OctoInformBeta

OpenAI preprint, September 2026 · §4. Physical solutions realizing local transport frames

Proposition 4.2: Order-eight corrections and local bounds for amplitudes augmented by their actual divergence

Physical realization of transverse amplitude frames.

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

A transport amplitude prescribes a displacement and a scalar divergence variable. Physical realization asks for an exact elastic displacement whose actual divergence has the prescribed asymptotics.

Let X=ℝ3X=\mathbb R^3, let BRB_R denote the open ball about the origin, and let DR2⊂ℝ2D_R^2\subset\mathbb R^2 denote the planar disk. The smooth coefficient algebra is C∞(X;ℂ)C^\infty(X;\mathbb C); real coefficients are regarded as complex-valued functions. For real smooth Lamé coefficients, use the notation and ellipticity conditions in (1): m=μ>0,3λ+2μ>0,k=λ+μ,ℓ=λ+2μ.m=\mu>0,\qquad 3\lambda+2\mu>0,\qquad k=\lambda+\mu,\qquad \ell=\lambda+2\mu.(1) A smooth coefficient ff is constant outside a compact set if f−cf-c has compact support for some c∈ℂc\in\mathbb C.✦

Let eie_i be the coordinate vectors and ∂if(x)=dfx(ei)\partial_i f(x)=\mathrm df_x(e_i). The divergence and Laplacian are the sums of the first and second coordinate derivatives, respectively. A complex direction θ∈ℂ3\theta\in\mathbb C^3 is null when it is nonzero and θ⋅θ=0\theta\cdot\theta=0, with the dot product complex bilinear. Directional differentiation and the phase notation with parameter τ\tau are defined in (2), where α,β∈X\alpha,\beta\in X: θ=α+iβ,Dθ=∑i=13θi∂i,z=θ/h,h=τ−1,Eτ(x)=exp⁡(τθ⋅x),div⁡a=∑i∂iai,Δf=∑i∂i2f.\begin{gathered} \theta=\alpha+\mathrm i\beta,\qquad D_\theta=\sum_{i=1}^3\theta_i\partial_i,\qquad z=\theta/h,\qquad h=\tau^{-1},\qquad E_\tau(x)=\exp(\tau\theta\cdot x),\\ \operatorname{div}a=\sum_i\partial_i a_i,\qquad \Delta f=\sum_i\partial_i^2f. \end{gathered}(2) The phase functional associated with θ∈ℂ3\theta\in\mathbb C^3 is the continuous real-linear map x↦θ⋅xx\mapsto\theta\cdot x from XX to ℂ\mathbb C. For a continuous real linear functional L:X→ℂL:X\to\mathbb C, the smooth exponential is the smooth function x↦exp⁡(L(x))x\mapsto\exp(L(x)).

The Schwartz space 𝒮(X)\mathcal S(X) consists of smooth complex-valued functions whose derivatives decay faster than every inverse polynomial. The space 𝒮′(X)\mathcal S'(X) of tempered distributions is the continuous complex-linear dual of 𝒮(X)\mathcal S(X).✦ Schwartz functions are included in both C∞(X;ℂ)C^\infty(X;\mathbb C) and L2(X)L^2(X), with Lebesgue measure used throughout. The multiplication operator MfM_f associated with a coefficient ff is the map given by Mfv=fvM_fv=fv.✦ Shifted coordinate differentiation, shifted divergence and physical augmentation are the operators in (3); the derivative has its distributional meaning on 𝒮′\mathcal S': Tz,i=∂i+zi,div⁡zv=∑iTz,ivi,Aug⁡zv=(div⁡zv,v1,v2,v3).T_{z,i}=\partial_i+z_i,\qquad \operatorname{div}_zv=\sum_iT_{z,i}v_i,\qquad \operatorname{Aug}_zv=(\operatorname{div}_zv,v_1,v_2,v_3).(3) The conjugated elasticity operator ℒz\mathcal L_z is the shifted stress divergence defined by (4); write ℒ=ℒ0\mathcal L=\mathcal L_0: (ℒzv)i=Tz,i(λdiv⁡zv)+∑jTz,j(m(Tz,jvi+Tz,ivj)).(\mathcal L_zv)_i =T_{z,i}\bigl(\lambda\operatorname{div}_zv\bigr) +\sum_jT_{z,j}\bigl(m(T_{z,j}v_i+T_{z,i}v_j)\bigr).(4)✦

For real order ss and Fourier variable ξ∈X\xi\in X, let ℱ\mathcal F denote the Fourier transform; the Bessel weight whsw_h^s, its multiplication operator MhsM_h^s, the Bessel potential JhsJ_h^s and the semiclassical Sobolev norm are defined by (5). Here VV has finitely many Schwartz components, q∈L2(X)q\in L^2(X), and ∥f∥2\|f\|_2 is the norm of the Schwartz function in L2(X)L^2(X): $$\begin{gathered} w_h^s(\xi)=(1+|2\pi h\xi|^2)^{s/2},\qquad M_h^sf=w_h^sf,\qquad J_h^s=\mathcal F^{-1}M_h^s\mathcal F,\\ \|V\|_{H_h^s}^2=\sum_i\|J_h^sV_i\|_2^2,\qquad \iota_s q=J_1^{-s}q\quad\hbox{in }\mathcal S'(X). \end{gathered}$$(5) The Sobolev injection ιs:L2(X)→𝒮′(X)\iota_s:L^2(X)\to\mathcal S'(X) applies the Bessel potential to the canonical distribution associated with qq.✦ For displacement corrections use the Hilbert norm on L2(X;ℂ3)L^2(X;\mathbb C^3); for four-component L2L^2 families write ∥q∥∞,2=max⁡i∥qi∥2\|q\|_{\infty,2}=\max_i\|q_i\|_2.

Compatible Amplitudes and Physical Columns

An augmented amplitude A=(a,b)A=(a,b) is a pair in C∞(X;ℂ3)×C∞(X;ℂ)C^\infty(X;\mathbb C^3)\times C^\infty(X;\mathbb C).✦ The pointwise amplitude space is ℂ3×ℂ\mathbb C^3\times\mathbb C, consisting of pairs (a,b)(a,b) with a displacement vector and a scalar component. For θ∈ℂ3\theta\in\mathbb C^3, the normal amplitude map Nθ:ℂ3×ℂ→ℂN_\theta:\mathbb C^3\times\mathbb C\to\mathbb C is the complex-linear map (a,b)↦θ⋅a(a,b)\mapsto\theta\cdot a; on smooth amplitudes it acts pointwise.✦ The constrained transport fiber Eθ⊂ℂ3×ℂE_\theta\subset\mathbb C^3\times\mathbb C is the kernel of the normal amplitude map, as in (6): Hθ={a∈ℂ3:θ⋅a=0},Eθ=Hθ⊕ℂ.H_\theta=\{a\in\mathbb C^3:\theta\cdot a=0\},\qquad E_\theta=H_\theta\oplus\mathbb C.(6) For a nowhere-zero smooth 𝜚\varrho, the vector and scalar transformations of AA are 𝜚a\varrho a and −(kb/𝜚+2∇𝜚⋅a)/2-(kb/\varrho+2\nabla\varrho\cdot a)/2, respectively. For the real coefficients in (1), the normal form transformation of AA is (7). Here 𝜚=m\varrho=\sqrt m is the positive smooth square root, smooth reciprocals are pointwise reciprocals, and 2∇𝜚/𝜚=∇log⁡m2\nabla\varrho/\varrho=\nabla\log m: p=𝜚a,s=−kb2𝜚−12∇log⁡m⋅p.p=\varrho a,\qquad s=-\frac{k b}{2\varrho}-\frac12\nabla\log m\cdot p.(7) For a nowhere-zero smooth function 𝜚\varrho on XX, its logarithmic gradient is the smooth vector field 2∇𝜚/𝜚2\nabla\varrho/\varrho.

For complex smooth λ,m\lambda,m, the shear modulus gradient coupling CmC_m and the vector and scalar augmented residuals ℛ,𝒬\mathcal R,\mathcal Q are (8): Cm(a)i=∑j(∂jm)(∂jai+∂iaj),ℛi(A)=mΔai+k∂ib+Cm(a)i+(∂iλ)b,𝒬(A)=ℓΔb+2∑i(∂ik)∂ib+2∑i(∂im)Δai+2∑i,j(∂i∂jm)∂iaj+(Δλ)b.\begin{aligned} C_m(a)_i&=\sum_j(\partial_jm)(\partial_ja_i+\partial_ia_j),\\ \mathcal R_i(A)&=m\Delta a_i+k\partial_ib+C_m(a)_i+(\partial_i\lambda)b,\\ \mathcal Q(A)&=\ell\Delta b+2\sum_i(\partial_ik)\partial_ib +2\sum_i(\partial_im)\Delta a_i\\ &\quad+2\sum_{i,j}(\partial_i\partial_jm)\partial_ia_j+(\Delta\lambda)b. \end{aligned}(8) The scalar operator PθP_\theta and the vector and scalar transport operators ℛθ,𝒬θ\mathcal R_\theta,\mathcal Q_\theta are defined by (9): Pθf=2mDθf+(Dθm)f,(ℛθA)i=2mDθai+(Dθm)ai+θi(kb+∇m⋅a),𝒬θA=2ℓDθb+2(Dθk)b+4∑i(∂im)Dθai+2∑iDθ(∂im)ai.\begin{aligned} P_\theta f&=2mD_\theta f+(D_\theta m)f,\\ (\mathcal R_\theta A)_i &=2mD_\theta a_i+(D_\theta m)a_i +\theta_i(kb+\nabla m\cdot a),\\ \mathcal Q_\theta A &=2\ell D_\theta b+2(D_\theta k)b +4\sum_i(\partial_im)D_\theta a_i +2\sum_iD_\theta(\partial_im)a_i. \end{aligned}(9) A leading transport solution A=(a,b)A=(a,b) on XX is a smooth amplitude satisfying θ⋅a=0\theta\cdot a=0 and ℛθA=𝒬θA=0\mathcal R_\theta A=\mathcal Q_\theta A=0. The same formulas apply in a commutative complex coefficient algebra with derivations DiD_i, replacing ∂i\partial_i by DiD_i; a superscript DD will indicate this replacement.

An extendible local smooth function f:U→ℂf:U\to\mathbb C on an open set U⊂XU\subset X is the restriction to UU of a smooth function on XX.✦ Restricted amplitudes and coefficients are obtained componentwise, and local coordinate derivatives are the derivatives of these restrictions. A compatible amplitude A=(a,b)A=(a,b) on UU is a smooth pair satisfying (10), with δ(A)=div⁡a−b\delta(A)=\operatorname{div}a-b: $$\theta\cdot\mathcal R(A)=P_\theta\delta(A)\quad\hbox{on }U.$$(10) A recursive step A⇝UA′A\rightsquigarrow_U A' is a pair of smooth amplitudes A=(a,b)A=(a,b) and A′=(a′,b′)A'=(a',b') satisfying (11); the normal relation is imposed on all of XX: $$\mathcal R_\theta A'=-\mathcal R(A),\quad \mathcal Q_\theta A'=-\mathcal Q(A)\quad\hbox{on }U, \qquad \theta\cdot a'=-\delta(A)\quad\hbox{on }X.$$(11) For a nonnegative integer NN, the truncated amplitude Ah[N]A_h^{[N]} of a sequence A=(Aj)j≥0A=(A_j)_{j\geq0} with components Aj=(aj,bj)A_j=(a_j,b_j) is the finite sum (12); write δj=δ(Aj)\delta_j=\delta(A_j): Ah[N]=(ah[N],bh[N])=∑j=0NhjAj.A_h^{[N]}=(a_h^{[N]},b_h^{[N]})=\sum_{j=0}^Nh^jA_j.(12)

Fix R≥1R\geq1. Let χj\chi_j, 0≤j≤40\leq j\leq4, be the nested radial cutoffs with inner radius R+j+1R+j+1 and outer radius R+j+2R+j+2, and set χ=χ0\chi=\chi_0.✦ Compact cutoff multiplication is f↦χff\mapsto\chi f✦, and localized exponential multiplication is f↦χez⋅xff\mapsto\chi e^{z\cdot x}f.✦ The corrected amplitude Wh(v)W_h(v), physical complex geometrical optics field uh(v)u_h(v), localized formal amplitude Fh(g)F_h(g) and localized leading augmented amplitude G0(g)G_0(g) are (13), for v∈𝒮(X)3v\in\mathcal S(X)^3 and a smooth multiplier gg: Wh(v)=χah[8]−v,uh(v)=χE1/hWh(v),Fh(g)=gχah[8],G0(g)=(gχb0+∇g⋅(χa0),gχa0).\begin{gathered} W_h(v)=\chi a_h^{[8]}-v,\qquad u_h(v)=\chi E_{1/h}W_h(v),\qquad F_h(g)=g\chi a_h^{[8]},\\ G_0(g)=\bigl(g\chi b_0+\nabla g\cdot(\chi a_0),\;g\chi a_0\bigr). \end{gathered}(13) A smooth multiplier g∈C∞(X;ℂ)g\in C^\infty(X;\mathbb C) has temperate growth when each of its derivatives is bounded by a polynomial in |x||x|.

A physical column A=(Aj)j≥0A=(A_j)_{j\geq0} is a smooth amplitude sequence for which there are C,h0>0C,h_0>0 such that every 0<h≤h00<h\leq h_0 admits w∈L2(X;ℂ3)w\in L^2(X;\mathbb C^3), q∈L2(X)4q\in L^2(X)^4 and v∈𝒮(X)3v\in\mathcal S(X)^3 satisfying the correction conditions (14) and ℒθ/hWh(v)=0\mathcal L_{\theta/h}W_h(v)=0 on BRB_R: $$\begin{gathered} w=v\quad\hbox{almost everywhere on }B_R,\qquad \iota_1q=\chi\operatorname{Aug}_{\theta/h}w \quad\hbox{in }\mathcal S'(X)^4,\\ \|w\|_{L^2(X;\mathbb C^3)}\leq Ch^7,\qquad \|q\|_{\infty,2}\leq Ch. \end{gathered}$$(14) The local normal recurrence needed for amplitude estimates is (15): $$\theta\cdot a_0=0\quad\hbox{on }X,\qquad \theta\cdot a_{j+1}=-\delta_j\quad\hbox{on }B_R \quad(0\leq j<8).$$(15)

The Claim

Theorem 1. Let U⊂XU\subset X be open with compact closure, let θ\theta be nonzero and null, and let k,𝜚k,\varrho be complex smooth functions such that 𝜚\varrho, kk and k+𝜚2k+\varrho^2 are nowhere zero. Set λ=k−𝜚2\lambda=k-\varrho^2 and m=𝜚2m=\varrho^2. For every compatible amplitude A*A_* on UU, there is a smooth amplitude sequence satisfying (16): A0=A*,Aj⇝UAj+1,Aj is compatible on U(j≥0).A_0=A_*,\qquad A_j\rightsquigarrow_U A_{j+1},\qquad A_j\text{ is compatible on }U\quad(j\geq0).(16)✦

Let 𝒜\mathcal A be a commutative complex algebra with commuting derivations DiD_i, let λ,m∈𝒜\lambda,m\in\mathcal A, let θ∈ℂ3\theta\in\mathbb C^3 satisfy θ⋅θ=0\theta\cdot\theta=0, and let Aj=(aj,bj)∈𝒜3×𝒜A_j=(a_j,b_j)\in\mathcal A^3\times\mathcal A for 0≤j≤N0\leq j\leq N. Suppose ℛθDA0=0\mathcal R_\theta^D A_0=0 and θ⋅a0=0\theta\cdot a_0=0, and suppose that ℛθDAj+1=−ℛD(Aj)\mathcal R_\theta^D A_{j+1}=-\mathcal R^D(A_j) and θ⋅aj+1=−δj\theta\cdot a_{j+1}=-\delta_j for j<Nj<N, with δj=div⁡Daj−bj\delta_j=\operatorname{div}^D a_j-b_j. For every nonzero complex hh, the truncated physical residual is (17): (ℒh−1θDah[N])i=hN(ℛiD(AN)+kDiδN+(Diλ)δN+h−1θikδN).(\mathcal L^D_{h^{-1}\theta}a_h^{[N]})_i =h^N\left(\mathcal R_i^D(A_N)+kD_i\delta_N +(D_i\lambda)\delta_N+h^{-1}\theta_i k\delta_N\right).(17)✦

Let R≥1R\geq1 and θ=α+iβ\theta=\alpha+\mathrm i\beta with α≠0\alpha\ne0 and θ⋅θ=0\theta\cdot\theta=0. Let λ,μ\lambda,\mu be real smooth coefficients satisfying (1), each constant outside a compact set. There are three physical columns AνA^\nu and a basis B(x)=(B1(x),B2(x),B3(x))B(x)=(B^1(x),B^2(x),B^3(x)) of EθE_\theta at every x∈Xx\in X, such that (18) holds, with (p0ν,s0ν)(p_0^\nu,s_0^\nu) given by (7): θ⋅a0ν=0,Bν(x)=(p0ν(x),s0ν(x))(ν=1,2,3,x∈X).\theta\cdot a_0^\nu=0,\qquad B^\nu(x)=\bigl(p_0^\nu(x),s_0^\nu(x)\bigr) \quad(\nu=1,2,3,\ x\in X).(18)✦

For arbitrary R≥1R\geq1 and α,β∈X\alpha,\beta\in X, set θ=α+iβ\theta=\alpha+\mathrm i\beta. Let λ,m\lambda,m be complex smooth coefficients constant outside compact sets, and let AA be a physical column. There are C,h0>0C,h_0>0 such that every 0<h≤h00<h\leq h_0 admits corrections satisfying (14) and the physical equation (19): $$\mathcal L u_h(v)=0\quad\hbox{on }B_R.$$(19)✦

Assume also (15), and let g∈Cc∞(X)g\in C_c^\infty(X) have support contained in BRB_R. There are C,h1>0C,h_1>0 such that every 0<h≤h10<h\leq h_1 admits v∈𝒮(X)3v\in\mathcal S(X)^3 satisfying (19) and (20): ∥Aug⁡θ/h(gWh(v))∥Hh1≤C.\bigl\|\operatorname{Aug}_{\theta/h}\bigl(gW_h(v)\bigr)\bigr\|_{H_h^1} \leq C.(20)✦ If B⊂BRB\subset B_R is open and g=1g=1 on BB, there are C,D,h1>0C,D,h_1>0 such that every 0<h≤h10<h\leq h_1 admits vh∈𝒮(X)3v_h\in\mathcal S(X)^3 satisfying (21): $$\begin{gathered} \mathcal L_{\theta/h}v_h=0\quad\hbox{on }B,\qquad \|\operatorname{Aug}_{\theta/h}v_h\|_{H_h^1}\leq C,\\ \bigl\|[\operatorname{Aug}_{\theta/h}v_h -\operatorname{Aug}_{\theta/h}F_h(g)]_i\bigr\|_2\leq Dh \quad(i=0,1,2,3). \end{gathered}$$(21)✦

For any smooth amplitude sequence satisfying (15), let gg be a smooth multiplier of temperate growth supported in BRB_R. For every sequence of nonzero real numbers hn→0h_n\to0, the formal augmented amplitudes satisfy (22): ∥[Aug⁡θ/hnFhn(g)]i−[G0(g)]i∥2→0(i=0,1,2,3).\bigl\|[\operatorname{Aug}_{\theta/h_n}F_{h_n}(g)]_i -[G_0(g)]_i\bigr\|_2\longrightarrow0 \quad(i=0,1,2,3).(22)✦

In the paper, for a radius r>0r>0, the result reads: for every nonzero null θ\theta and t0∈(−2r,2r)t_0\in(-2r,2r), there are an open interval t0∈J⊂(−2r,2r)t_0\in J\subset(-2r,2r) and three smooth leading transport solutions (a0ν,b0ν)(a_0^\nu,b_0^\nu) on a neighborhood of B5r¯\overline{B_{5r}}, forming a frame of EθE_\theta at every point of D2r2×JD_{2r}^2\times J, such that for each column and all sufficiently large τ\tau there is a smooth exact physical solution uτνu_\tau^\nu on B5rB_{5r} with Uτν=(div⁡uτν,uτν)U_\tau^\nu=(\operatorname{div}u_\tau^\nu,u_\tau^\nu) satisfying Eτ−1Uτν→(b0ν,a0ν)E_\tau^{-1}U_\tau^\nu\to(b_0^\nu,a_0^\nu) strongly in L2(B4r)L^2(B_{4r}) and ∥Eτ−1∇Uτν∥L2(B4r)=O(τ)\|E_\tau^{-1}\nabla U_\tau^\nu\|_{L^2(B_{4r})}=O(\tau).

What Is Checked

The frame statement is over real smooth Lamé coefficients satisfying (1), constant outside compact sets, and nonzero null directions; physical columns are formulated using L2L^2 displacement corrections and their shifted distributional divergence through the Sobolev injection. The localized estimates use compactly supported smooth multipliers and the local normal recurrence, whereas the formal limit is taken along arbitrary nonzero real sequences tending to zero. Recursive amplitudes are stated over extendible local smooth functions, and the residual identity uses commuting derivations.

The positive smooth square root of mm is nowhere zero.✦ A smooth function vanishing on an open set has each coordinate derivative vanishing there.✦ The nested radial cutoffs are smooth✦ and compactly supported.✦