前面五块碑,其实都是同一种思想的不同侧面:把一个几何问题翻译成一个偏微分方程,用分析的力量解出它,再把答案翻译回几何。这种思想就是"几何分析"。
丘成桐不是第一个用方程研究几何的人,但他把它变成了一门系统的学科:发展出一整套通用工具(梯度估计、极大值原理、先验估计、极小曲面),培养了大批学生,并在 1982 年整理出一份影响深远的公开问题清单。
The other five steles are really facets of one idea: translate a geometric problem into a PDE, solve it with the power of analysis, and translate the answer back into geometry. That idea is "geometric analysis".
Yau was not the first to study geometry through equations, but he turned it into a systematic discipline — developing a general toolkit (gradient estimates, maximum principles, a priori estimates, minimal surfaces), training generations of students, and compiling in 1982 an enormously influential list of open problems.
核心思想:形状即方程The core idea: shape is an equation
几何对象的很多"最优"状态,都由一个方程刻画:
Many "optimal" states of geometric objects are described by an equation:
- 肥皂膜面积最小 ⟺ 平均曲率 $H=0$(极小曲面方程);
- Soap films minimize area ⟺ mean curvature $H=0$ (the minimal surface equation);
- 温度平衡 ⟺ 调和函数 $\Delta u=0$;
- Thermal equilibrium ⟺ harmonic functions $\Delta u=0$;
- 真空中的引力 ⟺ 里奇平坦 $\operatorname{Ric}=0$(卡拉比猜想正是在解它);
- Gravity in vacuum ⟺ Ricci-flat $\operatorname{Ric}=0$ (exactly what the Calabi conjecture solves);
- 规定曲率的凸体 ⟺ 实蒙日–安培方程 $\det(\nabla^2 u)=f$(闵可夫斯基问题)。
- Convex bodies with prescribed curvature ⟺ the real Monge–Ampère equation $\det(\nabla^2 u)=f$ (the Minkowski problem).
首屏那个不断变形的金色曲面,是从悬链面到螺旋面的"伴随族"。变形过程中的每一个瞬间都是极小曲面,而且度量保持不变,就像一张不可伸缩的肥皂膜在扭转。
The golden surface morphing at the top is the "associate family" from a catenoid to a helicoid: every instant is a minimal surface, and the metric never changes — like an unstretchable soap film twisting.
工具箱The toolkit
梯度估计与刘维尔定理(1975)Gradient estimates and Liouville (1975)
丘先生证明:在里奇曲率非负的完备流形上,正调和函数满足一个只依赖于维数的梯度估计。在半径为 $R$ 的测地球 $B(x,R)$ 上,若 $u>0$ 调和,则在 $B(x,R/2)$ 上
Yau proved that on a complete manifold with nonnegative Ricci curvature, positive harmonic functions obey a gradient estimate depending only on dimension: if $u>0$ is harmonic on the geodesic ball $B(x,R)$, then on $B(x,R/2)$
$$\frac{|\nabla u|}{u}\le\frac{C(n)}{R}.$$令 $R\to\infty$,立刻得到:这样的流形上,正调和函数一定是常数。这一方法后来发展成李–丘估计,以及大量梯度估计的工作。
Let $R\to\infty$: on such a manifold, every positive harmonic function is constant. The method grew into the Li–Yau estimate and a vast literature of gradient estimates.
广义极大值原理The generalized maximum principle
紧流形上,函数在最大值点梯度为零、拉普拉斯非正。非紧流形上最大值可能取不到。大森英树与丘成桐发展的极大值原理说:在里奇曲率有下界的完备流形上,有界函数总有一列点"近似"满足这些性质。这是把紧流形上的论证推广到非紧情形的标准工具,后来在丘先生的施瓦茨引理中起了关键作用。
On a compact manifold, a function's maximum point has zero gradient and nonpositive Laplacian. On a noncompact one the maximum may not be attained. The Omori–Yau maximum principle says: on a complete manifold with Ricci curvature bounded below, a bounded function has a sequence of points "approximately" satisfying these properties — the standard tool for carrying compact arguments to noncompact settings, and key to Yau's Schwarz lemma.
先验估计与连续性方法A priori estimates and the continuity method
卡拉比猜想的证明把这套工具推到极致:一个完全非线性方程,用一层层的先验估计($C^0$、$C^2$、$C^3$)配合连续性方法解出。这一模式后来被反复用于各种几何方程。
The proof of the Calabi conjecture pushed this toolkit to the limit: a fully nonlinear equation solved with layered a priori estimates ($C^0$, $C^2$, $C^3$) and the continuity method — a pattern reused for countless geometric equations.
极小曲面作为探针Minimal surfaces as probes
把极小曲面放进一个空间,看它被迫怎样弯曲,就能读出空间的曲率和拓扑。正质量定理、正数量曲率流形的结构,以及 Meeks–丘对三维拓扑的工作,都用了这一思想。
Place a minimal surface inside a space and see how it is forced to bend: it reveals the space's curvature and topology. The positive mass theorem, the structure of positive scalar curvature manifolds, and Meeks–Yau's work on 3-manifold topology all use this idea.
代表性成果地图A map of landmark results
上排是工具,下排是成果。点击任意方块查看它的内容。
Tools on top, results below. Click any box to read about it.
一门学科的建立Building a discipline
1979–80 年在普林斯顿高等研究院,丘先生组织了微分几何特别年。成果汇编成 1982 年出版的《Seminar on Differential Geometry》,其中的 "Problem Section" 列出了一百多个公开问题,影响了此后几十年的几何分析研究,其中许多问题已被他的学生和后继者解决。
In 1979–80 at the Institute for Advanced Study, Yau organized a special year in differential geometry. The resulting volume, Seminar on Differential Geometry (1982), contains a "Problem Section" of more than a hundred open problems that shaped geometric analysis for decades; many have since been solved by his students and successors.
他也积极推动 Hamilton 的里奇流纲领,鼓励学生进入这个方向。里奇流后来成为 Perelman 证明庞加莱猜想和几何化猜想的工具。
He also championed Hamilton's Ricci flow program and encouraged students to enter it; Ricci flow later became Perelman's tool for the Poincaré and geometrization conjectures.
今天,几何分析已经是数学的一个主要分支,从广义相对论、弦论到图像处理(例如丘先生学生顾险峰发展的计算共形几何),到处都能看到它的影子。主页的"薪火·师承"一章,就是这门学科的一部分家谱。
Today geometric analysis is a major branch of mathematics, visible from general relativity and string theory to image processing (e.g. the computational conformal geometry developed by Yau's student Xianfeng Gu). The "Lineage" chapter on the homepage is a piece of this discipline's family tree.
原文与延伸阅读Original papers & further reading
- 1975S.-T. Yau, Harmonic functions on complete Riemannian manifolds, Comm. Pure Appl. Math. 28, 201–228.
- 1975S. Y. Cheng, S.-T. Yau, Differential equations on Riemannian manifolds and their geometric applications, Comm. Pure Appl. Math. 28, 333–354.
- 1976S. Y. Cheng, S.-T. Yau, On the regularity of the solution of the n-dimensional Minkowski problem, Comm. Pure Appl. Math. 29, 495–516.
- 1978S.-T. Yau, A general Schwarz lemma for Kähler manifolds, Amer. J. Math. 100, 197–203.
- 1980W. H. Meeks III, S.-T. Yau, Topology of three dimensional manifolds and the embedding problems in minimal surface theory, Ann. of Math. 112, 441–484.
- 1982S.-T. Yau (ed.), Seminar on Differential Geometry, Annals of Mathematics Studies 102, Princeton Univ. Press.