3
$\begingroup$

I've been given the following puzzle

Let $a_1, a_{100}$ be given real numbers. Let $a_i=a_{i-1}a_{i+1}$ for $2\leq i \leq 99$. Further suppose that the product of the first $50$ is $27$, and the product of all the $100$ numbers is also $27$.

Find $a_1+a_2$.

I tried the following, looking at the sequence for a moment we see: $$ a_2=a_1\, a_3\\ a_3=a_2\,a_4\\ \vdots\\ a_{99}=a_{98}\,a_{100} $$ So $$a_2=\frac 1 {a_{99}} \prod_i a_i =\frac {27}{a_{99}}$$

Looking at the other elements, we find that $$a_3=\frac {a_2} {a_1}, a_4=\frac {a_2}{a_1 a_2}, a_5=\frac {a_2}{a_1a_2a_3},\dots , a_n=\frac {a_2}{\prod_{i=1}^{n-2} a_i}$$

Then, $$27=\prod_i a_i=\prod_{1\leq i\leq 100} \frac {a_2}{\prod_{k=1}^{i-1}a_i}$$

The last product I believe is much more complicated than what I should have gotten...

Has anyone ran into this puzzle before?

|cite|improve this question
$\endgroup$
  • 2
    $\begingroup$ Note $a_4=1/a_1$, $a_7=a_1$, .... I Think there will be more reductions if you always consider 3 consecutive equations. There is some kind of oscillation between $a_1$ and $1/a_1$. $\endgroup$ – MrYouMath Jun 1 '16 at 21:26
  • $\begingroup$ @MrYouMath Yes, I noted it, how does that help? $\endgroup$ – YoTengoUnLCD Jun 1 '16 at 21:28
  • $\begingroup$ You can also group other pairs of numbers and try to detect other patterns. $\endgroup$ – MrYouMath Jun 1 '16 at 21:35
8
$\begingroup$

Note that since $a_i=a_{i-1}a_{i+1}$, then, for $i\geqslant 3$, we also have $a_{i-1}=a_{i-2}a_i$. Plug this in the first formula, we have $a_i=a_{i-2}a_ia_{i+1}\Rightarrow a_{i-2}a_{i+1}=1\Rightarrow a_{i+3}=\frac{1}{a_i}.$

Then, the sequence is in the form of $a_1=a,a_2=ab,a_3=b,a_4=\frac{1}{a},a_5=\frac{1}{ab},a_6=\frac{1}{b},a_7=a,\dots$.

$a_{i+6}=a_i$. So $a_{50}=ab$ and $a_{100}=\frac{1}{a}$. Hence, $1=\prod_{i=51}^{100}=\frac{b}{a}\Rightarrow a=b$. Also, $27=\prod_{i=1}^{50}=a\cdot (ab)=a^3\Rightarrow a=3\Rightarrow a_1+a_2=3+9=12$.

|cite|improve this answer
$\endgroup$
  • $\begingroup$ To get a complete answer out of this, try considering the sequence $c_1=a_1,$ $c_2 = a_1a_2,$ $c_3=a_1 a_2 a_3$ etc... $\endgroup$ – Semiclassical Jun 1 '16 at 21:37
5
$\begingroup$

Such a sequence is periodic with period length 6, one cycle $$ A, B, \frac{B}{A}, \frac{1}{A}, \frac{1}{B}, \frac{A}{B} $$ Note that the product of all six is $1.$ Therefore the product of the first 96 is one, and 27 is $\frac{B^2}{A}.$ The product of the first 48 is one, and 27 is also $AB.$ That is $$ 27 A = B^2, $$ $$ 27 = AB.$$ From $$ 729 = 27AB = (27A)B = B^3$$ I get $729 = B^3 $ and $$ B = 9. $$ Then $$ A = 3. $$ The sum is $$ A + B =12. $$ The cycle of length six becomes $$ 3, \; 9, \; 3, \; \frac{1}{3}, \; \frac{1}{9}, \; \frac{1}{3}. $$ Putting two together shows $$ 3, \; 9, \; 3, \; \frac{1}{3}, \; \frac{1}{9}, \; \frac{1}{3}, \; 3, \; 9, \; 3, \; \frac{1}{3}, \; \frac{1}{9}, \; \frac{1}{3} $$

In the original notation, with cycles of six multiplying to one, we have $$ a_{49} a_{50} = a_1 a_2 = 3 \cdot 9 = 27. $$ Also $$ a_{97} a_{98} a_{99} a_{100} = a_1 a_2 a_3 a_4 = 3 \cdot 9 \cdot 3 \cdot \frac{1}{3} = 27. $$

|cite|improve this answer
$\endgroup$
  • $\begingroup$ This is very classy! Is there some kind of theory encompassing sequences like this? There's something similar in notes by Fomin and Reading, a sequence defined by $f_{n+1} = \dfrac{f_n + 1}{f_{n - 1}}$ that has period five; I'm just curious if these kinds of things fit into any particular genre, or are curiosities. $\endgroup$ – pjs36 Jun 1 '16 at 22:10
  • $\begingroup$ @pjs36 I got suspicious after taking logarithms $b_i = \log a_i$ and having the characteristic roots for the $b_i$ relation being cube roots of unity. When I simply wrote out the first few terms, beginning with unknown $A,B$ it fell into place. Indeed, the sequence is something similar to $$ 3^{2 \sin \left( \frac{(2n-1) \pi}{6} \right)} $$ $\endgroup$ – Will Jagy Jun 1 '16 at 22:29
2
$\begingroup$

Since the product of all the numbers is nonzero, the numbers are all nonzero, which means we can write them as $a_i=3^{b_i}$ (where $b_i$ is potentially complex, although we'll see it never is). The recursion translates to $b_{i+1}=b_i-b_{i-1}$, which in turn implies $b_{i+3}=-b_i$ for $i=1,2,\ldots,97$, so that the sequence of $100$ $b_i$'s is just

$$b_1,b_2,b_3,-b_1,-b_2,-b_3,b_1,b_2,b_3,\ldots,b_1,b_2,b_3,-b_1$$

The product of the first $50$ $a$'s corresponds to the sum of the first $50$ $b$'s, which, after cancellations, is just $b_1+b_2$, so we have

$$b_1+b_2=3$$

Similarly, the sum of all $100$ $b$'s reduces to $b_2+b_3$, so we have

$$b_2+b_3=3$$

Putting this together with $b_3=b_2-b_1$, we find $b_1=1$ and $b_2=2$, which means

$$a_1+a_2=3^1+3^2=12$$

|cite|improve this answer
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.