诱导联络

2020-07-25  本文已影响0人  xhje

这篇笔记里假设(M^m,g)是一个Riemann流形, \nabla是对应的Levi-Civita联络.

定义1(沿着光滑映射的向量场).N^n是一个光滑流形, \varphi:N\rightarrow M是光滑映射. 如果光滑映射X:N\rightarrow TM满足\forall p\in N, X(p)\in T_{\varphi(p)}M, 那么我们称X\varphi上的向量场(或者说沿着\varphi的向量场). 有时我们也记X(p)X_p.

\varphi上向量场的例子有很多, 最重要的情形是N=[a,b]\subset\mathbb{R}, \gamma:[a,b]\rightarrow M是光滑映射, 即M上的一条光滑道路. 那么此时沿着\gamma的向量场就是对每个t\in[a,b], (光滑地)赋予一个向量X(t)\in T_{\gamma(t)}M.

我们(跟随伍鸿熙)引入诱导联络如下:

定义2(诱导联络).X是沿着\varphi:N\rightarrow M的向量场, p\in N, v\in T_pN. 我们定义一个\widetilde\nabla_vX\in T_{\varphi(p)}M如下: 设U\ni \varphi(p)上有局部标架场e_1,\ldots,e_m, 那么存在\varphi^{-1}(U)上的光滑函数X^i使得\forall q\in\varphi^{-1}(U), 有X(q)=X^i(q)e_i|_{\varphi(q)}. 此时我们定义
\widetilde\nabla_vX\overset{def}{=}vX^ie_i|_{\varphi(p)}+X^i(p)\nabla_{d\varphi_p(v)}e_i

我们首先要做的就是说明这个定义和Ue_i的选取无关. 如果还有一个\varphi(p)附近的开集\widetilde U和局部标架\widetilde e_i, 那么存在\widetilde U\cap U上的m^2个光滑函数a_i^j使得e_i=a_i^j\widetilde e_j. 那么X(q)=\widetilde X^i(q)\widetilde e_i, 其中\widetilde X^i(q)=a_j^i(\varphi(q))X^j(q), 并且
vX^ie_i|_{\varphi(p)}+X^i(p)\nabla_{d\varphi_p(v)}e_i
=vX^i\cdot a_i^j(\varphi(p))\widetilde e_j|_{\varphi(p)}+X^i(p)\nabla_{d\varphi_p(v)}(a_i^j\widetilde e_j)
=vX^i\cdot a_i^j(\varphi(p))\widetilde e_j|_{\varphi(p)}+X^i(p)v(a_i^j\circ\varphi)\widetilde e_j+X^i(p)a_i^j(\varphi(p))\nabla_{d\varphi_p(v)}\widetilde e_j
=v(X^i\cdot(a_i^j\circ\varphi))\widetilde e_j|_{\varphi(p)}+X^i(p)a_i^j(\varphi(p))\nabla_{d\varphi_p(v)}\widetilde e_j
=v\widetilde X^j\widetilde e_j|_{\varphi(p)}+\widetilde X^j(p)\nabla_{d\varphi_p(v)}\widetilde e_j
这就说明了定义是良好的, 不依赖于Ue_i的选择.

如果VN上的向量场, X\varphi: N\rightarrow M上的向量场, 那么我们可以定义一个新的\varphi上向量场(\widetilde\nabla_VX)(p)=\widetilde\nabla_{V(p)}X.

接下来我们要说明诱导联络有的时候确实可以看作联络, 这样才和我们的几何直观相符.

命题3(诱导联络有时可以看作联络). 如果v\in T_pN, X\in\Gamma(TN)(此时d\varphi(X)\varphi上向量场), Y\in\Gamma(TM), 并且X,Y\varphi-相关的(即\forall q\in N, d\varphi_q(X_q)=Y_{\varphi(q)}), 那么\widetilde\nabla_vd\varphi(X)=\nabla_{d\varphi_p(v)}Y.
证明.U\ni\varphi(p), e_1,\ldots,e_mU上的局部标架场. 设d\varphi(X)可以展开为d\varphi_q(X_q)=X^i(q)e_i|_{\varphi(q)}, Y可以展开为Y_x=Y^i(x)e_i|_x.
由向量场\varphi-相关的定义立即可以得到X^i=Y^i\circ\varphi.
我们现在可以来计算诱导联络:
\widetilde\nabla_v(d\varphi(X))=v(X^i)e_i|_{\varphi(p)}+X^i(p)\nabla_{d\varphi_p(v)}e_i
=v(Y^i\circ\varphi)e_i|_{\varphi(p)}+Y^i(\varphi(p))\nabla_{d\varphi_p(v)}e_i
=d\varphi_p(v)(Y^i)e_i|_{\varphi(p)}+Y^i(\varphi(p))\nabla_{d\varphi_p(v)}e_i=\nabla_{d\varphi_p(v)}Y
\blacksquare

诱导联络有和联络类似的性质.

命题4. (1)(线性性)设u,v\in T_pM, \lambda\in\mathbb{R}, X\varphi上向量场, 则\widetilde\nabla_{u+v}X=\widetilde\nabla_uX+\widetilde\nabla_vX, \widetilde\nabla_{\lambda v}X=\lambda\widetilde\nabla_{v}X.
(2)(Leibniz法则)如果f\in C^\infty(N), X\varphi上向量场, 则\widetilde\nabla_v(fX)=v(f)X(p)+f(p)\widetilde\nabla_vX.
(3)(与度量相容)对任何\varphi上向量场X,Y, v(\langle X,Y\rangle_g)=\langle \widetilde\nabla_vX,Y(p)\rangle_g+\langle X(p),\widetilde\nabla_v Y\rangle_g.
(4)(无挠性)对任何N上向量场X,Y, 有\widetilde\nabla_Xd\varphi(Y)-\widetilde\nabla_Yd\varphi(X)=d\varphi([X,Y]).
证明. (1)这根据定义是真的很容易验证的, 所以这里略过.
(2)取标架直接计算:
\widetilde\nabla_v(fX)=v(fX^i)e_i|_{\varphi(p)}+f(p)X^i(p)\nabla_{d\varphi_p(v)}e_i
=v(f)X^i(p)e_i|_{\varphi(p)}+f(p)v(X^i)e_i|_{\varphi(p)}+f(p)X^i(p)\nabla_{d\varphi_p(v)}e_i
=v(f)X(p)+f(p)\widetilde\nabla_vX
(3)直接计算:
v(\langle X,Y\rangle_g)=v(\langle X^ie_i,Y^je_j\rangle_g)=v(X^iY^j\cdot\langle e_i,e_j\rangle_g\circ\varphi)
=v(X^i)Y^j(p)\langle e_i|_{\varphi(p)},e_j|_{\varphi(p)}\rangle_g+X^i(p)v(Y^j)\langle e_i|_{\varphi(p)},e_j|_{\varphi(p)}\rangle_g
+X^i(p)Y^j(p)d\varphi_p(v)(\langle e_i,e_j\rangle_g)
=v(X^i)Y^j(p)\langle e_i|_{\varphi(p)},e_j|_{\varphi(p)}\rangle_g+X^i(p)v(Y^j)\langle e_i|_{\varphi(p)},e_j|_{\varphi(p)}\rangle_g
+X^i(p)Y^j(p)\langle\nabla_{d\varphi_p(v)}e_i,e_j|_{\varphi(p)}\rangle_g+X^i(p)Y^j(p)\langle e_i|_{\varphi(p)},\nabla_{d\varphi_p(v)}e_j\rangle_g
=\langle v(X^i)e_i|_{\varphi(p)}+X^i(p)\nabla_{d\varphi_p(v)}e_i,Y^j(p)e_j|_{\varphi(p)}\rangle_g
+\langle X^i(p)e_i|_{\varphi(p)},v(Y^j)e_j|_{\varphi(p)}+Y^j(p)\nabla_{d\varphi_p(v)}e_j\rangle_g
=\langle \widetilde\nabla_vX,Y(p)\rangle_g+\langle X(p),\widetilde\nabla_v Y\rangle_g
(4)这个我没想出不用坐标系的方法, 这里的方法来自Induced connection is torsion free. 设x^1,\ldots,x^np附近的局部坐标, u^1,\ldots,u^m\varphi(p)附近的局部坐标, 并且X=X^i\frac{\partial}{\partial x^i}, Y=Y^j\frac{\partial}{\partial x^j}, 那么
\widetilde\nabla_Xd\varphi(Y)-\widetilde\nabla_Yd\varphi(X)=\widetilde\nabla_X\left(Y^j\frac{\partial u^k}{\partial x^j}\frac{\partial}{\partial u^k}\right)-\widetilde\nabla_Y\left(X^i\frac{\partial u^k}{\partial x^i}\frac{\partial}{\partial u^k}\right)
=X^i\frac{\partial}{\partial x^i}\left(Y^j\frac{\partial u^k}{\partial x^j}\right)\frac{\partial}{\partial u^k}+Y^j\frac{\partial u^k}{\partial x^j}\widetilde\nabla_X\frac{\partial}{\partial u^k}
-Y^j\frac{\partial}{\partial x^j}\left(X^i\frac{\partial u^k}{\partial x^i}\right)\frac{\partial}{\partial u^k}-X^i\frac{\partial u^k}{\partial x^i}\widetilde\nabla_Y\frac{\partial}{\partial u^k}
=X^i\frac{\partial Y^j}{\partial x^i}\frac{\partial u^k}{\partial x^j}\frac{\partial}{\partial u^k}+X^iY^j\left(\frac{\partial u^k}{\partial x^i\partial x^j}\right)\frac{\partial}{\partial u^k}+X^iY^j\frac{\partial u^k}{\partial x^j}\frac{\partial u^l}{\partial x^i}\nabla_{\frac{\partial}{\partial u^l}}\frac{\partial}{\partial u^k}
-Y^j\frac{\partial X^i}{\partial x^j}\frac{\partial u^k}{\partial x^i}\frac{\partial}{\partial u^k}-X^iY^j\left(\frac{\partial u^k}{\partial x^i\partial x^j}\right)\frac{\partial}{\partial u^k}-X^iY^j\frac{\partial u^k}{\partial x^i}\frac{\partial u^l}{\partial x^j}\nabla_{\frac{\partial}{\partial u^l}}\frac{\partial}{\partial u^k}
=\left(X^i\frac{\partial Y^j}{\partial x^i}-Y^i\frac{\partial X^j}{\partial x^i}\right)\frac{\partial u^k}{\partial x^j}\frac{\partial}{\partial u^k}+X^iY^j\frac{\partial u^k}{\partial x^j}\frac{\partial u^l}{\partial x^i}\left(\nabla_{\frac{\partial}{\partial u^l}}\frac{\partial}{\partial u^k}-\nabla_{\frac{\partial}{\partial u^k}}\frac{\partial}{\partial u^l}\right)
=d\varphi([X,Y])+0=d\varphi([X,Y])
\blacksquare

上一篇 下一篇

猜你喜欢

热点阅读