ScalingStacks

4.5 Refined local ansatz in the non-generic region [023W]

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

4.5 Refined local ansatz in the non-generic region

Recall the distance to the origin is O⁡(|x|n+22​n)O(|x|^{\frac{n+2}{2n}}). Thus the O⁡(x1−nn−1)O(x_{1}^{-\frac{n}{n-1}}) decay is slower than quadratic, and we need further correction terms to improve the ansatz. The linearization of the complex Monge-Ampère equation will naturally lead to a Poisson equation.

Using section 4.3, we can solve (a rescaled version of) the Poisson equation on the Tian-Yau space D1∖D2D_{1}\setminus D_{2}:

(n+2)​(n−1)2​π​n​d22​v0​an−2​(d1d2)n⁡(n−2)n−1​(d​dc​U)∧ωT​Yn−2=−ℐ1,\frac{(n+2)(n-1)}{2\pi nd_{2}^{2}}v_{0}a^{n-2}(\frac{d_{1}}{d_{2}})^{\frac{n(n-2)}{n-1}}(dd^{c}U)\wedge\omega_{TY}^{n-2}=-\mathcal{I}_{1}, (34)

where we recall from (33) that ℐ1\mathcal{I}_{1} is a top degree form on D1∖D2D_{1}\setminus D_{2}, with exponential decay O⁡(e−c​x21/2)O(e^{-cx_{2}^{1/2}}) for some c>0c>0 to all orders of derivatives. The caveat is that ∫D1∖D2ℐ1\int_{D_{1}\setminus D_{2}}\mathcal{I}_{1} is not guaranteed to be zero, so UU may not decay at infinity. Instead,

U=a′​u0+U~,U=a^{\prime}u_{0}+\tilde{U},

where |∇T​YkU~|=O⁡(e−c​x21/2)|\nabla^{k}_{TY}\tilde{U}|=O(e^{-cx_{2}^{1/2}}) for some possibly shrinked c>0c>0, and a′a^{\prime} is a constant.

We regard UU as a function in the region x2≪x1x_{2}\ll x_{1}, namely the tubular neighbourhood around D1∖D2D_{1}\setminus D_{2}, and let ϕD1(2)=ϕD1+x~1n+2n−2​nn−1​U\phi_{D_{1}}^{(2)}=\phi_{D_{1}}+\tilde{x}_{1}^{\frac{n+2}{n}-\frac{2n}{n-1}}U. From the leading term u0∼x21n−1u_{0}\sim x_{2}^{\frac{1}{n-1}} in UU, we can compute using Lemma 4.7 that the correction term x~1n+2n−2​nn−1​U\tilde{x}_{1}^{\frac{n+2}{n}-\frac{2n}{n-1}}U has local Ck,αC^{k,\alpha}-norm O⁡(x1−nn−1​x2−1)O(x_{1}^{-\frac{n}{n-1}}x_{2}^{-1}). This small correction term leads to better volume form decay:

Lemma 4.11.

In the region |x2|+1≪x1|x_{2}|+1\ll x_{1}, we have the improved decay

‖(d​dc​ϕD1(2))n−K0​−1n2​Ω∧Ω¯‖k,α,l​o​c=O⁡(x1−2​nn−1​x21n−1).\left\lVert(dd^{c}\phi_{D_{1}}^{(2)})^{n}-K_{0}\sqrt{-1}^{n^{2}}\Omega\wedge\overline{\Omega}\right\rVert_{k,\alpha,loc}=O(x_{1}^{-\frac{2n}{n-1}}x_{2}^{\frac{1}{n-1}}).
Proof.

We revisit the calculations in Lemma 4.10. The volume form (d​dc​ϕD1(2))n(dd^{c}\phi_{D_{1}}^{(2)})^{n} should be viewed as a perturbation of (d​dc​ϕD1)n(dd^{c}\phi_{D_{1}})^{n}. Again (d​dc​ϕD1(2))n(dd^{c}\phi_{D_{1}}^{(2)})^{n} is a convergent power series of x~1−nn−1\tilde{x}_{1}^{-\frac{n}{n-1}}, with coefficient in top degree forms on D1∖D2D_{1}\setminus D_{2}. The leading order contribution to (d​dc​ϕD1(2))n−(d​dc​ϕD1)n(dd^{c}\phi_{D_{1}}^{(2)})^{n}-(dd^{c}\phi_{D_{1}})^{n} is

n⁡(n−1)​v0​d​dc​x~1n+2n∧(a​(d1d2)nn−1​x~1n−2n⁡(n−1)​d​dc​ϕT​Y)n−2∧x~1n+2n−2​nn−1​d​dc​U,n(n-1)v_{0}dd^{c}\tilde{x}_{1}^{\frac{n+2}{n}}\wedge(a(\frac{d_{1}}{d_{2}})^{\frac{n}{n-1}}\tilde{x}_{1}^{\frac{n-2}{n(n-1)}}dd^{c}\phi_{TY})^{n-2}\wedge\tilde{x}_{1}^{\frac{n+2}{n}-\frac{2n}{n-1}}dd^{c}U,

which after some calculation gives

2​(n+2)​(n−1)n​d22​v0​an−2​(d1d2)n⁡(n−2)n−1​(d​dc​U)∧ωT​Yn−2∧x~1−nn−1​−14​π​d​log⁡ξ′∧d​log⁡ξ′¯.\frac{2(n+2)(n-1)}{nd_{2}^{2}}v_{0}a^{n-2}(\frac{d_{1}}{d_{2}})^{\frac{n(n-2)}{n-1}}(dd^{c}U)\wedge\omega_{TY}^{n-2}\wedge\tilde{x}_{1}^{-\frac{n}{n-1}}\frac{\sqrt{-1}}{4\pi}d\log\xi^{\prime}\wedge d\overline{\log\xi^{\prime}}.

This is by construction −−1​d​log⁡ξ′∧d​log⁡ξ′¯∧ℐ1,-\sqrt{-1}d\log\xi^{\prime}\wedge d\overline{\log\xi^{\prime}}\wedge\mathcal{I}_{1}, which is precisely designed to cancel the leading order error −1​d​log⁡ξ′∧d​log⁡ξ′¯∧ℐ1\sqrt{-1}d\log\xi^{\prime}\wedge d\overline{\log\xi^{\prime}}\wedge\mathcal{I}_{1} in (33).

The next order of error has x~1−2​nn−1\tilde{x}_{1}^{-\frac{2n}{n-1}} in front. Within the x2=O⁡(1)x_{2}=O(1) region, the volume error is now O⁡(x~1−2​nn−1)O(\tilde{x}_{1}^{-\frac{2n}{n-1}}). For x2≫1x_{2}\gg 1 one needs to be careful about the effect of UU having a growing term a′​u0a^{\prime}u_{0}, which will damage the exponential decay. Recall from section 4.3 that u0=x21n−1u_{0}=x_{2}^{\frac{1}{n-1}} outside some compact region in the Tian-Yau space. The largest new contributions to the volume forms error come from terms such as

(x~1n−2n⁡(n−1)​d​dc​ϕT​Y)n−2∧dc​x~1n−2n⁡(n−1)∧d​ϕT​Y∧d​x~1n+2n−2​nn−1∧dc​u0,(\tilde{x}_{1}^{\frac{n-2}{n(n-1)}}dd^{c}\phi_{TY})^{n-2}\wedge d^{c}\tilde{x}_{1}^{\frac{n-2}{n(n-1)}}\wedge d\phi_{TY}\wedge d\tilde{x}_{1}^{\frac{n+2}{n}-\frac{2n}{n-1}}\wedge d^{c}u_{0},

whose local Ck,αC^{k,\alpha}-norm is O⁡(x~1−2​nn−1​x21n−1)O(\tilde{x}_{1}^{-\frac{2n}{n-1}}x_{2}^{\frac{1}{n-1}}). ∎

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