03.06.23 · modern-geometry / characteristic-classes

Modularity of the elliptic genus

shipped3 tiersLean: none

Anchor (Master): Hirzebruch-Berger-Jung *Manifolds and Modular Forms* Lectures 4-6 + Appendix I (full proof of modularity for $\Gamma_0(2)$ and $\mathrm{SL}_2(\mathbb{Z})$ via the Jacobi-theta transformation laws); Ochanine 1987 *Topology* 26, 143-151 (originator paper); Hirzebruch 1988 *Differential Geometric Methods in Theoretical Physics* (NATO ASI C 250, Reidel) 37-63 (Elliptic genera of level $N$ for complex manifolds); Landweber-Stong 1988 *Topology* 27, 145-161 (Circle actions on spin manifolds and characteristic numbers); Zagier 1988 LNM 1326, 216-224 (Note on the Landweber-Stong elliptic genus)

Intuition [Beginner]

The elliptic genus is a recipe that turns a closed shape into a power series in a formal variable $q$ . The recipe runs the same multiplicative-sequence machine that already produces the signature and the Dirac index, but with a more elaborate power series on the input side, one whose coefficients depend on $q$ in a precise way. The output is no longer a single number but a whole formal Taylor expansion in $q$ , and the question is: what kind of function does that expansion describe?

The modularity theorem answers the question. For a closed spin shape of dimension $4 k$ , the power series produced by the Ochanine recipe is the $q$ -expansion of a modular form of weight $2 k$ for a small index-two subgroup of the modular group. For a closed string shape, the more refined Witten recipe produces the $q$ -expansion of a modular form of weight $2 k$ for the full modular group. In both cases, a topological invariant of a manifold turns out to be an arithmetic object: a function on the upper half-plane that transforms predictably under a discrete group of motions.

The reason this matters is that the same power series gathers all the classical Hirzebruch genera at once. At one cusp of the modular curve, the elliptic genus specialises to the Dirac index. At another cusp, it specialises to the signature. Modularity is the rule that lets you slide between the two specialisations through a one-parameter family of modular forms, and it lets you put the spin and string manifolds of the world inside a single ring of arithmetic invariants.

Visual [Beginner]

Picture the upper half-plane, the same fundamental triangle that organises modular forms, with two distinguished points: the cusp at infinity, marked with the label $\hat{A}$ , and the cusp at zero, marked with the label $L$ . The elliptic genus of a fixed manifold $M$ is a single function on this picture, taking modular-form values, that interpolates between the Dirac index at one cusp and the signature at the other.

$A schematic of the modular fundamental domain in the upper half-plane, with cusps at $\tau = i \infty$ labelled with the $\hat A$-genus and at $\tau = 0$ labelled with the $L$-genus, and an arrow indicating that the elliptic genus interpolates between the two.$

The picture's central feature is the interpolation. Without modularity, the Dirac index and the signature would be two separate invariants of two separate manifold structures. With modularity, they become two cuspidal values of one modular-form-valued invariant, the elliptic genus, and the entire upper half-plane in between is filled with intermediate $q$ -expansions, each of which is itself a manifold invariant in disguise.

Worked example [Beginner]

Pick a simple closed shape: a quaternionic projective line, the eight-real-dimensional shape that already plays a starring role in characteristic-class computations. Its Ochanine elliptic genus is the simplest informative example of a modular form arising from a manifold.

Step 1. The Ochanine power series, expanded out to first order, agrees with the Dirac power series. Plug in the Pontryagin classes of the quaternionic projective line and the $q^{0}$ coefficient of the answer is the Dirac index of that manifold, which is the integer $1$ .

Step 2. The next term, the $q^{1}$ coefficient, is a more elaborate Pontryagin polynomial that takes account of the twist coming from a copy of the tangent bundle. Plugging in the same numerical data, this coefficient comes out to $- 2$ in the standard normalisation.

Step 3. Continuing the expansion, the $q$ -series of the genus matches the first few Fourier coefficients of a known modular form of weight $4$ for the level- $2$ congruence subgroup of the modular group. This is the modularity statement: the $q$ -expansion of the Ochanine genus of the quaternionic projective line is not an arbitrary power series but the Fourier expansion of a specific modular form.

Step 4. Take the cuspidal limit, sending $q$ to zero. The whole series collapses to the leading coefficient, which is the Dirac index $1$ . Take the other cuspidal limit, sending $τ$ to zero, and the modular transformation rule converts the series into the signature of the manifold, which is also $1$ in this case.

What this tells us: the elliptic genus of a manifold is a single arithmetic object, a modular form, whose two cuspidal specialisations recover two classical topological invariants. The whole modular-form structure lives between them, and modularity is the rule guaranteeing that the interpolation is a function with strong symmetry properties.

Check your understanding [Beginner]

Formal definition [Intermediate+]

Let $M$ be a closed oriented smooth manifold of dimension $n = 4 k$ . Recall (from [03.06.15]) that a multiplicative sequence with characteristic power series $Q (x) = 1 + q_{1} x + q_{2} x^{2} + \dots \in Q [[x]]$ produces a genus $φ_{Q} : Ω_{*}^{S O} \otimes Q \to Q [[q]]$ when the coefficients of $Q$ are themselves power series in an auxiliary variable $q$ . Explicitly, for a closed oriented $4 k$ -manifold $M$ with formal Pontryagin roots $\pm x_{1}^{2}, \dots, \pm x_{k}^{2}$ of the tangent bundle (Chern roots after complexification), one defines $$ \varphi_Q(M) ;=; \big\langle \prod_{i=1}^{k} Q(x_i), , [M] \big\rangle ;\in; \mathbb{Q}[[q]], $$ where $[M]$ is the integer fundamental class. The genus is multiplicative on Cartesian products and additive on disjoint unions.

Definition (Ochanine elliptic genus). The Ochanine elliptic genus is the genus attached to the formal power series $$ Q_{\mathrm{Och}}(x, q) ;=; \frac{x/2}{\sinh(x/2)} , \prod_{n \geq 1} \frac{(1 - q^n)^2}{(1 - q^n e^x)(1 - q^n e^{-x})}. $$ Equivalently, in terms of the Jacobi theta function $θ_{1}$ on the upper half-plane, $$ Q_{\mathrm{Och}}(x, \tau) ;=; \frac{x \cdot \theta_1'(0, \tau)}{2 \pi i , \theta_1(x/(2 \pi i), \tau)}, $$ with the standard convention $q = e^{2 π i τ}$ . The resulting genus is denoted $φ_{Och} (M) \in Q [[q]]$ .

Definition (Witten genus). The Witten genus is the genus attached to the formal power series $$ Q_W(x, q) ;=; \frac{x}{e^{x/2} - e^{-x/2}} , \prod_{n \geq 1} \frac{(1 - q^n)^2}{(1 - q^n e^x)(1 - q^n e^{-x})}, $$ defined for $M$ a closed string manifold, i.e., a closed oriented manifold with $w_{1} (M) = 0$ , $w_{2} (M) = 0$ , and $\frac{p _{1} ( M )}{2} = 0$ in $H^{4} (M; Z)$ . The Witten genus is denoted $φ_{W} (M)$ .

Definition (modular form of weight $k$ for $Γ$ ). Let $Γ \subseteq SL_{2} (Z)$ be a finite-index subgroup. A holomorphic function $f : H \to C$ on the upper half-plane is a modular form of weight $k$ for $Γ$ if for every $γ = (a c b d) \in Γ$ , $$ f(\gamma \tau) ;=; (c \tau + d)^k , f(\tau), $$ and $f$ is holomorphic at every cusp of $Γ$ . The ring $M_{*} (Γ) = ⨁_{k} M_{k} (Γ)$ is graded by weight; the foundational example $M_{*} (SL_{2} (Z)) = C [E_{4}, E_{6}]$ is treated in [21.04.01].

Definition (congruence subgroup $Γ_{0} (N)$ ). For $N \geq 1$ , $$ \Gamma_0(N) ;=; \Big{ \begin{pmatrix} a & b \ c & d \end{pmatrix} \in \mathrm{SL}_2(\mathbb{Z}) ,:, c \equiv 0 \pmod N \Big}. $$ It is a finite-index subgroup of $SL_{2} (Z)$ with $[SL_{2} (Z) : Γ_{0} (N)] = N \prod_{p ∣ N} (1 + 1/ p)$ . The case $N = 2$ has index $3$ and three cusps; the ring $M_{*} (Γ_{0} (2))$ is freely generated as a $C$ -algebra by two elements $δ, ϵ$ of weights $2$ and $4$ respectively (Skoruppa appendix to Hirzebruch-Berger-Jung).

Counterexamples to common slips

Slip. "The Ochanine genus is a modular form for the full modular group." It is not; it is modular only for $Γ_{0} (2) ⊊ SL_{2} (Z)$ . The full-modular-form statement is reserved for the Witten genus, and only on string manifolds.
Slip. "The Witten genus is defined on every spin manifold." It is not. The defining product expansion of $Q_{W}$ converges to a formal series with integer coefficients only when the half-Pontryagin class $p_{1} /2$ vanishes; on a general spin manifold the resulting $q$ -series fails to be modular for $SL_{2} (Z)$ , only for $Γ_{0} (2)$ .
Slip. "Modularity gives an analytic continuation of the Pontryagin polynomial." It does not. Modularity is a statement about the $q$ -expansion: the formal power series $φ_{Och} (M) \in Q [[q]]$ is the $q$ -expansion of a modular form on $H$ , not the result of analytically continuing a polynomial.
Slip. "The two cuspidal limits give $\hat{A}$ and $L$ on every manifold." They do for the Ochanine genus, but the precise statement uses the transformation behaviour at the two cusps of $Γ_{0} (2)$ . At $τ = i \infty$ the $q \to 0$ limit gives $\hat{A} (M)$ ; the cusp at $τ = 0$ is reached by the involution $τ \mapsto - 1/ (2 τ)$ (a Fricke-type involution adapted to $Γ_{0} (2)$ ), and the $L$ -genus emerges only after this transformation.

Key theorem with proof [Intermediate+]

Theorem (modularity of the Ochanine elliptic genus; Ochanine 1987, Landweber-Stong 1988, Zagier 1988). Let $M$ be a closed oriented spin manifold of dimension $4 k$ . Then $φ_{Och} (M) (τ)$ is the $q$ -expansion of a holomorphic modular form of weight $2 k$ for the congruence subgroup $Γ_{0} (2) \subset SL_{2} (Z)$ .

Theorem (modularity of the Witten genus; Witten 1987, Hirzebruch 1988, Zagier 1988). Let $M$ be a closed string manifold of dimension $4 k$ . Then $φ_{W} (M) (τ)$ is the $q$ -expansion of a holomorphic modular form of weight $2 k$ for the full modular group $SL_{2} (Z)$ .

Proof (Ochanine modularity, sketch). The argument has three ingredients.

Ingredient 1 (Theta-function repackaging). By Zagier's identity, the Ochanine characteristic series factors as $$ Q_{\mathrm{Och}}(x, \tau) ;=; \frac{x \cdot \theta_1'(0, \tau)}{2 \pi i , \theta_1(x/(2 \pi i), \tau)}, $$ where $θ_{1} (z, τ) = 2 \sum_{n \geq 0} (- 1)^{n} q^{(n + 1/2)^{2} /2} sin ((2 n + 1) π z)$ is the standard Jacobi odd theta function. The product expansion of $θ_{1}$ matches the infinite product in $Q_{Och}$ up to the $sinh$ -prefactor that produces the $\hat{A}$ -class. This rewrites the Ochanine power series as a quotient of theta functions on the elliptic curve $C / (Z + τ Z)$ .

Ingredient 2 (Modular transformation of $θ_{1}$ ). The Jacobi theta function satisfies the transformation laws $$ \theta_1(z, \tau + 1) ;=; e^{i \pi / 4} \theta_1(z, \tau), \qquad \theta_1(z/\tau, -1/\tau) ;=; -i (-i \tau)^{1/2} , e^{i \pi z^2 / \tau} , \theta_1(z, \tau). $$ For elements $γ = (a c b d) \in Γ_{0} (2)$ , the composite transformation gives $$ \theta_1(z / (c \tau + d), \gamma \tau) ;=; \varepsilon(\gamma) (c \tau + d)^{1/2} e^{i \pi c z^2 / (c \tau + d)} \theta_1(z, \tau), $$ where $ε (γ) \in μ_{8}$ is a multiplier. The point of restricting to $Γ_{0} (2)$ is that $ε$ becomes a character (rather than a more general multiplier system) and the half-integer-weight transformation organises into the weight- $2 k$ rule when assembled across all $k$ factors.

Ingredient 3 (Modular invariance of the Pontryagin assembly). The Pontryagin polynomial $\prod_{i = 1}^{k} Q_{Och} (x_{i}, τ)$ is the product of $k$ copies of the quotient in Ingredient 1, one per Pontryagin root. Under $γ \in Γ_{0} (2)$ , each factor picks up the multiplier $ε (γ) (c τ + d)^{1/2}$ times a Gaussian factor $e^{iπ c x_{i}^{2} / (c τ + d)}$ . The product of $k$ multipliers contributes $ε (γ)^{k} (c τ + d)^{k /2}$ . The Gaussian factors combine to $e^{iπ c \sum x_{i}^{2} / (c τ + d)} = e^{iπ c p_{1} (M) / (c τ + d)}$ when one identifies $\sum x_{i}^{2}$ with the first Pontryagin class via the splitting principle. For a spin manifold, $p_{1} (M)$ is an integral class (and in fact $p_{1} /2$ is half-integral, with the parity of $w_{2}$ ); pairing against $[M]$ gives a Pontryagin number, but the Gaussian exponential is evaluated formally in cohomology, where the integration against $[M]$ leaves a polynomial of degree $\leq k$ in the Gaussian factor. A detailed calculation (carried out in Hirzebruch-Berger-Jung Lecture 6 and in Zagier's LNM 1326 note) shows that the Gaussian contributions cancel against the multiplier when the dimension is exactly $4 k$ and the spin condition holds. The remaining $(c τ + d)^{k}$ together with another $(c τ + d)^{k}$ from the differential factor $d x_{i}$ in $\int_{M}$ assemble to $(c τ + d)^{2 k}$ , the asserted weight.

Holomorphicity at cusps. The function $θ_{1} (z, τ) / θ_{1}^{'} (0, τ)$ is holomorphic in $τ$ in the interior of the fundamental domain. At each of the three cusps of $Γ_{0} (2)$ — namely $τ = i \infty$ , $τ = 0$ , and $τ = 1/2$ — the $q$ -expansion of $θ_{1}$ has the standard form $θ_{1} = 2 q^{1/8} sin (π z) \prod (1 - q^{n}) (1 - q^{n} e^{2 π i z}) (1 - q^{n} e^{- 2 π i z})$ , and the resulting $q$ -expansion of $φ_{Och} (M)$ has no negative-power terms. Holomorphicity at the cusps is automatic.

$□$

Proof (Witten modularity, sketch). The string condition $p_{1} /2 = 0$ kills the Gaussian factor $e^{iπ c p_{1} / (c τ + d)}$ that obstructed full modular invariance in the Ochanine case. The character $ε (γ)$ also extends to the full modular group under the string condition, becoming a character of $SL_{2} (Z)$ that turns out to be identically $1$ after one accounts for the half-integer weights from the $k$ theta factors. The resulting transformation law is the full weight- $2 k$ rule for $SL_{2} (Z)$ . $□$

Bridge. This presentation builds toward 03.06.24 pending (the Bott-Taubes rigidity theorem), where the $S^{1}$ -equivariant elliptic genus on a spin manifold with circle action turns out to be constant in the equivariant parameter — a statement that depends decisively on the modularity result above, because the equivariant genus is identified with a modular form on a thickened moduli space and the only modular forms with poles only at the equivariant origin are the constants. The foundational reason modularity works is exactly the theta-function identity in Ingredient 1: the Pontryagin polynomial of the elliptic genus is a section of a line bundle on an elliptic curve, and modular transformations are translations of the elliptic-curve modulus, so the elliptic genus is the cohomological shadow of an elliptic-curve-valued invariant. Putting these together, the elliptic genus becomes the bridge from the bordism ring of spin or string manifolds into the ring of modular forms, and the central insight is that the same multiplicative-sequence machinery that turned formal power series into Hirzebruch genera now turns elliptic-integral power series into elliptic genera valued in modular forms.

Exercises [Intermediate+]

Exercise 4 (medium, short-answer).

Explain why $φ_{Och} (M)$ is a modular form of weight $2 k$ rather than $k$ or $k /2$ , for $M$ of dimension $4 k$ .

Hint

The Pontryagin polynomial is a product of $k$ copies of $Q_{Och} (x_{i}, τ)$ , one per Pontryagin root. The theta-function quotient $θ_{1} (z, τ) / θ_{1}^{'} (0, τ)$ transforms with weight $- 1/2$ under $SL_{2}$ , so each factor contributes weight $1/2$ ; the integration $\int_{M}$ contributes additional weight. Track the assembly carefully.

Answer

The characteristic power series $Q_{Och} (x, τ) = x θ_{1}^{'} (0, τ) / (2 π i θ_{1} (x / (2 π i), τ))$ involves one factor of $θ_{1}^{'} (0, τ)$ (a modular form of weight $3/2$ for $Γ_{0} (2)$ ) in the numerator and one factor of $θ_{1} (z, τ)$ (weight $1/2$ ) in the denominator, giving net weight $1$ per Pontryagin root $x_{i}$ . The Pontryagin polynomial $\prod_{i = 1}^{k} Q_{Och} (x_{i}, τ)$ thus has weight $k$ from the theta factors alone. The integration $\int_{M}$ against $[M]$ contributes another factor of weight $k$ via the volume-form scaling, because in $4 k$ real dimensions the top characteristic-class pairing $\int_{M}$ scales by $(c τ + d)^{k}$ under the modular action on the Pontryagin roots (each $x_{i}$ is paired with a $d x_{i}$ , and $d x_{i}$ picks up weight $1$ ). Total weight $2 k$ . Rubric: full credit for the per-root weight- $1$ accounting plus the integration-doubling.

Exercise 5 (medium, symbolic).

Compute $φ_{Och} (HP^{2})$ at the cuspidal limit $q = 0$ in terms of the $\hat{A}$ -genus of the quaternionic projective plane.

Hint

At $q = 0$ the Ochanine series collapses to the $\hat{A}$ -power series $Q_{\hat{A}} (x) = (x /2) / sinh (x /2)$ . Use the known value $\hat{A} (HP^{2}) = 0$ .

Answer

At $q = 0$ , $Q_{Och} (x, 0) = Q_{\hat{A}} (x) = (x /2) / sinh (x /2)$ , so $φ_{Och} (M) (0) = \hat{A} (M)$ . For $M = HP^{2}$ (real dimension $8$ , $k = 2$ ), the $\hat{A}$ -genus is the coefficient of $x_{1}^{2} x_{2}^{2}$ in the expansion of $(x_{1} /2) / sinh (x_{1} /2) \cdot (x_{2} /2) / sinh (x_{2} /2)$ paired against $[M]$ . Direct computation (Hirzebruch Topological Methods §17, Lawson-Michelsohn §III.13) gives $\hat{A} (HP^{2}) = 0$ . So $φ_{Och} (HP^{2}) (0) = 0$ , and the constant term of the $q$ -expansion vanishes. The higher $q$ -coefficients are non-zero and assemble into a non-zero cusp form on $Γ_{0} (2)$ . Rubric: full credit for the $q \to 0$ identification with $\hat{A}$ and the $\hat{A} (HP^{2}) = 0$ computation.

Exercise 6 (medium, short-answer).

Show that on a string manifold of dimension $4 k$ , the Witten genus reduces to the Ochanine genus when one forgets the string structure and remembers only the spin structure.

Hint

Compare the two characteristic power series $Q_{W}$ and $Q_{Och}$ . The string condition $p_{1} /2 = 0$ kills the half-Pontryagin term and aligns the two formal power series at the level of the bordism invariant.

Answer

The characteristic power series $Q_{W} (x, q)$ differs from $Q_{Och} (x, q)$ by a multiplicative factor $exp (- x^{2} G_{2} (τ))$ , where $G_{2} (τ)$ is the weight- $2$ quasi-modular Eisenstein series and $- x^{2}$ is the first Pontryagin root in the splitting principle. When the manifold is string, the Pontryagin polynomial computed against the $exp (- p_{1} G_{2} /2)$ correction vanishes (since $p_{1} /2 = 0$ in cohomology), so the Witten and Ochanine characteristic-class evaluations agree as $q$ -series. The Witten genus is the modular-form completion of the Ochanine genus in the string-bordism setting, where the extra structure allows the genus to be modular for the full $SL_{2} (Z)$ . Rubric: full credit for identifying the $exp (- p_{1} G_{2} /2)$ correction factor and its vanishing on string manifolds.

Exercise 7 (hard, short-answer).

Use the modularity statement to prove that the value of $φ_{Och} (M)$ at the cusp $τ = 0$ (the other cusp of $Γ_{0} (2)$ , reached by $τ \mapsto - 1/ (2 τ)$ ) equals the signature $sig (M)$ of $M$ .

Hint

The Fricke involution $τ \mapsto - 1/ (2 τ)$ sends $Γ_{0} (2)$ to itself and exchanges its two cusps $i \infty$ and $0$ . Under this involution, the Jacobi theta quotient defining $Q_{Och}$ transforms into the $tanh$ -quotient defining $Q_{L}$ .

Answer

The Fricke involution $W_{2} : τ \mapsto - 1/ (2 τ)$ is an Atkin-Lehner involution on $Γ_{0} (2)$ that exchanges the two cusps. Under $W_{2}$ , the Jacobi theta function $θ_{1} (z, τ)$ transforms (via the modular transformation laws expressed earlier) into a multiple of $θ_{2} (z, τ)$ , the second Jacobi theta function, whose product expansion is $θ_{2} (z, τ) = 2 q^{1/8} cos (π z) \prod_{n \geq 1} (1 - q^{n}) (1 + q^{n} e^{2 π i z}) (1 + q^{n} e^{- 2 π i z})$ . Substituting into the Ochanine characteristic series and simplifying (using the trigonometric identity $cosh (x /2) \cdot sinh (x /2) = (1/2) sinh (x)$ ), the $W_{2}$ -transformed power series becomes $$ W_2 \cdot Q_{\mathrm{Och}}(x, \tau) \big|{q \to 0} ;=; \frac{x/2}{\tanh(x/2)} ;=; Q_L(x), $$ which is the $L$ -genus characteristic power series. So at the second cusp, $W_2 \cdot \varphi{\mathrm{Och}}(M)(0) = \varphi_L(M) = \mathrm{sig}(M) $b y t h eH i r z e b r u c h s i g na t u r e t h eor e m ‘ [03.06.11] ‘. T h e m o d u l a r i t y s t a t e m e n t g u a r an t ees t ha tt h e v a l u e a tt h eseco n d c u s p i s w e l l - d e f in e d an d ma t c h es t h e$ L $- g e n u sco m p u t a t i o n . R u b r i c : f u l l cr e d i t f or t h e F r i c k e - in v o l u t i o ni d e n t i f i c a t i o n o f t h e tw oc u s p s an d t h e l imi t co m p u t a t i o n s h o w in g t h e$ L$-genus emerges.

Exercise 8 (hard, short-answer).

Prove that the Witten genus of a $4$ -dimensional string manifold is a constant integer multiple of the discriminant $Δ = η^{24}$ . Identify the integer.

Hint

A modular form of weight $2$ for $SL_{2} (Z)$ is automatically a cusp form (since $M_{2} (SL_{2} (Z)) = 0$ ), and in fact $S_{2} (SL_{2} (Z)) = 0$ as well. So $φ_{W} (M) = 0$ identically for any closed string $4$ -manifold. Then check the dimension- $8$ case where $S_{4} (SL_{2} (Z)) = C \cdot Δ$ does not arise either; the dimension- $24$ case where weight $12 =$ that of $Δ$ is the first place a multiple of $Δ$ can appear.

Answer

For $M$ of real dimension $4 k$ string, $φ_{W} (M)$ is a modular form of weight $2 k$ for $SL_{2} (Z)$ . The ring $M_{*} (SL_{2} (Z)) = C [E_{4}, E_{6}]$ (from [21.04.01]) has $dim M_{2} = 0$ , $dim M_{4} = 1$ (generated by $E_{4}$ ), $dim M_{6} = 1$ ( $E_{6}$ ), $dim M_{8} = 1$ ( $E_{4}^{2}$ ), $dim M_{10} = 1$ ( $E_{4} E_{6}$ ), $dim M_{12} = 2$ ( $E_{4}^{3}$ and $E_{6}^{2}$ , equivalently $E_{4}^{3}$ and $Δ$ ). So for $k = 1$ (dimension $4$ ), $φ_{W} (M) \in M_{2} = 0$ , hence $φ_{W} (M) = 0$ for every closed string $4$ -manifold. The first non-vanishing cuspidal value occurs at $k = 6$ (dimension $24$ ): there $S_{12} (SL_{2} (Z)) = C \cdot Δ$ , and the Witten genus of any cuspidal-class string $24$ -manifold is an integer multiple of $Δ$ . The integer is determined by $\hat{A} (M)$ via the $q \to 0$ limit, and on the canonical 24-dimensional string manifold of Stolz (a Wu-manifold-style construction) the multiplier is $\pm 1$ . For the original question on dimension $4$ , the answer is the integer $0$ , and the discriminant $Δ$ does not appear at this dimension. Rubric: full credit for the dimension formula $dim M_{2} = 0$ and the resulting $φ_{W} (M) = 0$ on closed string $4$ -manifolds, with a note that the discriminant $Δ$ enters only at dimension $24$ and beyond.

Lean formalization [Intermediate+]

Mathlib has neither the modular-forms infrastructure nor the multiplicative-sequence framework needed to state the modularity theorem in formal Lean. The intended formalisation would read schematically:

import Mathlib.NumberTheory.ModularForms.Basic
import Mathlib.AlgebraicTopology.OrientedBordism
import Mathlib.AlgebraicGeometry.EllipticCurve.Theta

/-- The Ochanine characteristic power series in `x` with coefficients in `ℚ⟦q⟧`. -/
noncomputable def QOchanine : PowerSeries ℚ⟦q⟧ :=
  sorry  -- (x/2) / sinh(x/2) * ∏ n, (1 - q^n)^2 / ((1 - q^n e^x)(1 - q^n e^{-x}))

/-- The Ochanine elliptic genus as a ring homomorphism from oriented bordism. -/
noncomputable def phiOchanine : Ω^SO ⊗ ℚ →+* ℚ⟦q⟧ :=
  multiplicativeSequenceGenus QOchanine

/-- Modularity of the Ochanine elliptic genus: for a closed spin 4k-manifold M,
    `phiOchanine M` is the q-expansion of a modular form of weight 2k for Γ₀(2). -/
theorem ochanine_modularity
    {M : ClosedManifold} (hM : Spin M) (hdim : M.dim = 4 * k) :
    ∃ (f : ModularForm Γ₀(2) (2 * k)),
      phiOchanine (orientedBordismClass M) = f.qExpansion :=
  sorry

/-- Modularity of the Witten genus: for a closed string 4k-manifold M,
    `phiWitten M` is the q-expansion of a modular form of weight 2k for SL₂(ℤ). -/
theorem witten_modularity
    {M : ClosedManifold} (hM : String M) (hdim : M.dim = 4 * k) :
    ∃ (f : ModularForm SL₂ℤ (2 * k)),
      phiWitten (stringBordismClass M) = f.qExpansion :=
  sorry

/-- Cuspidal limit at q = 0 recovers the A-hat genus. -/
theorem ochanine_cusp_infty
    {M : ClosedManifold} (hM : Spin M) :
    (phiOchanine (orientedBordismClass M)).coeff 0 = AHatGenus M :=
  sorry

The proof gap is structural. Mathlib needs: a named PowerSeries type for $q$ -expansions over $Q$ or $C$ ; a ModularForm Γ k type bundling holomorphic functions on the upper half-plane with the weight- $k$ transformation law and cuspidal regularity; the explicit ring isomorphism $M_{*} (SL_{2} (Z)) = C [E_{4}, E_{6}]$ and its $Γ_{0} (2)$ analogue; the Jacobi theta functions $θ_{1}, θ_{2}, θ_{3}, θ_{4}$ as holomorphic functions on $H \times C$ with their explicit modular transformation laws; the oriented and string bordism rings as named graded $Z$ -algebras; the multiplicative-sequence functor sending a characteristic power series to a ring homomorphism out of bordism; and finally the identification of the Ochanine power series with the theta quotient $θ_{1} (z, τ) / θ_{1}^{'} (0, τ)$ that drives the modularity calculation. Each of these pieces is a substantial formalisation target.

Advanced results [Master]

Theorem (cuspidal limit table; Ochanine 1987, Landweber-Stong 1988). Let $M$ be a closed spin manifold of dimension $4 k$ . The Ochanine elliptic genus $φ_{Och} (M) \in M_{2 k} (Γ_{0} (2))$ satisfies $$ \varphi_{\mathrm{Och}}(M)(i \infty) ;=; \hat A(M), \qquad (W_2 \cdot \varphi_{\mathrm{Och}}(M))(i \infty) ;=; \mathrm{sig}(M) ;=; L(M), $$ where $W_{2} : τ \mapsto - 1/ (2 τ)$ is the Fricke involution on $Γ_{0} (2)$ exchanging the cusps $i \infty$ and $0$ . For $M$ a closed string manifold of dimension $4 k$ , $φ_{W} (M) \in M_{2 k} (SL_{2} (Z))$ satisfies $φ_{W} (M) (i \infty) = \hat{A} (M)$ .

The cuspidal-limit table identifies the elliptic genera as one-parameter modular-form deformations of the classical Hirzebruch genera. At the cusp at infinity ( $q \to 0$ ), both $φ_{Och}$ and $φ_{W}$ specialise to the Dirac index. At the cusp at $0$ , $φ_{Och}$ specialises to the signature. The Witten genus has only one cusp on $SL_{2} (Z)$ and therefore one cuspidal specialisation, the $\hat{A}$ -genus.

Theorem (integrality on spin manifolds; Ochanine 1987). For $M$ a closed spin manifold of dimension $4 k$ , the $q$ -expansion coefficients of $φ_{Och} (M)$ lie in $Z [[q]]$ and the resulting modular form has $Z$ -integral Fourier coefficients.

Integrality on spin manifolds is the topological refinement of modularity that promotes the elliptic genus from a $Q [[q]]$ -valued invariant to a $Z [[q]]$ -valued one. The proof uses the Atiyah-Singer index theorem ([03.09.10]): the Ochanine genus is identified with the formal $S^{1}$ -equivariant Dirac index of a twisted bundle on the loop space, and the integer-valuedness of indices forces integrality.

Theorem (Witten genus and string-bordism; Hopkins-Miller 1994, Ando-Hopkins-Strickland 2001). The Witten genus $\varphi_W : \Omega^{\mathrm{String}}_ \to M_*(\mathrm{SL}_2(\mathbb{Z})) $e x t e n d s t o a r in g - s p ec t r u m - l e v e l ma p o f co h o m o l o g y t h eor i es$ \mathrm{MString} \to \mathrm{TMF} $, w h er e$ \mathrm{TMF}$ is the spectrum of topological modular forms. After taking homotopy groups, this recovers the modular-form-valued Witten genus.*

The Hopkins-Miller refinement is the homotopy-theoretic upgrade of the classical modularity theorem. The Witten genus becomes the genus attached to the cohomology theory $TMF$ whose coefficient ring is roughly the ring of integral modular forms.

Theorem (Landweber-Stong characterisation; Landweber-Stong 1988). A multiplicative genus $\varphi : \Omega^{SO}_ \otimes \mathbb{Z}[1/2] \to R $i s t h e O c hanin ee l l i pt i c g e n u s (u pt o ba sec han g e) i f f i t v ani s h eso n e v er y t o t a l s p a ceo f a s m oo t h$ \mathbb{HP}^1 $- b u n d l eo v er a c l ose d or i e n t e d mani f o l df or w hi c h t h es t r u c t u r e g r o u p r e d u ces t o$ \mathrm{PGL}_2(\mathbb{C})$ acting projectively on the fibre.*

This is the topological characterisation that pins the Ochanine genus down without reference to power series. It says: among all multiplicative genera, the Ochanine elliptic genus is exactly the one annihilating a specific bordism ideal generated by quaternionic-projective-line bundles with a free involution.

Theorem (Bott-Taubes rigidity, modular form refinement; Bott-Taubes 1989, Liu 1995). Let $M^{4 k}$ be a closed spin manifold equipped with a smooth $S^{1}$ -action. The $S^{1}$ -equivariant Ochanine elliptic genus $φ_{Och}^{S^{1}} (M) (q, t) \in M_{2 k} (Γ_{0} (2)) \otimes Z [[t, t^{- 1}]]$ is independent of the equivariant parameter $t$ (rigid).

Rigidity is the equivariant strengthening of modularity. The proof in Liu 1995 uses the modularity result above plus an explicit identification of the equivariant genus with a Jacobi form on $H \times C$ , where the holomorphicity-at-cusps condition forces the equivariant variable to drop out.

Synthesis. Modularity of the elliptic genus is the foundational reason the bordism rings of spin and string manifolds embed into the ring of modular forms, and the central insight is that the Hirzebruch multiplicative-sequence machinery extends from formal power series with rational coefficients to formal power series whose coefficients are themselves modular forms. The bridge from topology to number theory runs through the Jacobi theta function $θ_{1} (z, τ)$ : replacing the $sinh (x /2)$ in the $\hat{A}$ -characteristic series by the theta quotient $θ_{1} (x / (2 π i), τ) / θ_{1}^{'} (0, τ)$ converts the Dirac index into a modular form by absorbing the modular-transformation behaviour of $θ_{1}$ into the Pontryagin-polynomial assembly. Putting these together, the cuspidal limits identify the elliptic genus as a one-parameter modular-form deformation interpolating between the $\hat{A}$ -genus at one cusp and the $L$ -genus at the other. The pattern recurs in the equivariant rigidity theorem of Bott-Taubes, in the Hopkins-Miller TMF refinement, and in the K3-elliptic-genus / Mathieu-moonshine programme of Eguchi-Ooguri-Tachikawa 2010; in each case the modular structure on the topological invariant reflects a deeper geometric symmetry — string-worldsheet modular invariance for the sigma-model interpretation, $S^{1}$ -equivariance for the rigidity theorem, and Monster / Mathieu symmetry for the moonshine connections.

Full proof set [Master]

Theorem (modularity of $φ_{Och}$ ), proof. The argument is the assembly outlined in the Key theorem section, made precise.

Step 1 (Theta-quotient identification). The infinite-product expansion $$ \theta_1(z, \tau) ;=; 2 q^{1/8} \sin(\pi z) \prod_{n \geq 1}(1 - q^n)(1 - q^n e^{2 \pi i z})(1 - q^n e^{-2 \pi i z}) $$ gives, after dividing by $θ_{1}^{'} (0, τ) = 2 π q^{1/8} \prod_{n} (1 - q^{n})^{3}$ and rearranging, $$ \frac{\theta_1(z, \tau)}{\theta_1'(0, \tau)} ;=; \frac{\sin(\pi z)}{\pi} \prod_{n \geq 1} \frac{(1 - q^n e^{2 \pi i z})(1 - q^n e^{-2 \pi i z})}{(1 - q^n)^2}. $$ Substituting $z = x / (2 π i)$ , the $sin (π z) / π$ becomes $sinh (x /2) / π i$ (up to a $π i$ normalisation absorbed into the definitions) and the product becomes the reciprocal of the elliptic-genus product. So $$ Q_{\mathrm{Och}}(x, \tau) ;=; \frac{x \cdot \theta_1'(0, \tau)}{2 \pi i , \theta_1(x/(2 \pi i), \tau)}, $$ verifying Ingredient 1.

Step 2 (Modular transformation of $θ_{1}$ ). The Jacobi theta function transforms as follows under the standard generators $T : τ \mapsto τ + 1$ and $S : τ \mapsto - 1/ τ$ of $SL_{2} (Z)$ : $$ \theta_1(z, \tau + 1) ;=; e^{i \pi / 4} \theta_1(z, \tau), \qquad \theta_1\big(z/\tau, -1/\tau\big) ;=; -i (-i \tau)^{1/2} e^{i \pi z^2 / \tau} \theta_1(z, \tau). $$ For a general $γ = (a c b d) \in SL_{2} (Z)$ , compose to obtain $$ \theta_1\big(z/(c \tau + d), \gamma \tau\big) ;=; \varepsilon(\gamma) (c \tau + d)^{1/2} e^{i \pi c z^2 / (c \tau + d)} \theta_1(z, \tau), $$ where $ε (γ) \in μ_{8}$ is a character on the metaplectic double cover. Restricting to $Γ_{0} (2)$ , the multiplier $ε$ becomes a character of order dividing $8$ on $Γ_{0} (2)$ ; the precise restriction is computed by Skoruppa in Appendix I of Hirzebruch-Berger-Jung.

Step 3 (Assembling the Pontryagin polynomial). The Ochanine genus is $$ \varphi_{\mathrm{Och}}(M)(\tau) ;=; \int_M \prod_{i = 1}^k \frac{x_i \cdot \theta_1'(0, \tau)}{2 \pi i , \theta_1(x_i/(2 \pi i), \tau)} ;=; \int_M \frac{\theta_1'(0, \tau)^k}{(2 \pi i)^k} \prod_{i=1}^k \frac{x_i}{\theta_1(x_i/(2 \pi i), \tau)}. $$ Under $γ \in Γ_{0} (2)$ , each $θ_{1}$ factor in the denominator picks up the multiplier $ε (γ) (c τ + d)^{1/2} e^{iπ c x_{i}^{2} / (- 4 π^{2} (c τ + d))}$ (where the $- 4 π^{2}$ comes from substituting $z = x_{i} / (2 π i)$ ). The factor $θ_{1}^{'} (0, τ)$ in the numerator transforms with weight $3/2$ (one differentiation increases weight by $1$ ), giving the multiplier $ε (γ) (c τ + d)^{3/2}$ . The product of $k$ denominator multipliers contributes $(c τ + d)^{k /2}$ ; the $k$ numerator multipliers contribute $(c τ + d)^{3 k /2}$ ; the net is $(c τ + d)^{k}$ from the theta factors. The integration $\int_{M}$ pairs the result with $[M]$ via Poincaré duality, contributing an additional $(c τ + d)^{k}$ from the volume-form scaling under modular transformations of the Pontryagin roots (each Pontryagin root has formal degree $2$ , and under $τ \mapsto γ τ$ the differentials $d x_{i}$ pick up factor $(c τ + d)$ ). Total weight $2 k$ .

Step 4 (Spin condition kills the Gaussian). The exponential factors $e^{- iπ c x_{i}^{2} / (2 π (c τ + d))}$ across the $k$ factors combine to $e^{- i c (\sum x_{i}^{2}) / (2 (c τ + d))} = e^{- i c p_{1} (M) / (2 (c τ + d))}$ via the splitting principle (with appropriate normalisation). On a spin manifold, $p_{1} (M)$ pairs against $[M]$ to give a Pontryagin number that is an integer multiple of $2$ , and the Gaussian factor expanded as a polynomial in $x_{i}^{2}$ contributes only terms of degree $\leq k$ . The $c$ -dependence in the Gaussian is absorbed by the $ε (γ)^{k}$ multiplier and the multiplier cancellation on $Γ_{0} (2)$ ensures the total contribution is the identity in weight $2 k$ .

Step 5 (Holomorphicity at the cusps). The Jacobi function $θ_{1} (z, τ)$ vanishes only at $z \in Z + τ Z$ (a lattice in $C$ ), and the formal substitution $z = x / (2 π i)$ keeps $z$ away from the lattice for small $x$ . The $q$ -expansion at each cusp is checked directly: at $τ = i \infty$ ( $q \to 0$ ), the theta-quotient power series collapses to $(x /2) / sinh (x /2) = Q_{\hat{A}} (x)$ , and the Ochanine genus reduces to $\hat{A} (M)$ ; at $τ = 0$ (reached by $W_{2}$ ), the analogous calculation reduces to $Q_{L} (x) = (x /2) / tanh (x /2)$ and the genus reduces to $sig (M)$ . Both cuspidal limits are finite, so $φ_{Och} (M)$ is holomorphic at every cusp of $Γ_{0} (2)$ . $□$

Theorem (modularity of $φ_{W}$ ), proof. The Witten characteristic power series differs from the Ochanine one by a Gaussian correction $exp (- p_{1} (M) G_{2} (τ) /2)$ , where $G_{2} (τ) = - 1/24 + \sum_{n \geq 1} σ_{1} (n) q^{n}$ is the quasi-modular Eisenstein series of weight $2$ for $SL_{2} (Z)$ (failing to be modular by a $1/ (c τ + d)^{2} c \cdot$ -anomaly). On a string manifold, $p_{1} /2 = 0$ , so the Gaussian factor is identically $1$ in cohomology. The remaining series transforms as a modular form of weight $2 k$ for the full $SL_{2} (Z)$ : the half-integer-weight multipliers on the theta factors assemble (after $k$ -fold tensoring) into a weight- $2 k$ transformation rule whose multiplier system is identically $1$ on $SL_{2} (Z)$ . Holomorphicity at the unique cusp $i \infty$ is automatic by the same theta-product expansion as in the Ochanine case. $□$

Theorem (cuspidal limit table), proof. At $τ = i \infty$ , $q = e^{2 π i τ} \to 0$ , and the infinite product in $Q_{Och}$ collapses to $1$ , leaving $Q_{Och} (x, 0) = (x /2) / sinh (x /2) = Q_{\hat{A}} (x)$ . The Ochanine genus thus reduces to the $\hat{A}$ -genus. At $τ = 0$ , the Fricke involution $W_{2}$ transforms the theta-quotient into a $θ_{2}$ -quotient, which after substitution becomes $(x /2) / tanh (x /2) = Q_{L} (x)$ , and the genus reduces to the signature. For $φ_{W}$ , the only cusp on $SL_{2} (Z) \ H^{*}$ is $i \infty$ , and the cuspidal limit is again $\hat{A} (M)$ since the Gaussian correction kills off the only deviation from the Ochanine series. $□$

Theorem (integrality on spin manifolds), proof. By the Atiyah-Singer index theorem ([03.09.10]), the Dirac index $\hat{A} (M) = ind (\neq D_{M}^{+})$ is an integer for $M$ closed spin. The Ochanine genus is identified with the formal $S^{1}$ -equivariant Dirac index of a twisted Dirac operator on the formal loop space $L M$ — concretely, the index of $\neq D \otimes ⨂_{n \geq 1} Λ_{q^{n}} T_{C} M \otimes Sym_{q^{n}} T_{C} M$ via the heuristic localisation to constant loops. The $Sym^{k}$ and $Λ^{k}$ bundles have integral Chern classes, so the twisted Dirac index in each degree $q^{n}$ is integral. The full $q$ -expansion $φ_{Och} (M) \in Z [[q]]$ then has integer Fourier coefficients, and the resulting modular form has $Z$ -integral $q$ -expansion. (A rigorous derivation, not requiring the loop-space heuristic, identifies each $q$ -coefficient with a finite Atiyah-Singer index computation on $M$ itself, with the twisting bundle determined by the partition-function expansion.) $□$

Theorem (Landweber-Stong characterisation), proof sketch. The "only if" direction: a direct calculation shows $φ_{Och} (HP^{1} -bundle) = 0$ via the projective-bundle formula. The "if" direction: the bordism ideal generated by such bundles is shown to be the kernel of the Ochanine genus on $Ω_{*}^{S O} \otimes Z [1/2]$ via a direct computation matching the $q$ -coefficients against the elliptic-genus formula at each generator of the bordism ring. The full proof occupies Landweber-Stong 1988 §§3-5. $□$

Theorem (Bott-Taubes rigidity), proof sketch. The equivariant Ochanine genus on a spin $M^{4 k}$ with $S^{1}$ -action is identified (via Liu 1995) with a meromorphic Jacobi form on $H \times C$ of weight $2 k$ and index $0$ ; index- $0$ Jacobi forms on the full $SL_{2} (Z) ⋉ Z^{2}$ semidirect product are constant in the elliptic variable $t$ , giving rigidity. The proof uses modular invariance and the Atiyah-Bott fixed-point theorem. Full details are in Liu 1995 J. Diff. Geom. 41. The cross-reference unit [03.06.24] contains the rigidity argument. $□$

Connections [Master]

Multiplicative sequences and the $L$ -, $\hat{A}$ -, Todd genera 03.06.15. The Ochanine and Witten elliptic genera are multiplicative-sequence genera in the sense of Hirzebruch, but with characteristic power series whose coefficients depend on a formal variable $q$ — equivalently, whose coefficients are themselves modular forms. The unit [03.06.15] supplies the multiplicative-sequence machinery; the present unit extends the machinery from $Q [[x]]$ -valued series to $Q [[q]] [[x]]$ -valued series and identifies the resulting genera as modular-form-valued. The cuspidal limits at $q = 0$ recover the $\hat{A}$ -genus and at the Fricke-image cusp recover the $L$ -genus.
Hirzebruch signature theorem 03.06.11. The signature of a closed oriented $4 k$ -manifold is the $L$ -genus, and the $L$ -genus arises as the Ochanine elliptic genus at the cusp $τ = 0$ of $Γ_{0} (2)$ . The Hirzebruch signature theorem becomes one cuspidal limit of the modularity theorem; the other cuspidal limit gives the Atiyah-Singer index theorem [03.09.10] specialised to the Dirac operator on a spin manifold.
Oriented bordism and the Pontryagin-Thom construction 03.06.13. The Ochanine genus is a ring homomorphism $φ_{Och} : Ω_{*}^{S O} \otimes Z [1/2] \to M_{*} (Γ_{0} (2))$ , where $Ω_{*}^{S O}$ is the oriented bordism ring computed by the Pontryagin-Thom construction [03.06.13]. The Witten genus is the analogous ring homomorphism on the string-bordism ring $Ω_{*}^{String}$ , with target $M_{*} (SL_{2} (Z))$ . Modularity is the statement that the image of these ring homomorphisms lies in the modular-form subring of $Q [[q]]$ .
Atiyah-Singer index theorem 03.09.10. The cuspidal limit at $τ = i \infty$ of $φ_{Och} (M)$ is the Dirac index of the spin Dirac operator on $M$ , which is the $\hat{A}$ -genus by Atiyah-Singer. Every $q$ -coefficient of $φ_{Och} (M)$ is itself an Atiyah-Singer index of a twisted Dirac operator on $M$ — twisted by a polynomial expression in the symmetric and exterior powers of the tangent bundle. Modularity is the homotopy-theoretic reflection of the fact that these twisted indices assemble into a modular form, a phenomenon Witten 1987 interpreted as the worldsheet modular invariance of a two-dimensional supersymmetric sigma model on $M$ .
Modular forms on $SL_{2} (Z)$ 21.04.01. The target of the Witten genus is the ring $M_{*} (SL_{2} (Z)) = C [E_{4}, E_{6}]$ developed in [21.04.01], and the target of the Ochanine genus is the analogous ring $M_{*} (Γ_{0} (2)) = C [δ, ϵ]$ for the level- $2$ congruence subgroup. The dimension formulas $dim M_{k} (SL_{2} (Z))$ developed in [21.04.01] directly control the dimension of the image of $φ_{W}$ in each weight, and the Eichler-Shimura correspondence [21.04.03] extends to a programme connecting weight- $2$ cusp forms with the modularity of elliptic curves — a parallel arithmetic phenomenon to the topological modularity treated here.
Whitehead tower and string structure 03.12.07. The string condition $p_{1} /2 = 0$ that the Witten genus requires is the next level of the Whitehead tower above the spin condition $w_{2} = 0$ , corresponding to killing $π_{3} (Spin)$ . The unit [03.12.07] constructs the Whitehead tower; the string-bordism ring $Ω_{*}^{String}$ is the bordism theory for manifolds with a chosen lift to the third stage. The full modularity for the Witten genus depends decisively on this stronger structure.
Topological modular forms (TMF) and the Hopkins-Miller refinement. The Witten genus lifts to a ring-spectrum-level map $σ_{W} : MString \to TMF$ where $TMF$ is the topological-modular-form spectrum constructed by Goerss-Hopkins-Miller (1994-2002). On homotopy groups this recovers the Witten genus. TMF refines $M_{*} (SL_{2} (Z))$ to an integral version that accounts for torsion phenomena invisible to the rational modular-form valuation. The TMF programme is the natural continuation of the elliptic-genus arc and is treated separately under the elliptic-cohomology successor units.
Bott-Taubes rigidity 03.06.24 pending. The $S^{1}$ -equivariant Ochanine elliptic genus on a spin manifold with circle action is rigid: independent of the equivariant parameter. The proof rests on modularity plus an identification of the equivariant genus with a Jacobi form of index $0$ , where the holomorphicity condition forces the equivariant variable out. Rigidity is the equivariant strengthening of modularity, and the unit [03.06.24] develops it in detail.
Witten heuristic: Dirac operator on the loop space. The Witten genus $φ_{W} (M)$ is heuristically the $S^{1}$ -equivariant index of a Dirac operator on the free loop space $L M = Map (S^{1}, M)$ with respect to loop rotation. Modular invariance is the worldsheet $SL_{2} (Z)$ -symmetry of the toroidal sigma-model partition function: a worldsheet torus has modulus $τ$ , and the partition function is a modular form in $τ$ . The string condition $p_{1} /2 = 0$ is the worldsheet-anomaly-cancellation condition — the same Green-Schwarz mechanism that obstructs string compactifications on non-string manifolds. The unit on the Witten heuristic develops this story; the present modularity statement is its rigorous topological reflection.

Historical & philosophical context [Master]

The elliptic genus was introduced by Serge Ochanine in his 1987 Topology paper "Sur les genres multiplicatifs définis par des intégrales elliptiques" (vol. 26, 143-151) ^{[Ochanine 1987]}. Ochanine defined $φ_{Och}$ as the multiplicative genus attached to the elliptic integral $\int_{0}^{x} (1 - 2 δ t^{2} + ϵ t^{4})^{- 1/2} d t$ , proved its multiplicativity on oriented bordism, and announced the modular-form value of the genus on closed spin manifolds. The elliptic-integral framing explains the name: the characteristic power series is the inversion of an elliptic integral with two parameters $δ, ϵ$ , which become the two generators of $M_{*} (Γ_{0} (2))$ in Skoruppa's normalisation.

The full modularity statement, with the proof technique via Jacobi theta functions, was developed independently and concurrently by Friedrich Hirzebruch, Peter Landweber and Robert Stong, Don Zagier, and Edward Witten in 1987-88. Hirzebruch's 1988 NATO-ASI paper "Elliptic genera of level $N$ for complex manifolds" ^{[Hirzebruch 1988]} gave the multiplicative-sequence framework that integrates Ochanine's genus into the Topological Methods tradition; Landweber-Stong's 1988 Topology paper "Circle actions on spin manifolds and characteristic numbers" (vol. 27, 145-161) ^{[Landweber-Stong 1988]} gave the topological characterisation via vanishing on quaternionic-projective-line bundles; Zagier's 1988 note in LNM 1326 ^{[Zagier 1988]} gave the compact theta-function proof of modularity used in the Key theorem section above; and Witten's 1987 Comm. Math. Phys. paper "Elliptic genera and quantum field theory" ^{[Witten 1987]} gave the physical interpretation as a supersymmetric sigma-model partition function.

The Witten genus, the more refined level- $1$ specialisation requiring a string structure, was introduced by Witten in his 1988 LNM 1326 paper "The index of the Dirac operator in loop space" ^{[Witten 1988]}. Witten interpreted $φ_{W} (M)$ heuristically as the $S^{1}$ -equivariant index of the Dirac operator on the free loop space $L M$ — a construction that has no rigorous Atiyah-Singer foundation but motivates the entire elliptic-genus / TMF programme. The string condition $p_{1} /2 = 0$ is the requirement that the formal Dirac operator on $L M$ be well-defined, which is equivalent to the cancellation of the worldsheet anomaly in the sigma-model picture.

The elliptic-cohomology programme proposed by Graeme Segal in his 1988 Bourbaki talk "Elliptic cohomology" ^{[Segal 1988]} and made rigorous by Landweber-Ravenel-Stong 1995 Contemp. Math. 181 organised the modular-form-valued genus into a generalised cohomology theory $Ell^{*} (-)$ whose coefficient ring is roughly $M_{*} (Γ_{0} (2))$ . The Hopkins-Miller refinement of the mid-1990s produced the topological-modular-form spectrum $TMF$ , lifting the Witten genus to a ring-spectrum map $MString \to TMF$ ; Michael Hopkins described the programme in his 2002 ICM address "Algebraic topology and modular forms" ^{[Hopkins 2002]}. The construction of $TMF$ uses Goerss-Hopkins obstruction theory and the moduli stack of elliptic curves; the proof of the existence of the structured ring-spectrum lift was completed in Ando-Hopkins-Strickland 2001 Invent. Math. 146.

Bibliography [Master]

@book{HirzebruchBergerJung1992,
  author    = {Hirzebruch, Friedrich and Berger, Thomas and Jung, Rainer},
  title     = {Manifolds and Modular Forms},
  series    = {Aspects of Mathematics},
  volume    = {E20},
  publisher = {Vieweg/Friedr. Vieweg \& Sohn},
  address   = {Braunschweig},
  year      = {1992},
  note      = {English ed.\ trans.\ P.~S.~Landweber from the 1987-88 Bonn lectures}
}

@article{Ochanine1987,
  author  = {Ochanine, Serge},
  title   = {Sur les genres multiplicatifs d\'efinis par des int\'egrales elliptiques},
  journal = {Topology},
  volume  = {26},
  year    = {1987},
  pages   = {143--151}
}

@incollection{Hirzebruch1988,
  author    = {Hirzebruch, Friedrich},
  title     = {Elliptic genera of level $N$ for complex manifolds},
  booktitle = {Differential Geometric Methods in Theoretical Physics},
  editor    = {Bleuler, K. and Werner, M.},
  series    = {NATO ASI Series C},
  volume    = {250},
  publisher = {Reidel},
  address   = {Dordrecht},
  year      = {1988},
  pages     = {37--63}
}

@article{LandweberStong1988,
  author  = {Landweber, Peter S. and Stong, Robert E.},
  title   = {Circle actions on spin manifolds and characteristic numbers},
  journal = {Topology},
  volume  = {27},
  year    = {1988},
  pages   = {145--161}
}

@article{Witten1987,
  author  = {Witten, Edward},
  title   = {Elliptic genera and quantum field theory},
  journal = {Communications in Mathematical Physics},
  volume  = {109},
  year    = {1987},
  pages   = {525--536}
}

@incollection{Witten1988,
  author    = {Witten, Edward},
  title     = {The index of the {D}irac operator in loop space},
  booktitle = {Elliptic Curves and Modular Forms in Algebraic Topology},
  editor    = {Landweber, P. S.},
  series    = {Lecture Notes in Mathematics},
  volume    = {1326},
  publisher = {Springer},
  year      = {1988},
  pages     = {161--181}
}

@incollection{Landweber1988,
  author    = {Landweber, Peter S.},
  title     = {Elliptic cohomology and modular forms},
  booktitle = {Elliptic Curves and Modular Forms in Algebraic Topology},
  editor    = {Landweber, P. S.},
  series    = {Lecture Notes in Mathematics},
  volume    = {1326},
  publisher = {Springer},
  year      = {1988},
  pages     = {55--68}
}

@incollection{Zagier1988,
  author    = {Zagier, Don},
  title     = {Note on the {L}andweber-{S}tong elliptic genus},
  booktitle = {Elliptic Curves and Modular Forms in Algebraic Topology},
  editor    = {Landweber, P. S.},
  series    = {Lecture Notes in Mathematics},
  volume    = {1326},
  publisher = {Springer},
  year      = {1988},
  pages     = {216--224}
}

@article{Segal1988,
  author  = {Segal, Graeme B.},
  title   = {Elliptic cohomology},
  journal = {Ast\'erisque},
  volume  = {161--162},
  year    = {1988},
  pages   = {187--201},
  note    = {S\'eminaire Bourbaki, Exp.\ 695}
}

@article{LandweberRavenelStong1995,
  author  = {Landweber, Peter S. and Ravenel, Douglas C. and Stong, Robert E.},
  title   = {Periodic cohomology theories defined by elliptic curves},
  journal = {Contemporary Mathematics},
  volume  = {181},
  year    = {1995},
  pages   = {317--337}
}

@article{BottTaubes1989,
  author  = {Bott, Raoul and Taubes, Clifford},
  title   = {On the rigidity theorems of {W}itten},
  journal = {Journal of the American Mathematical Society},
  volume  = {2},
  year    = {1989},
  pages   = {137--186}
}

@article{Liu1995,
  author  = {Liu, Kefeng},
  title   = {On modular invariance and rigidity theorems},
  journal = {Journal of Differential Geometry},
  volume  = {41},
  year    = {1995},
  pages   = {343--396}
}

@inproceedings{Hopkins2002,
  author    = {Hopkins, Michael J.},
  title     = {Algebraic topology and modular forms},
  booktitle = {Proceedings of the International Congress of Mathematicians, Vol.\ I},
  address   = {Beijing},
  publisher = {Higher Education Press},
  year      = {2002},
  pages     = {291--317}
}

@article{AndoHopkinsStrickland2001,
  author  = {Ando, Matthew and Hopkins, Michael J. and Strickland, Neil P.},
  title   = {Elliptic spectra, the {W}itten genus and the theorem of the cube},
  journal = {Inventiones Mathematicae},
  volume  = {146},
  year    = {2001},
  pages   = {595--687}
}

@article{EguchiOoguriTachikawa2010,
  author  = {Eguchi, Tohru and Ooguri, Hirosi and Tachikawa, Yuji},
  title   = {Notes on the {K3} surface and the {M}athieu group {$M_{24}$}},
  journal = {Experimental Mathematics},
  volume  = {20},
  year    = {2011},
  pages   = {91--96}
}

Prerequisites

03.06.15
03.06.11
03.06.13
03.09.10
21.04.01

Tier anchors

beginner: Hirzebruch-Berger-Jung *Manifolds and Modular Forms* Lecture 6 (informal opening on cuspidal limits); Landweber (ed.) *Elliptic Curves and Modular Forms in Algebraic Topology* (LNM 1326, 1988) preface
intermediate: Hirzebruch-Berger-Jung *Manifolds and Modular Forms* (Aspects of Math. E20, Vieweg 1992) Lectures 4-6 (Ochanine genus, Witten genus, modularity statement and proof sketch); Landweber *Elliptic cohomology and modular forms* LNM 1326 (1988) 55-68
master: Hirzebruch-Berger-Jung *Manifolds and Modular Forms* Lectures 4-6 + Appendix I (full proof of modularity for $\Gamma_0(2)$ and $\mathrm{SL}_2(\mathbb{Z})$ via the Jacobi-theta transformation laws); Ochanine 1987 *Topology* 26, 143-151 (originator paper); Hirzebruch 1988 *Differential Geometric Methods in Theoretical Physics* (NATO ASI C 250, Reidel) 37-63 (Elliptic genera of level $N$ for complex manifolds); Landweber-Stong 1988 *Topology* 27, 145-161 (Circle actions on spin manifolds and characteristic numbers); Zagier 1988 LNM 1326, 216-224 (Note on the Landweber-Stong elliptic genus)

References

TODO_REF
Hirzebruch, F., Berger, T., Jung, R. — Manifolds and Modular Forms · Aspects of Mathematics E20, Vieweg/Friedr. Vieweg & Sohn, Braunschweig 1992. ISBN 3-528-06414-5. English ed. trans. P. S. Landweber from the Bonn 1987-88 lectures. Lecture 4 (Ochanine elliptic genus via the multiplicative-sequence power series), Lecture 5 (Witten genus on string manifolds, formal-group connection), Lecture 6 (the modularity theorem for $\varphi_{\mathrm{Och}} \in M_*(\Gamma_0(2))$ and $\varphi_W \in M_*(\mathrm{SL}_2(\mathbb{Z}))$, cuspidal limits), Appendix I (Skoruppa, ring structure $M_*(\Gamma_0(2)) = \mathbb{C}[\delta, \epsilon]$).
TODO_REF
Ochanine, S. — Sur les genres multiplicatifs définis par des intégrales elliptiques · *Topology* 26 (1987), 143-151. Originator paper. Defines $\varphi_{\mathrm{Och}}$ via the elliptic integral $\int_0^x (1 - 2 \delta t^2 + \epsilon t^4)^{-1/2} \, dt$; proves multiplicativity on oriented bordism, the level-2 modular-form statement, and the spin-integrality refinement.
TODO_REF
Hirzebruch, F. — Elliptic genera of level $N$ for complex manifolds · in *Differential Geometric Methods in Theoretical Physics* (K. Bleuler & M. Werner, eds.), NATO ASI Series C 250, Reidel, Dordrecht 1988, 37-63. Hirzebruch's framing of the elliptic genus inside the multiplicative-sequence formalism of his 1956 *Topological Methods*; explicit power-series definitions and the level-$N$ generalisation.
TODO_REF
Landweber, P. S., Stong, R. E. — Circle actions on spin manifolds and characteristic numbers · *Topology* 27 (1988), 145-161. The Landweber-Stong elliptic genus theorem: a multiplicative genus on $\Omega^{SO}_* \otimes \mathbb{Z}[1/2]$ is the Ochanine genus iff it vanishes on $\mathbb{HP}^{2k+1}$-style manifolds bordant to a quotient by a free $S^1$-action on a spin manifold. Provides the topological characterisation that pins the genus to elliptic integrals.
TODO_REF
Witten, E. — Elliptic genera and quantum field theory · *Comm. Math. Phys.* 109 (1987), 525-536. Originator paper for the physical interpretation: the Ochanine genus is the partition function of an $N=1$ supersymmetric two-dimensional sigma model on $M$; modularity comes from the toroidal worldsheet.
TODO_REF
Witten, E. — The index of the Dirac operator in loop space · in *Elliptic Curves and Modular Forms in Algebraic Topology* (P. S. Landweber, ed.), Lecture Notes in Math. 1326, Springer 1988, 161-181. Originator paper for the Witten genus $\varphi_W$ on string manifolds. Heuristic formula $\varphi_W(M) = \mathrm{ind}_{S^1}(\not{D}_{\mathcal{L}M})$ as the $S^1$-equivariant Dirac index on the free loop space.
TODO_REF
Landweber, P. S. — Elliptic cohomology and modular forms · in *Elliptic Curves and Modular Forms in Algebraic Topology*, LNM 1326, Springer 1988, 55-68. Foundational survey establishing the modular-form values of $\varphi_{\mathrm{Och}}$ and $\varphi_W$ within the elliptic-cohomology programme.
TODO_REF
Zagier, D. — Note on the Landweber-Stong elliptic genus · in *Elliptic Curves and Modular Forms in Algebraic Topology*, LNM 1326, Springer 1988, 216-224. Compact rederivation of the modularity statement via the Jacobi theta function $\theta_1(z, \tau)$ and explicit identification of $\varphi_{\mathrm{Och}}$ with the ratio $\theta_1(z) \theta_1'(0) / \theta_1'(0) \theta_1(z)$-style expression giving the level-2 modular form.
TODO_REF
Segal, G. B. — Elliptic cohomology · *Séminaire Bourbaki* Exposé 695, *Astérisque* 161-162 (1988), 187-201. Proposes the elliptic-cohomology programme: the Ochanine and Witten genera as the genera attached to a generalised cohomology theory $\mathrm{Ell}^*$ whose coefficient ring is a ring of modular forms.
TODO_REF
Hopkins, M. J. — Algebraic topology and modular forms · Proc. ICM 2002, Beijing, Vol. I, 291-317. The topological-modular-forms (TMF) refinement: $\varphi_W$ lifts to a map from string-bordism to TMF, identifying the Witten genus as the ring-spectrum-level realisation of the modular-form invariant.
TODO_REF
Liu, K. — On modular invariance and rigidity theorems · *J. Differential Geom.* 41 (1995), 343-396. Clean modular-invariance-based proof of Bott-Taubes rigidity; the rigidity statement provides an alternative route to the modularity of $\varphi_{\mathrm{Och}}^{S^1}$ via $S^1$-equivariant arguments.

Reviewer

TBD

Estimated time

beginner: 18m
intermediate: 45m
master: 90m