|
Author |
hbghlyj
Posted 2025-5-21 05:28
The Ricci curvature tensor $\mathrm{Ric}$ of a Riemannian manifold $(M, g)$ is defined as the trace of the Riemann curvature tensor $R$ on its first and third indices:$$\mathrm{Ric}(X, Y) = \mathrm{Tr}(Z \mapsto R(Z, X)Y)$$In local coordinates:$$\mathrm{Ric}_{ij} = R^k_{ikj}$$The scalar curvature $S$ is the trace of the Ricci tensor with respect to the metric:$$S = g^{ij} \mathrm{Ric}_{ij}$$
- Completeness of $g$:
The metric $g = g_1 + g_2$ on $M_1 \times M_2$ is Riemannian (symmetric, positive definite, smooth).
A geodesic in $M_1 \times M_2$ is a pair of geodesics in $M_1$ and $M_2$. Since both $(M_1, g_1)$ and $(M_2, g_2)$ are complete, their geodesics exist for all time, by Hopf–Rinow theorem $(M, g)$ is complete. - Ricci and Scalar Curvatures:
The Ricci tensor of $(M, g)$ splits as:$$\mathrm{Ric}_g((X_1, X_2), (Y_1, Y_2)) = \mathrm{Ric}_{g_1}(X_1, Y_1) + \mathrm{Ric}_{g_2}(X_2, Y_2)$$The scalar curvature is additive:$$S_g = S_{g_1} + S_{g_2}$$
- Now suppose that $(M_j,g_j)$ has constant sectional curvature $K_j$ for $j=1,2$. Recall that a manifold of dimension $n$ and sectional curvature $K$ has Ricci tensor
$$
\operatorname{Ric}_{g_j}=(n_j-1)K_jg_j,
$$
and its scalar curvature is
$$
\operatorname{scal}_{g_j}= n_j(n_j-1)K_j.
$$
- Einstein Condition When $n_1=n_2$
A Riemannian metric $g$ is said to be Einstein if there is a constant $\lambda$ such that
$$
\operatorname{Ric}_g = \lambda g.
$$
In our product situation, if we evaluate on vectors tangent to $M_1$ (and similarly for $M_2$) we have:
$$
\operatorname{Ric}_g((X_1,0),(Y_1,0)) = \operatorname{Ric}_{g_1}(X_1,Y_1) = (n_1-1)K_1 g_1(X_1,Y_1).
$$
On the other hand, Einstein's condition requires that
$$
\operatorname{Ric}_g((X_1,0),(Y_1,0)) = \lambda\, g\bigl((X_1,0),(Y_1,0)\bigr) = \lambda g_1(X_1,Y_1).
$$
Thus, for all $X_1,Y_1$ we must have
$$
(n_1-1)K_1 = \lambda.
$$
A similar computation using vectors tangent to $M_2$ yields
$$
(n_2-1)K_2 = \lambda.
$$
If we assume $n_1=n_2$ (say, equal to $n$), the two conditions force
$$
(n-1)K_1 = (n-1)K_2 \quad \Longrightarrow \quad K_1 = K_2.
$$Thus, when $n_1=n_2$ the product $(M,g)$ is Einstein if and only if $K_1=K_2$. - Scalar–Flat and Einstein Metrics
A metric is scalar-flat if $\operatorname{scal}_g = 0$. In our situation, we have:
$$
\operatorname{scal}_g = n_1(n_1-1)K_1 + n_2(n_2-1)K_2.
$$
If in addition $(M,g)$ is Einstein (so that as above $\lambda = (n_1-1)K_1 = (n_2-1)K_2$), then its Einstein constant is given by
$$
\lambda=\frac{\operatorname{scal}_g}{n_1+n_2}.
$$
Thus, if $\operatorname{scal}_g=0$ then $\lambda=0$ and consequently
$$
(n_1-1)K_1 =0\quad \text{and}\quad (n_2-1)K_2=0.
$$
Since each $n_j\ge2$ (hence $n_j-1 > 0$) it follows that $K_1=K_2=0$. Conversely if $K_1=K_2=0$ then clearly $\operatorname{Ric}_{g_j}=0$ and the product metric is Ricci-flat (thus Einstein with $\lambda=0$) and scalar flat. Hence, $(M,g)$ is scalar–flat and Einstein if and only if $K_1=K_2=0$. - Noncompactness When Scalar–Flat and Factors are Simply Connected
Under the assumption that $(M,g)$ is scalar–flat and that each $M_j$ is simply connected, we already saw in part (ii) that scalar-flat together with the Einstein condition forces $K_1=K_2=0$. But a complete, simply connected Riemannian manifold with constant sectional curvature $0$ is isometric to Euclidean space $\mathbb{R}^{n_j}$ (by the standard classification of space forms). Therefore, each $M_j$ is isometric to $\mathbb{R}^{n_j}$ and in particular is non–compact. Finally, the product of non–compact manifolds is non–compact. (In fact, one may observe that the metric $g$ is isometric to the Euclidean metric on $\mathbb{R}^{n_1+n_2}$.) Thus, $M=M_1\times M_2$ is non–compact.
|
|