Given Legendre symbol: $(\frac{y^2 - b^{-1}}{p})$. Please show a detailed proof of $\sum_{y=0}^{p-1 } (\frac{y^2 - b^{-1}}{p}) =-1$
Asked
Active
Viewed 130 times
0
-
There should be a typo in some LaTeX command, I don't see your formulas correctly. – Taroccoesbrocco Feb 06 '18 at 13:10
-
@i707107 sir, can you please help me with this – User432477438 Feb 06 '18 at 14:10
1 Answers
2
This question shows that for(1)(2) $c\neq0$ we have $$\sum_{y}\left(\frac{y(y+c)}{p}\right)=-1$$ This is done by replacing(3) $y(y+c)$ by $(y+c)/y$ which parametrizes(4) $\mathbb F_p\setminus\{1\}$, which reduces the problem to the well known(5) $$\sum_{y}\left(\frac{y}{p}\right)=0$$
In this case the question is different in that $y^2-b$ only factors as above if $b$ is a square mod $p$.(6) What we can do is invert the order of summation: $$\sum_{y}\left(\frac{y^2-b}{p}\right)=\sum_{x}\left(\frac{x}{p}\right) \left(1+\left(\frac{x+b}{p}\right)\right)$$ because $1+\left(\frac{x+b}{p}\right)$ is the number of solutions to $y^2-b\equiv x$.
This reduces to problem to the identity above.(7)
More details
- Strictly speaking only for $c=1$; the general case is proved analoguously.
- For $c=0$ all terms (except for $y=0$) are $1$, and the sum equals $p-1$
- Note that $\left(\frac{ab}p\right) = \left(\frac{ab/a^2}p\right) = \left(\frac{b/a}p\right)$ for $a\neq 0$.
- When $y$ runs through $1, \ldots, p-1$, $1+c/y$ runs through $2, 3, \ldots, p$: it cannot equal $1$ mod $p$, and $a = 1+c/y$ iff $y=(a-1)/c$ shows that $a$ can take all values from $2$ to $p$ (mod $p$), the only restriction being that $y \neq0$.
- Because there are as many QR's as non QR's mod $p$ (not counting $0$)
- It factors as $(y-b)(y+b)$, but making the change of variables $z=y-b$ gives the problem from the beginning, with $c=2b$.
- We are left with $\sum_{x}\left(\frac{x}{p}\right)=0$ and $\sum_{x}\left(\frac{x}{p}\right)\left(\frac{x+b}{p}\right)=0$, note that $\left(\frac{ab}p\right) = \left(\frac{a}p\right)\left(\frac{b}p\right)$.
Bart Michels
- 26,985
- 6
- 59
- 123
-
What is that identity? Or suggest any link that I can find that identity. – User432477438 Feb 07 '18 at 08:46
-
-
-
The question I link to in the beginning is the case $c=1$, the proof for the general case is the same. The key ideas are in my answer. – Bart Michels Feb 07 '18 at 09:14
-
How it come up that it is equal to -1? Sorry im not that familiar with legendre symbol – User432477438 Feb 07 '18 at 09:18
-
Because a change of variables reduces it to proving $\sum_{y \neq 1}\left(\frac{y}{p}\right)=-1$, which follows from $\sum_{y}\left(\frac{y}{p}\right)=0$ – Bart Michels Feb 07 '18 at 09:37
-
Can it be $\sum_{x=0}^{p-1} (1 + (\frac {x}{p})) (\frac {x+b}{p})= \sum_{x=0}^{p-1} (\frac {x}{p}) (1+ (\frac {x+b}{p})) $ ?? – User432477438 Feb 07 '18 at 09:52
-
Yes, both are $0$ using $\sum_{y}\left(\frac{y}{p}\right)=0$ and $\sum_{x=0}^{p-1} (\frac {x}{p}) (\frac {x+b}{p})=0$ for $b\not\equiv 0$ – Bart Michels Feb 07 '18 at 10:24
-
Sir, can you please show a step by step proof of your answer. Thank you so much! – User432477438 Feb 07 '18 at 11:13
-