Call a non-empty set V of non-zero integers victorious if there exists a polynomial P ( x ) with integer coefficients such that P ( 0 ) = 3 3 0 and P ( v ) = 2 ∣ v ∣ for all elements v ∈ V .
How many victorious sets exist?
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.
Great proof. Wanted to mention that in the second half of the proof, the products should read ∏ v ∈ V + rather than ∏ x ∈ V + .
Wow, great proof and great problem!!
A very nice proof -- I learned a lot. Thank you very much, Mark.
After I finally got my solution, I was excited to see if there was a better one. Unfortunately your solution is almost exactly the same as mine! At least there's no need for me to write it up. Very nice.
I wonder: is there a general formula for the quantity 1 0 3 that appears in the proof? Or maybe just in the special case of products of distinct primes? It should only depend on the number of distinct primes in that case...
Log in to reply
Well if N is square-free, being a product of n distinct primes, then the number of positive integer N -victorious sets is 2 k = 1 ∑ n ( k n ) B k + 1 where B k is the k th Bell number. To see this, if N = p 1 p 2 . . . p n , then for any subcollection of k of the prime factors of N there are B k partitions, and each set V obtained by multiplying together the primes in each component of the partition gives us an N -victorious set V that does not contain 1 . We can double the number of N -victorious sets by adding 1 to each set. Since there is also the singleton N -victorious set { 1 } , we obtain the above formula.
Life would be more complicated for a number N that is not square-free.
Problem Loading...
Note Loading...
Set Loading...
If V is a victorious set containing positive integers only, and if P ( x ) is a polynomial with the defining property for V , then we can find a polynomial A ( x ) with integer coefficients such that P ( x ) − 2 x = A ( x ) v ∈ V ∏ ( x − v ) and we must have (putting x = 0 ) 3 3 0 = A ( 0 ) v ∈ V ∏ ( − v ) which means that ∏ v ∈ V v must divide 3 3 0 . Thus V must be a subset of the set V = { 1 , 2 , 3 , 5 , 6 , 1 0 , 1 1 , 1 5 , 2 2 , 3 0 , 3 3 , 5 0 , 6 6 , 1 1 0 , 1 6 5 , 3 3 0 } of positive divisors of 3 3 0 , and the product of all the elements of V must also divide 3 3 0 . Without listing them all, there are 1 0 3 such sets. Thus there are 1 0 3 victorious sets consisting of positive integers only, and (by symmetry) there are 1 0 3 victorious sets consisting of negative integers only.
Suppose that V is a victorious set that contains both positive and negative integers, and let P ( x ) be its associated polynomial. Let V + and V − be the positive and negative elements of V respectively. It follows that we can find polynomials A ( x ) , B ( x ) such that P ( x ) − 2 x = A ( x ) v ∈ V + ∏ ( x − v ) P ( x ) + 2 x = B ( x ) v ∈ V − ∏ ( x − v ) and hence 2 P ( x ) = A ( x ) v ∈ V + ∏ ( x − v ) + B ( x ) v ∈ V − ∏ ( x − v ) If w ∈ V − then − 4 w = A ( w ) v ∈ V + ∏ ( w − v ) so that ∏ v ∈ V + ( w − v ) divides 4 w for any w ∈ V − . Similarly, ∏ v ∈ V − ( w − v ) divides 4 w for any w ∈ V + . Checking the possibilities, it follows that there are only three types of victorious set V that contain both positive and negative integers: { v , − v } , { v , − 3 v } and { 3 v , − v } .
Thus there are 1 0 3 + 1 0 3 + 3 + 4 + 4 = 2 1 7 different victorious sets.