2DFF070BC43F
(Immortal Lagrangian connected sums) Let $m\geq 3$ and let $X_1,X_2$ be special Lagrangian $m$-tori of the same phase in the flat Calabi–Yau torus $\mathbb C^ m/\Gamma $. Assume they intersect transversely at exactly one point, with tangent planes satisfying the Lawlor angle criterion (a type-one intersection). By [SuTsaiWood2026], the Main Theorem, Theorem 9.1, Proposition 9.5, and Corollary 9.6, a sufficiently small Lawlor-neck desingularization can be chosen to start a smooth embedded immortal Lagrangian flow $N_ t$.
After translating the initial time, take $t\geq 0$. Every finite-time slice is a compact complete proper embedded $m$-manifold of codimension $m$, with topology $\mathbb T^ m\mathbin {\# }\mathbb T^ m$ and no ends. As $t\to \infty $, the flow converges smoothly away from the intersection point to $X_1\cup X_2$. Its neck scale and curvature satisfy, for sufficiently large $t$,
Rescaling about the intersection point by $\varepsilon (t)^{-1}$ gives the Lawlor neck locally smoothly, and the curvature becomes unbounded. These are asymptotic estimates from a parabolic gluing construction, not an exact formula for $N_ t$. The singular limiting union does not make the regular slices nonembedded. No symmetry, uniqueness, or uniform mean-curvature decay is asserted. Euclidean Gaussian quantities are not assigned in this torus ambient space. For the illustration we choose equal characteristic angles $\pi /m$. Up to a common unitary rotation, the local Lawlor model from [SuTsaiWood2026], Section 2.3, has the exact profile
using the principal root. Its waist radius is $\varepsilon $. Opposite points in the signed profile $\pm \gamma _\varepsilon $ represent the same spherical fiber; the full neck is connected. This local symmetry is a choice of model, not an assertion about the entire compact flow.
Equation (1.1) of the source gives $d(\varepsilon ^2)/dt=-c\varepsilon ^ m+o(\varepsilon ^ m)$, with $c>0$. Dropping the remainder yields the dimensionless illustration clock
The time plate uses $m=3$ and one spatial scale; the separate magnified view divides lengths by the positive neck scale. At infinite time it displays the rescaled limit from Proposition 9.5. Dilating a minimal Lawlor neck alone does not solve MCF: the compact gluing and parabolic correction are essential and are not computed here.
Comments
Comments are not enabled yet.