Super Gromov-Witten Invariants via torus localization

The Big Picture

Imagine trying to count something inherently slippery: not apples or electrons, but the number of ways a surface can be mapped into a curved, high-dimensional space while satisfying specific constraints. That’s what Gromov-Witten invariants do. These numbers encode information about a space’s shape and structure. Since the 1980s and 90s they’ve become one of geometry’s most powerful tools, tying together string theory, algebraic geometry, and combinatorics.

Now imagine the surfaces themselves become “super.” In superstring theory, the mathematical spaces physicists work with carry not just ordinary coordinates but an additional layer governed by strange arithmetic: two of these new coordinates multiplied in either order give opposite signs, and squaring any one of them gives zero. These are Grassmann coordinates, the mathematical language of fermions (electrons, quarks, the particles that make up ordinary matter). For superstring theory, they aren’t optional extras. They’re woven into the foundations.

A supersymmetric version of Gromov-Witten theory should exist, but building it has stalled for years on a basic obstacle. The standard toolkit for “counting” in geometry, intersection theory, doesn’t yet work in the extended supersymmetric setting. You can’t count things you can’t integrate over.

Enno Keßler, Artan Sheshmani, and Shing-Tung Yau have found a way around that wall. Their paper defines super Gromov-Witten invariants by completely sidestepping the missing machinery, then shows the resulting invariants are well-behaved, calculable, and carry information beyond what classical invariants can access.

Key Insight: By assuming that a classical technique called “torus localization” extends to supergeometry, the authors convert an impossibly hard supergeometric calculation into a classical one that existing tools can handle.

How It Works

The trick at the heart of this paper comes from equivariant geometry, the study of spaces with a built-in symmetry like a circle rotating around an axis. In that setting, torus localization is a standard workhorse: when a torus (a group of rotations) acts on a space, an integral over the whole space can be reduced to a sum over just the fixed points of that action, corrected by a geometric factor. Think of computing the average temperature across a rotating planet by measuring only at the poles and dividing by a correction term.

The moduli space of super stable maps, which catalogs every valid supersymmetric curve mapping into a target (one point per configuration), contains an embedded classical subspace. These are stable spin maps: ordinary curves equipped with a spin structure, a geometric object that captures a consistent notion of “square root” of the curve’s geometry. The fermionic directions of the super moduli space form a vector bundle over this classical subspace, called the SUSY normal bundle, which records the fermionic degrees of freedom that have been factored out.

For a convex target scheme $X$ and degree $\beta$, the SUSY normal bundle $N_{k,\beta}$ is built from three ingredients:

• The spinor bundle $S = \mathcal{O}(1)$ extended over the curve
• The tangent bundle of $X$, pulled back via the map
• Evaluation terms at the marked points

The super Gromov-Witten invariant is then an integral over the classical moduli space of stable spin maps, with cohomology classes pulled back from the target, divided by the equivariant Euler class of the SUSY normal bundle. The Euler class measures the “size” of the fermionic directions at each point. The numerator is the ordinary Gromov-Witten-style integral; the denominator accounts for the surrounding fermionic space.

What makes this construction work? No supergeometric intersection theory is required, since the entire computation lives in classical algebraic geometry. When the cohomology classes have the right total degree, the super invariants reduce to ordinary Gromov-Witten invariants up to a polynomial prefactor. And when the degree mismatch is positive, the inverse Euler class contributes characteristic classes of the SUSY normal bundle, producing information that cannot be recovered from classical descendants. That last point is what makes these invariants genuinely new, not just a repackaging of known quantities.

To calculate these invariants for projective space $\mathbb{P}^n$, the authors apply torus localization a second time, now using a torus action on $\mathbb{P}^n$ itself (the standard engine for computing Gromov-Witten invariants via graphs). This yields explicit numbers for genus zero, degree one maps into low-dimensional projective spaces with few marked points.

Why It Matters

The authors prove that super Gromov-Witten invariants satisfy a generalized Kontsevich-Manin axiom system, the structural rules that make classical Gromov-Witten invariants so computable. These axioms govern what happens when you add a marked point, change the degree, or split a curve along a node. Satisfying them means the invariants fit into a real algebraic framework, not just a one-off definition.

This lets the invariants be assembled into a super small quantum cohomology ring, extending classical quantum cohomology to encode how geometric objects in the target space intersect in the supersymmetric setting.

The physics stakes are high. Super Gromov-Witten invariants are natural candidates for computing scattering amplitudes in superstring theory on curved backgrounds, where the worldsheet is a super Riemann surface rather than an ordinary one. Such amplitudes have historically been computed through a patchwork of methods, without the clean algebraic geometry available for bosonic strings. This paper offers a route toward that missing machinery: a well-defined, explicitly calculable invariant that incorporates real supergeometric data.

The assumed torus localization theorem still needs a proof in full generality. But the authors have shown that the answer is mathematically consistent even before that proof is complete.

Bottom Line: Keßler, Sheshmani, and Yau define super Gromov-Witten invariants by making a precise assumption about torus localization, then proving those invariants satisfy all the structural axioms needed for a supergeometric enumerative theory at the interface of superstring physics and algebraic geometry.