form a t + 1, p − good sequence.
1
Proof: Let the common difference be d.
According to the first lemma, in the first p terms, there is a term pi0 such that
vp (ai0 ) ≥ t + 1
vp (ai0 +wp ) = vp (ai0 + wpd) ≥ min(vp (ai0 ), vp (wpd)) = t + 1
So, the terms ai0 +wp are divisible by pt+1
Moreover, they form an arithmetic progression with common difference pd that
also has vp greater than or equal to t + 1.
Main part:
For a prime p ≤ k, we will use lemma 2 in the main sequence n + m, n + 2m, n +
3m, as the common difference m is co-prime to p.
Claim: ⌊ pkt ⌋ terms from the main sequence form a t, p-good sequence.
Proof using induction:
Base case: All pk0 = k terms have vp at least 0 and form an arithmetic pro-
gression.
Inductive step:
Given that there is a t, p-good sequence of ⌊ pkt ⌋ terms prove that there is a
k
t + 1, p-good sequence of ⌊ pt+1 ⌋ terms among them.
Proof: This is the result of the 2nd lemma. According to Lemma 2, in the
⌊ pkt ⌋
given sequence, there are ⌊ p ⌋ = terms that form such an arithmetic progress.
⌊ pkt ⌋ k
But, ⌊ p ⌋ = ⌊ pt+1 ⌋ as p is an integer.
So, at least ⌊ pkt ⌋ terms of the main sequence have vp = t for all t. Using
this,
vp (f (n, m)) ≥ ⌊ kp ⌋ + ⌊ pk2 ⌋ + ⌊ pk3 ⌋... = x>0 ⌊ pkx ⌋
P
But, vp (k!) = x>0 ⌊ pkx ⌋
P
So, vp (f (n, m)) ≥ vp (k!)
Hence, k! | f (n, m)
Key points:
• The answer is all m such that gcd(m, k!) = 1.
• Prove that m doesn’t work otherwise.
• vp (a) ≥ vp (b) for all prime p implies b | a. Use this to prove that all m
with gcd(m, k!) = 1 works.
• vp (k!) = x>0 ⌊ pkx ⌋. Find vp (f (n, m)).
P
• ”Among any consecutive p integers, one of them is divisible p”- extend
this so that it works for any arithmetic progression.
2
2 vp
vp (a) means the highest power of p that divides a.
Some basic things about vp :
vp (ab) = vp (a) + vp (b)
vp (a + b) = min(vp (a), vp (b)) when vp (a) ̸= vp (b)
vp (a + b) ≥ vp (a) when vp (a) = vp (b)
Another useful one is
vp (n!) = k=1 ⌊ pnk ⌋
P
Which is also used in the last part of the solution.
It is true because exactly pnt of the first n integers are divisible by pt .
So, ⌊ pnt ⌋ − ⌊ pt+1
n
⌋ of the first n integers have vp exactly t.
So, vp (n!) = k=1 k(⌊ pnk ⌋ − ⌊ pk+1 n
⌋) = k=1 ⌊ pnk ⌋(k + 1 − k) = k=1 ⌊ pnk ⌋
P P P
The same idea was used to determine vp (f (n, m))
