OctoInformBeta

OpenAI preprint, September 2026 · §4. Geodesic passages and the index form

Lemma 4.4 [Elementary passages]: Endpoint conditions and height restrictions for intrinsic and normal index forms

Index estimates on finite neck passages.

Hover over a blue or green term in the text, or a node in the graph, to find it in the other. Select it to read its declaration both in Lean and in words.

The Setting

Elementary passage estimates concern the index of a vector field during a finite interval in a neck. The variable tt denotes neck height. Fix a neck parameter 0<δ≤120<\delta\le\tfrac12 and an endpoint ε∈{L,R}\varepsilon\in\{\mathrm L,\mathrm R\}. In endpoint coordinates a curve has the form γ(s)=(τ(s),z(s))∈ℝ×ℝ2\gamma(s)=(\tau(s),z(s))\in\mathbb R\times\mathbb R^2, where ss is geodesic time, τ\tau is inward endpoint height, and zz is transverse position. Write u=(uτ,u⟂)u=(u_\tau,u_\perp) for the axial and transverse splitting of a vector in ℝ3\mathbb R^3.

The one-sided cutoff α:ℝ→ℝ\alpha:\mathbb R\to\mathbb R and its integrated cutoff A:ℝ→ℝA:\mathbb R\to\mathbb R are the functions in (1). Here θ\theta denotes the standard smooth transition from zero to one; χ\chi is the shear transition and λ\lambda the collar transition. α(t)={exp⁡(−1/(2/3−t)),t<2/3,0,t≥2/3,A(t)=∫0tα(u)du,χ(t)=θ(6(t−5/12)),λ(t)=θ(2t+3/2).\begin{gathered} \alpha(t)= \begin{cases} \exp\!\bigl(-1/(2/3-t)\bigr),&t<2/3,\\ 0,&t\ge2/3, \end{cases} \qquad A(t)=\int_0^t\alpha(u)\,du,\\ \chi(t)=\theta\!\left(6(t-5/12)\right), \qquad \lambda(t)=\theta(2t+3/2). \end{gathered}(1)

The collar denominator dδd_\delta is also the first diagonal coefficient d1d_1; d2d_2 is the second diagonal coefficient. The collar terminal decay rate ηδ\eta_\delta, logarithmic decay kδk_\delta, and warping function ℓδ\ell_\delta are specified by (2), using the cutoffs in (1). dδ(t)=d1(δ,t)=1+δA(t),d2(δ,t)=1+δA(1−t),ηδ(t)=δα(t)2dδ(t),kδ(t)=1−λ(t)+λ(t)ηδ(t),ℓδ(t)=exp⁡(−∫0tkδ(u)du).\begin{gathered} d_\delta(t)=d_1(\delta,t)=1+\delta A(t), \qquad d_2(\delta,t)=1+\delta A(1-t),\\ \eta_\delta(t)=\frac{\delta\alpha(t)}{2d_\delta(t)},\qquad k_\delta(t)=1-\lambda(t)+\lambda(t)\eta_\delta(t),\\ \ell_\delta(t)=\exp\!\left(-\int_0^t k_\delta(u)\,du\right). \end{gathered}(2)

The cometric shear σδ\sigma_\delta is the scalar function defined by σδ(t)=δ2χ(t)\sigma_\delta(t)=\delta^2\chi(t).✦ The real symmetric 2×22\times2 cometric matrix Pδ(t)P_\delta(t) is defined by (3), with the diagonal coefficients of (2) evaluated at (δ,t)(\delta,t). Pδ(t)=(d1+d2σδ(t)2d2σδ(t)d2σδ(t)d2).P_\delta(t)= \begin{pmatrix} d_1+d_2\sigma_\delta(t)^2&d_2\sigma_\delta(t)\\ d_2\sigma_\delta(t)&d_2 \end{pmatrix}.(3)✦

The endpoint frame matrix C2(δ)C_2(\delta) determines the terminal cometric from (3). The endpoint coordinate cometrics Pε,δP_{\varepsilon,\delta} and their transverse metric matrices Qε,δQ_{\varepsilon,\delta} are the matrix-valued functions in (4). C2(δ)=(011−δ2),PL,δ(τ)=Pδ(τ),PR,δ(τ)=C2(δ)Pδ(1−τ)C2(δ)𝖳,Qε,δ(τ)=Pε,δ(τ)−1.\begin{gathered} C_2(\delta)=\begin{pmatrix}0&1\\1&-\delta^2\end{pmatrix},\qquad P_{\mathrm L,\delta}(\tau)=P_\delta(\tau),\\ P_{\mathrm R,\delta}(\tau) =C_2(\delta)P_\delta(1-\tau)C_2(\delta)^{\mathsf T}, \qquad Q_{\varepsilon,\delta}(\tau) =P_{\varepsilon,\delta}(\tau)^{-1}. \end{gathered}(4)

A smooth Riemannian metric gg on a smooth manifold is a smooth field of positive definite symmetric bilinear forms on its tangent spaces.✦ On a vector space, a smooth field qq of positive definite symmetric bilinear forms with bounded unit balls defines the metric whose tangent inner product at xx is q(x)q(x).✦ The pullback A*hA^*h of a bilinear form hh by a linear map AA is the bilinear form (u,w)↦h(Au,Aw)(u,w)\mapsto h(Au,Aw).✦ The Cartesian bilinear form hh on a finite coordinate space is the bilinear form given by the Euclidean dot product.✦ A moving frame metric gg on a vector space is the metric gx(u,w)=K(x)u⋅K(x)wg_x(u,w)=K(x)u\cdot K(x)w defined by a smooth family of linear automorphisms K(x)K(x).✦ For nonzero radii r,ρr,\rho and shear cc, the triangular frame transformation and its inverse are given in (5). Kr,ρ,cu=(u0,r(u1−cu2),ρu2),Kr,ρ,c−1w=(w0,w1/r+cw2/ρ,w2/ρ).\begin{aligned} K_{r,\rho,c}u&=(u_0,r(u_1-cu_2),\rho u_2),\\ K_{r,\rho,c}^{-1}w&=(w_0,w_1/r+cw_2/\rho,w_2/\rho). \end{aligned}(5)✦

The triangular frame metric gg determined by smooth radius functions r,ρr,\rho and a smooth shear function cc on ℝ3\mathbb R^3, with radii nowhere zero, is the moving frame metric from (5).✦ In coordinates (t,x,y)(t,x,y) on ℝ3\mathbb R^3, the cylinder metric gg determined by radius functions r,ρr,\rho and a shear function cc depending only on height is the triangular frame metric with line element dt2+r(t)2(dx−c(t)dy)2+ρ(t)2dy2dt^2+r(t)^2(dx-c(t)\,dy)^2+\rho(t)^2dy^2.✦ The normalized neck metric gL,δg_{\mathrm L,\delta} and the right framed neck metric gR,δg_{\mathrm R,\delta} are defined by (6), with the warping function of (2). Here LδL_\delta is the right collar linear map and FδF_\delta the right collar coordinate map. The pullback metric f*gf^*g of a smooth metric gg by a smooth map ff with invertible differential is the metric with tangent inner product (f*g)x(u,w)=gf(x)(dfxu,dfxw)(f^*g)_x(u,w)=g_{f(x)}(df_xu,df_xw).✦ gL,δ=dt2+ℓδ(t)2(dx−δ2χ(t)dy)2+ℓδ(1−t)2dy2,Lδ(y)=(−y0,δ2y1+y2,y1),Lδ−1(x)=(−x0,x2,x1−δ2x2),Fδ(y)=Lδ(y)+(1,0,0),gR,δ=Fδ*gL,δ.\begin{gathered} g_{\mathrm L,\delta} =dt^2+\ell_\delta(t)^2(dx-\delta^2\chi(t)\,dy)^2 +\ell_\delta(1-t)^2dy^2,\\ L_\delta(y)=(-y_0,\delta^2y_1+y_2,y_1),\qquad L_\delta^{-1}(x)=(-x_0,x_2,x_1-\delta^2x_2),\\ F_\delta(y)=L_\delta(y)+(1,0,0),\qquad g_{\mathrm R,\delta}=F_\delta^*g_{\mathrm L,\delta}. \end{gathered}(6) The endpoint framed neck metric gε,δg_{\varepsilon,\delta} on ℝ3\mathbb R^3 is gL,δg_{\mathrm L,\delta} or gR,δg_{\mathrm R,\delta} according to the chosen endpoint ε\varepsilon.✦

Fields and Index Forms

For a smooth metric gg, a field VV along a curve γ\gamma has coordinates w(s)=dcγ(s)V(s)w(s)=dc_{\gamma(s)}V(s) in a chart cc.✦ A field VV along γ\gamma is smooth on a time set UU when its coordinates are smooth on U∩γ−1(dom⁡c)U\cap\gamma^{-1}(\operatorname{dom}c) in every atlas chart.✦ The metric matrix gijg_{ij} is the matrix of inner products of the coordinate basis vectors, obtained from the standard basis eie_i by the inverse chart differential; gijg^{ij} denotes the inverse matrix entries.✦ For a smooth scalar function ff in coordinates, the coordinate derivative ∂if\partial_i f is df(ei)df(e_i).✦ The Christoffel symbols and curvature components are defined by (7), with indices ranging from 00 to 22. Γjki=12∑lgil(∂jglk+∂kglj−∂lgjk),Rlkij=∂iΓjkl−∂jΓikl+∑m(ΓimlΓjkm−ΓjmlΓikm).\begin{aligned} \Gamma^i_{jk} &=\frac12\sum_l g^{il} (\partial_jg_{lk}+\partial_kg_{lj}-\partial_lg_{jk}),\\ R^l{}_{kij} &=\partial_i\Gamma^l_{jk}-\partial_j\Gamma^l_{ik} +\sum_m(\Gamma^l_{im}\Gamma^m_{jk} -\Gamma^l_{jm}\Gamma^m_{ik}). \end{aligned}(7)

For coordinate tangent vectors u,v,w∈ℝ3u,v,w\in\mathbb R^3, the connection term has components Γ(u,w)i=∑j,kΓjkiujwk\Gamma(u,w)^i=\sum_{j,k}\Gamma^i_{jk}u^jw^k and the curvature operator has components [R(u,v)w]l=∑k,i,jRlkijwkuivj[R(u,v)w]^l=\sum_{k,i,j}R^l{}_{kij}w^ku^iv^j, using (7). The covariant derivative DsVD_sV is the inverse chart differential applied to w′+Γ((c∘γ)′,w)w'+\Gamma((c\circ\gamma)',w).✦ The velocity γ̇\dot\gamma of a curve γ\gamma is the vector γ̇=dγs(1)\dot\gamma=d\gamma_s(1), and a geodesic γ\gamma on a time set UU is a smooth curve with vanishing covariant acceleration there. The curvature term along γ\gamma is R(V,γ̇)γ̇R(V,\dot\gamma)\dot\gamma.✦ For fields V,HV,H along γ\gamma and a finite interval [a,b][a,b], the intrinsic index density ig(V,H)i_g(V,H) and its integral I[a,b]g(V)I^g_{[a,b]}(V) are defined by (8). ig(V,H)(s)=g(DsV,DsH)−g(R(V,γ̇)γ̇,H),I[a,b]g(V)=∫abig(V,V)(s)ds.\begin{aligned} i_g(V,H)(s) &=g(D_sV,D_sH)-g(R(V,\dot\gamma)\dot\gamma,H),\\ I^g_{[a,b]}(V)&=\int_a^b i_g(V,V)(s)\,ds. \end{aligned}(8)✦ For a coordinate curve x(s)x(s) and a coordinate vector field w(s)w(s), the coordinate covariant derivative is Dsw=w′+Γ(x′,w)D_sw=w'+\Gamma(x',w), with the connection evaluated at x(s)x(s).✦

For a transverse matrix HH and transverse vectors u,w∈ℝ2u,w\in\mathbb R^2, the matrix bilinear pairing is ℬH(u,w)=u𝖳Hw\mathcal B_H(u,w)=u^{\mathsf T}Hw, and the quadratic form is qH(u)=ℬH(u,u)q_H(u)=\mathcal B_H(u,u). The split neck metric gˆQ\widehat g_Q associated with a transverse metric matrix QQ is given, for split vectors U,V∈ℝ3U,V\in\mathbb R^3, by gˆQ(U,V)=UτVτ+ℬQ(U⟂,V⟂)\widehat g_Q(U,V)=U_\tau V_\tau+\mathcal B_Q(U_\perp,V_\perp).✦ For transverse matrices P,H,KP,H,K, the neck Christoffel operator Γ=ΓP,H\Gamma=\Gamma_{P,H}, its axial derivative Γτ=Γτ;P,H,K\Gamma_\tau=\Gamma_{\tau;P,H,K}, and the coordinate curvature tensor ℛ=ℛP,H,K\mathcal R=\mathcal R_{P,H,K} are defined in (9). Γ(U,V)τ=−12ℬH(U⟂,V⟂),Γ(U,V)⟂=12(UτPHV⟂+VτPHU⟂),Γτ(U,V)τ=−12ℬK(U⟂,V⟂),Γτ(U,V)⟂=12(Uτ(PK−PHPH)V⟂+Vτ(PK−PHPH)U⟂),ℛ(U,V)W=UτΓτ(V,W)−VτΓτ(U,W)+Γ(U,Γ(V,W))−Γ(V,Γ(U,W)).\begin{aligned} \Gamma(U,V)_\tau&=-\tfrac12\mathcal B_H(U_\perp,V_\perp),\\ \Gamma(U,V)_\perp&=\tfrac12(U_\tau PHV_\perp+V_\tau PHU_\perp),\\ \Gamma_\tau(U,V)_\tau&=-\tfrac12\mathcal B_K(U_\perp,V_\perp),\\ \Gamma_\tau(U,V)_\perp &=\tfrac12\bigl(U_\tau(PK-PHPH)V_\perp +V_\tau(PK-PHPH)U_\perp\bigr),\\ \mathcal R(U,V)W &=U_\tau\Gamma_\tau(V,W)-V_\tau\Gamma_\tau(U,W)\\ &\quad+\Gamma(U,\Gamma(V,W))-\Gamma(V,\Gamma(U,W)). \end{aligned}(9)

For transverse matrices P,Q,H,KP,Q,H,K and split vectors T,V,D∈ℝ3T,V,D\in\mathbb R^3, the normal index density ν\nu is the scalar function in (10), where Z=D+ΓP,H(T,V)Z=D+\Gamma_{P,H}(T,V) and the operators are those of (9). ν(P,Q,H,K;T,V,D)=gˆQ(Z,Z)−gˆQ(V,ℛP,H,K(V,T)T)−gˆQ(Z,T)2.\nu(P,Q,H,K;T,V,D) =\widehat g_Q(Z,Z)-\widehat g_Q(V,\mathcal R_{P,H,K}(V,T)T) -\widehat g_Q(Z,T)^2.(10)✦ For a coordinate vector field VV along γ\gamma on [a,b][a,b], the coordinate normal index I[a,b]⟂(V)I^\perp_{[a,b]}(V) for the endpoint matrices of (4) is the integral in (11), with the axial and transverse splitting understood for each vector. Primes on matrices denote derivatives in τ\tau; primes on curves and fields denote derivatives in ss. I[a,b]⟂(V)=∫abν(Pε,δ(τ(s)),Qε,δ(τ(s)),Qε,δ′(τ(s)),Qε,δ″(τ(s));γ′(s),V(s),V′(s))ds.\begin{aligned} I^\perp_{[a,b]}(V) &=\int_a^b\nu\!\Bigl( P_{\varepsilon,\delta}(\tau(s)),Q_{\varepsilon,\delta}(\tau(s)),\\ &\qquad Q'_{\varepsilon,\delta}(\tau(s)), Q''_{\varepsilon,\delta}(\tau(s));\gamma'(s),V(s),V'(s)\Bigr)\,ds. \end{aligned}(11) The boundary correction BB is the scalar function of a transverse matrix HH and split tangent vectors T,VT,V defined by (12). B(H;T,V)=−12TτℬH(V⟂,V⟂)+VτℬH(T⟂,V⟂).B(H;T,V)=-\tfrac12T_\tau\mathcal B_H(V_\perp,V_\perp) +V_\tau\mathcal B_H(T_\perp,V_\perp).(12)✦

The Claim

Theorem 1 (Elementary passage estimates). Fix 0<δ≤120<\delta\le\tfrac12 and an endpoint ε\varepsilon, and put g=gε,δg=g_{\varepsilon,\delta}.

For the fixed endpoint estimate, suppose p𝖳[−Pε,δ″(τ)]p≥0p^{\mathsf T}[-P''_{\varepsilon,\delta}(\tau)]p\ge0 for every τ∈[0,1]\tau\in[0,1] and every p∈ℝ2p\in\mathbb R^2. Let UU be open, let a≤ba\le b with [a,b]⊂U[a,b]\subset U, and let γ\gamma be a geodesic on UU whose inward height stays in [0,1][0,1] on [a,b][a,b]. If V:ℝ→ℝ3V:\mathbb R\to\mathbb R^3 is smooth on UU and V(a)=V(b)=0V(a)=V(b)=0, then I[a,b]g(V)≥0.I^g_{[a,b]}(V)\ge0.(13)✦

For the monotone estimate, let UU be open and a<ba<b with [a,b]⊂U[a,b]\subset U. Let γ=(τ,z)\gamma=(\tau,z) be a geodesic on UU, unit-speed on [a,b][a,b], and let VV be smooth along γ\gamma on UU. Suppose there is a conserved transverse momentum p∈ℝ2p\in\mathbb R^2 with Qε,δ(τ(s))z′(s)=pQ_{\varepsilon,\delta}(\tau(s))z'(s)=p on [a,b][a,b]. Assume τ∈[0,1]\tau\in[0,1] on [a,b][a,b], while 0<τ<10<\tau<1 and the height velocity v=τ′v=\tau' never vanishes on (a,b)(a,b). At each of a,ba,b, require either that the height is 00 or 11, or that VV vanishes on a neighborhood of that time. Then I[a,b]g(V)≥0.I^g_{[a,b]}(V)\ge0.(14)✦

In the transverse normal estimate, take an open set UU and a≤ba\le b with [a,b]⊂U[a,b]\subset U. Let γ=(τ,z)\gamma=(\tau,z) be a geodesic on UU, unit-speed on [a,b][a,b], with τ∈[0,1]\tau\in[0,1] there. Let W:ℝ→ℝ3W:\mathbb R\to\mathbb R^3 be smooth on UU with Wτ(s)=0W_\tau(s)=0 for every s∈Us\in U. Suppose the boundary correction (12) satisfies B(Qε,δ′(τ(b));γ′(b),W(b))≤B(Qε,δ′(τ(a));γ′(a),W(a))B(Q'_{\varepsilon,\delta}(\tau(b));\gamma'(b),W(b)) \le B(Q'_{\varepsilon,\delta}(\tau(a));\gamma'(a),W(a)). For the normal index of (11), I[a,b]⟂(W)≥0.I^\perp_{[a,b]}(W)\ge0.(15)✦

For the shallow estimate, let a≤ba\le b and let γ:ℝ→ℝ3\gamma:\mathbb R\to\mathbb R^3 be smooth at every time of [a,b][a,b], with 0≤τ(s)≤140\le\tau(s)\le\tfrac14 there. For any field V:ℝ→ℝ3V:\mathbb R\to\mathbb R^3 along γ\gamma, I[a,b]g(V)≥0.I^g_{[a,b]}(V)\ge0.(16)✦

In the paper, the result reads: if VV is the restriction of a smooth compactly supported field along a complete unit-speed geodesic γ\gamma to an entire neck passage JJ, a maximal interval spent in the neck, and either J=ℝJ=\mathbb R, or the height velocity vv never vanishes on JJ, or JJ is a turn returning to the endpoint at which it began and contained in an endpoint portion 0≤τ≤140\le\tau\le\tfrac14, then IJ⟂(V)≥0I_J^\perp(V)\ge0.

What Is Checked

The estimates (13) and (14) concern the intrinsic index (8) of fields along geodesics for the endpoint framed metrics (6); (15) concerns the coordinate normal density (10) with the boundary condition (12), and (16) is stated for curves smooth at the times of a finite shallow interval.

Smooth moving frames have smooth pullback bilinear forms✦, and the triangular frame in (5) depends smoothly on smooth radius and shear functions when the radii never vanish.✦ The collar warping functions in (2) are smooth and positive. For every smooth map, the pullback tangent inner product is a smooth section of the bilinear-form bundle.✦