Collapsing sequences of solutions to the Ricci flow on 3-manifolds with almost nonnegative curvature (Q2498485)

Ricci-flows on Riemannian compact \(3\)-dimensional manifolds are focused on by the mathematical community, since they are related to the solution of the \textit{Poincaré conjecture}. This states that a simply-connected closed compact \(3\)-dimensional manifold is topologically equivalent to the three-sphere \(S^3\). In fact, starting with a simply-connected closed compact \(3\)-dimensional manifold \(M\), endowed with an arbitrary Riemannian metric, solutions of the Ricci flow equation \((\partial_tg)=-2\text{\,Ric}(g(t))\), (introduced by Hamilton in 1982), just represent the possible geometric evolutions of such a manifold. In particular, if among such solutions there are metrics with constant positive sectional curvature, one could prove that \(M\) is topologically equivalent to a \(3\)-sphere. In 1982 Hamilton proved that this is the case if the initial metric has positive Ricci curvature. Successively some further improvements have been obtained, starting with a general metric, by adopting some surgery techniques applied first by Hamilton, and more recently by Perelman (2002). The last one, starting with a, non-necessarily simply connected, manifold \(M\), finds solutions topologically equivalent to a connected sum: \((S^3/\Gamma_1)\sharp\cdots\sharp(S^3/\Gamma_k)\sharp(S^1\times S^2)\sharp\cdots\sharp (S^1\times S^2)\), with \(\Gamma_i\) a finite subgroup of \(SO(4)\) acting freely on \(S^3\). By van Kampen's theorem one has that \(\pi_1(M)=\Gamma_1\star\cdots\star\Gamma_k\star{\mathbb Z}\star\cdots\star{\mathbb Z}\), (here \(\star\) denotes the free product). So if \(\pi_1(M)=0\), then \(M=S^3\sharp\cdots\sharp S^3=S^3\). In this way the Poincaré conjecture should end its existence in Mathematics, and should become a theorem finally. The present paper is just placed in this mathematical framework. In fact a way to prove existence to such solutions is to study collapsing sequences of solutions of the Ricci-flow equation. The claimed solution that solves the Poincaré conjecture is a Ricci-flow-with-surgery solution. The surgery is made if one encounters a singularity by curvature blowup. Thus we need to know that singularity is caused by tiny-necks collapsing. Then a way to understand that this is the case is to take a sequence of points and times approaching the singularity. In other words, if the solution (\textit{origin-solution}) exists on the time interval \([0,T)\), but no further, then one chooses a sequence of times \(t_i\) and points \(x_i\) in \(M\) so that: (i) \(\lim_{i\to\infty}t_i=T\); (ii) \(\lim_{i\to\infty}| \text{Riem}(x_i,t_i)| =\infty\). Now one blows-up the scale to obtain a unit scale. More precisely define the \textit{intrinsic scale \(r_i>0\) at the point \((x_i,t_i)\)}, by \(r_i^{-2}=| \text{Riem}(x_i,t_i)| \). Since \(\lim_{i\to\infty}r_i=0\), we spatially expand \(M\) by a factor \(r_i^{-1}\) so that \(| \text{Riem}(x_i,t_i)| \) becomes \(1\). In this way we still get a Ricci flow solution provided that we also expand time by \(r^{-2}\). The rescaled solution lives for a time of order \(T| \text{Riem}(x_i,t_i)| \) going to \(\infty\) as \(i\to\infty\). Therefore, we can take a sequence of Ricci-flow solutions \(g_i(t)\) on \(M\), defined with shifted time parameter in such a way that all ends at time \(0\). The hope is to obtain a convergent subsequence as \(i\to\infty\). Suppose \((M_\infty,g_\infty(t))\) is this limit, and call it \textit{ancient-solution}. It lives in an infinite time interval \((-\infty,0]\). If we can find a cylinder (\textit{near-cylinder}) \(S^2\times[L,-L]\) in \((M_\infty,g_\infty(0))\), then there are tiny necks in the original solution near the space-time point \((x_i,t_i)\) where we can proceed for the surgery. In order to assure existence of a convergent subsequence of the Ricci-flow solutions \((M,g_i(t))\), some technicality must be satisfied, related to results by Cheeger-Gromov and Hamilton. More precisely, we need to know that (i) the curvature is uniformly bounded in neighborhoods of the points \((x_i,0)\); and that (ii) the unit balls around \(x_i\) in the rescaled metrics \((M,g_i(0))\) have uniformly positive volume. Then, the \textit{Hamilton compactness theorem} assures the existence of a blowup limit \((M_\infty,g_\infty(t))\) that is an ancient Ricci-flow solution. A problem is whether the ancient solution is like \({\mathbb R}\times\text{cigar-soliton}\). In fact, in such a case we cannot recognize a near-cylinder. The authors study in this paper collapsing sequences of \(3\)-dimensional solutions to the Ricci-flow equation with almost non-negative sectional curvature and diameters tending to infinity. They classify the possible Gromov-Hausdorff limits and construct \(2\)-dimensional virtual limits. These last are constructed using Fukeaya's theory of limits of local covers. They prove that virtual limits arising from appropriate dilations of some particular type of singularity is always Hamilton's cigar-soliton solution.