Mirror Symmetry

Mirror symmetry is a duality, originating in string theory, asserting that for many Calabi–Yau manifolds

X

there exists a "mirror" Calabi–Yau

X^{\lor}

whose complex (algebraic) geometry encodes the symplectic (enumerative) geometry of

X

, and vice versa. Key result in words: counting rational curves on

X

(a hard symplectic problem) is computed by classical period integrals (variation of Hodge structure) on

X^{\lor}

. Where it applies: Calabi–Yau threefolds, Fano/toric varieties via Landau–Ginzburg models, and broadly across algebraic geometry, symplectic topology, and theoretical physics. Who/when: conjectured by physicists (Dixon; Lerche–Vafa–Warner; Greene–Plesser) in the late 1980s; spectacularly confirmed by Candelas–de la Ossa–Green–Parkes (1991) for the quintic threefold; formalized by Kontsevich (1994) and Strominger–Yau–Zaslow (1996).

Mirror symmetry relates two a priori unrelated branches of geometry on a Calabi–Yau manifold: the A-model, governed by symplectic structure and Gromov–Witten invariants (curve counts), and the B-model, governed by complex structure and periods of holomorphic forms. The conjecture pairs each Calabi–Yau $X$ with a mirror $X^{\lor}$ so that the A-model on $X$ equals the B-model on $X^{\lor}$ and conversely; in particular the Hodge numbers are swapped, $h^{p, q} (X) = h^{n - p, q} (X^{\lor})$ . It matters because it converts intractable, transcendental enumerative computations into solvable systems of ODEs (Picard–Fuchs equations), and because it knits together symplectic topology, derived categories, and integrable systems into one conceptual framework. The surprising fact: a physical "stringy" symmetry predicted the number of rational curves of every degree on the quintic — numbers mathematicians could not even compute for degree 4 at the time — and was later proven correct.

The Core Insight

A string propagating on a Calabi–Yau $X$ "sees" only the conformal field theory (CFT) attached to $X$ , not $X$ itself. That CFT carries two natural $N = 2$ supersymmetry algebras whose roles can be exchanged by a trivial relabeling $(Q_{+} \leftrightarrow \overset{ˉ}{Q}_{+})$ . On the geometric side this relabeling swaps the symplectic data (Kähler moduli, $H^{1, 1}$ ) with the complex-structure data (complex moduli, $H^{n - 1, 1}$ ). A second manifold $X^{\lor}$ realizing the relabeled CFT is the mirror. Hence two different geometries give the same physics, and quantities that are deep on one side become elementary on the other.

Formal Statement

Calabi–Yau Manifold

A Calabi–Yau $n$ -fold is a compact Kähler manifold $X$ of complex dimension $n$ with trivial canonical bundle, $K_{X} ≅ O_{X}$ , equivalently with a nowhere-vanishing holomorphic $n$ -form $Ω \in H^{0} (X, Ω_{X}^{n})$ . By Yau's theorem it admits a unique Ricci-flat Kähler metric in each Kähler class. The relevant moduli are:

Complex-structure moduli, dimension $h^{n - 1, 1} (X)$ , deformations of $Ω$ ;
Complexified Kähler moduli, dimension $h^{1, 1} (X)$ , parametrizing $B + i J$ where $J$ is the Kähler form and $B$ the B-field.

Mirror Pair (Hodge-Theoretic Form)

Two Calabi–Yau $n$ -folds $X$ and $X^{\lor}$ form a mirror pair if their Hodge diamonds are reflected across the diagonal:

h^{p, q} (X) = h^{n - p, q} (X^{\lor}), 0 \leq p, q \leq n .

⟳the CFT relabeling

Q_{+} \leftrightarrow \overset{ˉ}{Q}_{+}

acts on the U(1) R-charges

(q, \overset{q}{ˉ})

(q, \overset{q}{ˉ}) \mapsto (- q, \overset{q}{ˉ})

; geometric (a,c)-ring elements of charge

(p, q)

map to (c,c)-ring elements, exchanging

H^{p, q}

with

H^{n - p, q}

In particular for threefolds $(n = 3)$ the Euler characteristics satisfy

χ (X^{\lor}) = - χ (X), χ = 2 (h^{1, 1} - h^{2, 1}) .

⟳substitute

h^{1, 1} \leftrightarrow h^{2, 1}

into

χ = 2 (h^{1, 1} - h^{2, 1})

valid for simply-connected CY3.

The Mirror Theorem (Givental; Lian–Liu–Yau, 1996–97)

Let $X \subset P^{4}$ be the quintic threefold. Define the A-model generating function of genus-0 Gromov–Witten invariants $n_{d}$ (instanton numbers) through the prepotential, and the B-model Yukawa coupling $Y (q)$ obtained from periods of the mirror $X^{\lor}$ . Then the two coincide after the mirror map $q = q (t)$ :

Y_{A} (q) = 5 + d \geq 1 \sum n_{d} d^{3} \frac{q ^{d}}{1 - q ^{d}} = Y_{B} (q (τ)),

where the B-side

Y_{B}

is the normalized triple period and

q (τ)

is the exponentiated mirror map. The first instanton numbers are

n_{1} = 2875, n_{2} = 609250, n_{3} = 317206375, \dots

Picard–Fuchs Equation of the Mirror Quintic

The mirror quintic family $X_{ψ}^{\lor}$ (the Greene–Plesser orbifold of $\sum x_{i}^{5} - 5 ψ \prod x_{i} = 0$ ) has its holomorphic period $ϖ_{0} (z)$ , with $z = (5 ψ)^{- 5}$ , annihilated by the hypergeometric operator

L = θ^{4} - 5 z (5 θ + 1) (5 θ + 2) (5 θ + 3) (5 θ + 4), θ = z \frac{d}{d z} .

⟳

L

is the Gel'fand–Kapranov–Zelevinsky (GKZ) hypergeometric operator for the toric ambient; alternatively expand

ϖ_{0} = \sum_{m} \frac{( 5 m )!}{( m ! ) ^{5}} z^{m}

and check

θ^{4} ϖ_{0} = 5 z (5 θ + 1) \dots (5 θ + 4) ϖ_{0}

Why It Holds / Derivation

From Periods to Curve Counts (Outline of the CDOGP Computation)

Step 1 — Solve the Picard–Fuchs equation. Near

z = 0

the operator

L

of [4] has a maximally unipotent monodromy point. Its four solutions take Frobenius form

ϖ_{0}, ϖ_{1} = ϖ_{0} lo g z + \dots, ϖ_{2}, ϖ_{3},

with

ϖ_{0} (z) = \sum_{m \geq 0} \frac{( 5 m )!}{( m ! ) ^{5}} z^{m}

⟳apply the Frobenius method to a fourth-order ODE with a regular singular point whose indicial equation has a quadruple root

0

, producing logarithms of degrees

0, 1, 2, 3

Step 2 — Define the mirror map. The flat coordinate on Kähler moduli is the ratio of the single-log to the holomorphic period:

t (z) = \frac{ϖ _{1} ( z )}{ϖ _{0} ( z )} = lo g z + \frac{770}{1} z + \dots, q = e^{2 π i t} (suitably normalized) .

⟳the asymptotics

t \sim lo g z

identify

t

with the complexified Kähler parameter

B + i J

; invert the series

z (q)

by Lagrange inversion.

Step 3 — Compute the B-model Yukawa coupling. Using

Ω_{ψ}

and Griffiths transversality, the triple coupling is

Y_{z z z} = - \int_{X^{\lor}} Ω \land \partial_{z}^{3} Ω = \frac{5}{( 2 π i ) ^{3}} \frac{1}{1 - 5 ^{5} z} \frac{1}{ϖ _{0}^{2}} \cdot (Wronskian normalization) .

⟳differentiate

\int Ω \land \partial_{z}^{k} Ω = 0

for

k < 3

(Griffiths transversality), reduce

\partial_{z}^{4} Ω

via

L

, and integrate the resulting first-order ODE for

Y_{z z z}

Step 4 — Change to flat coordinates. Convert to the

t

-coordinate and normalize by

ϖ_{0}

Y_{ttt} = \frac{1}{ϖ _{0}^{2}} (\frac{d z}{d t})^{3} Y_{z z z} .

⟳the Yukawa coupling is a section of (line bundle)

^{\otim)}

; the factor

ϖ_{0}^{- 2}

removes the Kähler ambiguity and

(d z / d t)^{3}

is the chain rule for the third covariant derivative.

Step 5 — Match to the A-model and extract $n_{d}$ . Expanding [8] in

q

and equating with [3] gives the integers

n_{d}

. The classical term

5 = \int_{X} c_{1} (O (1))^{3}

is the degree of the quintic; the rest is the instanton sum.

⟳read off coefficients of

q^{d} / (1 - q^{d})

, the multiple-cover formula

\sum_{k ∣ d} n_{d / k} / k^{3}

being already incorporated by the Aspinwall–Morrison factor

d^{3}

The agreement of [3], originally a physical prediction, was later proven rigorously by Givental and by Lian–Liu–Yau using equivariant localization on moduli of stable maps. $■$

Brian Greene & Ronen Plesser | 1990

Greene and Plesser gave the first systematic construction of mirror pairs by quotienting a Calabi–Yau by a discrete group of phase symmetries that act trivially on the worldsheet CFT (the Greene–Plesser orbifold). For the Fermat quintic they identified the mirror as the resolution of ${\sum x_{i}^{5} = 5 ψ x_{0} x_{1} x_{2} x_{3} x_{4}} / (Z_{5})^{3}$ . This explicit recipe transformed mirror symmetry from a numerological coincidence (matching Hodge numbers across Calabi–Yau databases) into a usable computational engine, and it set the stage for the 1991 quintic prediction.

Building Intuition

Think of a Calabi–Yau as carrying two independent "shape knobs": one for size/symplectic form (how big the holes are, controlled by Kähler moduli) and one for complex shape (how the holes are tilted, controlled by complex moduli). Quantum-mechanically, a string is too dumb to tell which knob is which — it only feels the resulting energy spectrum. Mirror symmetry is the statement that there's a second manifold whose two knobs are wired in the opposite order, yet produces the identical spectrum. Now here is the leverage: quantum corrections (worldsheet instantons = holomorphic curves) appear in the symplectic sector but the complex sector is exact at tree level (no instantons, because complex deformations don't wrap curves). So if you want the hard, instanton-corrected answer on $X$ , you compute the easy, classical answer on $X^{\lor}$ . It is like discovering that the value of a divergent, instanton-summed series equals a simple ratio of hypergeometric functions on a mirror world.

A-model ↔ B-model exchange across a mirror pair

<svg viewBox="0 0 760 380" xmlns="http://www.w3.org/2000/svg">
<rect width="760" height="380" fill="#0a0e17"/>
<text x="380" y="30" fill="#e2e8f0" font-size="18" text-anchor="middle" font-family="serif">Mirror Symmetry: exchange of geometric data</text>

<!-- Left box: X -->
<rect x="60" y="70" width="250" height="220" rx="10" fill="none" stroke="#94a3b8" stroke-width="1.5"/>
<text x="185" y="100" fill="#e2e8f0" font-size="16" text-anchor="middle">Calabi–Yau X</text>
<text x="185" y="135" fill="#d4a853" font-size="14" text-anchor="middle">A-model (symplectic)</text>
<text x="185" y="160" fill="#e2e8f0" font-size="12" text-anchor="middle">Kähler moduli  h^{1,1}(X)</text>
<text x="185" y="182" fill="#e2e8f0" font-size="12" text-anchor="middle">Gromov–Witten n_d</text>
<text x="185" y="204" fill="#e2e8f0" font-size="12" text-anchor="middle">curve counts (quantum)</text>
<text x="185" y="240" fill="#94a3b8" font-size="12" text-anchor="middle">Fukaya category</text>
<text x="185" y="262" fill="#94a3b8" font-size="12" text-anchor="middle">Lagrangians</text>

<!-- Right box: X^v -->
<rect x="450" y="70" width="250" height="220" rx="10" fill="none" stroke="#94a3b8" stroke-width="1.5"/>
<text x="575" y="100" fill="#e2e8f0" font-size="16" text-anchor="middle">Mirror X^∨</text>
<text x="575" y="135" fill="#d4a853" font-size="14" text-anchor="middle">B-model (complex)</text>
<text x="575" y="160" fill="#e2e8f0" font-size="12" text-anchor="middle">complex moduli  h^{n-1,1}(X^∨)</text>
<text x="575" y="182" fill="#e2e8f0" font-size="12" text-anchor="middle">periods of Ω</text>
<text x="575" y="204" fill="#e2e8f0" font-size="12" text-anchor="middle">Picard–Fuchs (classical)</text>
<text x="575" y="240" fill="#94a3b8" font-size="12" text-anchor="middle">D^b Coh</text>
<text x="575" y="262" fill="#94a3b8" font-size="12" text-anchor="middle">coherent sheaves</text>

<!-- Arrows -->
<path d="M310 150 C 380 110, 380 110, 450 150" fill="none" stroke="#d4a853" stroke-width="2" marker-end="url(#a)"/>
<path d="M450 220 C 380 260, 380 260, 310 220" fill="none" stroke="#d4a853" stroke-width="2" marker-end="url(#a)"/>
<defs>
<marker id="a" markerWidth="10" markerHeight="10" refX="8" refY="3" orient="auto">
<path d="M0,0 L8,3 L0,6 Z" fill="#d4a853"/>
</marker>
</defs>
<text x="380" y="105" fill="#d4a853" font-size="12" text-anchor="middle">mirror map  q=q(t)</text>
<text x="380" y="285" fill="#d4a853" font-size="12" text-anchor="middle">h^{p,q} ↔ h^{n-p,q}</text>

<text x="380" y="345" fill="#e2e8f0" font-size="13" text-anchor="middle">Hard quantum counts on X  =  easy classical periods on X^∨</text>
</svg>

Worked Examples

Instanton Numbers of the Quintic from the Picard–Fuchs Equation

Given. The operator

L = θ^{4} - 5 z (5 θ + 1) (5 θ + 2) (5 θ + 3) (5 θ + 4)

of [4],

z = (5 ψ)^{- 5}

. Step 1. Holomorphic period:

ϖ_{0} = \sum_{m} \frac{( 5 m )!}{( m ! ) ^{5}} z^{m} = 1 + 120 z + 113400 z^{2} + \dots

. Step 2. Log period and mirror map:

t = lo g z + 5 (120) z + \dots

, giving the inverse series

z (q) = q - 770 q^{2} + \dots

with

q = e^{t}

(normalized). Step 3. B-model Yukawa from [7]–[8]:

Y_{ttt} = \frac{5}{1 - 5 ^{5} z} \frac{1}{ϖ _{0}^{2}} (\frac{z ^{'}}{1})^{3}_{z = z (q)} = 5 + 2875 q + 4876875 q^{2} + \dots

Step 4. Match to the A-model formula [3],

Y = 5 + \sum n_{d} d^{3} \frac{q ^{d}}{1 - q ^{d}}

Coefficient of $q$ : $n_{1} \cdot 1^{3} = 2875 \Rightarrow n_{1} = 2875$ .
Coefficient of $q^{2}$ : $n_{1} \cdot 1^{3} + n_{2} \cdot 2^{3} = 4876875 \Rightarrow 8 n_{2} = 4876875 - 2875 = 4874000 \Rightarrow n_{2} = 609250$ .

Answer.

n_{1} = 2875, n_{2} = 609250, n_{3} = 317206375 (rational curves of given degree)

These are dimensionless counts (genus-0 Gromov–Witten / BPS instanton numbers).

Hodge Diamond Reflection for the Quintic Pair

Given. The quintic threefold

X

has

h^{1, 1} (X) = 1

h^{2, 1} (X) = 101

. Step 1. Apply [1] with

n = 3

h^{1, 1} (X^{\lor}) = h^{2, 1} (X) = 101

, and

h^{2, 1} (X^{\lor}) = h^{1, 1} (X) = 1

. Step 2. Euler characteristics from [2]:

χ (X) = 2 (1 - 101) = - 200

χ (X^{\lor}) = 2 (101 - 1) = + 200

. Step 3. Check the reflection:

χ (X^{\lor}) = - χ (X)

✓. The mirror has one complex-structure modulus

ψ

(matching the single Kähler modulus of

X

) — exactly the one-parameter family used in [4]. Answer.

(h^{1, 1}, h^{2, 1}) (X) = (1, 101) ⟷ (h^{1, 1}, h^{2, 1}) (X^{\lor}) = (101, 1), χ = \mp 200

Common Misconceptions

Mirror symmetry is not a homeomorphism or diffeomorphism.

X

and

X^{\lor}

are generally topologically distinct manifolds — their Betti numbers

b_{2}, b_{4}

are swapped, so they are not even homotopy equivalent in general. The statement is an equivalence of physical theories / categorical structures, not of underlying spaces. Likewise, the "mirror map"

q = q (t)

is a transcendental change of coordinates between moduli spaces, not an isomorphism of varieties; naively identifying

z

with

q

gives the wrong instanton numbers.

Not Every Calabi–Yau Has a (Geometric) Mirror

The mirror of a rigid Calabi–Yau threefold $X$ (one with $h^{2, 1} (X) = 0$ , hence no complex deformations) would need $h^{1, 1} (X^{\lor}) = 0$ — but a Kähler manifold must have $h^{1, 1} \geq 1$ . So a rigid Calabi–Yau cannot have a Calabi–Yau mirror in the ordinary sense. The resolution is that its "mirror" lives outside ordinary geometry: it is a Landau–Ginzburg model or a non-geometric CFT. This shows mirror symmetry is fundamentally a statement about CFTs/categories, with manifolds being only special, geometric realizations.

Connections

Mirror symmetry sits at the crossroads of several theories. Kontsevich's Homological Mirror Symmetry reformulates it as an equivalence of triangulated categories, $D^{b} Coh (X^{\lor}) ≃ D^{π} Fuk (X)$ , replacing numerical curve counts with a categorical statement and tying it to Derived Categories and the Fukaya Category. The Strominger–Yau–Zaslow Conjecture gives a geometric mechanism: Calabi–Yau manifolds admit special Lagrangian $T^{n}$ -fibrations, and mirroring is fiberwise dualization of these tori (T-duality), linking the subject to Special Lagrangian Submanifolds and integrable systems. On the computational side it is inseparable from Gromov–Witten Invariants, Variation of Hodge Structure, and the Picard–Fuchs Equations coming from Toric Geometry via the Batyrev construction over reflexive polytopes. Physically it descends from $N = (2, 2)$ Superconformal Field Theory and the Topological String.

Real-World Applications

Enumerative algebraic geometry. Mirror symmetry computed previously inaccessible numbers of rational curves on the quintic and other Calabi–Yau hypersurfaces — the Candelas–de la Ossa–Green–Parkes (1991) instanton numbers — launching the entire field of modern Gromov–Witten theory.
Batyrev–Borisov toric constructions. Reflexive polytopes and nef partitions produce explicit mirror pairs for thousands of toric Calabi–Yau complete intersections, used to populate string vacua databases (e.g. the Kreuzer–Skarke list of $\sim$ 473 million reflexive 4-polytopes).
String compactification & particle physics. Mirror symmetry determines the Yukawa couplings and moduli-space metrics of the four-dimensional effective field theory obtained by compactifying the heterotic/Type II string on a Calabi–Yau, central to attempts at the Standard Model from strings.
Knot theory and Chern–Simons. Large- $N$ duality (Gopakumar–Vafa, Ooguri–Vafa) relates open mirror symmetry on the resolved conifold to Chern–Simons invariants and HOMFLY polynomials of knots.
Wall-crossing and Donaldson–Thomas theory. The Kontsevich–Soibelman wall-crossing formula, organizing BPS state counts (DT invariants), is governed by mirror-symmetric structures and stability conditions on $D^{b} Coh$ .
Symplectic topology / Floer theory. Homological mirror symmetry produces concrete computations of Fukaya categories and Floer cohomology of Lagrangian tori, e.g. for $P^{2}$ and toric Fanos via their Landau–Ginzburg mirrors.
Cluster algebras & scattering diagrams. The Gross–Siebert program reconstructs mirror families combinatorially from tropical/scattering data, with downstream impact on the theory of cluster varieties and canonical bases.

Historical Development

1985. Candelas–Horowitz–Strominger–Witten show that $N = 1$ supersymmetric string compactification requires a Calabi–Yau threefold, focusing attention on these spaces.
1987–88. L. Dixon, and independently Lerche–Vafa–Warner, observe that the $N = 2$ worldsheet algebra admits an automorphism exchanging the (a,c)- and (c,c)-rings, predicting paired CFTs — the first hint of mirror symmetry.
1989. Candelas–Lynker–Schimmrich scan large databases of Calabi–Yau hypersurfaces in weighted projective space and find a striking symmetry of the $(h^{1, 1}, h^{2, 1})$ plot under reflection.
1990. Greene and Plesser construct mirror pairs explicitly via orbifolding by phase symmetries, proving the duality at the level of CFTs for Fermat-type families.
1991. Candelas–de la Ossa–Green–Parkes compute the mirror quintic periods, predict the instanton numbers $n_{1} = 2875, n_{2} = 609250, \dots$ , and verify $n_{1}, n_{2}$ against then-known algebraic geometry — a landmark prediction.
1993. Batyrev gives the general toric construction of mirror pairs via reflexive polytopes and polar duality; Batyrev–Borisov extend it to complete intersections (1994).
1994. Kontsevich, in his ICM address, formulates Homological Mirror Symmetry as an equivalence of derived/Fukaya categories.
1996. Strominger, Yau, and Zaslow propose the SYZ picture: mirror symmetry as T-duality along special Lagrangian torus fibrations.
1996–97. Givental, and Lian–Liu–Yau, give rigorous proofs of the mirror theorem (the CDOGP prediction) for toric complete intersections, turning the physics prediction into a theorem.
2000s–2010s. Gross–Siebert develop an algebraic-geometric reconstruction program; Seidel, Abouzaid, and others prove HMS in many cases (quartic K3, projective spaces, genus-2 curves).

Subtleties & Limits

Despite enormous progress, the conjecture is not fully proven in general. (1) Homological mirror symmetry is established only for special families (abelian varieties, certain toric and Fano cases, some hypersurfaces); a general proof for arbitrary Calabi–Yau pairs is open, and the precise definition of the Fukaya category (convergence, $A_{\infty}$ -obstructions, bulk deformations) remains technically delicate. (2) The SYZ conjecture's analytic core — existence of special Lagrangian $T^{n}$ -fibrations and control of their singular fibers — is largely unproven; instanton corrections to the naive fiberwise dualization require the full Gross–Siebert machinery of scattering diagrams. (3) Higher-genus mirror symmetry (BCOV theory and the holomorphic anomaly equations) lacks a complete mathematical foundation, though Costello–Si Li and Fan–Jarvis–Ruan have made deep progress. (4) Mirror symmetry for non-Calabi–Yau spaces (Fano via Landau–Ginzburg, general type) is an active frontier with no unified statement. (5) Identifying the actual mirror manifold from intrinsic data (beyond toric/orbifold recipes) is not algorithmic in general.

Mirror Symmetry

The Core Insight

Formal Statement

Calabi–Yau Manifold

Mirror Pair (Hodge-Theoretic Form)

The Mirror Theorem (Givental; Lian–Liu–Yau, 1996–97)

Picard–Fuchs Equation of the Mirror Quintic

Why It Holds / Derivation

From Periods to Curve Counts (Outline of the CDOGP Computation)

Brian Greene & Ronen Plesser | 1990

Building Intuition

A-model ↔ B-model exchange across a mirror pair

Worked Examples

Instanton Numbers of the Quintic from the Picard–Fuchs Equation

Hodge Diamond Reflection for the Quintic Pair

Common Misconceptions

Not Every Calabi–Yau Has a (Geometric) Mirror

Connections

Real-World Applications

Historical Development

Subtleties & Limits

See Also

Before this

Where this leads