Arithmetic Frobenius

I have an unlimited supply of 17-cent, 18-cent, and 19-cent stamps. I'd like to put exactly N N cents' worth of postage on an envelope, but no matter which combination of stamps I try, I am unable to accomplish this.

What is the largest possible value of N N ?


Bonus 1: Generalize to any three consecutive integers.
Bonus 2: Generalize to any three-term arithmetic progression .
Bonus 3: Generalize to any arithmetic progression.


The answer is 135.

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.

1 solution

Patrick Corn
Jul 25, 2016

Let's do Bonus 1. The number we're looking for is called the Frobenius number g ( n , n + 1 , n + 2 ) g(n,n+1,n+2) . The claim is that this number equals n n / 2 1 , n\lfloor n/2 \rfloor - 1, so that for n = 17 n=17 the answer is 17 8 1 = 135 . 17 \cdot 8 - 1 = \fbox{135}.

To prove this, note that it is possible to find a sequence of n n representable numbers starting at n n / 2 , n \lfloor n/2 \rfloor, as follows. Each of the given 3-tuples ( a , b , c ) (a,b,c) represents n a + ( n + 1 ) b + ( n + 2 ) c . n \cdot a + (n+1) \cdot b + (n+2) \cdot c. n n / 2 = ( n / 2 , 0 , 0 ) n n / 2 + 1 = ( n / 2 1 , 1 , 0 ) n n / 2 + 2 = ( n / 2 1 , 0 , 1 ) n n / 2 + 3 = ( n / 2 2 , 1 , 1 ) n n / 2 + n 1 = { ( 0 , 0 , n / 2 ) if n is odd ( 0 , 1 , n / 2 1 ) if n is even \begin{aligned} n\lfloor n/2 \rfloor &= (\lfloor n/2 \rfloor,0,0) \\ n\lfloor n/2 \rfloor +1 &= (\lfloor n/2 \rfloor - 1,1,0) \\ n\lfloor n/2 \rfloor + 2 &= (\lfloor n/2 \rfloor - 1,0,1) \\ n\lfloor n/2 \rfloor + 3 &= (\lfloor n/2 \rfloor - 2,1,1) \\ &\vdots \\ n\lfloor n/2 \rfloor + n-1 &= \begin{cases} (0,0,\lfloor n/2 \rfloor) & \text{ if } n \text{ is odd} \\ (0,1,\lfloor n/2\rfloor - 1) & \text{ if } n \text{ is even} \end{cases} \end{aligned} and so we can represent any number m m larger than these by noting that m n q m-nq is one of the numbers in this range, so we can represent m n q m-nq and then add q q to the first coordinate of the representation to get a representation for m . m.

So this shows that g ( n , n + 1 , n + 2 ) n n / 2 1. g(n,n+1,n+2) \le n\lfloor n/2 \rfloor - 1. The proof concludes by showing that n n / 2 1 n\lfloor n/2 \rfloor - 1 is not representable. To see this, consider the equation modulo n . n. Note that n n / 2 1 n\lfloor n/2 \rfloor - 1 is congruent to n 1 n-1 mod n . n. And a number represented by a 3-tuple ( a , b , c ) (a,b,c) is congruent to b + 2 c b+2c mod n . n. Also note that for a representation ( a , b , c ) (a,b,c) of n n / 2 1 , n\lfloor n/2 \rfloor -1, we would have a + b + c < n / 2 . a+b+c < \lfloor n/2 \rfloor. So b + 2 c b+2c is a nonnegative integer which is (because of this last restriction) 2 ( n / 2 1 ) . \le 2(\lfloor n/2 \rfloor - 1). This is strictly smaller than n 1 , n-1, so there is no way to represent a number that small which is congruent to n 1 n-1 mod n . n.

(If this is confusing, it might help to consider the example n = 17. n=17. The point is that 135 135 is one less than a multiple of 17 , 17, and a representation ( a , b , c ) (a,b,c) of 135 135 would have to have a + b + c < 8 , a+b+c < 8, but there is no way to have b + 2 c = 16 b+2c=16 if a + b + c < 8 ; a+b+c<8; the best we could do would be to take a = b = 0 a=b=0 and c = 7 c=7 which would give b + 2 c = 14. b+2c=14. )

For Bonus 2 and 3, there is a general formula that is proved in much the same way as the above. I leave the details to you. :) g ( n , n + d , , n + s d ) = ( n 2 s + 1 ) n + ( n 1 ) ( d 1 ) 1. g(n,n+d,\ldots,n+sd) = \left( \left\lfloor \frac{n-2}{s} \right\rfloor +1\right) n + (n-1)(d-1) - 1.

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...