ScalingStacks

3.2. Statement and first reductions [015H]

Original official author HTML, exact retained edition. Historical TeX conversion verdicts remain unchanged. Cited-edition alignment and mathematical self-containment are not assessed.

Complete original source context · Original author HTML

3.2. Statement and first reductions

It will be convenient to introduce the quantity

λ⁡(t):=(log⁡|t|−1)−1,\lambda(t):=(\log|t|^{-1})^{-1},

for t∈𝔻∗t\in{\mathbb{D}}^{*}. Note that λ⁡(t)→0\lambda(t)\to 0 as t→0t\to 0.

Let 𝒳hyb:=X​∐Δ⁡(𝒳){\mathcal{X}}^{\mathrm{hyb}}:=X\coprod\Delta({\mathcal{X}}) be the locally compact hybrid space constructed in §2. It comes with a proper map π:𝒳hyb→Δ\pi\colon{\mathcal{X}}^{\mathrm{hyb}}\to\Delta extending π:X→𝔻∗\pi\colon X\to{\mathbb{D}}^{*} and such that Δ​(𝒳)=π−1​(0)\Delta({\mathcal{X}})=\pi^{-1}(0). The next result implies Theorem A in the introduction.

Theorem 3.4.

Let π:𝒳→𝔻\pi\colon{\mathcal{X}}\to{\mathbb{D}} be an snc degeneration, ℒ{\mathcal{L}} a ℚ{\mathbb{Q}}-line bundle on 𝒳{\mathcal{X}} extending KX/𝔻∗K_{X/{\mathbb{D}}^{*}}, and ψ\psi a continuous metric on ℒ{\mathcal{L}}. Define κmin\kappa_{\min} as above, and set d:=dimΔ⁡(ℒ)d:=\dim\Delta({\mathcal{L}}). Then, viewed as measures on 𝒳hyb{\mathcal{X}}^{\mathrm{hyb}},

μt:=λ​(t)d(2​π)d​|t|2​κmin​e2​ψt\mu_{t}:=\frac{\lambda(t)^{d}}{(2\pi)^{d}|t|^{2\kappa_{\min}}}e^{2\psi_{t}}

converges weakly to

μ0:=∑σ(∫YσResYσ⁡(ψ))​bσ−1​λσ,\mu_{0}:=\sum_{\sigma}\left(\int_{Y_{\sigma}}\operatorname{Res}_{Y_{\sigma}}(\psi)\right)b_{\sigma}^{-1}\lambda_{\sigma},

where σ\sigma ranges over the dd-dimensional faces of Δ⁡(ℒ)\Delta({\mathcal{L}}). Here λσ\lambda_{\sigma} denotes normalized Lebesgue measure on σ\sigma and bσ=gcdi∈J⁡bib_{\sigma}=\gcd_{i\in J}b_{i}, where 𝒳0=∑ibi​Ei{\mathcal{X}}_{0}=\sum_{i}b_{i}E_{i} and EiE_{i}, i∈Ji\in J are the divisors defining σ\sigma.

We start by making a few reductions. First, we may—and will—assume in what follows that κmin=0\kappa_{\min}=0. Indeed, tt defines a nonvanishing section of 𝒪𝒳​(𝒳0){\mathcal{O}}_{\mathcal{X}}({\mathcal{X}}_{0}), and hence a smooth metric log⁡|t|\log|t|, so we may replace ℒ{\mathcal{L}} and ψ\psi with ℒ−κmin​𝒳0{\mathcal{L}}-\kappa_{\min}{\mathcal{X}}_{0} and ψ−κmin​log⁡|t|\psi-\kappa_{\min}\log|t|, respectively, and end up with κmin=0\kappa_{\min}=0.

Since mini⁡ai/bi=κmin=0\min_{i}a_{i}/b_{i}=\kappa_{\min}=0, we then have ai≥0a_{i}\geq 0, with equality if and only if EiE_{i} corresponds to a vertex of Δ⁡(ℒ)\Delta({\mathcal{L}}).

Next we reduce the assertion of Theorem 3.4 to a local problem. Let Y⊂𝒳0Y\subset{\mathcal{X}}_{0} be the stratum of an arbitrary face σ\sigma of Δ⁡(𝒳)\Delta({\mathcal{X}}), and denote by E0,…,EpE_{0},\dots,E_{p} the components of 𝒳0{\mathcal{X}}_{0} cutting out YY, ordered so that

κ0=⋯=κq<κq+1≤⋯≤κp.\kappa_{0}=\dots=\kappa_{q}<\kappa_{q+1}\leq\dots\leq\kappa_{p}.

We can then make the identification

σ={w∈ℝ+p+1∣b⋅w=1}\sigma=\left\{w\in{\mathbb{R}}_{+}^{p+1}\mid b\cdot w=1\right\}

with b=(b0,…,bp)∈ℤ>0p+1b=(b_{0},\dots,b_{p})\in{\mathbb{Z}}_{>0}^{p+1}. Set b′=(b0,…,bq)∈ℤ>0q+1b^{\prime}=(b_{0},\dots,b_{q})\in{\mathbb{Z}}_{>0}^{q+1} and

σ′:={w′∈ℝ+q+1∣b′⋅w′=1}.\sigma^{\prime}:=\left\{w^{\prime}\in{\mathbb{R}}_{+}^{q+1}\mid b^{\prime}\cdot w^{\prime}=1\right\}.

Then σ′\sigma^{\prime} is a face of σ\sigma under the embedding ℝ+q+1↪ℝ+p+1{\mathbb{R}}_{+}^{q+1}\hookrightarrow{\mathbb{R}}_{+}^{p+1} given by w′→(w′,0)w^{\prime}\to(w^{\prime},0). Let Y′⊃YY^{\prime}\supset Y be the corresponding stratum of 𝒳0{\mathcal{X}}_{0}.

Note that σ\sigma contains a face of Δ⁡(ℒ)\Delta({\mathcal{L}}) if and only if κ0=0\kappa_{0}=0; in that case, the face is unique, equal to σ′\sigma^{\prime} (which then implies q≤dq\leq d).

Pick x∈Y̊x\in\mathring{Y}, and choose local coordinates z=(z0,…,zn)z=(z_{0},\dots,z_{n}) at xx such that ziz_{i} is a local equation of EiE_{i} for 0≤i≤p0\leq i\leq p and

t=∏i=0pzibit=\prod_{i=0}^{p}z_{i}^{b_{i}}

We may assume that zz is defined on a polydisc 𝒰≃𝔻​(r)p+1×𝔻n−p{\mathcal{U}}\simeq{\mathbb{D}}(r)^{p+1}\times{\mathbb{D}}^{n-p} with 0<r≪10<r\ll 1. Decompose

z=(z0,…,zn)∈𝒰≃𝔻​(r)p+1×𝔻n−pz=(z_{0},\dots,z_{n})\in{\mathcal{U}}\simeq{\mathbb{D}}(r)^{p+1}\times{\mathbb{D}}^{n-p}

as

z=(z′,z′′,y)∈𝔻​(r)q+1×𝔻​(r)p−q×𝔻n−p,z=(z^{\prime},z^{\prime\prime},y)\in{\mathbb{D}}(r)^{q+1}\times{\mathbb{D}}(r)^{p-q}\times{\mathbb{D}}^{n-p},

where we view yy as a point of 𝒰∩Y≃𝔻n−p{\mathcal{U}}\cap Y\simeq{\mathbb{D}}^{n-p}, and (z′′,y)(z^{\prime\prime},y) as a point of 𝒰∩Y′≃𝔻​(r)p−q×𝔻n−p{\mathcal{U}}\cap Y^{\prime}\simeq{\mathbb{D}}(r)^{p-q}\times{\mathbb{D}}^{n-p}.

The coordinate chart (𝒰,z)({\mathcal{U}},z) is adapted to 𝒳0{\mathcal{X}}_{0} in the sense of §2.2, with

Log𝒰:𝒰∖𝒳0→σ\operatorname{Log}_{{\mathcal{U}}}\colon{\mathcal{U}}\setminus{\mathcal{X}}_{0}\to\sigma

given by

Log𝒰=(log⁡|zi|log⁡|t|)0≤i≤p.\operatorname{Log}_{{\mathcal{U}}}=\left(\frac{\log|z_{i}|}{\log|t|}\right)_{0\leq i\leq p}.

We aim to establish the following result.

Lemma 3.5.

Pick χ∈Cc0​(𝒰)\chi\in C^{0}_{c}({\mathcal{U}}). If κ0=0\kappa_{0}=0 and q=dq=d, then

limt→0(Log𝒰)∗​(χ​μt)=(∫Y′χ​ResY′⁡(ψ))​bσ′−1​λσ′\lim_{t\to 0}(\operatorname{Log}_{{\mathcal{U}}})_{*}(\chi\mu_{t})=\left(\int_{Y^{\prime}}\chi\operatorname{Res}_{Y^{\prime}}(\psi)\right)b_{\sigma^{\prime}}^{-1}\lambda_{\sigma}^{\prime}

in the weak topology of measures on σ\sigma, with σ′\sigma^{\prime} the unique dd-dimensional face of Δ⁡(ℒ)\Delta({\mathcal{L}}) contained in σ\sigma. Otherwise (i.e. if κ0>0\kappa_{0}>0 or q<dq<d) (Log𝒰)∗​(χ​μt)→0(\operatorname{Log}_{{\mathcal{U}}})_{*}(\chi\mu_{t})\to 0.

Granted this result, let us show how to prove Theorem 3.4. For 0<r≪10<r\ll 1, 𝒱:=π−1​(𝔻¯r)⊂𝒳{\mathcal{V}}:=\pi^{-1}(\overline{{\mathbb{D}}}_{r})\subset{\mathcal{X}} is an compact neighborhood of 𝒳0{\mathcal{X}}_{0} with a map Log𝒱:𝒱hyb→Δ⁡(𝒳)\operatorname{Log}_{\mathcal{V}}\colon{\mathcal{V}}^{\mathrm{hyb}}\to\Delta({\mathcal{X}}) as in Proposition 2.1. We will use

Lemma 3.6.

Let μt\mu_{t}, t∈𝔻rt\in{\mathbb{D}}_{r} be a family of probability measures on 𝒳hyb{\mathcal{X}}^{\mathrm{hyb}} such that μt\mu_{t} is supported on 𝒳t{\mathcal{X}}_{t}. Then limt→0μt=μ0\lim_{t\to 0}\mu_{t}=\mu_{0} if and only if limt→0(Log𝒱)∗​μt=μ0\lim_{t\to 0}(\operatorname{Log}_{\mathcal{V}})_{*}\mu_{t}=\mu_{0}. Here the limits are in the sense of weak convergence of measures on 𝒳hyb{\mathcal{X}}^{\mathrm{hyb}} and Δ⁡(𝒳)\Delta({\mathcal{X}}), respectively.

By Lemma 3.6 we must show that

(Log𝒱)∗​μt→μ0=∑σ′(∫Y′ResY′⁡(ψ))​bσ′−1​λσ′,(\operatorname{Log}_{\mathcal{V}})_{*}\mu_{t}\to\mu_{0}=\sum_{\sigma^{\prime}}\left(\int_{Y^{\prime}}\operatorname{Res}_{Y^{\prime}}(\psi)\right)b_{\sigma^{\prime}}^{-1}\lambda_{\sigma^{\prime}},

where σ′\sigma^{\prime} ranges over dd-dimensional simplices in Δ⁡(ℒ)\Delta({\mathcal{L}}). But this is easily seen to follow from Lemma 3.5, using a partition of unity argument as in the proof of Proposition 2.1.

Proof of Lemma 3.6.

The direct implication follows from the continuity of Log𝒱\operatorname{Log}_{\mathcal{V}}. For the reverse implication, assume that limt→0(Log𝒱)∗​μt=μ0\lim_{t\to 0}(\operatorname{Log}_{\mathcal{V}})_{*}\mu_{t}=\mu_{0} and consider the following three subsets of C0​(𝒱)C^{0}({\mathcal{V}}): A1A_{1} is the set of functions of the form Log𝒱∗​φ\operatorname{Log}_{\mathcal{V}}^{*}\varphi, where φ∈C0​(Δ​(𝒳))\varphi\in C^{0}(\Delta({\mathcal{X}})); A2A_{2} is the set of functions of the form π∗​g\pi^{*}g, where g∈C0​(𝔻r)g\in C^{0}({\mathbb{D}}_{r}); and A3=Cc0​(𝒱∖Δ⁡(𝒳))A_{3}=C^{0}_{c}({\mathcal{V}}\setminus\Delta({\mathcal{X}})) together with the constant function 1. Then the real vector space A⊂C0​(𝒱)A\subset C^{0}({\mathcal{V}}) spanned by functions of the form f1​f2​f3f_{1}f_{2}f_{3}, with fi∈Aif_{i}\in A_{i} is easily seen to be an ℝ{\mathbb{R}}-algebra that separates points and contains all constant functions. By the Stone-Weierstrass Theorem, AA is dense in C0​(𝒱)C^{0}({\mathcal{V}}), so it suffices to prove that lim∫⁡f​μt=∫f​μ0\lim\int f\mu_{t}=\int f\mu_{0} for f∈Af\in A. By linearity, we may assume f=f1​f2​f3f=f_{1}f_{2}f_{3} with fi∈Aif_{i}\in A_{i}. We may further assume f3=1f_{3}=1. Write f1=Log𝒱∗​φf_{1}=\operatorname{Log}_{\mathcal{V}}^{*}\varphi and f2=π∗​gf_{2}=\pi^{*}g. Then

limt→0∫Xtf​μt=limt→0g⁡(t)​∫Xtφ∘Log𝒱⁡μt=limt→0g⁡(t)​∫Δ⁡(𝒳)φ​(Log𝒱)∗​μt=g⁡(0)​∫Δ⁡(𝒳)φ​μ0=∫f​μ0,\lim_{t\to 0}\int_{X_{t}}f\mu_{t}=\lim_{t\to 0}g(t)\int_{X_{t}}\varphi\circ\operatorname{Log}_{\mathcal{V}}\mu_{t}\\ =\lim_{t\to 0}g(t)\int_{\Delta({\mathcal{X}})}\varphi\ (\operatorname{Log}_{\mathcal{V}})_{*}\mu_{t}=g(0)\int_{\Delta({\mathcal{X}})}\varphi\mu_{0}=\int f\mu_{0},

which completes the proof. ∎

Original mathematics by the credited authors. Source-backed reader collection; mathematical self-containment is not assessed.