Rigorous realization of Witten's sigma-model interpretation of elliptic genera: the loop-space Dirac operator, its fixed-point theorem, and geometric elliptic cohomology

Witten's 1987 paper 'Elliptic genera and quantum field theory' identified the elliptic genus of a compact spin manifold with the character-valued index of the right-moving supercharge of the supersymmetric nonlinear sigma model, and announced a Dirac-operator-on-loop-space interpretation of elliptic cohomology while explicitly leaving its mathematical foundations open: a rigorous proof of the equivariant constancy (rigidity) of the genus, a fixed-point theorem for the circle action on a Dirac-type operator on free loop space, Jacobi-form transformation laws in equivariantly anomalous cases, a 'properly tailored' elliptic cohomology incorporating the quantum-field-theoretic prefactors and anomaly conditions, and speculative mod-2 refinements. Later work resolved several components by other means: rigidity was proved by Taubes and Bott-Taubes and re-proved via equivariant elliptic cohomology (Grojnowski, Rosu, Ando-Basterra, Vergne); the tailored cohomology theory was constructed as topological modular forms with its string orientation (Ando-Hopkins-Strickland); and mod-2 elliptic genera were characterized as mod-2 reductions of integral modular forms (Ochanine, Liu, and conditionally Tachikawa-Yamashita-Yonekura). The surviving open core is the rigorous realization of the quantum-field-theoretic picture itself: a mathematically well-defined two-dimensional supersymmetric sigma model - equivalently a Dirac-type operator on free loop space carrying its loop-rotation circle action - whose graded equivariant index recovers the elliptic genus and its refinements, together with the corresponding fixed-point theorem and quantization map to topological modular forms (the loop-space sector of the Stolz-Teichner conjecture). That core remains open, with active partial progress on spinor bundles over loop space and on conditional mod-2 statements.

Background

Let $M$ be a closed smooth manifold. A multiplicative genus is a ring homomorphism from a bordism ring of manifolds into a coefficient ring, computed by characteristic numbers. The classical example is the $\widehat A$-genus which, by the Atiyah-Singer index theorem, equals the index of the Dirac operator (the first-order elliptic operator on spinors whose square is essentially the Laplacian); this theorem is what ties complex K-theory to elliptic operators. In the mid-1980s Ochanine, and independently Landweber and Stong, discovered a new multiplicative genus, the elliptic genus: for an oriented manifold of dimension $4n$ it is a modular form of weight $2n$ for the congruence subgroup $\Gamma_0(2)\subset SL(2,\mathbb Z)$ (a level-2 modular form is one transforming correctly only under matrices congruent to a triangular matrix mod 2), built from theta-type functions of the formal Chern roots. Landweber, Ravenel and Stong then constructed from the universal elliptic curve a generalized cohomology theory, elliptic cohomology, of which the elliptic genus is the orientation - the hoped-for analogue of the relation between spin bordism, the $\widehat A$-genus and K-theory. The obvious question was what plays the role of the Dirac operator and its index theorem for elliptic cohomology.

Witten's 1987 paper proposed a physical answer. The supersymmetric nonlinear sigma model is the two-dimensional quantum field theory of maps from a Riemann surface (the worldsheet) into $M$; its right-moving supercharge $Q$ is an odd operator with $Q^2$ a Hamiltonian, and supersymmetry pairs states of nonzero energy with opposite fermion parity $(-1)^{F_R}$, so the difference of zero-energy state counts - the Witten index, refined here by grading with the momentum operator $P=\bar L_0-L_0$ (the difference of right- and left-moving Hamiltonians) and by powers of $q=e^{2\pi i\tau}$ recording the worldsheet torus modulus - is invariant under deformation. Taking right-moving fermions periodic (Ramond sector) and left-moving fermions antiperiodic (Neveu-Schwarz sector), Witten showed that this character-valued index $F(q)$ of $Q$ is, up to an explicit factor formed from the Dedekind eta function, the elliptic genus of $M$, computing in the large-radius limit $q^{-d/16}\widehat A(M)\operatorname{ch}(\bigotimes_{k\geq 1/2}\Lambda_{q^k}T\otimes\bigotimes_{l\geq 1}S_{q^l}T)[M]$, where $T$ is the complexified tangent bundle and $\Lambda_t$, $S_t$ denote exterior- and symmetric-power generating series. A variant sigma model with only the right-moving fermions present, taken periodic in both world-sheet directions, gives a genus $G(q)$ which, when $M$ is spin with vanishing first rational Pontryagin class $p_1(M)=0$ (a string manifold), is $\eta(q)^{-d}$ times a modular form of weight $d/2$ for the full group $SL(2,\mathbb Z)$ - the Witten genus; here $\eta$ is the Dedekind eta function, and the $\eta$-power prefactor is the quantum determinant (vacuum) contribution. The paper also: (i) argued, via the integers $b_{n,k}$ (indices of $Q$ graded simultaneously by $P$ and by the generator $K$ of a circle action on $M$) and continuity in a boundary-condition twist parameter, that the $S^1$-character-valued elliptic genus is constant in the angle - rigidity - generalizing earlier vanishing theorems of Atiyah-Hirzebruch type; (ii) computed global anomalies of the fermion determinant, forcing phases $e^{-i\pi d/8}$ resp. $e^{-i\pi d/12}$ in the operator identity for $e^{2\pi iP}$ and hence congruences of the dimension $d$ modulo 8 resp. 24 that obstruct equivalences of sigma models on manifolds of different dimension; (iii) observed that in equivariantly anomalous cases, where the anomaly class $p_1(T)-p_1(V)$ ($V$ an auxiliary bundle coupled to left-moving fermions) does not vanish but is pulled back from the classifying space $BS^1$ of the acting circle, the correct transformation behavior should be that of a Jacobi form - a modular form with an additional elliptic variable, in the sense of Eichler and Zagier; (iv) discussed the mod-2 index of Dirac-type operators in dimensions $8k+1$ and $8k+2$ (the $\mathbb Z/2$-valued index arising from a real or quaternionic structure) and proposed momentum-resolved mod-2 refinements, speculating that these might assemble into some kind of mod-2 analog of a modular form; and (v) proposed that elliptic cohomology should be 'properly tailored' - with suitable modular elements inverted - so as to accommodate the quantum-field-theoretic prefactors and anomaly conditions and to unify the observed invariants. The natural geometric home of the whole picture would be a Dirac-type operator on the free loop space $LM$ (the space of smooth maps from the circle into $M$) with the circle acting by rotating loops; Witten promised a mathematical account elsewhere, which appeared as his 1988 paper 'The index of the Dirac operator in loop space' but remained heuristic.

None of the physics arguments is mathematically rigorous: the two-dimensional sigma model is an interacting quantum field theory with no constructive definition, and free loop space is infinite-dimensional. The program was nevertheless enormously influential. Rigidity became a theorem by pure topology and analysis (Taubes; Bott-Taubes; later Liu), and was eventually understood conceptually through equivariant elliptic cohomology (Grojnowski's delocalized theory; Rosu; Ando-Basterra; Vergne). The 'properly tailored' elliptic cohomology was constructed as topological modular forms, tmf - a cohomology theory built from elliptic curves whose associated genus is the Witten genus on string manifolds - together with its string orientation (Ando-Hopkins-Strickland), lifting the Witten genus to a map of ring spectra. Stolz and Teichner conjectured that tmf is represented by two-dimensional supersymmetric field theories: that there should be a 'quantization' map from (suitably twisted) supersymmetric field theories over a manifold $X$ to $\mathrm{TMF}(X)$, reducing on the point to the grading by gravitational anomaly (the integer measuring the mismatch between the numbers of left- and right-moving degrees of freedom of the theory). Mod-2 elliptic genera were studied by Ochanine and Liu, and recently by Tachikawa-Yamashita-Yonekura, who identified them with mod-2 reductions of integral modular forms via the homomorphism $\pi_n\mathrm{TMF}\to\pi_n\mathrm{KO}((q))$ for $n=8k+1,8k+2$ - but their field-theoretic interpretation is conditional on the Stolz-Teichner conjecture. According to Berwick-Evans's 2024 survey, identifying the complete mathematical content of a two-dimensional supersymmetric quantum field theory 'remains an open problem', and the corresponding loop-space index theory 'remains out of reach': rigorous spinor bundles on loop space now exist (Ludewig), and a stochastic refinement of the Duistermaat-Heckman localization formula for the loop-rotation action on the loop space of a spin manifold has been obtained (Güneysu-Miehe), but this localization is in the ungraded $\widehat A$-genus setting and does not settle a fixed-point theorem for a $q$-graded elliptic-genus operator; no Dirac-type operator on loop space with a finite graded index, no index theorem identifying that index with the elliptic genus, and no Lefschetz fixed-point theorem for the circle action on such an operator is known.

Problem Statement

Construct, within rigorous mathematics, the two-dimensional supersymmetric nonlinear sigma model on a compact spin target manifold (on a string manifold, i.e. a spin manifold with vanishing first rational Pontryagin class $p_1(T)=0$, and in the bundle-twisted case where $p_1(T)-p_1(V)$ is nonzero but pulls back from $BS^1$ along the classifying map of the circle action - the equivariantly anomalous case - for the corresponding Witten-type genera), equivalently a Dirac-type operator on a geometrically structured version of the free loop space $LM$ together with the natural circle action by rotation of loops, so that Witten's quantum-field-theoretic picture of elliptic genera becomes a theorem. Resolving this open core requires:

(i) define the supercharge / Dirac-type operator as a mathematically well-posed object with a graded, character-valued index (graded by the momentum operator and by powers of $q$);

(ii) prove that this index equals the elliptic genus up to its explicit eta-function prefactor, respectively the Witten genus with its Dedekind-eta prefactor and its modular transformation behavior, via a rigorous index theorem for the construction (in the anomalous equivariant cases, with the Jacobi-form transformation laws predicted by Witten);

(iii) prove the corresponding Lefschetz fixed-point theorem for the circle action on the loop-space operator - a formula expressing the equivariant index as a sum of contributions from the fixed loops - implying the equivariant constancy (rigidity) of the genus and its twisted variants;

and (iv) show that the construction quantizes to the 'properly tailored' elliptic cohomology - producing the string orientation of topological modular forms - so that the momentum-resolved mod-2 indices arise as the mod-2 modular forms of the theory (identified in recent work with mod-2 reductions of integral modular forms via $\pi_n\mathrm{TMF}\to\pi_n\mathrm{KO}((q))$ for $n=8k+1,8k+2$) and the anomaly phases and dimension congruences $d\bmod 8$ and $d\bmod 24$ follow from the operator identity for $e^{2\pi iP}$.

The following parts of the originating 1987 program are already theorems by other methods and may be used as ingredients or as consistency checks rather than re-proved from scratch: the rigidity theorem itself (Taubes; Bott-Taubes; Liu; and the equivariant elliptic cohomology proofs of Rosu, Ando-Basterra and Vergne), and tmf with its string orientation as a cohomology theory. The open core being asked for is the analytic/geometric object - the rigorous sigma model, equivalently the loop-space Dirac operator - together with its index and fixed-point theorems and its quantization map to tmf. This is the sigma-model / loop-space sector of the Stolz-Teichner conjecture that supersymmetric two-dimensional field theories represent elliptic cohomology; resolving that sector (constructing the quantization map and proving the index-theoretic statements above within it) would complete Witten's program.

References

1. Edward Witten, 'Elliptic genera and quantum field theory', Communications in Mathematical Physics 109 (1987) 525-536. DOI 10.1007/BF01208956. https://doi.org/10.1007/BF01208956
2. Peter S. Landweber, review of E. Witten, 'Elliptic genera and quantum field theory', zbMATH Open, Zbl 0625.57008. https://zbmath.org/0625.57008
3. Edward Witten, 'The index of the Dirac operator in loop space', in: Elliptic Curves and Modular Forms in Algebraic Topology (Proc. Princeton 1986), Lecture Notes in Mathematics 1326, Springer, Berlin, 1988, pp. 161-181. DOI 10.1007/BFB0078045. https://doi.org/10.1007/bfb0078045
4. Serge Ochanine, 'Sur les genres multiplicatifs definis par des integrales elliptiques', Topology 26 (1987) 143-151. DOI 10.1016/0040-9383(87)90055-3. https://doi.org/10.1016/0040-9383(87)90055-3
5. Peter S. Landweber, Douglas C. Ravenel, Robert E. Stong, 'Periodic cohomology theories defined by elliptic curves', in: The Cech Centennial (Contemporary Mathematics 181), American Mathematical Society, Providence, 1995, pp. 317-337. DOI 10.1090/conm/181/02040. https://doi.org/10.1090/conm/181/02040
6. Martin Eichler, Don Zagier, 'The Theory of Jacobi Forms', Progress in Mathematics 55, Birkhauser, Boston, 1985. DOI 10.1007/978-1-4684-9162-3. https://link.springer.com/book/10.1007/978-1-4684-9162-3
7. Clifford Henry Taubes, 'S^1 actions and elliptic genera', Communications in Mathematical Physics 122 (1989) 455-526. DOI 10.1007/BF01238437. https://doi.org/10.1007/BF01238437
8. Raoul Bott, Clifford Taubes, 'On the rigidity theorems of Witten', Journal of the American Mathematical Society 2 (1989) 137-186. DOI 10.1090/S0894-0347-1989-0954493-5. https://doi.org/10.1090/s0894-0347-1989-0954493-5
9. Kefeng Liu, 'On modular invariance and rigidity theorems', Journal of Differential Geometry 41 (1995) 343-396. DOI 10.4310/jdg/1214456221. https://doi.org/10.4310/jdg/1214456221
10. Kefeng Liu, 'On elliptic genera and theta-functions', Topology 35 (1996) 617-640. DOI 10.1016/0040-9383(95)00042-9. https://doi.org/10.1016/0040-9383(95)00042-9
11. Serge Ochanine, 'Elliptic genera, modular forms over KO*, and the Brown-Kervaire invariant', Mathematische Zeitschrift 206 (1991) 277-291. DOI 10.1007/BF02571343. https://doi.org/10.1007/BF02571343
12. Kefeng Liu, 'On mod 2 and higher elliptic genera', Communications in Mathematical Physics 149 (1992) 71-95. DOI 10.1007/BF02096624. https://doi.org/10.1007/BF02096624
13. Matthew Ando, Michael J. Hopkins, Neil P. Strickland, 'Elliptic spectra, the Witten genus and the theorem of the cube', Inventiones Mathematicae 146 (2001) 595-687. DOI 10.1007/s002220100175. https://doi.org/10.1007/s002220100175
14. Ioanid Rosu, 'Equivariant elliptic cohomology and rigidity', American Journal of Mathematics 123 (2001) 647-677. DOI 10.1353/ajm.2001.0027. arXiv:math/9912089. https://doi.org/10.1353/ajm.2001.0027
15. Matthew Ando, Maria Basterra, 'The Witten genus and equivariant elliptic cohomology', Mathematische Zeitschrift 240 (2002) 787-822. DOI 10.1007/s002090100399. arXiv:math/0008192. https://doi.org/10.1007/s002090100399
16. Ian Grojnowski, 'Delocalised equivariant elliptic cohomology', in: Elliptic Cohomology: Geometry, Applications, and Higher Chromatic Analogues, London Mathematical Society Lecture Note Series 342, Cambridge University Press, Cambridge, 2007, pp. 114-121 (preprint circulated 1994). DOI 10.1017/CBO9780511721489.007. https://doi.org/10.1017/cbo9780511721489.007
17. Stephan Stolz, Peter Teichner, 'Supersymmetric field theories and generalized cohomology', in: Mathematical Foundations of Quantum Field Theory and Perturbative String Theory, Proceedings of Symposia in Pure Mathematics 83, American Mathematical Society, Providence, 2011, pp. 279-340. DOI 10.1090/pspum/083/2742432. arXiv:1108.0189. https://doi.org/10.1090/pspum/083/2742432
18. Michele Vergne, 'Bouquets revisited and equivariant elliptic cohomology', International Journal of Mathematics 32 (2021), Article 2140012. DOI 10.1142/S0129167X21400127. arXiv:2005.00312. https://doi.org/10.1142/s0129167x21400127
19. Matthias Ludewig, 'The spinor bundle on loop space', arXiv:2305.12521 (2023). https://arxiv.org/abs/2305.12521
20. Yuji Tachikawa, Mayuko Yamashita, Kazuya Yonekura, 'Remarks on mod-2 elliptic genus', Communications in Mathematical Physics 406 (2025), Article 16. arXiv:2302.07548. DOI 10.1007/s00220-024-05202-4. https://arxiv.org/abs/2302.07548
21. Daniel Berwick-Evans, 'Elliptic cohomology and quantum field theory', arXiv:2408.07693 (2024). https://arxiv.org/abs/2408.07693
22. Batu Güneysu, Jonas Miehe, 'Fermionic Dyson expansions and stochastic Duistermaat-Heckman localization on loop spaces', arXiv:2410.14034 (2024). https://arxiv.org/abs/2410.14034