On a complete Riemann manifold $(M,g)$, we know if $v \in T_pM$ is a cut locus of $p$, i.e. $\exp_p(sv)$ is length minimizing for $s \in [0,1]$ and is not length minimizing for $s \in [0,1+\epsilon]$, then either there are two distinct length minimizing geodesics connecting $p$ to $q=\exp_p(v)$, or $q$ is the first conjugate point of $p$ along the geodesic $(s \mapsto \exp_p(sv))$.

Also if we have two disctinct length minimizing geodesics connecting $p$ to $q$, then $v$ is a cut locus of $p$.

My question is, if $q$ is the first conjugate point of $p$ along $(s \mapsto \exp_p(sv))$, is $v$ necessarily a cut locus of $p$?


No, this is false. Take the quotient of the round $S^3$ by a finite cyclic group of isometries acting freely and having order $n$ sufficiently high. Then for any pair of conjugate points their distance along the geodesic is $\pi$, which is greater than the injectivity radius (if $n$ is high enough, but maybe even $n=3$ will suffice).

Edit: Indeed, $n=2$ actually suffices. The easiest example is the projective plane $RP^2$ with the constant curvature ($+1$) metric. Then for each unit speed geodesic $\gamma(t)$ on $RP^2$, the distance along $\gamma$ between consecutive conjugate points is $\pi$, while the distance between points connected by a nonunique geodesic is $\pi/2$. Hence, for the constant curvature projective plane, the first conjugate point always occurs after the cut locus point.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.