The Smallest Possible 2014th Term

Consider an infinite sequence of positive integers which satisfies the property that the greatest common divisor of any two consecutive terms is strictly greater than the preceding term. Let N N be the smallest possible value of the 201 4 th 2014^{\text{th}} term of this sequence. Find the last three digits of 2 N . 2N.

Details and assumptions
- In other words, if { a i } i = 1 \{a_i\}_{i=1}^{\infty} is the sequence, we have gcd ( a i , a i + 1 ) > a i 1 i N > 1 , \gcd (a_i, a_{i+1}) > a_{i-1} \quad \forall \ i \in \mathbb{N_{> 1}}, and you have to find the smallest possible value of a 2014 . a_{2014}.
- This problem is not original.


The answer is 384.

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

We start with the following lemma.


Lemma: a n + 1 > a n + a n 1 n N a_{n+1} > a_n + a_{n-1} \quad \forall \ n \in \mathbb{N}

Proof: Note that gcd ( a n + 1 , a n ) = gcd ( a n + 1 a n , a n ) , \gcd (a_{n+1}, a_n) = \gcd (a_{n+1} - a_n, a_n), and gcd ( a n + 1 a n , a n ) a n + 1 a n . \gcd (a_{n+1}-a_n, a_n) \leq a_{n+1}-a_n. Since gcd ( a n + 1 , a n ) > a n 1 , \gcd (a_{n+1}, a_n) > a_{n-1}, we have that a n + 1 a n > a n 1 a n + 1 > a n + a n 1 , a_{n+1} - a_n > a_{n-1} \implies a_{n+1} > a_n + a_{n-1}, which is what we wanted. \blacksquare


Now we shall prove by strong induction that a n 2 n 1 a_n \geq 2^{n-1} for all n N . n \in \mathbb{N}.

Base case: n = 1 , 2 , 3 , 4 n= 1, 2, 3, 4

  • n = 1 : n=1: We trivially have a 1 1 = 2 0 , a_1 \geq 1= 2^0, since all a i a_i 's are defined to be positive.

  • n = 2 : n=2: Note that if a 2 = 1 , a_2=1, we would have gcd ( a 3 , a 2 ) = 1 = a 1 , \gcd (a_3, a_2) = 1 = a_1, which is not allowed. Thus, a 2 2 = 2 1 . a_2 \geq 2 = 2^1.

  • n = 3 : n=3: By the lemma, a 3 > a 2 + a 1 2 + 1 = 3 , a_3 > a_2 + a_1 \geq 2+1= 3, so a 3 4 = 2 2 . a_3 \geq 4 = 2^2.

  • n = 4 : n=4: Again, by the lemma, a 4 > a 3 + a 2 4 + 2 = 6 , a_4 > a_3 + a_2 \geq 4+2 = 6, so a 4 7. a_4 \geq 7. However, if a 4 = 7 , a_4=7, we would have gcd ( a 4 , a 3 ) = 1 = a 1 , \gcd (a_4, a_3) = 1 = a_1, since a 4 > a 3 a_4 > a_3 and 7 7 is a prime.

Our base cases are complete.

Inductive step

Assume our assertion is true for 1 , 2 , , k 1 , k 1, 2, \cdots , k-1, k for some k N , k \in \mathbb{N,} so a n 2 n 1 a_n \geq 2^{n-1} for all n k . n \leq k. On the contrary, assume a k + 1 < 2 k . a_{k+1} < 2^k. Note that, by the lemma, a k + 1 > a k + a k 1 > 2 k 1 + 2 k 2 = 3 2 2 k 1 . a_{k+1} > a_k + a_{k-1} > 2^{k-1} + 2^{k-2} = \dfrac{3}{2} \cdot 2^{k-1}. Let a k + 1 = j gcd ( a k + 1 , a k ) . a_{k+1} = j \cdot \gcd (a_{k+1}, a_k). Since gcd ( a k + 1 , a k ) > 2 k 2 , \gcd (a_{k+1}, a_k) > 2^{k-2}, j a k + 1 2 k 2 < 2 k 2 k 2 = 4. j \leq \dfrac{a_{k+1}}{2^{k-2}} < \dfrac{2^k}{2^{k-2}} = 4.

  • Case 1: j = 1 j=1

This implies a k + 1 = gcd ( a k + 1 , a k ) a k + 1 a k a k + 1 a k , a_{k+1} = \gcd (a_{k+1}, a_k) \implies a_{k+1} \mid a_k \implies a_{k+1} \leq a_k, contradiction.

  • Case 2: j = 2 j=2

This implies a k + 1 = 2 gcd ( a k + 1 , a k ) , a_{k+1} = 2 \gcd (a_{k+1}, a_k), so a k + 1 2 \dfrac{a_{k+1}}{2} divides a k . a_k. Note that a k 2 < a k + 1 2 < a k , \dfrac{a_k}{2} < \dfrac{a_{k+1}}{2} < a_k, which shows that this isn't possible (no integer x x can have a divisor in the range ( x 2 , x ) \left( \dfrac{x}{2}, x \right) ).

  • Case 3: j = 3 j=3

This implies a k + 1 = 3 gcd ( a k + 1 , a k ) . a_{k+1} = 3 \gcd (a_{k+1}, a_k). Note that a k < a k + 1 , a_k < a_{k+1}, so a k a_k is either equal to gcd ( a k + 1 , a k ) \gcd (a_{k+1}, a_k) or 2 gcd ( a k + 1 , a k ) . 2 \gcd (a_{k+1}, a_k). However, if a k = gcd ( a k + 1 , a k ) , a_k = \gcd (a_{k+1}, a_k), a k + 1 = 3 a k > 3 2 k 1 > 2 k , a_{k+1} = 3a_k > 3 \cdot 2^{k-1} > 2^k, contradiction. Hence a k = 2 gcd ( a k + 1 , a k ) a_k= 2 \gcd (a_{k+1}, a_k) and a k + 1 = 3 2 a k . a_{k+1} = \dfrac{3}{2}a_k.

Let a k = m gcd ( a k , a k 1 ) . a_k = m \gcd (a_k, a_{k-1}). We can similarly reject the cases m = 1 , 2. m= 1, 2.

  • Subcase 1: m 6 m \geq 6

Then, a k = 2 3 a k + 1 > 6 2 k 3 a k + 1 > 9 2 k 3 > 8 2 k 1 = 2 k , a_k = \dfrac{2}{3} a_{k+1} > 6 \cdot 2^{k-3} \implies a_{k+1} > 9 \cdot 2^{k-3} > 8 \cdot 2^{k-1} = 2^k, contradiction.

  • Subcase 2: 3 m 4 3 \leq m \leq 4

Then, a k 1 < 4 2 gcd ( a k , a k 1 ) = 2 gcd ( a k , a k 1 ) , a_{k-1} < \dfrac{4}{2} \gcd (a_k, a_{k-1}) = 2 \gcd (a_k, a_{k-1}), which implies a k 1 = gcd ( a k 1 , a k ) , a_{k-1} = \gcd (a_{k-1}, a_k), or a k 1 a k . a_{k-1} \mid a_k. Since a k 1 < a k , a_{k-1} < a_k, a k 3 a k 1 a_k \geq 3 a_{k-1} and a k + 1 = 3 2 a k 3 2 3 a k 1 > 9 2 a k 1 > 4 a k 1 > 4 2 k 2 = 2 k , a_{k+1} = \dfrac{3}{2} a_k \geq \dfrac{3}{2} \cdot 3 a_{k-1} > \dfrac{9}{2} a_{k-1} > 4 a_{k-1} > 4 \cdot 2^{k-2} = 2^k, contradiction.

  • Subcase 3: m = 5 m=5

This gives a k + 1 = 15 2 gcd ( a k , a k 1 ) , a_{k+1}= \dfrac{15}{2} \gcd (a_k, a_{k-1}), and a k 1 a k + 1 3 = 5 2 gcd ( a k , a k 1 ) , a_{k-1} \leq \dfrac{a_{k+1}}{3} = \dfrac{5}{2} \gcd (a_k, a_{k-1}), so a k 1 a_{k-1} is either gcd ( a k , a k 1 ) \gcd (a_k, a_{k-1}) or 2 gcd ( a k , a k 1 ) . 2 \gcd (a_k, a_{k-1}). Let a k 1 = i gcd ( a k 2 , a k 3 ) . a_{k-1} = i \cdot \gcd (a_{k-2}, a_{k-3}). Note that a k 1 a_{k-1} divides 2 gcd ( a k , a k 1 ) , 2 \gcd (a_k, a_{k-1}), so 2 gcd ( a k , a k 1 ) = g gcd ( a k 2 , a k 3 ) 2 \gcd (a_k, a_{k-1}) = g \cdot \gcd (a_{k-2}, a_{k-3}) for some g N . g \in \mathbb{N}. Once again, we do casework on g . g. We can similarly reject the cases g = 1 , 2. g= 1,2.

  • Sub-sub case 1: g 5 g \geq 5

Then, a k + 1 75 4 gcd ( a k 1 , a k 2 ) > 75 4 2 k 4 = 75 2 k 6 > 64 2 k 6 = 2 k , a_{k+1} \geq \dfrac{75}{4} \gcd (a_{k-1}, a_{k-2}) > \dfrac{75}{4} \cdot 2^{k-4} = 75 \cdot 2^{k-6} > 64 \cdot 2^{k-6} = 2^k, contradiction.

  • Sub-sub case 2: 3 g 4 3 \leq g \leq 4

Then, similar to subcase 2, a k 2 = gcd ( a k 1 , a k 2 ) , a_{k-2} = \gcd (a_{k-1}, a_{k-2}), and a k + 1 3 × 15 4 a k 2 > 45 2 k 4 > 16 2 k 4 = 2 k , a_{k+1} \geq 3 \times \dfrac{15}{4} a_{k-2} > 45 \cdot 2^{k-4} > 16 \cdot 2^{k-4} = 2^k, contradiction.

Now that we have covered all cases, our inductive step is complete. \blacksquare


We then have a n 2 n 1 a_n \geq 2^{n-1} for all n N . n \in \mathbb{N}. The sequence { 2 i 1 } i = 1 \{2^{i-1}\}_{i=1}^{\infty} obviously satisfies the given condition, since gcd ( 2 i , 2 i + 1 ) = 2 i > 2 i 1 i N . \gcd (2^i, 2^{i+1}) = 2^i > 2^{i-1} \quad \forall \ i \in \mathbb{N}. Thus, the minimum possible value of a 2014 a_{2014} is 2 2013 , 2^{2013}, so N = 2 2013 2 N = 2 2014 . N= 2^{2013} \implies 2N = 2^{2014}. Evaluating this modulo 1000 , 1000, we find out that the last three digits of 2 N 2N are 384 . \boxed{384}.


This problem is taken from ISL 2008 N3 .

Fantastic problem with beautiful solution.Thanks for posting it. Upvoted & liked!!

rajdeep brahma - 4 years, 1 month ago
Finn Hulse
Apr 19, 2014

FANTASTIC PROBLEM! I found this sequence to satisfy the given conditions: 1 , 2 , 4 , 8 , 2 n 1 1, 2, 4, 8, \dots 2^{n-1} where a n = 2 n 1 a_n=2^{n-1} . This is obviously the smallest possible sequence, another example would be 1 , 3 , 9 , 27 , 81 , 1, 3, 9, 27, 81, \dots . Since the sequence has to have the least possible common ratio, and the least integral first term, the first sequence ( 1 , 2 , 4 , 8 , 1, 2, 4, 8, \dots ) is indeed the smallest. Thus we want to calculate 2 × 2 2014 1 2 × 2 2013 2 2014 2 \times 2^{2014-1} \Longrightarrow 2 \times 2^{2013} \Longrightarrow 2^{2014} . Now, evaluating 2 2014 ( m o d 1000 ) 2^{2014} \pmod{1000} , we get 384 \boxed{384} .

Your solution is not rigorous enough. I agree that the sequence { 2 i 1 } i = 1 \{2^{i-1}\}_{i=1}^{\infty} is indeed the minimal sequence, but you haven't proven it. The result is sort of intuitive, though.

Hint: Use strong induction. The base cases n = 1 , 2 n= 1,2 are easy to handle, and the inductive step is just a bunch of casework.

I'll be posting my full solution shortly.

Sreejato Bhattacharya - 7 years, 1 month ago

Log in to reply

Okay. Yeah tomorrow I'll go trough and do a rigorous proof. :D

Finn Hulse - 7 years, 1 month ago

Hey, Streejato Bhattacharya, there's a typo you will hate harshly. At the beginning of the inductive step, we have <p> a k + 1 > a k + a k 1 2 k 1 + 2 k 2 = 3 2 k 2 a_{k+1} > a_k+a_{k-1} \ge 2^{k-1} + 2^{k-2} = 3 \cdot 2^{k-2} </p> and not 3 2 k 2 / 2 3 \cdot 2^{k-2} /2 as you wrote. This ends easily the problem, because j 4 / 3 j \le 4/3 reduce the cases to the first. Am I wrong?

Andrea Marino - 7 years, 1 month ago

Log in to reply

Thanks for pointing the typo. However, this does does not end the problem, since 3 2 k 2 < 4 2 k 2 = 2 k . 3 \cdot 2^{k-2} < 4 \cdot 2^{k-2}= 2^k.

Sreejato Bhattacharya - 7 years, 1 month ago

i by mistake took the 2014th term to be 2^2014

A Former Brilliant Member - 7 years, 1 month ago

Log in to reply

You still had two tries left, though. ;)

Sreejato Bhattacharya - 7 years, 1 month ago

I did the same mistake at first. Later I got it correct. :)

Fahim Shahriar Shakkhor - 7 years, 1 month ago

Yeah, that sucks. :O

Finn Hulse - 7 years, 1 month ago

@Calvin Lin Now that this is a Level 5 NT, can I cash in my points for answering it right as soon as it was posted but before it got a rating? Thanks. :D

Finn Hulse - 7 years, 1 month ago

Good intuition but a rigorous proof would have been better.

rajdeep brahma - 4 years, 1 month ago

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...