---
title: Toric Volume Form in Kähler & Calabi–Yau Geometry
url: https://www.emergentmind.com/topics/toric-volume-form
type: topic
---

# Toric Volume Form in Kähler & Calabi–Yau Geometry

In toric Kähler and toric Calabi–Yau geometry, the natural volume form on the link \(Y_{2n+1}\) of a cone \(C(Y_{2n+1})\) is
\[
\eta \wedge \omega^n / n!,
\]
where \(\eta\) is the contact \(1\)-form and \(\omega\) is the transverse Kähler form. Its integral defines a “master volume” \(\mathcal V(b;\lambda)\) depending on the Reeb vector \(b\) and transverse Kähler class parameters \(\lambda_a\), while in the toric Calabi–Yau \(3\)-fold case it specializes to a volume functional \(V(b;Y_5)\) on Sasaki–Einstein \(5\)-manifolds. These constructions are central both to geometric extremization and to holography: for toric Calabi–Yau \(3\)-folds, the minimum volume of the Sasaki–Einstein base is inversely proportional to the central charge of the corresponding \(4d\ \mathcal N=1\) superconformal field theories, and recent work has also produced explicit machine-learning-regularized approximations for that minimum volume in terms of toric-diagram invariants [1904.04282] [2310.19276].

## 1. Geometric setting of the toric volume form

Let \(C(Y_{2n+1})\) be a Gorenstein toric Kähler cone of complex dimension \(n=d\), with link \(Y_{2n+1}\). The geometry carries a \(U(1)^n\)-action generated by angular coordinates \(\phi_i\), \(i=1,\dots,d\), and a Reeb vector
\[
\xi = \sum_{i=1}^d b_i \partial_{\phi_i},
\]
with \(b \in \mathbb R^d\) in the interior of the dual cone. Moment-map coordinates are introduced as
\[
y_i = \tfrac12 r^2 (\partial_{\phi_i}\!\lrcorner\, \eta),
\]
where \(\eta\) satisfies \(\eta(\xi)=1\) and \(d\eta=2\omega_{\text{Sasakian}}\) [1904.04282].

The toric cone is the polyhedral cone
\[
\mathcal C=\{y\in \mathbb R^d \mid \langle y,v_a\rangle \ge 0,\ a=1,\dots,D\},
\]
with inward-pointing primitive facet normals \(v_a\in \mathbb Z^d\). The link at \(r=1\) projects to a compact polytope obtained by intersecting \(\mathcal C\) with the Reeb hyperplane
\[
H(b)=\{y\in \mathbb R^d \mid \langle y,b\rangle=\tfrac12\}.
\]
After deforming the transverse Kähler class by parameters \(\lambda_a\), the relevant polytope becomes
\[
P(b;\lambda)=\mathcal C\cap H(b;\lambda)\subset H(b),
\]
or equivalently
\[
P(b;\lambda)=\{y\in H(b)\mid \langle y-y_0,v_a\rangle \ge \lambda_a\},
\]
with \(y_0=(\tfrac12 b_1^{-1},0,\dots,0)\in H(b)\) [1904.04282].

The Kähler deformation is encoded by
\[
[\omega] = -2\pi \sum_a \lambda_a [c_a],
\]
where the \(c_a\) form a basis of invariant \(2\)-cycles on \(Y_{2n+1}\) lifting the toric divisors. Three linear relations among the \(\lambda_a\), coming from \(\langle v_a,b\rangle\), imply that only \(D-d\) of them are independent [1904.04282].

## 2. Integral, polytope, and facet-sum realizations

The master volume is defined by
\[
\mathcal V(b;\lambda)\equiv \int_{Y_{2n+1}} \eta \wedge \omega^n/n!.
\]
In symplectic \((y,\phi)\) coordinates one has
\[
\eta\wedge \omega^n/n! = (2\pi)^n \,\partial(S^1\cdots S^1)\wedge dy_1\cdots dy_n / |b|,
\]
and integrating out the \(U(1)^n\) angles yields
\[
\mathcal V(b;\lambda)=\frac{(2\pi)^n}{|b|}\,\mathrm{Vol}_n[P(b;\lambda)].
\]
Thus the toric volume form reduces the computation of \(\mathcal V\) to Euclidean polytope volume [1904.04282].

An equivalent description uses a symplectic potential \(g(y)\) on the cone:
\[
ds_C^2=\sum_{i,j}\frac{\partial^2 g}{\partial y_i\partial y_j}dy_i dy_j + (\text{inverse})\, d\phi_i d\phi_j.
\]
In this description one can show directly that \(\int \eta\wedge \omega^n/n!\) is proportional to \(\int_P d^n y\) [1904.04282].

A closed-form expression is obtained by triangulating \(P(b;\lambda)\) into simplices. The resulting “facet-sum” formula expresses \(\mathcal V(b;\{\lambda_a\})\) as a sum over facets and cyclically labelled vertices, with determinants built from \((v_a,v_{a,k-1},v_{a,k},b)\) and the \(\lambda_a\). The same quantity can also be written more compactly in terms of triple-intersection numbers
\[
I_{abc}=\int_{Y_{2n+1}} \eta\wedge c_a\wedge c_b\wedge c_c
=-\frac1{(2\pi)^n}\frac{\partial^3\mathcal V}{\partial\lambda_a\partial\lambda_b\partial\lambda_c},
\]
so that
\[
\mathcal V(b;\lambda)=-(2\pi)^n\frac1{n!}\sum_{a,b,c} I_{abc}\lambda_a\lambda_b\lambda_c.
\]
In the special case \(\lambda_a=\lambda\) for all \(a\), these expressions simplify to the well-known toric Sasaki volume formula [1904.04282].

## 3. Toric Calabi–Yau \(3\)-folds and the Hilbert-series volume functional

For any non-compact toric Calabi–Yau three-fold \(\mathcal X\), \(\mathcal X\) is the affine cone over a five-dimensional Sasaki–Einstein manifold \(Y_5\), and the volume of \(Y_5\) can be written in closed form as a rational function of the Reeb vector \(b\equiv (b_1,b_2,b_3)\). In practice one often fixes the normalization \(b_3=3\), so that \(b\) lives in the interior of the toric diagram \(\Delta\subset \mathbb Z^2\) at height \(z=1\) [2310.19276].

Two standard approaches to compute the volume function \(V(b;Y_5)\) are the Duistermaat–Heckman formula or equivariant index, and the Hilbert series of \(\mathcal X\) [2310.19276]. If \(\Delta\) admits a fine triangulation into \(r\) unimodular triangles \(\Delta_i\), each with outward-pointing normal vectors
\[
u_{i,1},u_{i,2},u_{i,3}\in \mathbb Z^3,
\]
the Hilbert series is
\[
g(t_1,t_2,t_3;\mathcal X)=\sum_{i=1}^r \prod_{j=1}^3 \frac{1}{1-t^{u_{i,j}}},
\]
where \(t^u\equiv t_1^{u(1)} t_2^{u(2)} t_3^{u(3)}\). The volume function is extracted as
\[
V(b;Y_5)=\lim_{\mu\to 0}\mu^3 g(e^{-\mu b_1},e^{-\mu b_2},e^{-\mu b_3};\mathcal X),
\]
and takes the explicit form
\[
V(b;Y_5)=\sum_{i=1}^r \frac{1}{(b\cdot u_{i,1})(b\cdot u_{i,2})(b\cdot u_{i,3})}
\]
with the normalization \(b_3=3\) [2310.19276].

This specialization exhibits the toric volume form in a particularly computable way: combinatorial data of a triangulated toric diagram determine a rational function of the Reeb vector, and the geometric problem becomes one of extremization inside the Reeb cone.

## 4. Extremization, derivatives, and holographic role

The minimum-volume problem is
\[
V_{\min}(\Delta)=\min_b V(b;Y_5(\Delta)).
\]
For a Sasaki–Einstein metric, one fixes
\[
\lambda_a=-1/(2b_1),
\]
then minimizes \(\mathrm{Vol}(Y(b))\) over \(b\) in the Reeb cone subject to \(b_1=2n\) [1904.04282]. In the toric Calabi–Yau \(3\)-fold setting, the minimum volume of the Sasaki–Einstein base is inversely proportional to the central charge of the corresponding \(4d\ \mathcal N=1\) superconformal field theories under the AdS/CFT correspondence [2310.19276].

Derivatives of the master volume encode additional geometric data. One has
\[
\frac{\partial \mathcal V}{\partial \lambda_a}
= \int_{T_a} \eta\wedge \omega^{n-1}/(n-1)!,
\]
which gives the volume of the toric divisor \(T_a\), while
\[
\frac{\partial^2 \mathcal V}{\partial \lambda_a \partial \lambda_b}
=-(2\pi)^n I_{abc}\lambda_c
\]
and \(\partial \mathcal V/\partial b_i\) are obtained by varying the Reeb hyperplane \(H(b)\) [1904.04282]. These derivatives enter directly into geometric extremization and flux-quantization constraints.

For “GK” AdS\(_3\times Y_7\) or AdS\(_2\times Y_9\) geometries, one extremizes a suitable functional \(S(b,\lambda)\), built from linear combinations of \(\partial\mathcal V/\partial\lambda_a\) and \(\partial^2\mathcal V/\partial\lambda_a\partial\lambda_b\), subject to linear flux constraints such as \(\partial^2\mathcal V/\partial\lambda_a\partial\lambda_b=N_{ab}\). In applications to black-hole entropy or \(c\)-extremization, one uses these derivatives to impose flux-quantization and transversality constraints and then extremizes the remaining variables. The gauge-invariance identity
\[
\sum_a \left(v_{ai}-\frac{b_i}{b_1}\right)\frac{\partial \mathcal V}{\partial \lambda_a}\equiv 0
\]
shows that \(\mathcal V\), \(\partial\mathcal V/\partial\lambda_a\), and \(\partial^2\mathcal V/\partial\lambda_a\partial\lambda_b\) depend only on \(D-d\) independent combinations of the \(\lambda_a\) [1904.04282].

## 5. Toric-diagram invariants and machine-learning regularization

For toric Calabi–Yau \(3\)-folds, the data of the toric diagram \(\Delta\) can be encoded in integer-valued geometric invariants. These include the area
\[
A=\mathrm{area}(\Delta)= I + E/2 -1,
\]
computed in \(\mathbb Z^2\) by Pick’s theorem, where \(I\) is the number of interior lattice points and \(E\) is the number of boundary lattice points; the number of vertices \(V\); and, for any positive integer \(n\), the \(n\)-enlarged polytope
\[
\Delta_n=\{\, n\cdot v \mid v\in \Delta\subset \mathbb Z^2\},
\]
with interior and boundary lattice counts \(I_n\) and \(E_n\). In particular, \(I_n\) grows roughly like \(n^2A\) but carries global shape information beyond \((I,E)\) [2310.19276].

Instead of minimizing the exact Hilbert-series volume each time, one can approximate the inverse minimum volume
\[
y\equiv 1/V_{\min}
\]
as a linear combination of features \(x_a\),
\[
\hat y = \beta_0 + \sum_{a=1}^{N_x} \beta_a x_a,
\]
fitted by minimizing the ordinary least-squares loss
\[
L_0(\beta)=\tfrac{1}{2|S|}\sum_{j\in S}(y_j-\hat y_j)^2
\]
over a dataset \(S\) of toric diagrams. Sparsity is then imposed through the \(L^1\)-penalized loss
\[
L(\beta)=L_0(\beta)+\alpha\sum_{a=1}^{N_x} |\beta_a|,
\]
with hyperparameter \(\alpha\) chosen to maximize
\[
R^2(\alpha)-\lambda \,\tfrac{N_{nz}(\alpha)}{N_x},
\]
where \(R^2\) is the coefficient of determination and \(N_{nz}\) is the number of nonzero coefficients [2310.19276].

After training on four large datasets of toric diagrams, up to \(2\times 10^5\) examples, the best compromise was obtained by \(L^1\)-regularized logarithmic regression using only three features,
\[
x_1=A,\qquad x_2=V,\qquad x_3=I_3,
\]
with
\[
\log \hat y=\gamma_0+\gamma_1 \log A+\gamma_2 \log V+\gamma_3 \log I_3.
\]
For the largest training set \(S_{2a}\), the coefficients are
\[
\gamma_0=\log(2.50772)\simeq 0.9195,\qquad
\gamma_1=0.95411,\qquad
\gamma_2=-0.21992,\qquad
\gamma_3=0.02867,
\]
so that
\[
\hat y \equiv 1/V_{\min}\simeq 2.50772\, A^{0.95411}\,V^{-0.21992}\,I_3^{0.02867},
\]
or equivalently
\[
V_{\min}(\Delta)\simeq \bigl[2.50772\, A^{0.95411}\,V^{-0.21992}\,I_3^{0.02867}\bigr]^{-1}.
\]
This yields a closed-form approximation involving only three toric-diagram invariants [2310.19276].

## 6. Interpretation, accuracy, and scope

The exponents in the three-feature approximation admit a geometric interpretation. The factor \(A^{0.95411}\) reflects that “barycentric” Reeb vectors tend to spread weight uniformly across the \(r\approx 2A\) triangles in the triangulation. The factor \(V^{-0.21992}\) reflects that corners of \(\Delta\) contribute slightly more to the singularity of the volume integral, hence more vertices \(V\) lower the volume. The factor \(I_3^{0.02867}\) reflects that the interior lattice structure of the three-times enlarged diagram \(\Delta_3\) refines subleading shape information about \(\Delta\), such as holes or indentations, which still has a small effect on \(V_{\min}\) [2310.19276].

On the largest test set \(S_{2a}\), approximately \(2.0\times 10^5\) toric diagrams, this three-parameter formula achieves \(R^2\approx 0.9928\), with average relative error
\[
\bigl\langle |V_{\min}^{\text{exact}}-V_{\min}^{\text{pred}}|/V_{\min}^{\text{exact}}\bigr\rangle \simeq 3.58\%
\]
and standard deviation \(\simeq 2.40\%\). For the smaller box \(S_{1a}\), the error falls to \(\simeq 2.16\%\) with \(R^2\approx 0.9893\) [2310.19276].

These results separate two levels of description. The master-volume and Hilbert-series formulas give exact combinatorial expressions for toric volumes in terms of \(b\), \(\lambda_a\), and toric data [1904.04282] [2310.19276]. The machine-learning-regularized expression gives a compact and interpretable approximation to the minimized volume \(V_{\min}\) in terms of a small set of toric invariants [2310.19276]. A plausible implication is that the toric volume form occupies a productive intermediate position between differential geometry and combinatorics: it is sufficiently rigid to admit exact polyhedral formulas, but sufficiently structured that low-dimensional invariant summaries can approximate extremal quantities with high accuracy.

Source: https://www.emergentmind.com/topics/toric-volume-form