广义相对论里有一个看似显然、却几十年没人能证明的问题:一个孤立的引力系统,它的总质量(总能量)会不会是负的?引力能本身是负的,相互抵消之后会不会把总量拉到零以下?
1979 年,舍恩(Richard Schoen)与丘成桐证明:不会。只要物质满足合理的能量条件,总质量 $m\ge0$;并且 $m=0$ 当且仅当时空是平直的。他们的工具不是物理,而是几何中的极小曲面。
General relativity had a question that looked obvious yet resisted proof for decades: can the total mass (energy) of an isolated gravitating system be negative? Gravitational binding energy is negative — could it drag the total below zero?
In 1979 Richard Schoen and Shing-Tung Yau proved: no. If matter satisfies a reasonable energy condition, the total mass satisfies $m\ge0$, and $m=0$ only for flat space. Their tool was not physics but geometry — minimal surfaces.
引力的能量问题The energy problem of gravity
在牛顿力学里,能量可以在每一点上定义并加起来。可在广义相对论中,引力本身就是时空的弯曲。根据等效原理,在任何一点附近都可以选一个自由下落的参考系,让引力"局部消失"。于是引力场的能量密度没法在一点上定义。
In Newtonian mechanics, energy can be defined at every point and summed. In general relativity, gravity is the curvature of spacetime, and by the equivalence principle you can always choose a freely falling frame where gravity "vanishes" locally. So the energy density of the gravitational field cannot be defined at a point.
但对一个孤立系统(比如一颗星、一个星系,远处的时空越来越接近平直)来说,可以站在无穷远处,看时空度量偏离平直的程度,读出一个总质量。这就是 1961 年 Arnowitt、Deser、Misner 定义的 ADM 质量。
But for an isolated system (a star, a galaxy — spacetime becoming flat far away) one can stand at infinity, look at how the metric deviates from flat, and read off a total mass. This is the ADM mass of Arnowitt, Deser and Misner (1961).
就像你不必数清一座城市里每个人的体重,只要站在远处,看它让周围的地面陷下去多少,就能称出它的总重量。
Like weighing a city without counting each person: stand far away and see how much it makes the ground sag around it.
精确的陈述The precise statement
取时空在某一时刻的空间切片。在"时间对称"的情形下,它是一个三维黎曼流形 $(M^3,g)$,在无穷远处渐近于欧氏空间:在某个紧集外,$g_{ij}=\delta_{ij}+O(r^{-1})$,导数衰减得更快。这时 ADM 质量为
Take a spatial slice of spacetime at one moment. In the "time-symmetric" case it is a Riemannian 3-manifold $(M^3,g)$, asymptotically Euclidean: outside a compact set, $g_{ij}=\delta_{ij}+O(r^{-1})$ with faster-decaying derivatives. The ADM mass is
$$m=\frac{1}{16\pi}\lim_{r\to\infty}\oint_{S_r}\big(\partial_i g_{ij}-\partial_j g_{ii}\big)\,\nu^j\,dA .$$在时间对称情形下,爱因斯坦方程把"物质能量密度非负"翻译成一个纯几何条件:数量曲率 $R_g\ge0$。
In the time-symmetric case, Einstein's equations translate "matter energy density is nonnegative" into a purely geometric condition: scalar curvature $R_g\ge0$.
设 $(M^3,g)$ 是完备、渐近平坦的三维黎曼流形,且数量曲率 $R_g\ge0$。则其 ADM 质量 $m\ge0$;并且 $m=0$ 当且仅当 $(M,g)$ 等距于欧氏空间 $\mathbb{R}^3$。
Let $(M^3,g)$ be a complete, asymptotically flat Riemannian 3-manifold with scalar curvature $R_g\ge0$. Then its ADM mass satisfies $m\ge0$, and $m=0$ if and only if $(M,g)$ is isometric to Euclidean $\mathbb{R}^3$.
1981 年,他们利用 Jang 方程把一般(非时间对称)情形归结为这一情形,证明了在主导能量条件下的完整结论。同年 Witten 用旋量给出了另一个完全不同的证明。
In 1981 they reduced the general (non-time-symmetric) case to this one using Jang's equation, proving the full statement under the dominant energy condition. The same year Witten gave an entirely different proof using spinors.
工具:极小曲面The tool: minimal surfaces
把铁丝圈浸进肥皂水,拿出来就是一张膜。表面张力让它面积尽量小,这就是极小曲面:平均曲率 $H=0$。如果一张极小曲面在微小扰动下面积不会减小,就称为稳定的。
Dip a wire loop into soapy water and lift it out: surface tension makes the film as small as possible. That is a minimal surface, with mean curvature $H=0$. If small perturbations never decrease its area, it is stable.
稳定性可以写成一个积分不等式(第二变分公式):对曲面 $\Sigma$ 上任何紧支函数 $f$,
Stability can be written as an integral inequality (the second variation formula): for every compactly supported function $f$ on $\Sigma$,
$$\int_\Sigma |\nabla f|^2 \;\ge\; \int_\Sigma\big(\operatorname{Ric}(\nu,\nu)+|A|^2\big)f^2 ,$$其中 $\nu$ 是单位法向量,$A$ 是第二基本形式。舍恩和丘先生的关键观察,是用高斯方程把右边改写,让数量曲率和曲面自身的高斯曲率 $K$ 出现:
where $\nu$ is the unit normal and $A$ the second fundamental form. Schoen and Yau's key observation was to rewrite the right side using the Gauss equation so that the scalar curvature and the surface's own Gauss curvature $K$ appear:
$$\operatorname{Ric}(\nu,\nu)+|A|^2=\tfrac12 R_g-K+\tfrac12|A|^2 \qquad(\text{on a minimal surface}).$$于是稳定性变成:
So stability becomes:
$$\int_\Sigma |\nabla f|^2+K f^2\;\ge\;\tfrac12\int_\Sigma\big(R_g+|A|^2\big)f^2\;\ge\;0 .$$数量曲率非负,会迫使稳定极小曲面"整体上偏正弯曲"。而拓扑(高斯–博内定理)又严格限制了一张曲面的总弯曲。两者的冲突就是证明的引擎。
Nonnegative scalar curvature forces a stable minimal surface to be "positively curved on balance". Topology (Gauss–Bonnet) strictly limits the total curvature a surface can have. The clash between the two drives the proof.
证明的思路:反证法The proof: by contradiction
第一步:化简Step 1: simplify
假设 $m<0$。通过一个小的共形变换,可以让度量在无穷远处变成"调和平坦"的标准形式 $g=u^4\delta$($u$ 是调和函数,$u=1+\tfrac{m}{2r}+\dots$),同时保持 $R_g\ge0$、$m<0$,并在适当情形下让 $R_g>0$。
Suppose $m<0$. A small conformal change makes the metric "harmonically flat" near infinity, $g=u^4\delta$ with $u$ harmonic and $u=1+\tfrac{m}{2r}+\dots$, while keeping $R_g\ge0$ and $m<0$, and arranging $R_g>0$ where needed.
第二步:负质量造出一张"平面状"的极小曲面Step 2: negative mass produces a plane-like minimal surface
在无穷远处看坐标平面 $\{x^3=\pm h\}$。当 $m<0$ 时,它们相对度量 $g$ 的平均曲率恰好指向"合适"的方向,可以当作障碍,把面积极小化的曲面夹在中间。在越来越大的圆周上解普拉托问题,再取极限,就得到一张完备、稳定、渐近于平面的面积极小曲面 $\Sigma$。
Look at the coordinate planes $\{x^3=\pm h\}$ near infinity. When $m<0$, their mean curvature with respect to $g$ points in just the right direction for them to act as barriers, trapping area-minimizing surfaces between them. Solving Plateau problems on ever larger circles and passing to the limit yields a complete, stable, asymptotically planar area-minimizing surface $\Sigma$.
第三步:曲率与拓扑的冲突Step 3: curvature versus topology
在稳定性不等式里取 $f$ 为对数截断函数(在半径 $\rho$ 内为 1,到 $\rho^2$ 处降为 0)。令 $\rho\to\infty$,$\int|\nabla f|^2\to0$,于是
In the stability inequality take $f$ to be a logarithmic cutoff (equal to 1 inside radius $\rho$, decaying to 0 at $\rho^2$). As $\rho\to\infty$, $\int|\nabla f|^2\to0$, so
$$\int_\Sigma K\,dA\;\ge\;\tfrac12\int_\Sigma\big(R_g+|A|^2\big)\,dA\;>\;0 .$$但 $\Sigma$ 渐近于平面。由高斯–博内定理,大圆盘边界的测地曲率积分趋近 $2\pi$,所以总曲率 $\int_\Sigma K\,dA\le 0$。两个不等式矛盾,所以 $m\ge0$。
But $\Sigma$ is asymptotic to a plane. By Gauss–Bonnet, the geodesic curvature integral over the boundary of large discs tends to $2\pi$, so the total curvature satisfies $\int_\Sigma K\,dA\le 0$. Contradiction — hence $m\ge0$.
刚性:$m=0$ 时为什么一定平直Rigidity: why $m=0$ forces flatness
如果 $m=0$ 但度量不平直,可以对度量做一个小扰动,使数量曲率仍然非负、而质量变成负的,这又与刚证明的结论矛盾。所以 $m=0$ 时 $(M,g)$ 只能是 $\mathbb{R}^3$。
If $m=0$ but the metric is not flat, a small perturbation keeps the scalar curvature nonnegative while making the mass negative — contradicting what was just proved. So $m=0$ forces $(M,g)=\mathbb{R}^3$.
意义与影响Meaning and legacy
- 物理:平直的闵可夫斯基时空是引力的"基态",不可能通过释放能量衰变到更低的状态。这是广义相对论自洽性的一块基石。
- Physics: flat Minkowski spacetime is the "ground state" of gravity — it cannot decay to anything lower by releasing energy. A cornerstone of the consistency of general relativity.
- 几何:同一时期,舍恩和丘先生用同样的极小曲面方法研究正数量曲率流形的拓扑,例如证明三维环面 $T^3$ 上不存在正数量曲率度量。用极小曲面逐维下降的思想,至今仍是这个领域的核心技术。
- Geometry: in the same period Schoen and Yau used the minimal-surface method to study the topology of manifolds with positive scalar curvature, e.g. showing that the 3-torus $T^3$ admits no metric of positive scalar curvature. Dimension descent via minimal surfaces remains a core technique.
- 后续:彭罗斯不等式(把质量与黑洞视界面积联系起来)、准局部质量(包括刘秋菊–丘、王慕道–丘的质量定义)等方向,都是沿着这条路发展起来的。
- Afterwards: the Penrose inequality (relating mass to black-hole horizon area) and quasi-local mass (including the Liu–Yau and Wang–Yau definitions) grew along this road.
首屏那个金色的形状是史瓦西黑洞在某一时刻的空间切片(Flamm 抛物面):两张渐近平坦的"薄片"被一个喉部连起来,也就是爱因斯坦–罗森桥。它的 ADM 质量为正,正是定理所说的情形。
The golden shape at the top is a spatial slice of the Schwarzschild black hole (Flamm's paraboloid): two asymptotically flat sheets joined by a throat — the Einstein–Rosen bridge. Its ADM mass is positive, just as the theorem says.
原文与延伸阅读Original papers & further reading
- 1979R. Schoen, S.-T. Yau, On the proof of the positive mass conjecture in general relativity, Comm. Math. Phys. 65, 45–76. DOI ↗
- 1981R. Schoen, S.-T. Yau, Proof of the positive mass theorem. II, Comm. Math. Phys. 79, 231–260. DOI ↗
- 1979R. Schoen, S.-T. Yau, On the structure of manifolds with positive scalar curvature, Manuscripta Math. 28, 159–183.
- 1981E. Witten, A new proof of the positive energy theorem, Comm. Math. Phys. 80, 381–402.
- 1961R. Arnowitt, S. Deser, C. Misner, Coordinate invariance and energy expressions in general relativity, Phys. Rev. 122, 997–1006.