# Schnorr Ring Signatures - wrong hash results

I don't know if I've placed the question right, It is half maths, half programming. I'm writing Schnorr Ring Signatures on Elliptic Curves in Java and I have a problem. I've found scheme on integers and translated it to elliptic curves.

In point 3 on page 11

$R_s = g^a \prod_{i \neq s}(y_i^{-H(m, R_i)}) \ mod \ p$

and in my elliptic curve version

$R_s = aG + \sum_{i\neq s}(Y_i*-H(m,R_i))$

where $G$ is a generator, $Y_i$ is a public key of i-th user, $m$ is a message and $R_i$ is $a_iG$ where $a_i$ is random.

I have

• basepoint $G$
• cofactor $h$ to the order of $G$
• order $n$ of $G$
• points $A$ and $B$

My question is how to perform a modular inverse of $H(m,R_i)$. In Java I have a modInverse method on BigIntegers but I have to know value of $p$. How to find value of $p$? Or how to calculate $-H(m,R_i)$?

EDIT: OK, thakns. I've found $p$ value. But still I'm getting wrong results. If someone would be so nice to look at it here is my sample code. Thanks in advance for help

• Are you sure you haven't got $p$ as well? It's part of the curve definition, e.g. in Bouncy you can do params.getCurve().getField().getCharacteristic(); – Maarten - reinstate Monica Dec 28 '15 at 12:32

If $g^a$ in the modular version translates to $aG$ or $a*G$ in the elliptic curve version, then $y_i^{-H(m, R_i)}$ should translate to \begin{align*} (-H(m, R_i))*Y_i&=-(H(m, R_i)*Y_i)&\text{(1)}&\text{ or equivalently}\\ &=H(m, R_i)*(-Y_i)&\text{(2)}&\text{ or equivalently}\\ &=((-H(m, R_i))\bmod n)*Y_i&\text{(3)}&\\ \end{align*}
The form (1) leads to an expression requiring one point subtraction or inversion: $$R_s=a*G-\sum_{i\neq s}(H(m,R_i)*Y_i)$$
The form (3) leads to an expression that requires none: $$R_s=a*G+\sum_{i\neq s}(((-H(m,R_i))\bmod n)*Y_i)$$
Note: By definition, $u\bmod n$ is the uniquely defined $v$ with $0\le v<n$ and $u-v$ a multiple of $n$. When $u\ge0$, $(-u)\bmod n$ can be computed as $n-1-r$, where $r$ is the remainder of the division of $(n-1+u)$ by $n$.