There was this dog named Rita

Consider the equation

1 a + 1 b + 1 c = 1 d \dfrac{1}{a} + \dfrac{1}{b} + \dfrac{1}{c} = \dfrac{1}{d}

where a , b , c , d a, b, c, d are distinct positive integers, at least 3 3 of which are prime.

Up to a permutation of the variables, there is a unique solution to this problem. Find a + b + c + d a + b + c + d .


The answer is 54.

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.

4 solutions

Michael Mendrin
Sep 13, 2014

Yes. Finding the numbers a , b , c , d = 3 , 7 , 42 , 2 {a,b,c,d}={3,7,42,2} wasn't too difficult, but proving that this is unique is really hard.

Yeah, 'existence and uniqueness' proofs have always been my favorites. :P

Brian Charlesworth - 6 years, 9 months ago

Log in to reply

It's not too hard.

First, show that a , b , c a, b, c can't all be prime. Suppose not, then we have d ( a b + b c + c a ) = a b c d ( ab + bc + ca) = abc . Taking mod a a , we get d b c 0 ( m o d a ) dbc \equiv 0 \pmod{a} , and so a d a \mid d . By similarity, a b c d abc \mid d . But 1 a + 1 b + 1 c = 1 a b c d \frac{1}{a} + \frac{1}{b} + \frac{1}{c} = \frac{1}{abcd^*} is absurd.

WLOG, let c c be the non-prime. Clearly, c 1 c \neq 1 , so c c is composite. Let's denote it by x x , and a , b , d a, b, d are all prime. We have

d ( a b + b x + x a ) = a b x d ( ab + bx + xa) = abx

Since LHS is divisible by d d , hence d x d \mid x . Since RHS is divisible by a a , hence a d b x a x a \mid dbx \Rightarrow a \mid x . Since RHS is divisible by b b , hence b d x a b x b \mid dxa \Rightarrow b \mid x . Since a , b , d a,b,d are distinct primes, thus a b d x abd \mid x . Let x = a b d x x = abdx^* .

This gives us d ( a b + b a b d x + a a b d x ) = a b a b d x d(ab + babdx^* + aabdx^*) = ababdx^* . Factoring out a b d abd , we get 1 + d b x + d a x = a b x 1 + dbx^* + dax^* = abx^* . Taking modulus x x^* , we conclude that x = 1 x^* = 1 , so x = a b d x = abd .

Now, we have 1 + d b + d a = a b 1 + db + da = ab , or that ( a d ) ( b d ) = d 2 + 1 (a-d)(b-d) = d^2 + 1 .
Case 1: If d = 2 d = 2 , then we have ( a 2 ) ( b 2 ) = 5 (a-2)(b-2) = 5 , which gives us the solution { a , b } = { 3 , 7 } \{ a, b \} = \{3, 7 \} . Check that 2 , 3 , 7 2, 3, 7 are all primes.
Case 2: If d 2 d \neq 2 , then since d d is prime, so d d is odd. Clearly, a a and b b have to be primes larger than d d , so they are also odd. Thus a d a-d and b d b-d are both even, so 4 ( a d ) ( b d ) 4 \mid (a-d)(b-d) . But we know that 4 ∤ d 2 + 1 4 \not \mid d^2 + 1 , hence there are no solutions in this case.

Calvin Lin Staff - 6 years, 9 months ago

Log in to reply

This looks great, Calvin. For the most part it is the same approach as what I had in mind but your presentation is far more concise than mine would have been.

Brian Charlesworth - 6 years, 9 months ago

I just made a program.

Ashu Dablo - 6 years, 8 months ago

Yeah, numbers will be 3,7,42,2

Hint: Consider Sylvester's sequence ....

As for a proof of the uniqueness of the solution, that's the tough part. When I have time I will type that out, but in the meantime, any elegant proofs would be welcome.

Why is the problem named like this?

Bogdan Simeonov - 6 years, 9 months ago

Log in to reply

Maybe Rita's theorem can help us prove this...? (what's wrong with my brain)

Samuraiwarm Tsunayoshi - 6 years, 9 months ago

Log in to reply

Haha. Well, Rita the dog came up with many interesting questions but no theorems that I can recall. :)

Brian Charlesworth - 6 years, 9 months ago

"Rita the dog" is the name of a participant on another math site I used to spend time on. The avatar was a picture of a rather old and sleepy pit bull dog. I think the actual participant was a young man from Mexico, but it was always fun to think that it was the dog who was posing all the interesting questions. :)

Brian Charlesworth - 6 years, 9 months ago

Log in to reply

What site was that ? :) Can you please share it?

Sanjana Nedunchezian - 6 years, 9 months ago

Log in to reply

@Sanjana Nedunchezian The site is Yahoo! Answers. It's not as interesting a site as it used to be; Brilliant is the better site now. :)

Brian Charlesworth - 6 years, 9 months ago

It is easy to see all a,b,c are not primes so b,c,d are primes also, as a factor of bcd so doing case work (5cases) gives the unique solution 3,7,2,42

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...