弦论学家在 1990 年前后发现了一件怪事:卡拉比–丘流形似乎总是成对出现,两个形状截然不同的空间,给出完全相同的物理。这就是镜像对称。它甚至能用来解决古老的计数问题,可是没人知道这两个空间在几何上到底是什么关系。
1996 年,Strominger、丘成桐与 Zaslow 给出了一个几何解释:卡拉比–丘流形可以看成一族特殊拉格朗日环面铺在一个底空间上;把每个环面换成它的对偶环面,就得到镜像。一句话:镜像对称就是 T-对偶。
Around 1990 string theorists noticed something strange: Calabi–Yau manifolds seemed to come in pairs — two very different shapes giving identical physics. This is mirror symmetry. It could even solve classical counting problems, yet nobody knew what the two spaces had to do with each other geometrically.
In 1996 Strominger, Yau and Zaslow proposed a geometric answer: a Calabi–Yau manifold is a family of special Lagrangian tori spread over a base; replace each torus by its dual torus and you get the mirror. In one line: mirror symmetry is T-duality.
镜像对称是什么What mirror symmetry is
一个复三维卡拉比–丘流形 $X$ 有两个重要的数:$h^{1,1}(X)$ 大致数"能改变大小的方式"(凯勒形变),$h^{2,1}(X)$ 数"能改变形状的方式"(复结构形变)。物理学家发现,常常存在另一个卡拉比–丘流形 $\check X$,使得这两个数恰好互换:
A complex three-dimensional Calabi–Yau manifold $X$ has two important numbers: $h^{1,1}(X)$ roughly counts ways to change its size (Kähler deformations), $h^{2,1}(X)$ counts ways to change its shape (complex deformations). Physicists found that there is often another Calabi–Yau $\check X$ for which these numbers are exactly swapped:
$$h^{1,1}(X)=h^{2,1}(\check X),\qquad h^{2,1}(X)=h^{1,1}(\check X).$$更深的层面上,$X$ 的"辛几何"(大小、曲线的计数)等价于 $\check X$ 的"复几何"(形状、周期积分)。一边很难算的东西,在另一边可能很容易。
At a deeper level, the "symplectic geometry" of $X$ (sizes, curve counts) is equivalent to the "complex geometry" of $\check X$ (shapes, period integrals). What is hard to compute on one side may be easy on the other.
五次三维流形上有 2875 条直线、609250 条二次曲线……代数几何学家花了很大力气才算出前几项。1991 年,Candelas、de la Ossa、Green、Parkes 在镜像那一侧算一个周期积分,一次性给出了所有次数的有理曲线数目的预言,后来被数学家证明。
The quintic threefold contains 2875 lines and 609,250 conics… algebraic geometers struggled to compute even the first terms. In 1991 Candelas, de la Ossa, Green and Parkes computed a period integral on the mirror side and predicted the number of rational curves of every degree at once — later proved by mathematicians.
T-对偶:最简单的例子T-duality: the simplest example
在弦论里,一根弦在半径为 $R$ 的圆周上运动,和一根弦在半径为 $1/R$ 的圆周上运动,给出完全相同的物理:动量模式和缠绕模式互换了角色。这叫 T-对偶。
In string theory, a string moving on a circle of radius $R$ gives exactly the same physics as one on a circle of radius $1/R$: momentum modes and winding modes swap roles. This is T-duality.
一维复卡拉比–丘流形就是椭圆曲线,也就是环面 $T^2=S^1\times S^1$。把其中一个圆周换成对偶圆周,得到的仍然是一个环面,但"大小"和"形状"的参数互换了。这就是最简单的镜像对称。
A one-dimensional Calabi–Yau is an elliptic curve — a torus $T^2=S^1\times S^1$. Dualize one circle and you get another torus with its "size" and "shape" parameters exchanged. That is the simplest mirror symmetry.
SYZ 的想法就是:高维的情形,其实是把这个最简单的例子"一族一族"地做。
The SYZ idea: the higher-dimensional case is this simplest example done "in families".
猜想的陈述The statement
卡拉比–丘流形 $X$ 上有凯勒形式 $\omega$ 和处处非零的全纯体积形式 $\Omega$。一个实 $n$ 维子流形 $L$ 称为特殊拉格朗日的,如果
A Calabi–Yau manifold $X$ carries a Kähler form $\omega$ and a nowhere-vanishing holomorphic volume form $\Omega$. A real $n$-dimensional submanifold $L$ is special Lagrangian if
$$\omega|_L=0,\qquad \operatorname{Im}\Omega|_L=0 .$$这样的子流形在它的同调类里体积最小,是高维的"极小曲面"。McLean 证明,紧特殊拉格朗日子流形 $L$ 的形变空间维数等于 $b_1(L)$;对三维环面 $T^3$,恰好是 3。
Such submanifolds minimize volume in their homology class — higher-dimensional "minimal surfaces". McLean proved that compact special Lagrangians $L$ deform in a family of dimension $b_1(L)$; for a three-torus $T^3$, exactly 3.
若 $X$ 和 $\check X$ 是一对镜像卡拉比–丘三维流形(在适当的极限下),则
If $X$ and $\check X$ are a mirror pair of Calabi–Yau threefolds (near a suitable limit), then
- $X$ 有一个以特殊拉格朗日 $T^3$ 为一般纤维的纤维化 $f:X\to B$;$\check X$ 也有一个,$\check f:\check X\to B$,底空间相同;
- $X$ admits a fibration $f:X\to B$ whose generic fibres are special Lagrangian $T^3$s; so does $\check X$, $\check f:\check X\to B$, over the same base;
- 在一般点 $b\in B$ 上,纤维 $\check f^{-1}(b)$ 是 $f^{-1}(b)$ 的对偶环面;
- over a generic point $b\in B$, the fibre $\check f^{-1}(b)$ is the dual torus of $f^{-1}(b)$;
- $\check X$ 的复结构,由 $X$ 的辛几何加上全纯圆盘带来的"瞬子修正"决定。
- the complex structure of $\check X$ is determined by the symplectic geometry of $X$ together with "instanton corrections" from holomorphic discs.
首屏的金色图形是这种纤维化的示意:沿着底空间排开的一串环面纤维,到两端时退化收缩,就像真正的纤维化在奇异纤维处发生的那样。
The golden figure at the top sketches such a fibration: a row of torus fibres along the base, degenerating at the ends as real fibrations do at singular fibres.
为什么应该成立:D-膜论证Why it should hold: the D-brane argument
论文的论证来自物理,却有一个漂亮的几何内核。镜像对称应当把 $\check X$ 上的物体一一对应到 $X$ 上的物体。看最简单的物体:$\check X$ 上的一个点(D0-膜)。所有点构成的模空间就是 $\check X$ 本身。
The paper's argument comes from physics but has a beautiful geometric core. Mirror symmetry should match objects on $\check X$ with objects on $X$. Take the simplest object: a point of $\check X$ (a D0-brane). The moduli space of all points is $\check X$ itself.
在 $X$ 一侧,它应当对应一个特殊拉格朗日子流形 $L$,再配上 $L$ 上一个平坦 $U(1)$ 联络(D3-膜)。这个模空间要和 $\check X$ 一样是复三维(实六维)的,所以:
On the $X$ side, it should correspond to a special Lagrangian $L$ with a flat $U(1)$ connection on it (a D3-brane). This moduli space must be complex three-dimensional (real six), like $\check X$. Hence:
- $L$ 的形变给出 $b_1(L)$ 个实维度,平坦联络又给出 $b_1(L)$ 个,所以 $2b_1(L)=6$,$b_1(L)=3$;最自然的候选是 $L\cong T^3$;
- deformations of $L$ give $b_1(L)$ real dimensions and flat connections another $b_1(L)$, so $2b_1(L)=6$ and $b_1(L)=3$; the natural candidate is $L\cong T^3$;
- $L$ 的形变扫出 $X$,即 $X$ 被这些 $T^3$ 纤维化,底空间就是 $L$ 的模空间 $B$;
- deformations of $L$ sweep out $X$, so $X$ is fibred by these $T^3$s over their moduli space $B$;
- 固定 $L$ 时,平坦 $U(1)$ 联络构成的空间正是对偶环面 $\check T^3$,于是 $\check X$ 被对偶环面纤维化。
- for fixed $L$, the flat $U(1)$ connections form exactly the dual torus $\check T^3$, so $\check X$ is fibred by dual tori.
后来的故事What happened next
SYZ 猜想在完全一般的形式下仍然是开放问题,主要困难在于奇异纤维和特殊拉格朗日子流形的存在性。但它已经深刻地改变了这个领域:
In full generality the SYZ conjecture remains open — singular fibres and the existence of special Lagrangians are the main difficulties. But it has reshaped the field:
- 拓扑与辛版本:Gross 等人对五次三维流形等例子构造了拓扑和辛意义下的 SYZ 纤维化。
- Topological and symplectic versions: Gross and others constructed topological and symplectic SYZ fibrations for examples such as the quintic threefold.
- Gross–Siebert 纲领与 Kontsevich–Soibelman 的工作,把 SYZ 的思想代数化,发展出用"热带几何"和"墙穿越"系统构造镜像的方法。
- The Gross–Siebert program and the work of Kontsevich–Soibelman algebraized the SYZ idea, building mirrors systematically from tropical geometry and wall-crossing.
- 瞬子修正:全纯圆盘如何修正镜像的复结构,成为 Fukaya、Auroux 等人工作的核心问题,也连接到 Kontsevich 的同调镜像对称。
- Instanton corrections: how holomorphic discs correct the mirror's complex structure became central in the work of Fukaya, Auroux and others, linking to Kontsevich's homological mirror symmetry.
原文与延伸阅读Original papers & further reading
- 1996A. Strominger, S.-T. Yau, E. Zaslow, Mirror symmetry is T-duality, Nuclear Phys. B 479, 243–259. arXiv ↗
- 1991P. Candelas, X. de la Ossa, P. Green, L. Parkes, A pair of Calabi–Yau manifolds as an exactly soluble superconformal theory, Nuclear Phys. B 359, 21–74.
- 1998R. C. McLean, Deformations of calibrated submanifolds, Comm. Anal. Geom. 6, 705–747.
- 1994M. Kontsevich, Homological algebra of mirror symmetry, Proc. ICM Zürich, arXiv:alg-geom/9411018.