Berkovich Analytification, Non-Archimedean Skeletons, and the Dual Complex¶
Job: develop the Berkovich analytification (multiplicative seminorms), the skeleton / dual complex of an sncd model (the retraction, the homotopy type), the essential skeleton / K-stability / SYZ connection, and the Berkovich vs tropicalization comparison — with the Berkovich of \(\mathbb{G}_m\) / \(\mathbb{P}^1\) / the skeleton as the interval as the worked example.
Core¶
Berkovich analytification and the analytification functor¶
For a variety \(X\) over a non-Archimedean field \(K\) (e.g. \(\mathbb{C}\{\{t\}\}\) the Puiseux series, or \(\mathbb{Q}_p\)), the Berkovich analytification \(X^{\mathrm{an}}\) (the Berkovich spectrum / the multiplicative seminorms extending the valuation on \(K\)). The \(X^{\mathrm{an}}\) is a locally ringed space / a non-Archimedean analytic space, with the points the multiplicative seminorms \(|\cdot|_x:K[X]\to\mathbb{R}_{\ge0}\) extending the base valuation.
| Berkovich analytification | Description |
|---|---|
| \(X^{\mathrm{an}}\) | the Berkovich spectrum (the multiplicative seminorms extending the base valuation) |
| Points | the multiplicative seminorms $ |
| Berkovich vs rigid | the Berkovich is the "thickening" / the compactification of the rigid (the generic points / the valuations over all extensions) |
| Gelfand spectrum | the max-spectrum / the rigid points are a subset; the Berkovich adds the generic / the higher-rank valuations |
The skeleton and the dual complex¶
For a sncd (simple normal crossing divisor) model \(\bar X\setminus D\) of \(X\) (a compactification \(\bar X\) with \(D=\bar X\setminus X\) a simple normal crossing divisor), the skeleton \(\mathrm{Sk}(X)=\mathrm{Sk}(\bar X,D)\subset X^{\mathrm{an}}\) is the dual complex \(\Delta(D)\) of the divisor \(D\) (the simplex complex encoding the intersections of the components of \(D\), embedded in \(X^{\mathrm{an}}\) via the monomial valuations).
| Skeleton / dual complex | Description |
|---|---|
| Dual complex \(\Delta(D)\) | the combinatorial complex: vertices = irreducible components \(D_i\), edges = intersections \(D_i\cap D_j\), etc. |
| Skeleton \(\mathrm{Sk}(X)\) | \(\Delta(D)\subset X^{\mathrm{an}}\) (the monomial valuations, the "combinatorial core") |
| Retraction \(X^{\mathrm{an}}\to\mathrm{Sk}(X)\) | the deformation retraction onto the skeleton (the Berkovich / the Thuillier retraction) |
| Homotopy type | \(X^{\mathrm{an}}\simeq\mathrm{Sk}(X)\) for the sncd (the dual complex as the homotopy type of the analytification) |
The Kontsevich–Soibelman / the essential skeleton and the K-stability¶
| Essential skeleton / K-stability | Description |
|---|---|
| Essential skeleton | the Kontsevich–Soibelman / the minimal skeleton over all sncd models (the invariant skeleton) |
| Calabi–Yau theorem | for a CY variety, the essential skeleton is the dual complex of the canonical / the sncd (the SYZ base) |
| K-stability (Odaka–Song) | the dual complex and the K-moduli (the relation of the dual complex to the CY / the Kähler–Einstein metric) |
| Berkovich and SYZ | the skeleton as the SYZ base (the non-Archimedean / the tropical mirror symmetry) |
The Berkovich and the tropicalization¶
| Berkovich vs tropicalization | Description |
|---|---|
| Tropicalization \(\mathrm{trop}(V(I))\subset\mathbb{R}^n\) | the closure of the valuation of the \(K^\times\)-points (recalled from foundations) |
| Partial tropicalization | the skeleton maps to the tropicalization (the coordinate valuations \(X^{\mathrm{an}}\to\mathbb{R}^n\)) |
| Berkovich vs tropical | the Berkovich is the full / intrinsic non-Archimedean space, the tropicalization is the projection / the coordinate shadow |
| Kapranov (recalled) | the tropicalization of a hypersurface = the tropical hypersurface (the non-Archimedean amoeba), the Berkovich / the coordinate-tropicalization |
See it¶
Berkovich analytification of \(\mathbb{G}_m\) / \(\mathbb{P}^1\) and the skeleton. For \(X=\mathbb{G}_m=\mathrm{Spec}\,K[t,t^{-1}]\) over the Puiseux \(K\), the Berkovich analytification \(\mathbb{G}_m^{\mathrm{an}}\) is the "line with the generic points" (the multiplicative seminorms, the valuations on \(K[t,t^{-1}]\) extending the base). The skeleton \(\mathrm{Sk}(\mathbb{G}_m)=\mathbb{R}\) (the real line, the valuations \(v_a(f)=\mathrm{val}(a)+\mathrm{ord}_0(f)\) the monomial valuations at \(0\), \(v_a(t)=a\) for \(a\in\mathbb{R}\)) — the skeleton is the real line, the tropical line. The tropicalization \(\mathrm{trop}(\mathbb{G}_m)=\mathbb{R}\) (the valuation of the \(K^\times\)-points). The retraction \(\mathbb{G}_m^{\mathrm{an}}\to\mathbb{R}\) (the Berkovich / the skeleton).
For \(X=\mathbb{P}^1\setminus\{0,\infty\}=\mathbb{G}_m\), the skeleton is the interval \([0,1]\) (the dual complex of \(\{0,\infty\}\), the two endpoints), and the \(\mathbb{P}^1\) Berkovich retracts to the interval. The key point: the Berkovich analytification is the "thickening" of the tropicalization, the skeleton is the dual complex (the combinatorial core), and the retraction gives the homotopy type — the non-Archimedean / Berkovich is the intrinsic lift of the tropical.
Check¶
Show that the skeleton of \(\mathbb{G}_m\) over a Puiseux field is \(\mathbb{R}\), and identify the retraction.
\(\mathbb{G}_m=\mathrm{Spec}\,K[t,t^{-1}]\), the Berkovich points are the multiplicative seminorms \(|\cdot|_x\) on \(K[t,t^{-1}]\) extending the base valuation. The monomial valuations \(v_a\) (parametrized by \(a\in\mathbb{R}\)) defined by \(v_a(\sum c_i t^i)=\min_i(\mathrm{val}(c_i)+ai)\) (the Gauss valuations / the monomial valuations at \(0\), the \(v_a(t)=a\)) form a copy of \(\mathbb{R}\subset\mathbb{G}_m^{\mathrm{an}}\) — the skeleton. The retraction: every point \(x\in\mathbb{G}_m^{\mathrm{an}}\) retracts to the skeleton point \(v_a\) where \(a=v_x(t)\) (the valuation of \(t\) at \(x\)), the deformation retraction \(\mathbb{G}_m^{\mathrm{an}}\to\mathbb{R}=\mathrm{Sk}(\mathbb{G}_m)\). The homotopy type: \(\mathbb{G}_m^{\mathrm{an}}\simeq\mathbb{R}\) (the Berkovich is homotopy-equivalent to the skeleton, the real line).
Why is the essential skeleton (the Kontsevich–Soibelman) the invariant, and what is the role of the sncd model?
The skeleton \(\mathrm{Sk}(\bar X,D)\) depends on the choice of the sncd model \((\bar X,D)\) — different compactifications give different skeletons. The essential skeleton (Kontsevich–Soibelman) is the minimal / the invariant skeleton: the union (or the inverse limit) over all sncd models, the skeleton that is independent of the choice. For a Calabi–Yau variety, the essential skeleton is the dual complex of the canonical / the sncd model (the canonical divisor), the SYZ base. The role of the sncd: the skeleton is defined for each sncd model (the dual complex \(\Delta(D)\)), and the essential skeleton is the invariant over all models — the sncd is the input, the essential the output / the invariant.
Pitfalls¶
- The Berkovich analytification is the "thickening" / the intrinsic (the tropicalization is the projection, not the full space) — the Berkovich has the generic points / the higher-rank valuations, the tropicalization is the coordinate shadow.
- The skeleton requires the sncd model (the essential skeleton is the invariant / the minimal, not the arbitrary model) — the choice of the model matters, the essential is the canonical.
- The homotopy type \(X^{\mathrm{an}}\simeq\mathrm{Sk}(X)\) holds for the sncd model (the retraction, the deformation), but the essential skeleton may be smaller (the invariant).
- The Berkovich vs the rigid: the Berkovich is the compactification / the thickening of the rigid (the rigid points are a subset, the Berkovich adds the generic / the boundary points).
- The K-stability / the Odaka–Song is the relation of the dual complex to the K-moduli (the Calabi–Yau / the Kähler–Einstein metric), the deep / the recent connection, not a formal consequence.
Refs¶
- Berkovich, Spectral Theory and Analytic Geometry over Non-Archimedean Fields (1990); Baker–Rumely, Potential Theory on the Berkovich Projective Line
- Thuillier, "Géométrie toroïdale et géométrie analytique non archimédienne" (2007); Kontsevich–Soibelman, "Affine structures and non-archimedean analytic spaces" (2006)
- Payne, "Analytification is the limit of all tropicalizations" (2009); Abramchenko–Caporaso–Payne–Melo, "The tropicalization of the moduli space of curves" (2013)
- Odaka–Song, "The Calabi–Yau theorem and the dual complex" (2018); the essential skeleton / the K-stability
Next¶
TG index · Foundations · Matroids & Linear Spaces · Moduli & Enumerative · Scattering & Mirror · Mirror symmetry / SYZ · Algebraic geometry