This sequence looks familiar

Algebra Level 5

Let a 1 , a 2 , a 3 , a_1, a_2, a_3, \ldots be a sequence of non-negative integers defined as follows: { a 1 = 0 a k + 1 = max i = 1 , 2 , , k { i + a i + a k + 1 i } for k 1. \begin{cases} \begin{aligned} a_1 &= 0 \\ a_{k+1} &= \displaystyle \max_{i=1,2,\ldots,k} \left \{i + a_i + a_{k+1-i} \right \} \ \text{for }k \ge 1.\end{aligned} \end{cases} Find the value of a 200 a_{200} .


The answer is 19900.

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

Sharky Kesa
Feb 11, 2017

Relevant wiki: Strong Induction

Exploring this sequence, you would find that this sequence is actually just triangular numbers. Let's try and prove that a n = 0 + 1 + 2 + + ( n 1 ) a_n = 0 + 1 + 2 + \ldots + (n-1) via strong induction.

The base case, n = 1 n=1 is already given.

For the inductive part, assume a n = 0 + 1 + 2 + 3 + + ( n 1 ) a_n = 0 + 1 + 2 + 3 + \cdots + (n-1) for n k n \leq k . We know

a k + 1 = max i = 1 , 2 , , k { i + a i + a k + 1 i } ( 1 ) a_{k+1} = \displaystyle \max_{i=1,2,\ldots,k} \{i + a_i + a_{k+1-i}\} \quad (1)

We claim that the max in ( 1 ) (1) occurs when i = k i=k . For this, it is sufficient to prove that whenever i < k i < k , we have

k + a k + a 1 > i + a i + a k + 1 i k + a_k + a_1 > i + a_i + a_{k+1-i}

But since a 1 = 0 a_1 = 0 and k > i k>i , it suffices to show

a k a i a k + i 1 a_k - a_i \geq a_{k+i-1}

Using the inductive assumption, we have

a k a i = ( 0 + 1 + 2 + + ( k 1 ) ) ( 0 + 1 + 2 + + ( i 1 ) ) = i + ( i + 1 ) + + ( k 1 ) ( 2 ) \begin{aligned} a_k - a_i &= (0 + 1 + 2 + \ldots + (k-1)) - (0 + 1 + 2 + \ldots + (i-1))\\ &= i + (i+1) + \cdots + (k-1) \quad (2)\\ \end{aligned}

and

a k + 1 i = 0 + 1 + 2 + + ( k i ) = 1 + 2 + + ( k i ) ( 3 ) \begin{aligned} a_{k+1-i} &= 0 + 1 + 2 + \cdots + (k-i)\\ &= 1+ 2+ \cdots + (k-i) \quad (3)\\ \end{aligned}

Observe that both RHS of ( 2 ) (2) and ( 3 ) (3) consist of k i k-i consecutive terms. Since the first term in ( 2 ) (2) must be at least as big as ( 3 ) (3) , our claim must hold true.

Therefore, we have

a k + 1 = max i = 1 , 2 , , k { i + a i + a k + 1 i } = k + a k + a 1 = k + ( 0 + 1 + 2 + + ( k 1 ) ) + 0 = 0 + 1 + 2 + + k \begin{aligned} a_{k+1} &= \displaystyle \max_{i=1,2,\ldots,k} \{i + a_i + a_{k+1-i}\}\\ &= k + a_k + a_1\\ &= k + (0 + 1 + 2 + \ldots + (k-1)) + 0\\ &= 0 + 1 + 2 + \cdots + k\\ \end{aligned}

This completes our induction.

Now we can apply the well-known triangular numbers formula a n = n ( n 1 ) 2 a_n = \frac {n(n-1)}{2} to get

a 200 = 200 × 199 2 = 19900 . a_{200} = \dfrac{200 \times 199}{2} = \boxed{19900}.

The induction is a little easier than this, since ( i + ( i 2 ) + ( k + 1 i 2 ) = ( k + 1 2 ) + i ( i k ) 1 i k i + \binom{i}{2} + \binom{k+1-i}{2} \; = \; \binom{k+1}{2} + i(i-k) \hspace{1cm} 1 \le i \le k and so the maximum of m a x 1 i k ( i + a i + a k + 1 i ) \mathrm{max}_{1 \le i \le k}\big(i + a_i + a_{k+1-i}\big) is always achieved for i = k i=k .

Mark Hennings - 4 years, 3 months ago

Log in to reply

I tried to keep the proof as simple as possible. :P

However, you are still correct.

Sharky Kesa - 4 years, 3 months ago

Log in to reply

What motivates you to start with STRONG induction? Obviously Mark Henning's approach is (always) the best, but why didn't you use "normal induction" instead? Or is "STRONG induction" a more systematic way to prove it?

Pi Han Goh - 4 years, 3 months ago

Log in to reply

@Pi Han Goh I am using strong induction too, since I need to assume that a j = 1 2 j ( j 1 ) a_j=\tfrac12j(j-1) for 1 j k 1\le j\le k to use my formula (which is basically the same as Arjen's.

My solutions are not always the best...

Mark Hennings - 4 years, 3 months ago

Log in to reply

@Mark Hennings Wait, Arjen's solution IS strong induction? I must have misread it. Let me read it up again....

Pi Han Goh - 4 years, 3 months ago
Arjen Vreugdenhil
Feb 15, 2017

Claim : a k = 1 2 k ( k 1 ) a_k = \tfrac12 k(k-1) .

Proof : The claim is obviously true for k = 1 k = 1 .

Suppose the claim is true for all k n k \leq n . Let 1 i n 1 \leq i \leq n , and z = i 1 2 ( n + 1 ) z = i - \tfrac12(n+1) . Then i + a i + a n + 1 i = 1 2 ( n + 1 ) + z + 1 2 ( 1 2 ( n + 1 ) z ) ( 1 2 ( n + 1 ) z 1 ) + 1 2 ( 1 2 ( n + 1 ) + z ) ( 1 2 ( n + 1 ) + z 1 ) = 1 4 n 2 + 1 2 n + ( z + 1 2 ) 2 , i + a_i + a_{n+1-i} = \tfrac12(n+1) + z + \tfrac12 (\tfrac12(n+1) - z)(\tfrac12(n+1) - z - 1) + \tfrac12 (\tfrac12(n+1) + z)(\tfrac12(n+1) + z - 1) \\ = \tfrac14n^2 + \tfrac12n + (z+\tfrac12)^2, which is maximal when z + 1 2 |z + \tfrac12| is greatest; this happens for i = n i = n , with z = 1 2 ( n 1 ) z = \tfrac12(n-1) .

The recursion reduces to a n + 1 = 1 4 n 2 + 1 2 n + ( 1 2 n ) 2 = 1 2 n ( n + 1 ) , a_{n+1} = \tfrac14n^2 + \tfrac12n + (\tfrac12n)^2 = \tfrac12n(n+1), showing that the claim also holds for k = n + 1 k = n+1 . This finishes the induction.

Oh this is much simpler than the author's solution. And I thought I need to force myself to apply a "harder induction version" to solve this problem. I wonder:

Is it true that we can solve all "strong induction problems" using "standard induction" alone?

Pi Han Goh - 4 years, 3 months ago

Log in to reply

I did use "strong induction", though, by applying the assumption to all a i a_i and a n + 1 i a_{n+1-i} for i k n i \leq k \leq n .

There is no real formal difference. Proving " P ( n ) P(n) " with strong induction is the same as proving " k n , P ( k ) \forall k \leq n,\ P(k) " with standard induction.

Arjen Vreugdenhil - 4 years, 3 months ago

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...