You are currently on the new version of our website. Access the old version .
MathematicsMathematics
  • Article
  • Open Access

8 June 2022

On Certain Sum Involving Quadratic Residue

and
1
School of Mathematical Sciences, Sunway University, Petaling Jaya 47500, Malaysia
2
Institute of Mathematical Sciences, Universiti Malaya, Kuala Lumpur 50603, Malaysia
*
Author to whom correspondence should be addressed.
These authors contributed equally to this work.

Abstract

Let p be a prime and F p be the set of integers modulo p. Let χ p be a function defined on F p such that χ p ( 0 ) = 0 and for a F p \ { 0 } , set χ p ( a ) = 1 if a is a quadratic residue modulo p and χ p ( a ) = 1 if a is a quadratic non-residue modulo p. Note that χ p ( a ) = a p is indeed the Legendre symbol. The image of χ p in the set of real numbers. In this paper, we consider the following sum x F p χ p ( ( x a 1 ) ( x a 2 ) ( x a t ) ) where a 1 , a 2 , , a t are distinct elements in F p .
MSC:
11P70; 11B99

1. Introduction

Let p be a prime and F p be the set of integers modulo p. Let Q p F p be the set of all quadratic residues modulo p. For A 1 , A 2 , , A l F p , set
A 1 + A 2 + + A l = 1 i l a i : a i A i for all i .
The set Q p is said to have an l-decomposition if
Q p = A 1 + A 2 + + A l .
It is shown by Sárközy [1] that for sufficiently large p, Q p has no 3-decomposition A 1 + A 2 + A 3 = Q p with | A i | 2 for i = 1 , 2 , 3 . In the same paper, he conjectured that for sufficiently large p, Q p has no 2-decomposition A 1 + A 2 = Q p with | A i | 2 for i = 1 , 2 . He also showed that if p is a very large prime and A 1 + A 2 = Q p with | A i | 2 , then,
p 3 ln p | A i | p ln p , for i = 1 , 2 .
The bounds are improved by Shparlinski [2]. In fact, Shparlinski showed that for any prime p, there exist positive constants c 1 and c 2 such that
c 1 p | A i | c 2 p , for i = 1 , 2 .
Later, Shkredov [3] showed that for any prime p,
1 6 + o ( 1 ) p | A i | 3 + o ( 1 ) p , for i = 1 , 2 .
Here, o ( 1 ) can be considered as a function h ( p ) with the property lim p h ( p ) = 0 . Recently, Chen and Yan [4] showed that for any prime p,
7 17 16 p + 1 | A i | 7 + 17 4 p 6.63 , for i = 1 , 2 .
They also showed that for any prime p, Q p has no 3-decomposition A 1 + A 2 + A 3 = Q p with | A i | 2 for i = 1 , 2 , 3 . In their proof, they considered the sum
x F p χ p ( ( x a 1 ) ( x a 2 ) ( x a t ) ) ,
where a 1 , a 2 , , a t are distinct elements in F p . Here, χ p is a function defined on F p such that χ p ( 0 ) = 0 , and for a F p \ { 0 } , χ p ( a ) = 1 if a is a quadratic residue modulo p and χ p ( a ) = 1 if a is a quadratic non-residue modulo p. Note that χ p ( a ) = a p is indeed the Legendre symbol. Note that x y is a quadratic residue modulo p if both x and y are quadratic residue modulo p or both x and y are quadratic non-residue modulo p. Furthermore, x y is a quadratic non-residue modulo p if x is a quadratic residue modulo p and y is quadratic non-residue modulo p. Therefore, χ p ( x y ) = χ p ( x ) χ p ( y ) for all x , y F p \ { 0 } . Since χ p ( 0 ) = 0 , it is not hard to see that χ p ( x y ) = χ p ( x ) χ p ( y ) for all x , y F p . Chen and Yan [4] also conjectured that for distinct elements a 1 , a 2 , a 3 , a 4 ,
x F p χ p ( ( x a 1 ) ( x a 2 ) ( x a 3 ) ( x a 4 ) ) < 2 p .
They claimed that if the above equation is true, then they can show that for any prime p,
1 4 p + O ( 1 ) | A i | 2 p + O ( 1 ) , for i = 1 , 2 .
Here, O ( 1 ) can be considered as a function h ( p ) with the property that there is a positive integer n 0 such that | h ( p ) | c for all p n 0 . Chen and Yan gave the exact values of the sum for 1 t 2 .
In this paper, we consider the sum x F p χ p ( ( x a 1 ) ( x a 2 ) ( x a t ) ) where a 1 , a 2 , , a t are distinct elements in F p . We proved some relations involving the sum (see Theorems 1, 2, 4, 5 and 6). We also give the exact values of the sum for p 3 t p (see Lemmas 2 and 3, and Corollaries 1 and 2).

2. Main Results

Let A F p and A . We set
f p ( A ) = x F p χ p a A ( x a ) .
If A = F p , then, a A ( x a ) = 0 for all x F p . Therefore, f p ( A ) = 0 .
For p = 2 , we have F 2 = { 0 , 1 } . Note that
f 2 ( { 0 } ) = x F 2 χ 2 x = 1 .
Similarly, f 2 ( { 1 } ) = 1 . Hence, the following proposition follows.
Proposition 1.
f 2 ( F p ) = 0 and f 2 ( { 0 } ) = f 2 ( { 1 } ) = 1 .
From now onwards, we shall assume that p is an odd prime. For convenience, we set f p ( ) = p . We shall need the following lemma which is a well-known result.
Lemma 1.
f p ( A ) = 0 , if | A | = 1 ; 1 , if | A | = 2 .
Lemma 2.
If | A | = p , then, f p ( A ) = 0 .
Proof. 
This follows from a A ( x a ) = 0 for all x F p . □
Lemma 3.
If | A | = p 1 , then,
f p ( A ) = 1 , if p 1 mod 4 ; 1 , if p 3 mod 4 .
Proof. 
Let b F p \ A . Now, a A ( x a ) = 0 for all x A . Let r be a primitive element in F p . Note that the set { b a : a A } has p 1 elements and none of them is zero. So,
{ b a : a A } = { r i : 1 i p 1 } .
Since χ p ( r i ) = ( 1 ) i ,
f p ( A ) = x F p χ p a A ( x a ) = χ p a A ( b a ) = χ p 1 i p 1 r i = χ p r p ( p 1 ) 2 = ( 1 ) p ( p 1 ) 2 = ( 1 ) p 1 2 ,
where the last equality follows from the fact that p is odd. Hence, f p ( A ) = 1 if p 1 mod 4 and f p ( A ) = 1 if p 3 mod 4 . □
We introduce the following notation: Let A F p and b F p \ A . We set
f p ( b ; A ) = x F p χ p ( x b ) 2 a A ( x a ) .
Clearly, χ p ( x b ) 2 a A ( x a ) = χ p a A ( x a ) if and only if x b .
Lemma 4.
Let A F p and b F p \ A . Then,
f p ( A ) = f p ( b ; A ) + χ p a A ( b a ) .
Proof. 
Note that
f p ( b ; A ) = x F p χ p ( x b ) 2 a A ( x a ) = x F p \ ( A { b } ) χ p ( x b ) 2 a A ( x a ) = x F p \ ( A { b } ) χ p a A ( x a ) = x F p \ A χ p a A ( x a ) χ p a A ( b a ) .
The lemma follows by noting that
f p ( A ) = x F p χ p a A ( x a ) = x F p \ A χ p a A ( x a ) .
Lemma 5.
Let B F p and 1 | B | p 1 . Then,
a F p \ B f p ( B { a } ) + a B f p ( a ; B \ a ) = 0 .
Proof. 
Let y F p . By Lemma 1,
f p ( { y } ) = a F p χ p a y = 0 .
Note that χ p ( 1 ) = ( 1 ) p 1 2 as 1 is a quadratic residue modulo p if and only if p 1 mod 4 . Therefore,
0 = χ p ( 1 ) a F p χ p a y = a F p χ p y a .
This implies that
0 = χ p b B ( y b ) a F p χ p y a = a F p χ p y a b B ( y b ) = a F p \ B χ p y a b B ( y b ) + a B χ p y a b B ( y b )
Note that for each a F p \ B ,
f p ( B { a } ) = y F p χ p y a b B ( y b ) .
Also, for each a B ,
f p ( a ; B \ a ) = y F p χ p ( y a ) 2 b B \ { a } ( y b ) = y F p χ p y a b B ( y b ) .
The lemma follows from Equation (3), by summing up all the y F p . □
Lemma 6.
Let 1 k p 1 . Then
( k + 1 ) B F p , | B | = k + 1 f p ( B ) + ( p k ) B F p , | B | = k 1 f p ( B ) = 0 .
Proof. 
Consider the sum
A F p , | A | = k a F p \ A f p ( A { a } ) + a A f p ( a ; A \ a ) .
By Lemma 5, the sum in the bracket is zero. Hence, the above sum is zero.
Let B F p with | B | = k + 1 . For each b B , let A ( b ) = B \ { b } . Then | A ( b ) | = k and B = A ( b ) { b } . Thus, f p ( B ) is counted k + 1 times in
A F p , | A | = k a F p \ A f p ( A { a } ) .
so,
A F p , | A | = k a F p \ A f p ( A { a } ) = ( k + 1 ) B F p , | B | = k + 1 f p ( B ) .
On the other hand, let C F p with | C | = k 1 . For each b F p \ C , let A ( b ) = C { b } . Then | A ( b ) | = k . Thus,
A F p , | A | = k a A f p ( a ; A \ a ) = C F p , | C | = k 1 a F p \ C f p ( a ; C ) .
By Lemma 4,
f p ( a ; C ) = f p ( C ) χ p c C ( a c ) .
Therefore,
a F p \ C f p ( a ; C ) = ( p k + 1 ) f p ( C ) a F p \ C χ p c C ( a c ) = ( p k + 1 ) f p ( C ) a F p χ p c C ( a c ) = ( p k + 1 ) f p ( C ) f p ( C ) = ( p k ) f p ( C ) .
Hence,
0 = A F p , | A | = k a F p \ A f p ( A { a } ) + a A f b ( a ; A \ a ) = ( k + 1 ) B F p , | B | = k + 1 f p ( B ) + ( p k ) C F p , | C | = k 1 f p ( C ) .
Theorem 1.
B F p , B f p ( B ) = p .
Proof. 
By Lemma 6,
1 k p 1 ( k + 1 ) B F p , | B | = k + 1 f p ( B ) + ( p k ) B F p , | B | = k 1 f p ( B ) = 0 .
By Lemma 2, f ( A ) = 0 if | A | = p . Therefore,
1 k p 1 ( k + 1 ) B F p , | B | = k + 1 f p ( B ) = 1 k p 2 ( k + 1 ) B F p , | B | = k + 1 f p ( B ) = ( p 1 ) B F p , | B | = p 1 f p ( B ) + 1 k p 3 ( k + 1 ) B F p , | B | = k + 1 f p ( B ) .
By Lemma 1, f ( A ) = 0 if | A | = 1 . Recall that f ( ) = p . So,
1 k p 1 ( p k ) B F p , | B | = k 1 f b ( B ) = ( p 1 ) p + 2 k p 1 ( p k ) B F p , | B | = k 1 f b ( B ) = ( p 1 ) p + 3 k p 1 ( p k ) B F p , | B | = k 1 f b ( B ) = ( p 1 ) p + 1 k p 3 ( p k 2 ) B F p , | B | = k + 1 f b ( B ) .
Hence,
0 = ( p 1 ) p + ( p 1 ) B F p , | B | = p 1 f p ( B ) + 1 k p 3 ( k + 1 ) B F p , | B | = k + 1 f p ( B ) + 1 k p 3 ( p k 2 ) B F p , | B | = k + 1 f b ( B ) = ( p 1 ) p + ( p 1 ) B F p , | B | = p 1 f p ( B ) + 1 k p 3 ( p 1 ) B F p , | B | = k + 1 f p ( B ) = ( p 1 ) p + ( p 1 ) 2 k p 1 B F p , | B | = k f p ( B ) .
It follows from Lemmas 1 and 2 that
B F p , B f p ( B ) = 2 k p 1 B F p , | B | = k f p ( B ) = p .
Theorem 2.
For 0 l p 1 2 ,
B F p , | B | = p 2 l f p ( B ) = 0 .
Proof. 
By Lemma 2,
B F p , | B | = p f p ( B ) = 0 .
So, the theorem holds for l = 0 . Suppose l > 0 . Assume that the theorem holds for l 1 . By Lemma 6,
( p 2 l + 2 ) B F p , | B | = p 2 l + 2 f p ( B ) + ( 2 l 1 ) B F p , | B | = p 2 l f b ( B ) = 0 .
By induction,
B F p , | B | = p 2 l + 2 f p ( B ) = 0 .
Thus,
B F p , | B | = p 2 l f p ( B ) = 0 .
This completes the proof of the theorem. □
Theorem 3.
For 1 l p 1 2 ,
B F p , | B | = p 2 l + 1 f p ( B ) = ( 1 ) p 2 l + 1 2 p p 1 2 l 1 .
Proof. 
By Lemma 3,
B F p , | B | = p 1 f p ( B ) = ( 1 ) p 1 2 p .
So, the theorem holds for l = 1 . Suppose l > 1 . Assume that the theorem holds for l 1 . By Lemma 6,
( p 2 l + 3 ) B F p , | B | = p 2 l + 3 f p ( B ) + ( 2 l 2 ) B F p , | B | = p 2 l + 1 f p ( B ) = 0 .
By induction,
B F p , | B | = p 2 l + 3 f p ( B ) = ( 1 ) p 2 l + 3 2 p p 1 2 l 2 .
Thus,
B F p , | B | = p 2 l + 1 f p ( B ) = ( 1 ) p 2 l + 1 2 p p 2 l + 3 2 l 2 p 1 2 l 2 = ( 1 ) p 2 l + 1 2 p p 2 l + 3 2 l 1 p 1 2 ! p 2 l + 3 2 ! ( l 2 ) ! = ( 1 ) p 2 l + 1 2 p p 1 2 ! p 2 l + 1 2 ! ( l 1 ) ! = ( 1 ) p 2 l + 1 2 p p 1 2 l 1 .
This completes the proof of the theorem. □
Theorem 4.
Let 1 k p 2 and B F p with | B | = p k . Let A = F p \ B . Then
f p ( B ) = ( 1 ) p 1 2 a A χ p b ( A \ { a } ) ( a b ) .
In particular, when k = 1 , f p ( B ) = ( 1 ) p 1 2 .
Proof. 
Note that
f p ( B ) = x F p χ p b B ( x b ) = a A χ p b B ( a b ) .
For a fixed a 0 A , by Lemma 3,
f p ( B ( A \ { a 0 } ) ) = ( 1 ) p 1 2 .
On the other hand,
f p ( B ( A \ { a 0 } ) ) = x F p χ p b B ( A \ { a 0 } ) ( x b ) = χ p b B ( A \ { a 0 } ) ( a 0 b ) .
Therefore,
χ p b B ( A \ { a 0 } ) ( a 0 b ) = ( 1 ) p 1 2 ,
which is equivalent to
χ p b B ( a 0 b ) χ p b ( A \ { a 0 } ) ( a 0 b ) = ( 1 ) p 1 2 .
Since χ p ( x ) = ± 1 for x F p \ { 0 } , we have
χ p b B ( a 0 b ) = ( 1 ) p 1 2 χ p b ( A \ { a 0 } ) ( a 0 b ) .
Hence,
f p ( B ) = ( 1 ) p 1 2 a A χ p b ( A \ { a } ) ( a b ) .
Corollary 1.
If | A | = p 2 , then
f p ( A ) = ± 2 , if p 1 mod 4 ; 0 , if p 3 mod 4 .
Furthermore, if p 1 mod 4 and F p \ A = { b 1 , b 2 } , then f p ( A ) = 2 if b 1 b 2 is a quadratic residue modulo p and f p ( A ) = 2 , otherwise.
Proof. 
Let F p \ A = { b 1 , b 2 } . By Theorem 4,
f p ( A ) = ( 1 ) p 1 2 χ p ( b 1 b 2 ) + χ p ( b 2 b 1 ) .
Suppose p 3 mod 4 . Then χ p ( 1 ) = 1 . Therefore, χ p ( b 1 b 2 ) + χ p ( b 2 b 1 ) = χ p ( b 1 b 2 ) + χ p ( 1 ) χ p ( b 1 b 2 ) = χ p ( b 1 b 2 ) χ p ( b 1 b 2 ) = 0 . Thus, f p ( A ) = 0 .
Suppose p 1 mod 4 . Then χ p ( 1 ) = 1 and
f p ( A ) = χ p ( b 1 b 2 ) + χ p ( b 2 b 1 ) = 2 χ p ( b 1 b 2 ) .
Hence, the lemma follows. □
Corollary 2.
If | A | = p 3 , then f p ( A ) = 1 or 3 .
Proof. 
Let F p \ A = { b 1 , b 2 , b 3 } . By Theorem 4,
f p ( A ) = ( 1 ) p 1 2 χ p ( b 1 b 2 ) χ p ( b 1 b 3 ) + χ p ( b 2 b 1 ) χ p ( b 2 b 3 ) + χ p ( b 3 b 1 ) χ p ( b 3 b 2 ) .
Case 1. Suppose p 1 mod 4 . Then
f p ( A ) = χ p ( b 1 b 2 ) χ p ( b 1 b 3 ) + χ p ( b 2 b 1 ) χ p ( b 2 b 3 ) + χ p ( b 3 b 1 ) χ p ( b 3 b 2 ) ,
χ p ( 1 ) = 1 and χ p ( x ) = χ p ( x ) for all x F p . We distinguish several cases.
Case 1.1. χ p ( b 1 b 2 ) and χ p ( b 1 b 3 ) are of the same sign. This implies that χ p ( b 2 b 1 ) and χ p ( b 3 b 1 ) are of the same sign as χ p ( b 1 b 2 ) . Now, if χ p ( b 1 b 2 ) = 1 , then χ p ( b 1 b 3 ) = χ p ( b 3 b 1 ) = χ p ( b 2 b 1 ) = 1 . If χ p ( b 3 b 2 ) = 1 , then f p ( A ) = 3 . If χ p ( b 3 b 2 ) = 1 , then f p ( A ) = 1 .
Next, if χ p ( b 1 b 2 ) = 1 , then χ p ( b 1 b 3 ) = χ p ( b 3 b 1 ) = χ p ( b 2 b 1 ) = 1 . If χ p ( b 3 b 2 ) = 1 , then f p ( A ) = 1 . If χ p ( b 3 b 2 ) = 1 , then f p ( A ) = 3 .
Case 1.2. χ p ( b 1 b 2 ) and χ p ( b 1 b 3 ) are of different sign. This means χ p ( b 1 b 3 ) = χ p ( b 1 b 2 ) . So,
χ p ( b 2 b 1 ) χ p ( b 2 b 3 ) + χ p ( b 3 b 1 ) χ p ( b 3 b 2 ) = χ p ( b 1 b 2 ) χ p ( b 2 b 3 ) χ p ( b 1 b 2 ) χ p ( b 3 b 2 ) = χ p ( b 1 b 2 ) χ p ( b 2 b 3 ) χ p ( b 3 b 2 ) = 0 .
Hence, f p ( A ) = χ p ( b 1 b 2 ) χ p ( b 1 b 3 ) = 1 .
Case 2. Suppose p 3 mod 4 . Then
f p ( A ) = χ p ( b 1 b 2 ) χ p ( b 1 b 3 ) χ p ( b 2 b 1 ) χ p ( b 2 b 3 ) χ p ( b 3 b 1 ) χ p ( b 3 b 2 ) ,
χ p ( 1 ) = 1 and χ p ( x ) = χ p ( x ) for all x F p . We distinguish several cases.
Case 2.1. χ p ( b 1 b 2 ) and χ p ( b 1 b 3 ) are of the same sign. This implies that χ p ( b 2 b 1 ) and χ p ( b 3 b 1 ) are of opposite sign from χ p ( b 1 b 2 ) . So, χ p ( b 2 b 1 ) = χ p ( b 3 b 1 ) and
χ p ( b 2 b 1 ) χ p ( b 2 b 3 ) + χ p ( b 3 b 1 ) χ p ( b 3 b 2 ) = χ p ( b 2 b 1 ) χ p ( b 2 b 3 ) + χ p ( b 2 b 1 ) χ p ( b 3 b 2 ) = χ p ( b 2 b 1 ) χ p ( b 2 b 3 ) + χ p ( b 3 b 2 ) = 0 ,
where the last equation follows from χ p ( b 2 b 3 ) = χ p ( b 3 b 2 ) . Hence, f p ( A ) = χ p ( b 1 b 2 ) χ p ( b 1 b 3 ) = 1 .
Case 2.2. χ p ( b 1 b 2 ) and χ p ( b 1 b 3 ) are of different sign. Suppose χ p ( b 1 b 2 ) = 1 . Then χ p ( b 1 b 3 ) = 1 , χ p ( b 3 b 1 ) = 1 and χ p ( b 2 b 1 ) = 1 . If χ p ( b 2 b 3 ) = 1 , then χ p ( b 3 b 2 ) = 1 and f p ( A ) = 3 . If χ p ( b 2 b 3 ) = 1 , then f p ( A ) = 1 .
Suppose χ p ( b 1 b 2 ) = 1 . Then χ p ( b 1 b 3 ) = 1 , χ p ( b 3 b 1 ) = 1 and χ p ( b 2 b 1 ) = 1 . If χ p ( b 2 b 3 ) = 1 , then χ p ( b 3 b 2 ) = 1 and f p ( A ) = 1 . If χ p ( b 2 b 3 ) = 1 , then f p ( A ) = 3 .
This completes the proof of the corollary. □
Theorem 5.
Let 1 k p 2 and B F p with | B | = p k . If A F p and | A | = p k , then
f p ( A ) f p ( B ) mod 4 .
Proof. 
Let B c = F p \ B = { b 1 , b 2 , , b k } and A c = F p \ A = { a 1 , a 2 , , a k } . By Theorem 4,
f p ( B ) = ( 1 ) p 1 2 a B c χ p b B c \ { a } ( a b ) = ( 1 ) p 1 2 1 i k χ p 1 j k , j i ( b i b j ) = ( 1 ) p 1 2 1 i k 1 j k , j i χ p ( b i b j ) .
Similarly,
f p ( A ) = ( 1 ) p 1 2 1 i k 1 j k , j i χ p ( a i a j ) .
For i j , we set c i j = χ p ( b i b j ) and d i j = χ p ( a i a j ) . Then
f p ( B ) = ( 1 ) p 1 2 1 i k 1 j k , j i c i j ; f p ( A ) = ( 1 ) p 1 2 1 i k 1 j k , j i d i j .
Note that c i j = χ p ( 1 ) c j i = ( 1 ) p 1 2 c j i . Similarly, d i j = ( 1 ) p 1 2 d j i . If c i j = d i j for all i , j with i j , then f p ( A ) = f p ( B ) and the theorem holds. So, we may assume that c i 0 j 0 d i 0 j 0 for some i 0 , j 0 with i 0 j 0 .
Note that
f p ( B ) = ( 1 ) p 1 2 1 i k , i i 0 , j 0 1 j k , j i c i j + ( 1 ) p 1 2 c i 0 j 0 1 j k , j i 0 , j 0 c i 0 j + ( 1 ) p 1 2 c j 0 i 0 1 j k , j i 0 , j 0 c j 0 j
Now, set
f p ( B ) 1 = ( 1 ) p 1 2 1 i k 1 j k , j i c i j ,
where c i j = c i j for all i , j such that ( i , j ) ( i 0 , j 0 ) , ( j 0 , i 0 ) , c i 0 j 0 = d i 0 j 0 and c j 0 i 0 = d j 0 i 0 . Then
f p ( B ) 1 = ( 1 ) p 1 2 1 i k , i i 0 , j 0 1 j k , j i c i j + ( 1 ) p 1 2 d i 0 j 0 1 j k , j i 0 , j 0 c i 0 j + ( 1 ) p 1 2 d j 0 i 0 1 j k , j i 0 , j 0 c j 0 j ,
and
f p ( B ) f p ( B ) 1 = ( 1 ) p 1 2 c i 0 j 0 d i 0 j 0 1 j k , j i 0 , j 0 c i 0 j + c j 0 i 0 d j 0 i 0 1 j k , j i 0 , j 0 c j 0 j .
Now, c i 0 j 0 d i 0 j 0 = ± 2 and c j 0 i 0 d j 0 i 0 = ( 1 ) p 1 2 c i 0 j 0 d i 0 j 0 = ± 2 . Since 1 j k , j i 0 , j 0 c i 0 j = ± 1 and 1 j k , j i 0 , j 0 c j 0 j = ± 1 , we have
f p ( B ) f p ( B ) 1 = ± 4 o r 0 ,
i.e., f p ( B ) f p ( B ) 1 0 mod 4 . If c i j = d i j for all i , j with i j , then f p ( A ) = f p ( B ) 1 and the theorem holds. If c i 1 j 1 d i 1 j 1 for some i 1 , j 1 with i 1 j 1 , then we set
f p ( B ) 2 = ( 1 ) p 1 2 1 i k 1 j k , j i c i j ,
where c i j = c i j for all i , j such that ( i , j ) ( i 1 , j 1 ) , ( j 1 , i 1 ) , c i 1 j 1 = d i 1 j 1 and c j 1 i 1 = d j 1 i 1 . By using similar argument, it can be shown that f p ( B ) 1 f p ( B ) 2 0 mod 4 . So, by continuing this process, we conclude that f p ( A ) f p ( B ) mod 4 . □
Corollary 3.
Let 1 k p 3 2 and A max , A min F p be such that
f p ( A max ) = max A F p , | A | = 2 k + 1 f p ( A ) ; f p ( A min ) = min A F p , | A | = 2 k + 1 f p ( A ) .
Then f p ( A max ) = f p ( A min ) and f p ( A min ) 0 f p ( A max ) mod 2 .
Proof. 
Let A max = { a 1 , a 2 , , a 2 k + 1 } . Then
f p ( A max ) = x F p χ p 1 i 2 k + 1 ( x a i ) .
Let r be a primitive element in F p . Then χ p ( r ) = 1 and
f p ( A max ) = x F p χ p ( r ) χ p 1 i 2 k + 1 ( x a i ) = x F p χ p 1 i 2 k + 1 ( r x r a i ) = x F p χ p 1 i 2 k + 1 ( x r a i ) f p ( A min ) .
Thus, f p ( A max ) f p ( A min ) . Let A min = { b 1 , b 2 , , b 2 k + 1 } . Then
f p ( A min ) = x F p χ p 1 i 2 k + 1 ( x b i )
and
f p ( A min ) = x F p χ p ( r ) χ p 1 i 2 k + 1 ( x b i ) = x F p χ p 1 i 2 k + 1 ( r x r b i ) = x F p χ p 1 i 2 k + 1 ( x r b i ) f p ( A max ) .
Hence, f p ( A max ) f p ( A min ) and f p ( A max ) = f p ( A min ) . It follows from Theorem 5 that f p ( A max ) f p ( A min ) 0 mod 4 . So, 2 f p ( A max ) 0 mod 4 and thus f p ( A max ) 0 mod 2 . Finally, f p ( A min ) = f p ( A max ) 0 mod 2 . □
Theorem 6.
Let 1 k p 2 and B F p with | B | = p k . Let A = F p \ B . Then,
f p ( B ) = ( 1 ) p 1 2 a A f p ( A \ { a } ) + b B f p ( A { b } ) .
Proof. 
By Theorem 4,
f p ( B ) = ( 1 ) p 1 2 a A χ p b A \ { a } ( a b ) .
On the other hand,
f p ( A \ { a } ) = x F p χ p b A \ { a } ( x b ) = χ p b A \ { a } ( a b ) + x F p \ { a } χ p b A \ { a } ( x b ) = χ p b A \ { a } ( a b ) + x F p χ p ( x a ) 2 b A \ { a } ( x b ) .
Therefore,
f p ( B ) = ( 1 ) p 1 2 a A f p ( A \ { a } ) x F p χ p ( x a ) 2 b ( A \ { a } ) ( x b ) = ( 1 ) p 1 2 a A f p ( A \ { a } ) f p ( a ; A \ { a } ) .
By Lemma 5,
a F p \ A f p ( A { a } ) + a A f p ( a ; A \ a ) = 0 .
Hence,
f p ( B ) = ( 1 ) p 1 2 a A f p ( A \ { a } ) + a F p \ A f p ( A { a } ) ,
and the lemma follows by noting that B = F p \ A . □

3. Concluding Remark

In this paper, we proved some relations involving the sum in Equation (1) (see Theorems 1, 2, 4, 5 and 6). We hope that these relations may give some insights on how to tackle the conjecture of Chen and Yan [4] (see Equation (2)).

Author Contributions

Conceptualization, K.A.S. and K.B.W.; writing—original draft preparation, K.A.S. and K.B.W.; writing—review and editing, K.A.S. and K.B.W.; funding acquisition, K.A.S. and K.B.W. All authors read and agreed to the published version of the manuscript.

Funding

This research was funded by the Fundamental Research Grant Scheme (FRGS) Grant Number FRGS/1/2020/STG06/SYUC/03/1 by the Malaysia Ministry of Higher Education and Publication Support Scheme by Sunway University, Malaysia.

Institutional Review Board Statement

Not applicable.

Data Availability Statement

Not applicable.

Conflicts of Interest

The authors declare no conflict of interest.

References

  1. Sárközy, A. On additive decompositions of the set of quadratic residues modulo p. Acta Arith. 2012, 155, 41–51. [Google Scholar] [CrossRef]
  2. Shparlinski, I.E. Additive decompositions of subgroups of finite fields. SIAM J. Discret. Math. 2013, 27, 1870–1879. [Google Scholar] [CrossRef]
  3. Shkredov, I.D. Sumsets in quadratic residues. Acta Arith. 2014, 164, 221–243. [Google Scholar] [CrossRef][Green Version]
  4. Chen, Y.G.; Yan, X.H. A conjecture of Sárközy on quadratic residues. J. Number Theory 2021, 229, 100–124. [Google Scholar] [CrossRef]
Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Article Metrics

Citations

Article Access Statistics

Multiple requests from the same IP address are counted as one view.