The heat kernel and generalized $\zeta$ -functions

12.7 The heat kernel and generalized $\zeta$ -functions

As we mentioned in section 12.2, there are several methods of proving the AS index theorem. The heat kernel is relatively accessible to physicists and it also has many applications to other problems in physics. The generalized $\zeta$ -function is related to the heat kernel and also has relevance in physics.

12.7.1 The heat kernel and index theorem

Let $E$ be a complex vector bundle over an $m$ -dimensional compact manifold $M$ . Let $\Delta :\Gamma (M,E)\to \Gamma (M,E)$ be an elliptic operator with eigenvectors $|n⟩$ such that

$$\Delta |n⟩={\lambda }_n|n⟩.\text{\hspace{1em}(12.91)}$$

We denote the set of eigenvalues of $\Delta$ by $\mathrm{Spec}\Delta$ . We assume that $\Delta$ is non-negative, i.e. all the eigenvalues are non-negative. Suppose there are $n_0$ modes $|0,i⟩$ , $1\le i\le n_0$ with vanishing eigenvalue. In other words,

$$\dim \ker \Delta =n_0.\text{\hspace{1em}(12.92)}$$

These modes are called the zero modes. Define the heat kernel $h(t)$ by

$$h(t)\equiv {\mathrm{e}}^{-t\Delta }.\text{\hspace{1em}(12.93)}$$

It is convenient to represent $h(t)$ in the coordinate basis as

$$\begin{array}{rl}h(x,y;t)\equiv ⟨x|h(t)|y⟩& =⟨x|\sum _n{\mathrm{e}}^{-t\Delta }|n⟩⟨n|y⟩\\ & =\sum _n{\mathrm{e}}^{-t{\lambda }_n}⟨x|n⟩⟨n|y⟩.\end{array}\text{\hspace{1em}(12.94)}$$

Multiple eigenstates should be counted as many times as they appear. We assume $⟨x|n⟩$ is orthonormal: $\int ⟨n|x⟩⟨x|m⟩\mathrm{d}x={\delta }_{mn}$ . The convergence of (12.93) for $t>0$ is guaranteed since $\Delta$ is non-negative. Taking the limit $t\to \infty$ , we have

$$\underset{t\to \infty }{\lim }h(x,y;t)=\sum _{i=1}^{n_0}⟨x|0,i⟩⟨0,i|y⟩\text{\hspace{1em}(12.95)}$$

where the summation is over the zero modes $|0,i⟩$ only. Thus, $h={\mathrm{e}}^{-t\Delta }$ tends to be the projection operator onto the space of zero modes as

$${\mathrm{e}}^{-t\Delta }\stackrel{t\to \infty }{⟶}\sum _{i=1}^{n_0}|0,i⟩⟨0,i|.\text{\hspace{1em}(12.96)}$$

Define

$$\tilde{h}(t)\equiv \int h(x,x;t)dx=\sum _n{e}^{-t{\lambda }_n}.\text{\hspace{1em}(12.97)}$$

Then it follows from (12.95) that

$$n_0=\underset{t\to \infty }{\lim }\tilde{h}(t).\text{\hspace{1em}(12.98)}$$

It is easy to verify that $h$ satisfies the heat equation,

$$\left(\frac{\partial }{\partial t}+{\Delta }_x\right)h(x,y;t)=0.\text{\hspace{1em}(12.99)}$$

If $\Delta$ is the conventional Laplacian, (12.99) reduces to the ordinary heat equation. The initial condition is

$$h(x,y;0)=\sum _n⟨x|n⟩⟨n|y⟩=\delta (x-y)\text{\hspace{1em}(12.100)}$$

where the last equality follows from the completeness of the eigenvectors.

Exercise 12.4

Let $u(x,t)$ be a solution of (12.99) such that $u(x,0)=u(x)$ . Show that

$$u(x,t)=\int h(x,y;t)u(y)dy.\text{\hspace{1em}(12.101)}$$

[Hint: First verify that (12.101) satisfies the initial condition, next that it is a solution of the heat equation.]

It is known that the solution of (12.99) has an asymptotic expansion for $t\to \epsilon$ given by

$$h(x,x;\epsilon )=\sum _i{a}_i(x){\epsilon }^i\text{\hspace{1em}(12.102)}$$

see Gilkey (1984). Similarly, $h(t)$ has an expansion

$$\tilde{h}(ϵ)\equiv \sum _i{a}_i{\epsilon }^i\text{\hspace{1em}(12.103)}$$

where $a_i=\int a_i(x)\mathrm{d}x$

Let $E$ and $F$ be complex vector bundles over $M$ and $D:\Gamma (M,E)\to \Gamma (M,F)$ be an elliptic operator. We define two Laplacians

$${\Delta }_E\equiv D^{†}D:\Gamma (M,E)\to \Gamma (M,E)\text{\hspace{1em}(12.104a)}$$

$${\Delta }_F\equiv D{D}^{†}:\Gamma (M,F)\to \Gamma (M,F).\text{\hspace{1em}(12.104b)}$$

It is important to note that they have the same non-vanishing eigenvalues including the degeneracy. To see this, let ${\Delta }_E|\lambda ⟩=\lambda |\lambda ⟩$ . Then there is a vector $D|\lambda ⟩\in \Gamma (M,F)$ such that

$${\Delta }_F(D|\lambda ⟩)=D{D}^{†}D|\lambda ⟩=D{\Delta }_E|\lambda ⟩=\lambda (D|\lambda ⟩).$$

Note that $D|\lambda ⟩\ne 0$ since $\ker {\Delta }_E=\ker D$ . Conversely, if $|\mu ⟩\in \Gamma (M,F)$ satisfies ${\Delta }_F|\mu ⟩=\mu |\mu ⟩$ , then $D^{†}|\mu ⟩\in \Gamma (M,E)$ is an eigenvector of ${\Delta }_E$ with the same eigenvalue $\mu$ . Thus, we have found the symmetry³

$${\mathrm{Spec}}^{\prime }{\Delta }_E={\mathrm{Spec}}^{\prime }{\Delta }_F\text{\hspace{1em}(12.105)}$$

where the prime denotes that the zero eigenmodes are omitted.

Define two heat kernels $h_E$ and $h_F$ by

$$h_E(x,y,t)=\sum {\mathrm{e}}^{-{\lambda }_n}⟨x|n⟩⟨n|y⟩\text{\hspace{1em}(12.106a)}$$

$$h_F(x,y,t)=\sum {\mathrm{e}}^{-{\mu }_m}⟨x|m)(m|y⟩.\text{\hspace{1em}(12.106b)}$$

We have

$$\underset{t\to \infty }{\lim }{\tilde{h}}_E(t)=\dim \ker {\Delta }_E=\dim \ker D\text{\hspace{1em}(12.107a)}$$

$$\underset{t\to \infty }{\lim }{\tilde{h}}_F(t)=\dim \ker {\Delta }_F=\dim \ker D^{†}.\text{\hspace{1em}(12.107b)}$$

What is more interesting is the index of $D$ . Since $\ker D=\ker {\Delta }_E$ and $\ker D^{†}=\ker {\Delta }_F$ , we have

$$\begin{array}{rl}\mathrm{ind}D& =\dim \ker D-\dim \ker D^{†}=\dim \ker {\Delta }_E-\dim \ker {\Delta }_F\\ & =\underset{t\to \infty }{\lim }\left[{\tilde{h}}_E(t)-{\tilde{h}}_F(t)\right]={\tilde{h}}_E(t)-{\tilde{h}}_F(t).\end{array}\text{\hspace{1em}(12.108)}$$

The final equality follows since the $t$ -dependent part of ${\tilde{h}}_E(t)-{\tilde{h}}_F(t)$ cancels out by the symmetry (12.105). We expand ${\tilde{h}}_E(t)$ and ${\tilde{h}}_F(t)$ as

$${\tilde{h}}_E(t)=\sum a_i^E{t}^i{\tilde{h}}_F(t)=\sum a_i^F{t}^i.$$

This is a kind of ‘supersymmetry’, see section 12.10.

Picking up $t$ -independent terms, we have

$$\mathrm{ind}D=a_0^E-a_0^F=\int \mathrm{d}x\left[a_0^E(x)-a_0^F(x)\right]\mathrm{d}x\text{\hspace{1em}(12.109)}$$

where $a_0^{E,F}(x)$ are defined in (12.102).

In general, $a_0^{E,F}(x)$ are local invariants written in terms of curvature two-forms. In section 13.2, we use the heat kernel to prove the index theorem

$$\mathrm{ind}D=v_{+}-v_{-}={\int }_M\mathrm{ch}(\mathcal{F}){|}_{\text{v o l}}$$

for the twisted spin complex over a manifold with $\hat{A}(TM)=1$ .

Exercise 12.5

Let $D,D^{†},{\Delta }_E$ and ${\Delta }_F$ be as before. Show that

$$I(s)\equiv \mathrm{tr}\left[\frac{s}{{\Delta }_E+s}-\frac{s}{{\Delta }_F+s}\right]\mathrm{Re}s>0\text{\hspace{1em}(12.110)}$$

is independent of $s$ . Show also that $I(s)=\mathrm{ind}D$ .

12.7.2 Spectral $\zeta$ -functions

Let $E$ and $F$ be vector bundles over $M$ . Define a new function

$${\zeta }_E(x,y;s)\equiv \sum ^{\prime }⟨x|n⟩⟨n|y⟩{\lambda }_n^{-s}\mathrm{Re}s>0\text{\hspace{1em}(12.111)}$$

where ${\Delta }_E|n⟩={\lambda }_n|n⟩$ and the prime denotes the omission of the zero modes $({\lambda }_n=0)$ . A function ${\zeta }_F(x,y;s)$ may similarly be defined for ${\Delta }_F$ . The functions $h_E$ and ${\zeta }_E$ are related by the Mellin transformation. To see this, we recall the definition of the $\Gamma$ -function,

$$\Gamma (s)\equiv {\int }_0^{\infty }{t}^{s-1}{\mathrm{e}}^{-t}\mathrm{d}t={\lambda }^s{\int }_0^{\infty }{t}^{s-1}{\mathrm{e}}^{-\lambda t}\mathrm{d}t$$

where $\lambda$ is taken to be strictly positive. From this we find

$$\begin{array}{rl}\Gamma (s)\zeta (x,y;s)& =\sum _n^{\prime }{\int }_0^{\infty }{t}^{s-1}{\mathrm{e}}^{-{\lambda }_{n}t}⟨x|n⟩⟨n|y⟩\mathrm{d}t\\ & ={\int }_0^{\infty }{t}^{s-1}\left[h(x,y;t)-\sum _i⟨x|0,i⟩⟨0,i|y⟩\right]dt.\end{array}\text{\hspace{1em}(12.112)}$$

We also note that

$${\zeta }_{\Delta }(s)\equiv {\int }_M\zeta (x,x;s)dx=\sum _n^{\prime }{\lambda }_n^{-s}\text{\hspace{1em}(12.113)}$$

is the spectral $\zeta$ -function defined in (1.158).

Exercise 12.6

Verify that

$${\Delta }^{-s}f(x)=\int \zeta (x,y;s)f(y)dy\text{\hspace{1em}(12.114)}$$

where the general power of an operator may be defined in the sense of an eigenvalue, namely we put ${\Delta }^{-s}|n⟩={\lambda }_n^{-s}|n⟩$ . Re $s$ is assumed to be sufficiently large so that (12.114) is well defined. [Hint: Use the completeness of the eigenvectors.]

Example 12.7

The following example is taken from Kulkarni (1975). Let $M=S^1=\{{\mathrm{e}}^{\mathrm{i}\theta }\}$ and $E=F=$ a trivial line bundle over $S^1$ (a cylinder). Take an operator $\Delta \equiv -{\partial }^2/\partial {\theta }^2$ . From the eigenvalue equation,

$$-\frac{{\partial }^2{\mathrm{e}}^{\mathrm{i}n\theta }}{\partial {\theta }^2}=n^2{\mathrm{e}}^{\mathrm{i}n\theta }n\in \mathbb{Z}$$

we find that

$${\lambda }_n=n^2⟨\theta |n⟩=(2\pi {)}^{-1/2}{\mathrm{e}}^{\mathrm{i}n\theta }.$$

The heat kernel is

$$\begin{array}{rl}h({\theta }_1,{\theta }_2;t)& =\sum {\mathrm{e}}^{-n^{2}t}⟨{\theta }_1|n⟩⟨n|{\theta }_2⟩\\ & =\frac{1}{2\pi }\left(1+\sum ^{\prime }{e}^{-n^{2}t}{e}^{in\left({\theta }_1-{\theta }_2\right)}\right)\end{array}\text{\hspace{1em}(12.115)}$$

while

$$\begin{array}{rl}\zeta \left({\theta }_1,{\theta }_2;s\right)& =\sum ^{\prime }{n}^{-2s}⟨{\theta }_1|n⟩⟨n|{\theta }_2⟩\\ & =\frac{1}{2\pi }\sum ^{\prime }{n}^{-2s}{\mathrm{e}}^{\mathrm{i}n\left({\theta }_1-{\theta }_2\right)}.\end{array}\text{\hspace{1em}(12.116)}$$

We easily verify that $\tilde{h}(t)=1+{\sum }^{\prime }{\mathrm{e}}^{-n^{2}t}$ satisfies

$$1+2{\int }_1^{\infty }{e}^{-x^{2}t}dx<\tilde{h}(t)<1+2{\int }_0^{\infty }{e}^{-x^{2}t}dx.$$

We then find from these inequalities that

$${\int }_{-\infty }^{+\infty }{\mathrm{e}}^{-x^{2}t}\mathrm{d}x-1<\tilde{h}(t)<{\int }_{-\infty }^{+\infty }{\mathrm{e}}^{-x^{2}t}\mathrm{d}x+1$$

or by putting the value

$$\int {\mathrm{e}}^{-x^{2}t}\mathrm{d}x=\sqrt{\pi }{t}^{-1/2}$$

we find

$$\sqrt{\pi }{t}^{-1/2}-1<\tilde{h}(t)<\sqrt{\pi }{t}^{-1/2}+1.$$

This shows that

$$\underset{t\to 0^{+}}{\lim }\tilde{h}(t)\sim \sqrt{\pi }{t}^{-1/2}.\text{\hspace{1em}(12.117)}$$

In general, the asymptotic series starts with $t^{-\dim M/2}$ .

来源与版本

正文：英文 · OCR 机器稿 · 待校对

核对状态：OCR 机器稿 · 待校对

原书 PDF

来源版本：second-edition-local

来源 PDF SHA-256：fcd24c3371d092a8c699ccf5cf0765462924b6c48412804231076e8084b1940d

OCR 来源 SHA-256：0b6324f01172a3548cd34382e82fd8bfae574602b6b2cfaa357ba96bc2c35478

OCR 产物 SHA-256：1b4342576beb39dc8b319cdd495b09a90627faa802df3588e10229ce34846095