Riemann-Roch notes / VII

The elliptic curve case

On an elliptic curve, the canonical divisor is trivial and the dimension count shifts by one from the genus zero case.

Let \(X\) be a genus one compact Riemann surface. Analytically,

\[X\cong \mathbb C/\Lambda.\]

After choosing a base point, it is an elliptic curve. The invariant differential \(dz\) descends to the torus and has no zeros or poles, so

\[K\sim 0.\]

Riemann-Roch becomes

\[\ell(D)-\ell(-D)=\deg(D).\]

Positive effective divisors

If \(D\) is effective and \(\deg(D)>0\), then \(-D\) has negative degree. Hence \(\ell(-D)=0\), and therefore

\[\ell(D)=\deg(D).\]

This is the basic genus one count.

A single pole is not enough

For one point \(P\),

\[\ell(P)=1.\]

Thus a meromorphic function on an elliptic curve cannot have only one simple pole unless it is constant.

For two points,

\[\ell(P+Q)=2,\]

so a nonconstant function exists with poles bounded by \(P+Q\).

On the concrete elliptic curve \(E:y^2=x^3-x\), this appears in the familiar functions

\[L(2O)=\operatorname{span}\{1,x\}, \qquad L(3O)=\operatorname{span}\{1,x,y\}.\]

The first space has dimension \(2\) because \(x\) has a double pole at \(O\). The second has dimension \(3\) because \(y\) has a triple pole at \(O\). These two computations are the genus one analogue of polynomial bases on \(\mathbb P^1\).

Geometric reason

A nonconstant meromorphic function with one simple pole would define a degree one map \(X\to\mathbb P^1\). A degree one map between smooth compact curves is an isomorphism. That would force a genus one curve to be the projective line, which is impossible.

The equality \(\ell(P)=1\) is the dimension-count version of that geometric obstruction.

The shift from genus zero

On \(\mathbb P^1\), a divisor of degree \(n\) gives dimension \(n+1\). On an elliptic curve, a positive effective divisor of degree \(n\) gives dimension \(n\). The missing one dimension is exactly the genus term in Riemann-Roch.