Exercise 11.6

Consider x 0 3 + x 1 3 + x 2 3 = 0 as an equation over F 4 , the field with four elements. Show that there are nine points on the curve in P 2 ( F 4 ) . Calculate the zeta function. [Answer: ( 1 + 2 u ) 2 ( ( 1 u ) ( 1 4 u ) ) .]

Answers

Proof. Since q = 4 1 ( mod 3 ) , we can apply Theorem 2 of Chapter 10. Let χ be a character of order 3 over F = 𝔽 4 . The only other character of order 3 is then χ 2 . Thus

N 1 = q + 1 + 1 q 1 i , j , k J 0 ( χ i , χ j , χ k ) , where the sum is over all ( i , j , k ) { 1 , 2 } 3 such that i + j + k 0 ( mod 3 ) , that is ( 1 , 1 , 1 ) and ( 2 , 2 , 2 ) . Thus N 1 = q + 1 + 1 q 1 ( J 0 ( χ , χ , χ ) + J 0 ( χ 2 , χ 2 , χ 2 ) ) .

Using 1 q 1 J 0 ( χ k , χ k , χ k ) = 1 q g ( χ k ) 3 for k = 1 , 2 , we obtain

N 1 = q + 1 + 1 q ( g ( χ ) 3 + g ( χ 2 ) 3 ) .

Consider 𝔽 4 = 𝔽 2 [ x ] ( x 2 + x + 1 ) , where a = x ¯ = x + ( x 2 + x + 1 ) is a generator of 𝔽 4 . Then 𝔽 4 = { 0 , 1 , a , a 2 = a + 1 } . We compute g ( χ ) for the character χ of order 3 defined by

t 0 1 a a 2 χ ( t ) 0 1 ω ω 2

where ω = e 2 3 .

for each t 𝔽 4 , tr ( a ) = a + a 2 𝔽 2 , so the traces are tr ( 1 ) = 1 + 1 = 0 , tr ( a ) = a + a 2 = 1 , tr ( a 2 ) = a 2 + a 4 = a 2 + a = 1 . Therefore

g ( χ ) = t 𝔽 4 χ ( t ) ζ 2 tr ( t ) = t 𝔽 4 χ ( t ) ( 1 ) tr ( t ) = 1 ω ω 2 = 2 .

(This is in accordance with | g ( χ ) | = q 1 2 = 2 .) Then g ( χ 2 ) = g ( χ 1 ) = χ ( 1 ) g ( χ ) ¯ = g ( χ ) = 2 . Therefore

N 1 = q + 1 + 1 q g ( χ ) 3 + 1 q g ( χ 2 ) 3 = 5 + 1 4 ( 8 + 8 ) = 9 .

There are nine points on the curve with equation x 0 3 + x 1 3 + x 2 3 = 0 in P 2 ( F 4 ) (this is verified with a naive program in Sage).

Now we compute N s . We must replace q = 4 by q s = 4 s , and χ by χ s , a character with order 3 on F s = 𝔽 4 s .

We obtain

N s = q s + 1 + 1 q s ( g ( χ s ) 3 + g ( χ s 2 ) 3 ) .

Now we compute g ( χ s ) 3 . By the generalization of Corollary of Proposition 8.3.3.,

g ( χ s ) 3 = q s J ( χ s , χ s ) ,

thus

N s = q s + 1 + J ( χ s , χ s ) + J ( χ s 2 , χ s 2 ) .

We know that | J ( χ s , χ s ) | 2 = q s = 4 s (generalization of Corollary of Theorem 1). Writing J ( χ s , χ s ) = a + , a , b , we search the solutions of

| a + | 2 = a 2 ab + b 2 = 4 s .

Since [ ω ] is a PID, the factorization in primes is unique. Here 2 is a prime element of [ ω ] , and ( a + ) ( a + b ω 2 ) = 2 2 s , therefore a + = 𝜀 2 k , a + b ω 2 = ζ 2 l , where l , k and 𝜀 , ζ are units. Moreover 2 k = | a + | = | a + b ω 2 | = 2 l , so k = l = s . This shows that every solution a + of | a + | 2 = 4 s is associated to 2 s :

| a + | 2 = 4 s a + { 2 s , 1 2 s ω , 2 s ω , 2 s , 1 + 2 s ω , 2 s ω } .

Moreover, we know that a 1 ( mod 3 ) , b 0 ( mod 3 ) (generalization of Proposition 8.3.4.). Therefore

J ( χ s , χ s ) = a + = ( 2 ) s ,

and similarly J ( χ s 2 , χ s 2 ) = ( 2 ) s (this proves particular cases of the Hasse-Davenport relation, which we have not used here). This gives

N s = 4 s + 1 2 ( 2 ) s .

For s = 1 , we find anew N 1 = 9 .

Then

s = 1 N s u s s = s = 1 ( 4 u ) s s + s = 1 u s s 2 s = 1 ( 2 u ) s s = ln ( 1 4 u ) ln ( 1 u ) + 2 ln ( 1 + 2 u ) .

This gives

Z f ( u ) = ( 1 + 2 u ) 2 ( 1 4 u ) ( 1 u ) .

This is the first example where Z f has a zero, which satisfies the Riemann hypothesis for curves. □

User profile picture
2022-07-19 00:00
Comments