Ricci curvature is trace of $T_p M \to T_p M$, $Z \mapsto R(X, Z)Y$ right
The second variation formula says if $\gamma$ is geodesic $dH(\gamma)(X, Y) = \int_\gamma g(X'' + R(X, \gamma')\gamma', Y)$
dimension of kernel of $dH$ measures the multiplicity of conjugacy of the endpoints of $\gamma$
So I wonder if I can swoosh something and get information about the diameter of $M$
Take any vector field $X$ along $\gamma$, write it in orthonormal parallel fields as $X = a_i e^i$ and plug it in $dH(\gamma)(X, X)$?
Oh maybe I should take an orthonormal basis of Jacobi fields
$X = a_i J_i$, plugging and chugging gives $dH(\gamma)(X, X) =$ uhhh
$dH(\gamma)(X, X) = \int_\gamma g(a_i'' J_i + 2 a'_i J_i' + a_i J_i'' + a_i R(J_i, \gamma')\gamma', a_i J_i)$
Garbage in garbage out the last bit goes away by Jacobi equation
$\int_\gamma g(a_i'' J_i + 2 a_i' J_i', a_i J_i)$ now what the hell is this nonsense
I need curvature term to survive
Yeah I don't have a basis of Jacobi fields vanishing at both endpoints; I don't know the multiplicity
I guess I want my basis to be $\sin(\pi t) e^i$. $X = a_i \sin(\pi t) e^i$
$g(-\sin(\pi t) e^i, R(\sin(\pi t) e^i, \gamma') \gamma'), \sin(\pi t) e^i) = \sin^2(\pi t) (-1 + g(R(e^i, \gamma')\gamma', e^i))$
I should sum this crap? $\sum_i g(R(\gamma', e_i)\gamma', e_i)$ is $Ric(\gamma', \gamma')$
That should be $-\pi^2 + g(R(e^i, \gamma')\gamma', e^i)$ by the way
$\sum_{i = 1}^{n-1} dH(\sin(\pi t)e_i, \sin(\pi t)e_i) = 2 \int_0^1 \sin^2(\pi t) ((n-1) \pi^2 - Ricc(\gamma', \gamma'))$
Getting close. If the geodesic was length $c$ it'd be the same formula but $\sin(\pi t/c)$ and $(n-1)\pi^2/c^2$ accordingly. Then having $Ricc(\gamma', \gamma') \leq (n-1)\pi^2/c^2$ would make the integral positive
OK, if $Ricc(\gamma', \gamma') \geq (n-1)\pi^2/c^2$ then the integral is negative so some $dH(\sin(\pi t/c) e_i, \sin(\pi t/c) e_i)$ is negative so $dH$ is semidefinite so $\ker dH$ has stuff in it so $\gamma$ has conjugate points
If $Ricc \geq (n - 1)\pi^2/c^2$ then $diam(M) \leq c$. Bonnet-Myers boom
Cor: If $Ricc$ is bounded below by a positive constant $\pi_1 M$ is finite
Pf: Lift to universal covering where $Ricc$ is also bounded below by the same constant. Universal cover has finite fibers, so rip rip