Let N be the sum of all positive integers q of the form q = p k with prime p , such that for at least four different integer values of x from 1 to q ,
x 3 − 3 x ≡ 1 2 3 ( m o d q ) .
What are the last 3 digits of N ?
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.
Denote f ( x ) = x 3 − 3 x − 1 2 3 . Because d e g ( f ) = 3 , if q is a prime, then f can have no more than three roots in the field of residues modulo q . So q = p k for k ≥ 2 .
The main step in the argument, that allows to rule out most possibilities for p , is the following lemma, which is a particular case of the famous Hensel's Lemma.
Lemma. Suppose p is a prime, different from 3 , 5 , and 1 1 . Suppose x is an integer, such that f ( x ) ≡ 0 ( m o d p k ) , for some k ≥ 1 . Then there is exactly one residue y modulo p k + 1 , congruent to x modulo p k , such that f ( y ) ≡ 0 ( m o d p k + 1 ) .
Proof. Consider all residues y modulo p k + 1 that are congruent to x modulo p k . Each such y can be written as x + p k z , where z is an integer; the class of y modulo p k + 1 is uniquely determined by the class of z modulo p . Plugging into f ( y ) ≡ 0 ( m o d p k + 1 ) congruence that we need to solve, we get ( x + p k z ) 3 − 3 ( x + p k z ) − 1 2 3 ≡ 0 ( m o d p k + 1 ) , which is equivalent to ( x 3 − 3 x − 1 2 3 ) + 3 x 2 p k z + 3 x p 2 k z 2 + p 3 k z 3 − 3 x p k z ≡ 0 ( m o d p k + 1 ) Suppose x 3 − 3 x − 1 2 3 = p k ⋅ v . Because k ≥ 1 , 2 k and 3 k are greater than or equal to k + 1 , so the above can be rewritten as p k ⋅ v + ( 3 x 2 − 3 ) p k z ≡ 0 ( m o d p k + 1 ) Dividing by p k , we get 3 ( x 2 − 1 ) z ≡ v ( m o d p ) If 3 ( x 2 − 1 ) is not divisible by p , this congruence has a unique solution modulo p , and we are done. If p ∣ 3 ( x 2 − 1 ) , then either p = 3 or x ≡ ± 1 ( m o d p ) . Finally, if f ( 1 ) ≡ 0 ( m o d p ) , then p ∣ − 1 2 5 , so p = 5 ; if f ( − 1 ) ≡ 0 ( m o d p ) , then p ∣ − 1 2 1 , so p = 1 1 .
Remark. The above argument also works if p is 5 or 1 1 and x is not 1 or − 1 modulo p . Indeed, in this case 3 ( x 2 − 1 ) = 0 ( m o d p ) .
Suppose for some q = p k , p = 3 , 5 , 1 1 , we have f ( x ) ≡ 0 ( m o d q ) for at least four residues modulo q . Then by the above lemma the same must be true for p k − 1 , p k − 2 , . . . , p 1 = p , which is impossible. Therefore, p must be 3 , 5 , or 1 1 .
Case 1. p = 3 . If f ( x ) ≡ 0 ( m o d 3 ) , then x 3 ≡ 0 ( m o d 3 ) , so x = 3 y for some y . But then f ( x ) = 2 7 y 3 − 9 y − 1 2 3 is never divisible by 9 . So, no q = 3 k can satisfy the condition of the problem.
Case 2. p = 1 1 . If f ( x ) ≡ 0 ( m o d 1 1 ) , then, by direct check, either x ≡ − 1 or x ≡ 2 modulo 1 1 . For x ≡ 2 ( m o d 1 1 ) , the Remark above implies that there is exactly one residue modulo 1 1 k , congruent to 2 modulo 1 1 , such that f ( x ) ≡ 0 ( m o d 1 1 k ) .
If x ≡ − 1 ( m o d 1 1 ) , then x = − 1 + 1 1 y , so
f ( x ) = ( − 1 + 1 1 y ) 3 − 3 ( − 1 + 1 1 y ) − 1 2 3 = − 1 2 1 − 3 ⋅ 1 1 2 y 2 + 1 1 3 y 3 = 1 1 2 ( − 1 − 3 y 2 + 1 1 y 3 )
So for every y = 1 , 2 , . . . , 1 1 , we get x such that f ( x ) is divisible by 1 2 1 , thus q = 1 2 1 works. On the other hand, 1 1 2 ( − 1 − 3 y 2 + 1 1 y 3 ) ≡ 0 ( m o d 1 1 3 ) if and only if − 1 − 3 y 2 ≡ 0 ( m o d 1 1 ) , which never happens. So for any k ≥ 3 there are no residues modulo 1 1 k , congruent to − 1 modulo 1 1 , such that f ( x ) ≡ 0 ( m o d 1 1 k ) . Together with the sub-case x ≡ 2 ( m o d 1 1 ) , considered above, we conclude that for p = 1 1 there is one solution: q = 1 2 1 .
Case 3. p = 5 . If f ( x ) ≡ 0 ( m o d 5 ) , then, by direct check, either x ≡ 1 or x ≡ 3 modulo 5 . For x ≡ 3 ( m o d 5 ) , the Remark above implies that there is exactly one residue modulo 5 k , congruent to 3 modulo 5 , such that f ( x ) ≡ 0 ( m o d 5 k ) .
If x ≡ 1 ( m o d 5 ) , then x = 1 + 5 y , so f ( x ) = ( 1 + 5 y ) 3 − 3 ( 1 + 5 y ) − 1 2 3 = − 1 2 5 + 3 ⋅ 2 5 y 2 + 1 2 5 y 3 = 2 5 ( − 5 + 3 y 2 + 5 y 3 )
So for every y = 0 , 1 , 2 , 3 , 4 we get x such that f ( x ) is divisible by 2 5 , thus q = 2 5 works.
Note that 2 5 ( − 5 + 3 y 2 + 5 y 3 ) ≡ 0 ( m o d 5 3 ) if and only if y ≡ 0 ( m o d 5 ) , that is y = 5 z . In this case x = 1 + 2 5 z and f ( x ) = 1 2 5 ( − 1 + 1 5 z 2 + 1 2 5 z 3 ) . This gives 5 different residues x modulo 5 3 , so q = 1 2 5 also works. On the other hand, f ( x ) is never divisible by 5 4 for x ≡ 1 ( m o d 5 ) . So for any k ≥ 4 there is only one residue x modulo 5 k , such that f ( x ) ≡ 0 ( m o d 5 k ) , with x ≡ 3 ( m o d 5 ) .
Putting this all together, q is 1 2 1 , 2 5 , or 1 2 5 , so the answer is 1 2 1 + 2 5 + 1 2 5 = 2 7 1 .
Problem Loading...
Note Loading...
Set Loading...
The congruence is equivalent to x 3 − 3 x − 1 2 3 = 0 ( m o d q ) . Let f ( x ) = x 3 − 3 x − 1 2 3 .
By Lagrange's Theorem, we know that f ( x ) ≡ 0 ( m o d p ) has at most 3 roots. So we want k > 1 . By Hensel's lifting lemma, every root r of f ( x ) ≡ 0 ( m o d p k − 1 ) will lift to exactly one unique solution for f ( x ) ≡ 0 ( m o d p k ) , if f ′ ( r ) ≡ 0 ( m o d p ) . It easily follows by induction that the congruence f ( x ) ≡ 0 ( m o d p k ) has at most 3 roots if f ′ ( r ) ≡ 0 ( m o d p ) for all the roots r of f ( x ) ≡ 0 ( m o d p ) .
Thus, there must exist a root r of f ( x ) ≡ 0 ( m o d p k ) such that f ′ ( r ) ≡ 0 ( m o d p ) . This gives 0 ≡ 3 r 2 − 3 ≡ 3 ( r − 1 ) ( r + 1 ) ( m o d p ) . Thus, we have either 3 ≡ 0 ( m o d p ) or r − 1 ≡ 0 ( m o d p ) or r + 1 ≡ 0 ( m o d p ) .
The first case gives p = 3 .
The second case gives r ≡ 1 ( m o d p ) . Note that f ( r ) ≡ 0 ( m o d p k ) ⇒ f ( r ) ≡ 0 ( m o d p ) , so we have 0 ≡ f ( 1 ) ≡ − 1 2 5 ( m o d p ) . This gives p = 5 as the only solution.
The third case gives r ≡ − 1 ( m o d p ) . Similarly, we have 0 ≡ f ( − 1 ) ≡ − 1 2 1 ( m o d p ) . This gives p = 1 1 as the only solution. Since f ( x ) ≡ 0 ( m o d p k ) ⇒ f ( x ) ≡ 0 ( m o d p k − 1 ) , then for each p and k , we can use Hensel's lifting lemma to find all the roots of f ( x ) ≡ 0 ( m o d p k ) given the roots of f ( x ) ≡ 0 ( m o d p k − 1 ) .
For p = 3 , f ( x ) ≡ 0 ( m o d 3 k ) ⇒ f ( x ) ≡ 0 ( m o d 3 ) ⇒ x 3 ≡ 0 ( m o d 3 ) ⇒ x ≡ 0 ( m o d 3 ) . But for all integers y , f ( 3 y ) ≡ 2 7 y 3 − 9 y − 1 2 3 ≡ 3 ( m o d 9 ) , so f ( 3 y ) ≡ 0 ( m o d 3 k ) for all k ≥ 2 . Therefore f ( x ) ≡ 0 ( m o d 3 k ) has no solution for all k ≥ 2 . So k = 1 gives only one root, while k ≥ 2 gives no roots. So there is no q such that p = 3 .
[Note: The rest of the proof establishes that q = 2 5 , 1 2 1 , 1 2 5 are the only possible solutions. There are slightly quicker ways of showing this, though it's all based on the same idea. - Calvin]
For p = 5 , 0 ≡ f ( x ) ≡ x 3 − 3 x + 2 ≡ ( x + 2 ) ( x − 1 ) 2 ( m o d 5 ) . So x ≡ 1 , 3 ( m o d 5 ) are the only solutions to f ( x ) ≡ 0 ( m o d 5 ) . Now we consider k = 2 . f ′ ( 3 ) ≡ 0 ( m o d 5 ) , so the root x ≡ 3 ( m o d 5 ) will lift to exactly one solution of f ( x ) ≡ 0 ( m o d 5 ) . Call this solution x 1 . For x ≡ 1 ( m o d 5 ) , since f ′ ( 1 ) ≡ 0 ( m o d 5 ) and f ( 1 ) ≡ 0 ( m o d 2 5 ) , then it will lift to multiple solutions of f ( x ) ≡ 0 ( m o d 2 5 ) , namely x ≡ 1 , 6 , 1 1 , 1 6 , 2 1 ( m o d 5 ) . So the congruence f ( x ) ≡ 0 ( m o d 2 5 ) has 6 solutions. Now, we consider k = 3 . Again, x 1 will lift to exactly one solution of f ( x ) ≡ 0 ( m o d 1 2 5 ) . Call this solution x 2 . For x ≡ 5 j + 1 ( m o d 2 5 ) where j = 0 , … , 4 , we have f ( x ) ≡ ( 5 j + 1 ) 3 − 3 ( 5 j + 1 ) − 1 2 3 ≡ 1 2 5 j 3 + 7 5 j 2 − 1 2 5 ≡ 7 5 j 2 ( m o d 1 2 5 ) . Only roots where f ( x ) ≡ 0 ( m o d 1 2 5 ) will be lifted, so we want 7 5 j 2 ≡ 0 ( m o d 1 2 5 ) ⇒ 3 j 2 ≡ 0 ( m o d 2 5 ) ⇒ j = 0 . So only x ≡ 1 ( m o d 2 5 ) will lift. Specifically, it lifts to the solutions x ≡ 1 , 2 6 , 5 1 , 7 6 , 1 0 1 ( m o d 1 2 5 ) . Thus, the congruence f ( x ) ≡ 0 ( m o d 1 2 5 ) has 6 solutions as well. Now consider k = 4 . Again, x 2 will lift to exactly one solution of f ( x ) ≡ 0 ( m o d 6 2 5 ) . For x ≡ 2 5 j + 1 ( m o d 1 2 5 ) where j = 0 , . . , 4 , we have f ( x ) ≡ ( 2 5 j + 1 ) 3 − 3 ( 2 5 j + 1 ) − 1 2 3 = 1 5 6 2 5 j 3 + 1 8 7 5 j 2 − 1 2 5 ≡ − 1 2 5 ≡ 0 ( m o d 6 2 5 ) . So none of these roots will lift and the congruence f ( x ) ≡ 0 ( m o d 6 2 5 ) only has 1 solution. Since this solution is the one that gives f ′ ( x ) ≡ 0 ( m o d 5 ) , then it will lift to exactly 1 solution for each k ≥ 5 . So for p = 5 , we have k = 1 giving 2 roots, k = 2 , 3 giving 6 roots, and k ≥ 4 giving 1 root. Thus, when p = 5 , q = 2 5 or 1 2 5 .
For p = 1 1 , 0 ≡ f ( x ) ≡ x 3 − 3 x − 2 ≡ ( x − 2 ) ( x + 1 ) 2 ( m o d 1 1 ) . So x ≡ 2 , 1 0 ( m o d 1 1 ) are the only solutions to f ( x ) ≡ 0 ( m o d 1 1 ) . Now we consider k = 2 . f ′ ( 2 ) ≡ 0 ( m o d 1 1 ) , so the root x ≡ 2 ( m o d 1 1 ) will lift to exactly one solution of f ( x ) ≡ 0 ( m o d 1 2 1 ) . Call this solution x 3 . For x ≡ 1 0 ( m o d 1 1 ) , since f ′ ( 1 0 ) ≡ 0 ( m o d 1 1 ) and f ( 1 0 ) ≡ 0 ( m o d 1 2 1 ) , then it will lift to multiple solutions of f ( x ) ≡ 0 ( m o d 1 2 1 ) , namely x ≡ 1 1 j − 1 ( m o d 1 2 1 ) , where j = 1 , … , 1 0 . So the congruence f ( x ) ≡ 0 ( m o d 1 2 1 ) has 1 2 solutions. Now consider k = 3 . Again, x 3 will lift to exactly one solution of f ( x ) ≡ 0 ( m o d 1 3 3 1 ) . For x ≡ 1 1 j − 1 ( m o d 1 2 1 ) where j = 1 , … , 1 1 , we have f ( x ) ≡ ( 1 1 k − 1 ) 3 − 3 ( 1 1 k − 1 ) − 1 2 3 ≡ 1 3 3 1 k 3 − 3 6 3 k 2 − 1 2 1 ≡ − 1 2 1 ( 3 k 2 + 1 ) ( m o d 1 3 3 1 ) . Only roots where f ( x ) ≡ 0 ( m o d 1 3 3 1 ) will be lifted, so we want − 1 2 1 ( 3 k 2 + 1 ) ≡ 0 ( m o d 1 3 3 1 ) ⇒ 3 k 2 + 1 ≡ 0 ( m o d 1 1 ) . But it is easy to check that k = 1 , . . . , 1 0 does not satisfy this congruence, so none of these roots will lift. Therefore, the congruence f ( x ) ≡ 0 ( m o d 1 3 3 1 ) only has 1 solution. Since this solution is the one that gives f ′ ( x ) ≡ 0 ( m o d 1 1 ) , then it will lift to exactly 1 solution for each k ≥ 4 . So for p = 1 1 , we have k = 1 giving 2 roots, k = 2 giving 1 2 roots, and k ≥ 3 giving 1 root. Thus, when p = 1 1 , q = 1 2 1 .
We have now exhausted all the cases, so the only possible q are 2 5 , 1 2 1 and 1 2 5 . Their sum is 2 7 1 , so N = 2 7 1 .