Reflexive polytope and Batyrev mirror duality (pointer)

Intuition [Beginner]

A reflexive polytope is a lattice shape with a special center. Place the origin inside the shape. Now build the polar shape by asking which lattice directions stay at distance at most one from every point of the first shape. If the polar shape also has lattice vertices, the original shape is reflexive.

The word "reflexive" means that the shape reflects perfectly through this polar-dual operation. Starting from $P$, the polar is $P^{\circ }$. Starting again from $P^{\circ }$, the polar comes back to $P$. In toric geometry this reflection is not only a feature of convex geometry. It produces a pair of Calabi-Yau hypersurfaces that mirror each other.

The simplest picture is the square with vertices $(1,0)$, $(0,1)$, $(-1,0)$, and $(0,-1)$. Its polar is the square with vertices $(1,1)$, $(-1,1)$, $(-1,-1)$, and $(1,-1)$. Both are lattice polygons, so each is reflexive.

Visual [Beginner]

Draw a lattice polygon around the origin. Each side is one lattice step away from the origin. Then draw the polar polygon in the dual lattice. Vertices of one shape correspond to side directions of the other.

The mirror-symmetry slogan is: a Calabi-Yau hypersurface built from $P$ has a mirror built from $P^{\circ }$. The combinatorial exchange of vertices and sides becomes a geometric exchange of shape data and complex-structure data.

Worked example [Beginner]

Take the diamond $$ P=\operatorname{conv}{(1,0),(0,1),(-1,0),(0,-1)}. $$

A point $(a,b)$ lies in the polar when its pairing with every point of the diamond is at least $-1$. Checking the four vertices gives four bounds: $a\ge -1$, $b\ge -1$, $a\le 1$, and $b\le 1$. The polar is the square $[-1,1]\times [-1,1]$.

The square has lattice vertices, so the diamond is reflexive. The square is reflexive too, because applying the same polar construction returns the diamond. This two-dimensional example is the visible model for Batyrev's higher-dimensional construction.

Check your understanding [Beginner]

Formal definition [Intermediate+]

Let $M\cong {\mathbb{Z}}^n$ be a lattice, let $N=\mathrm{Hom}(M,\mathbb{Z})$, and write $M_{\mathbb{R}}=M{\otimes }_{\mathbb{Z}}\mathbb{R}$, $N_{\mathbb{R}}=N{\otimes }_{\mathbb{Z}}\mathbb{R}$. A full-dimensional lattice polytope $P\subset M_{\mathbb{R}}$ with $0\in \mathrm{int}(P)$ has polar dual $$ P^\circ={u\in N_\mathbb{R}:\langle m,u\rangle\geq -1\text{ for every }m\in P}. $$ The polytope $P$ is reflexive if $P^{\circ }$ is a lattice polytope. The bipolar theorem gives $(P^{\circ }{)}^{\circ }=P$, so reflexivity is symmetric in $M$ and $N$.

There is an equivalent facet description. Write each facet of $P$ as $$ \langle m,u_F\rangle\geq -a_F $$ with $u_F\in N$ primitive and $a_F\in {\mathbb{Z}}_{>0}$. Then $P$ is reflexive iff every $a_F=1$. Thus every facet sits at lattice distance one from the origin. This is the numerical form used in toric divisor calculations.

The toric meaning is the following. Let ${\Sigma }_P$ be the normal fan of $P$, and let $X_P=X_{{\Sigma }_P}$. Reflexivity is equivalent to saying that $X_P$ is a Gorenstein Fano toric variety and that the anticanonical divisor $-K_{X_P}$ has polytope $P$. A generic section of $-K_{X_P}$ cuts out a Calabi-Yau hypersurface by adjunction whenever the ambient singularities are handled by the usual toric crepant-resolution or mild-singularity framework.

Key theorem with proof [Intermediate+]

Theorem (Batyrev reflexive-polytope mirror construction). Let $P\subset M_{\mathbb{R}}$ be a reflexive polytope of dimension $n+1$, and let $P^{\circ }\subset N_{\mathbb{R}}$ be its polar dual. The normal toric varieties $X_P$ and $X_{P^{\circ }}$ are Gorenstein Fano toric varieties. Generic anticanonical hypersurfaces $Y\subset X_P$ and $Y^{\circ }\subset X_{P^{\circ }}$ are Calabi-Yau $n$-folds in the toric sense. After choosing maximal projective crepant resolutions when they exist, Batyrev's combinatorial Hodge formulas exchange their Hodge numbers: $$ h^{p,q}(\widehat{Y})=h^{n-p,q}(\widehat{Y^\circ}). $$

Proof. Reflexivity gives a polar pair $P\leftrightarrow P^{\circ }$ with both polytopes lattice polytopes and with all facets at lattice distance one from the origin. By the divisor-polytope construction of 04.11.08, the toric divisor polytope of $-K_{X_P}={\sum }_{\rho \in {\Sigma }_P(1)}{D}_{\rho }$ is exactly $P$. The facet-distance-one condition is exactly the Cartier condition for the canonical divisor to be integral Cartier after sign change; ampleness comes from the strict convexity of the support function of the normal fan. Hence $X_P$ is Gorenstein Fano, and the same argument applies to $P^{\circ }$.

For a generic anticanonical section $s\in H^0(X_P,-K_{X_P})$, let $Y=V(s)$. Adjunction gives $$ K_Y=(K_{X_P}+Y)|Y=(K{X_P}-K_{X_P})|_Y\cong\mathcal{O}_Y, $$ so $Y$ has vanishing canonical class. Toric Bertini-type statements give smoothness on the dense torus and mild singular behaviour along boundary strata; a maximal projective crepant resolution preserves the canonical class. Thus $\hat{Y}$ is a Calabi-Yau variety in Batyrev's setup.

Batyrev computes the stringy Hodge numbers of $Y$ from face data of $P$ and $P^{\circ }$: lattice points in faces, interior lattice points of dual faces, and alternating sums arranged by codimension. The formula is symmetric under exchanging a face $\Theta \subset P$ with its dual face ${\Theta }^{\circ }\subset P^{\circ }$, while the degree index changes from $p$ to $n-p$. Therefore the resolved hypersurfaces have exchanged Hodge diamonds. This proves the mirror statement. ^{[Batyrev 1994]}

Bridge. Reflexive polytopes build toward the Gross-Siebert reconstruction theorem 04.12.09 and appear again in the period-integral mirror-map unit 04.12.13. The foundational reason is that the facet-distance-one condition identifies anticanonical geometry with lattice polar duality, and this is exactly the bridge from the polytope-fan dictionary 04.11.10 to Calabi-Yau mirror symmetry. Putting these together generalises the divisor-polytope calculus of 04.11.08, is dual to the tropical degeneration picture of 04.12.07, and identifies polarity with mirror exchange.

Exercises [Intermediate+]

Advanced results [Master]

Batyrev's construction is stronger than the slogan "dual polytopes give mirrors." The actual theorem computes Hodge numbers from combinatorial data. For a Calabi-Yau hypersurface of dimension $n$ in the toric Fano variety of a reflexive polytope $\Delta$, the correction terms involve lattice points in faces $\Theta \subset \Delta$ and interior lattice points in the dual faces ${\Theta }^{\circ }\subset {\Delta }^{\circ }$. The exchange $\Theta \leftrightarrow {\Theta }^{\circ }$ is what swaps $h^{1,1}$ with $h^{n-1,1}$ and, more generally, produces the Hodge-diamond reflection.

The crepant-resolution clause is not cosmetic. Reflexive polytopes often define singular Gorenstein Fano toric varieties. In dimension four, maximal projective crepant partial resolutions are sufficient for Calabi-Yau threefold hypersurfaces to have the expected smooth model, and Batyrev's stringy-Hodge formalism records the resolution-independent invariant. Kreuzer and Skarke's classification of 473,800,776 four-dimensional reflexive polytopes turned this theorem into a large explicit database of mirror Calabi-Yau threefolds. ^{[Kreuzer-Skarke 2000]}

Batyrev-Borisov extends the construction from hypersurfaces to complete intersections. A nef partition decomposes the vertex set or divisor class into pieces whose Minkowski data define a Calabi-Yau complete intersection. The dual nef partition produces the mirror. This generalises the hypersurface case while preserving the same structural idea: anticanonical positivity is encoded by an integral polytope, and mirror duality is expressed by a polar operation enriched by the partition. ^{[Batyrev-Borisov 1996]}

The Gross-Siebert programme recasts Batyrev mirror symmetry as the toric special case of a tropical reconstruction principle. Instead of starting only from a reflexive polytope, it starts from an integral affine manifold with singularities, a polyhedral decomposition, and slab functions. For toric hypersurfaces, the affine sphere obtained from the reflexive polytope is the dual intersection complex of a maximally degenerate Calabi-Yau family. The Batyrev polar duality then becomes the first explicit model for the broader tropical mirror operation. ^{[Strominger-Yau-Zaslow 1996]}

Synthesis. The central insight is that reflexivity identifies combinatorial polarity with anticanonical geometry. The foundational reason is the facet-distance-one condition: it identifies $P$ with the divisor polytope of $-K_{X_P}$, and this is exactly what makes adjunction produce Calabi-Yau hypersurfaces. Putting these together, polar duality identifies vertices with facets and identifies complex-structure variation with Kähler variation across the mirror pair. The construction generalises the polytope-fan dictionary 04.11.10, is dual to the toric degeneration picture 04.12.07, builds toward Gross-Siebert reconstruction 04.12.09, and appears again in GKZ period calculations and the mirror map 04.12.13.

Full proof set [Master]

Proposition (polar vertices are primitive facet normals), proof. Let $$ P={m\in M_\mathbb{R}:\langle m,u_F\rangle\geq -a_F\text{ for every facet }F} $$ with $u_F\in N$ primitive and $a_F>0$. The polar is determined by the same inequalities after normalisation: each facet inequality contributes the point $u_F/a_F$ to the vertex set of $P^{\circ }$, and the polar is the convex hull of these points.

Indeed, a supporting hyperplane of $P$ with equation $⟨m,u_F⟩=-a_F$ gives a linear functional whose minimum on $P$ is $-a_F$. Rescaling by $a_F$ gives $u_F/a_F$, whose pairing with every $m\in P$ is at least $-1$, so $u_F/a_F\in P^{\circ }$. Conversely, every vertex of $P^{\circ }$ is exposed by a facet of $P$ under the face-reversing correspondence between a convex body containing the origin and its polar. Hence the vertices of $P^{\circ }$ are exactly the normalised primitive facet normals. This proves that $P^{\circ }$ is a lattice polytope iff every $a_F=1$. $\square$

Proposition (reflexive polytope gives Gorenstein Fano toric variety), proof. Let $P$ be reflexive and let ${\Sigma }_P$ be its normal fan. For each ray $\rho \in {\Sigma }_P(1)$ with primitive generator $v_{\rho }$, the anticanonical divisor is $-K_{X_P}={\sum }_{\rho }{D}_{\rho }$. Its divisor polytope is $$ P_{-K}={m\in M_\mathbb{R}:\langle m,v_\rho\rangle\geq -1\text{ for all }\rho}. $$ The rays $v_{\rho }$ are precisely the primitive facet normals of $P$, and reflexivity says every facet inequality has right side $-1$. Therefore $P_{-K}=P$.

Since $P$ is full-dimensional and its normal fan is complete, the support function of $P$ is strictly convex on ${\Sigma }_P$, so the line bundle associated to $-K_{X_P}$ is ample. Since the coefficients in the local support functions are lattice characters rather than fractional characters, $-K_{X_P}$ is Cartier. Thus $X_P$ is Gorenstein and Fano. The same proof applies to $P^{\circ }$. $\square$

Proposition (adjunction for the toric anticanonical hypersurface), proof. Let $X$ be a Gorenstein toric Fano variety and let $Y\in |-K_X|$ be a generic anticanonical hypersurface meeting the torus and boundary strata with the expected codimensions. Since $K_X$ is Cartier, the adjunction formula applies on the smooth locus and extends across the mild toric singularities by reflexive extension of dualising sheaves. It gives $$ \omega_Y\cong(\omega_X\otimes\mathcal{O}_X(Y))|_Y. $$ Because ${\mathcal{O}}_X(Y)\cong {\mathcal{O}}_X(-K_X)$, the tensor product is ${\mathcal{O}}_X$, and therefore ${\omega }_Y\cong {\mathcal{O}}_Y$. A crepant resolution $\pi :\hat{Y}\to Y$ preserves the canonical class by definition, so ${\omega }_{\hat{Y}}\cong {\mathcal{O}}_{\hat{Y}}$. $\square$

Connections [Master]

• Polytope-fan dictionary 04.11.10. Reflexive polytopes are the anticanonical special case of the dictionary. The normal fan gives $X_P$, the lattice points of $P$ give anticanonical sections, and polar duality upgrades the ordinary polytope-to-line-bundle correspondence into a mirror-symmetry construction.

• Toric divisor and support function 04.11.08. The equality $P=P_{-K}$ is stated in divisor-polytope language. The support function of the anticanonical divisor has value $-1$ on every primitive ray generator, and the Cartier condition becomes the facet-distance-one criterion for reflexivity.

• Adjunction formula 04.05.07. The Calabi-Yau condition for the hypersurface is the anticanonical adjunction calculation $K_Y=(K_X+Y){|}_Y$. Reflexive polytopes supply exactly the toric combinatorics needed to make $Y$ anticanonical and to keep the canonical class zero after crepant resolution.

• Toric degeneration of a Calabi-Yau variety 04.12.07. Batyrev hypersurfaces provide the standard toric-degeneration examples. The dual intersection complex is built from the reflexive polytope, and the polar dual controls the mirror degeneration.

• Gross-Siebert reconstruction theorem 04.12.09. Gross-Siebert generalises the Batyrev construction beyond toric hypersurfaces. Reflexive polytopes become the model case where the affine base, polyhedral decomposition, and mirror operation are visible before the full scattering-diagram machinery enters.

Historical & philosophical context [Master]

Victor Batyrev's 1994 paper introduced reflexive polytopes as a precise algebro-geometric mechanism for mirror symmetry of Calabi-Yau hypersurfaces in toric varieties ^{[Batyrev 1994]}. The philosophical shift was decisive: mirror symmetry no longer depended only on physical heuristics or isolated examples such as the quintic. It became a computable operation on lattice polytopes, with Hodge-number exchange verified by explicit formulas. Batyrev-Borisov then extended the idea to complete intersections, while Kreuzer-Skarke showed its scale by classifying all four-dimensional reflexive polytopes. The later Gross-Siebert programme preserved the same intuition but moved from a single polytope to tropical affine manifolds with singularities and wall-crossing data.

Bibliography [Master]

@article{Batyrev1994,
  author  = {Batyrev, Victor V.},
  title   = {Dual polyhedra and mirror symmetry for Calabi-Yau hypersurfaces in toric varieties},
  journal = {Journal of Algebraic Geometry},
  volume  = {3},
  pages   = {493--535},
  year    = {1994}
}

@incollection{BatyrevBorisov1996,
  author    = {Batyrev, Victor V. and Borisov, Lev A.},
  title     = {On Calabi-Yau complete intersections in toric varieties},
  booktitle = {Higher-dimensional complex varieties},
  pages     = {39--65},
  publisher = {de Gruyter},
  year      = {1996}
}

@article{KreuzerSkarke2000,
  author  = {Kreuzer, Maximilian and Skarke, Harald},
  title   = {Complete classification of reflexive polyhedra in four dimensions},
  journal = {Advances in Theoretical and Mathematical Physics},
  volume  = {4},
  pages   = {1209--1230},
  year    = {2000}
}

@book{Fulton1993,
  author    = {Fulton, William},
  title     = {Introduction to Toric Varieties},
  publisher = {Princeton University Press},
  series    = {Annals of Mathematics Studies},
  number    = {131},
  year      = {1993}
}

@book{CoxLittleSchenck2011,
  author    = {Cox, David A. and Little, John B. and Schenck, Henry K.},
  title     = {Toric Varieties},
  publisher = {American Mathematical Society},
  series    = {Graduate Studies in Mathematics},
  volume    = {124},
  year      = {2011}
}

@article{StromingerYauZaslow1996,
  author  = {Strominger, Andrew and Yau, Shing-Tung and Zaslow, Eric},
  title   = {Mirror symmetry is T-duality},
  journal = {Nuclear Physics B},
  volume  = {479},
  pages   = {243--259},
  year    = {1996}
}