Calculating divisor of function on elliptic curve

divisors-algebraic-geometryelliptic-curves

I read Pairings for Beginners by Craig Costello.
In the example 3.1.1 at 37-th page we consider $ E/F_{103} : y^2 = x^3 + 20x + 20$, with
points $ P = (26, 20), Q = (63, 78), R = (59, 95), T = (77, 84)$ all on $E$. The author defines function $f = \frac{6y + 71x^2 + 91x + 91}{x^2 + 70x + 11} $ and states that it's divisor is $(f) = (P) + (Q) – (R) – (T)$. To check that there is no zero/pole at $\mathcal{O} = (0 : 1 : 0)$ we look at $f$ in projective space $f = \frac{6YZ + 71X^2 + 91XZ + 91Z^2}{X^2 + 70XZ + 11Z^2}$. At infinity both the numerator and the denominator are zero and the book says that they cancels out. But I see that if we substitute $Y=1$ and $X=Z \to 0$ the numerator will tend to zero linearly while the denominator quadraticly. And that should give us first order pole at $\mathcal{O}$.

My version of $f$ is $f' =\frac{y+4x+82}{y+75x+12}$ and it clearly evaluates to $1$ at $\mathcal{O}$.
Do I calculating divisors wrong?

Best Answer

The example problem concerns an elliptic curve $$ y^2 = x^3 + 20x + 20. \tag{1} $$ If $\,x\,$ goes to infinity, then let $\,x=1/t^2\,$ where $\,t\,$ goes to zero and then $$ y = \sqrt{x^3 + 20x + 20} = 1/t^3 + 10t + 10t^3 -50t^5 +O(t^7). \tag{2} $$ Thus, $\,x\,$ has a pole of order $2$ and $\,y\,$ has a pole of order $3$ at infinity. More easily seen by eliminating all but the leading terms of equation $(1)$ to get $\,y^2=x^3\,$ and then $\,y=1/t^3.$

The standard parameterization $$ x = X/Z,\quad y = Y/Z \tag{3} $$ obscures this. A better one often used in cryptography is "Jacobian coordinates" $$ x = X/T^2,\quad y = Y/T^3. \tag{4} $$ Now, given that $$ f = \frac{6y + 71x^2 + 91x + 91}{x^2 + 70x + 11}, \tag{5} $$ the new projectivization in equation $(4)$ gives $$ f = \frac{71X^2 + 6TY + 91T^2 + 91T^4}{X^2+70XT^2+11T^4} . \tag{6} $$ The elliptic curve equation $(1)$ in the new parametrization is $$ Y^2 = X^3+20XT^4+20T^6. \tag{7} $$ The substitution $\,X = Y = 1\,$ and $\,T = 0\,$ satisfies this equation and represents the point at infinity. It also gives $\,f = 71.\,$

In a precise algebraic sense, the poles "cancel out".


Note that the advantage of equation $(3)$ is that it is homogeneous. That is, $\,(X,Y,Z)\,$ represents the same point as $\,(cX,cY,cZ)\,$ for any nonzero $\,c.\,$ For equation $(4)$ the corresponding property is that the tuple $\,(X,Y,T)\,$ represents the same point as $\,(c^2X,c^3Y,cZ)\,$ and this difference must be clearly understood.

Related Question