WELCOME! A Conference in Honor of ROBERT BRYANT Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 1 / 36 WELCOME! A Conference in Honor of ROBERT BRYANT Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 2 / 36 A GEOMETRIC PERSPECTIVE ON NONLINEAR PARTIAL DIFFERENTIAL EQUATIONS Reese Harvey Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 3 / 36 A GEOMETRIC PERSPECTIVE ON NONLINEAR PARTIAL DIFFERENTIAL EQUATIONS Reese Harvey Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 4 / 36 Fully Nonlinear PDE’s Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 5 / 36 Fully Nonlinear PDE’s Standard Setting: Ωopen ⊂ Rn f (x, u, Du, D 2 u) = 0 Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 5 / 36 Fully Nonlinear PDE’s Standard Setting: Ωopen ⊂ Rn f (x, u, Du, D 2 u) = 0 f (x, r , p, A) Blaine Lawson on Ω × R × Rn × Sym2 (Rn ) A Geometric Perspective on Nonlinear PDE’s July 11, 2013 5 / 36 Fully Nonlinear PDE’s Ωopen ⊂ Rn Standard Setting: f (x, u, Du, D 2 u) = 0 f (x, r , p, A) on Ω × R × Rn × Sym2 (Rn ) Example : The Linear Case. f = X ij Blaine Lawson aij (x) ∂ 2 u(x) X ∂u(x) + bj (x) + c(x)u(x) ∂xi ∂xj ∂xj j A Geometric Perspective on Nonlinear PDE’s July 11, 2013 5 / 36 Fully Nonlinear PDE’s Ωopen ⊂ Rn Standard Setting: f (x, u, Du, D 2 u) = 0 f (x, r , p, A) on Ω × R × Rn × Sym2 (Rn ) Example : The Linear Case. f = X ij aij (x) ∂ 2 u(x) X ∂u(x) + bj (x) + c(x)u(x) ∂xi ∂xj ∂xj j f (x, r , p, A) = ha(x), Ai + hb(x), pi + c(x)r Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 5 / 36 Fully Nonlinear PDE’s Ωopen ⊂ Rn Standard Setting: f (x, u, Du, D 2 u) = 0 f (x, r , p, A) on Ω × R × Rn × Sym2 (Rn ) Example : The Linear Case. f = X ij aij (x) ∂ 2 u(x) X ∂u(x) + bj (x) + c(x)u(x) ∂xi ∂xj ∂xj j f (x, r , p, A) = ha(x), Ai + hb(x), pi + c(x)r Weak ellipticity: a ≥ 0 Blaine Lawson Properness: c ≤ 0 A Geometric Perspective on Nonlinear PDE’s July 11, 2013 5 / 36 Some Standard Examples Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 6 / 36 Some Standard Examples Example 1. (Laplace). tr(A) = 0 Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 6 / 36 Some Standard Examples Example 1. (Laplace). tr(A) = 0 Example 2. (Monge-Ampère). det(A) = 0 Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 6 / 36 Some Standard Examples Example 1. (Laplace). tr(A) = 0 Example 2. (Monge-Ampère). det(A) = 0 Example 3. (Elementary Symmetric Functions). σk (A) = 0 Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 6 / 36 Some Standard Examples Example 4. (Special Lagrangian Potential). tr(arctan A) = 0 Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 7 / 36 Some Standard Examples Example 4. (Special Lagrangian Potential). tr(arctan A) = 0 Example 5. (Minimal Surface Equation). (1 + |p|2 )trA − pt Ap = 0 Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 7 / 36 Some Standard Examples Example 4. (Special Lagrangian Potential). tr(arctan A) = 0 Example 5. (Minimal Surface Equation). (1 + |p|2 )trA − pt Ap = 0 Example 6. (k -Laplacian 1 ≤ k ≤ ∞). |p|2 tr(A) + (k − 2)pt Ap = 0 Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 7 / 36 Some Basic Points Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 8 / 36 Some Basic Points 1. Equations like det(A) = 0 can have many BRANCHES. Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 8 / 36 Some Basic Points 1. Equations like det(A) = 0 can have many BRANCHES. Given A ∈ Sym2 (Rn ) let λ1 (A) ≤ λ2 (A) ≤ · · · ≤ λn (A) denote the ordered eigenvalues of A. Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 8 / 36 Some Basic Points 1. Equations like det(A) = 0 can have many BRANCHES. Given A ∈ Sym2 (Rn ) let λ1 (A) ≤ λ2 (A) ≤ · · · ≤ λn (A) denote the ordered eigenvalues of A. Then detA = λ1 (A) · · · λn (A) = 0 Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 8 / 36 Some Basic Points 1. Equations like det(A) = 0 can have many BRANCHES. Given A ∈ Sym2 (Rn ) let λ1 (A) ≤ λ2 (A) ≤ · · · ≤ λn (A) denote the ordered eigenvalues of A. Then detA = λ1 (A) · · · λn (A) = 0 The k th branch of this equation is λk (A) = 0. Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 8 / 36 Some Basic Points 1. Equations like det(A) = 0 can have many BRANCHES. Given A ∈ Sym2 (Rn ) let λ1 (A) ≤ λ2 (A) ≤ · · · ≤ λn (A) denote the ordered eigenvalues of A. Then detA = λ1 (A) · · · λn (A) = 0 The k th branch of this equation is λk (A) = 0. (Classical one considers the primary branch λ1 (A) = 0, i.e., A ≥ 0.) Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 8 / 36 Some Basic Points 2. Equations like det(A) = 0, σk (A) = 0, etc. make sense in the complex and quaternionic worlds. Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 9 / 36 Some Basic Points 2. Equations like det(A) = 0, σk (A) = 0, etc. make sense in the complex and quaternionic worlds. 1. Cn = (R2n , J) and A ∈ Sym2 (R2n ). Set AC ≡ Blaine Lawson 1 2 (A + JAJ). A Geometric Perspective on Nonlinear PDE’s July 11, 2013 9 / 36 Some Basic Points 2. Equations like det(A) = 0, σk (A) = 0, etc. make sense in the complex and quaternionic worlds. 1. Cn = (R2n , J) and A ∈ Sym2 (R2n ). Set AC ≡ 1 2 (A + JAJ). Eigenspaces of AC are complex and there are ordered eigenvalues C λC 1 (A) ≤ · · · ≤ λn (A). Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 9 / 36 Some Basic Points 2. Equations like det(A) = 0, σk (A) = 0, etc. make sense in the complex and quaternionic worlds. 1. Cn = (R2n , J) and A ∈ Sym2 (R2n ). Set AC ≡ 1 2 (A + JAJ). Eigenspaces of AC are complex and there are ordered eigenvalues C λC 1 (A) ≤ · · · ≤ λn (A). 2. Hn = (R4n , I, J, K ) and A ∈ Sym2 (R4n ). Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 9 / 36 Some Basic Points 2. Equations like det(A) = 0, σk (A) = 0, etc. make sense in the complex and quaternionic worlds. 1. Cn = (R2n , J) and A ∈ Sym2 (R2n ). Set AC ≡ 1 2 (A + JAJ). Eigenspaces of AC are complex and there are ordered eigenvalues C λC 1 (A) ≤ · · · ≤ λn (A). 2. Hn = (R4n , I, J, K ) and A ∈ Sym2 (R4n ). Set AH ≡ Blaine Lawson 1 4 (A + IAI + JAJ + IKI). A Geometric Perspective on Nonlinear PDE’s July 11, 2013 9 / 36 Some Basic Points 2. Equations like det(A) = 0, σk (A) = 0, etc. make sense in the complex and quaternionic worlds. 1. Cn = (R2n , J) and A ∈ Sym2 (R2n ). Set AC ≡ 1 2 (A + JAJ). Eigenspaces of AC are complex and there are ordered eigenvalues C λC 1 (A) ≤ · · · ≤ λn (A). 2. Hn = (R4n , I, J, K ) and A ∈ Sym2 (R4n ). Set AH ≡ 1 4 (A + IAI + JAJ + IKI). Eigenspaces of AH are quaternionic and there are ordered eigenvalues H λH 1 (A) ≤ · · · ≤ λn (A). Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 9 / 36 All the examples above carry over to any riemannian manifold. Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 10 / 36 All the examples above carry over to any riemannian manifold. Suppose X is a riemannian manifold with Levi-Civita connection ∇. Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 10 / 36 All the examples above carry over to any riemannian manifold. Suppose X is a riemannian manifold with Levi-Civita connection ∇. Definition. For f ∈ C 2 (X ), the riemannian hessian of f is the section Hess f ∈ Sym2 (T ∗ X ) Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 10 / 36 All the examples above carry over to any riemannian manifold. Suppose X is a riemannian manifold with Levi-Civita connection ∇. Definition. For f ∈ C 2 (X ), the riemannian hessian of f is the section Hess f ∈ Sym2 (T ∗ X ) defined on vector fields V , W by (Hess f )(V , W ) ≡ V W f − (∇V W )f . Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 10 / 36 The Viscosity Approach (Crandall, Iishi, Lions, Evans). Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 11 / 36 The Viscosity Approach (Crandall, Iishi, Lions, Evans). The idea: Extend the notion of “solution” f (x, u, Du, D 2 u) = 0 to more general functions u. Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 11 / 36 The Viscosity Approach (Crandall, Iishi, Lions, Evans). The idea: Extend the notion of “solution” f (x, u, Du, D 2 u) = 0 to more general functions u. For this one introduces subsolutions f (x, u, Du, D 2 u) ≥ 0 Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 11 / 36 The Viscosity Approach (Crandall, Iishi, Lions, Evans). The idea: Extend the notion of “solution” f (x, u, Du, D 2 u) = 0 to more general functions u. For this one introduces subsolutions f (x, u, Du, D 2 u) ≥ 0 and supersolutions f (x, u, Du, D 2 u) ≤ 0 Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 11 / 36 We Concentrate on Subsolutions. Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 12 / 36 We Concentrate on Subsolutions. Replace the inequality f (x, u, Du, D 2 u) ≥ 0 Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 12 / 36 We Concentrate on Subsolutions. Replace the inequality f (x, u, Du, D 2 u) ≥ 0 by the subset F ≡ {f ≥ 0} Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 12 / 36 We Concentrate on Subsolutions. Replace the inequality f (x, u, Du, D 2 u) ≥ 0 by the subset F ≡ {f ≥ 0} Two basic assumptions. Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 12 / 36 We Concentrate on Subsolutions. Replace the inequality f (x, u, Du, D 2 u) ≥ 0 by the subset F ≡ {f ≥ 0} Two basic assumptions. Set P ≡ {(0, 0, A) : A ≥ 0} Blaine Lawson and N ≡ {(r , 0, 0) : r ≤ 0} A Geometric Perspective on Nonlinear PDE’s July 11, 2013 12 / 36 We Concentrate on Subsolutions. Replace the inequality f (x, u, Du, D 2 u) ≥ 0 by the subset F ≡ {f ≥ 0} Two basic assumptions. Set P ≡ {(0, 0, A) : A ≥ 0} and N ≡ {(r , 0, 0) : r ≤ 0} We will assume positivity (weak ellipticity) F +P ⊂ F Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 12 / 36 We Concentrate on Subsolutions. Replace the inequality f (x, u, Du, D 2 u) ≥ 0 by the subset F ≡ {f ≥ 0} Two basic assumptions. Set P ≡ {(0, 0, A) : A ≥ 0} and N ≡ {(r , 0, 0) : r ≤ 0} We will assume positivity (weak ellipticity) F +P ⊂ F and negativity (properness) F +N ⊂ F Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 12 / 36 We Concentrate on Subsolutions. Replace the inequality f (x, u, Du, D 2 u) ≥ 0 by the subset F ≡ {f ≥ 0} Two basic assumptions. Set P ≡ {(0, 0, A) : A ≥ 0} and N ≡ {(r , 0, 0) : r ≤ 0} We will assume positivity (weak ellipticity) F +P ⊂ F and negativity (properness) F +N ⊂ F F is called a SUBEQUATION. Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 12 / 36 Viscosity Subsolutions. Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 13 / 36 Viscosity Subsolutions. Given X open ⊂ Rn , set USC(X ) ≡ {u : X → [−∞, ∞) : u is upper-semicontinuous} Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 13 / 36 Viscosity Subsolutions. Given X open ⊂ Rn , set USC(X ) ≡ {u : X → [−∞, ∞) : u is upper-semicontinuous} Definition. A C 2 -function ϕ is a test function for u ∈ USC(X ) at a point x ∈ X if u ≤ ϕ near x u = ϕ Blaine Lawson at x A Geometric Perspective on Nonlinear PDE’s July 11, 2013 13 / 36 Viscosity Subsolutions. Given X open ⊂ Rn , set USC(X ) ≡ {u : X → [−∞, ∞) : u is upper-semicontinuous} Definition. A C 2 -function ϕ is a test function for u ∈ USC(X ) at a point x ∈ X if u ≤ ϕ near x u = ϕ at x Definition. A function u ∈ USC(X ) is F -subharmonic on X if for each x ∈ X and each test function ϕ for u at x, Jx2 (u) ∈ F . Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 13 / 36 Viscosity Subsolutions. Given X open ⊂ Rn , set USC(X ) ≡ {u : X → [−∞, ∞) : u is upper-semicontinuous} Definition. A C 2 -function ϕ is a test function for u ∈ USC(X ) at a point x ∈ X if u ≤ ϕ near x u = ϕ at x Definition. A function u ∈ USC(X ) is F -subharmonic on X if for each x ∈ X and each test function ϕ for u at x, Jx2 (u) ∈ F . F (X ) ≡ the set of such functions. Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 13 / 36 Properties. Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 14 / 36 Properties. F (X ) is closed under: • Uniform limits Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 14 / 36 Properties. F (X ) is closed under: • Uniform limits • Decreasing limits Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 14 / 36 Properties. F (X ) is closed under: • Uniform limits • Decreasing limits • Taking maximum (u, v ∈ F (X ) ⇒ max{u, v } ∈ F (X )). Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 14 / 36 Properties. F (X ) is closed under: • Uniform limits • Decreasing limits • Taking maximum (u, v ∈ F (X ) ⇒ max{u, v } ∈ F (X )). • Taking upper envelopes: F ⊂ F (X ) locally bounded above, then ∗ v ≡ sup u u∈F Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 14 / 36 Properties. F (X ) is closed under: • Uniform limits • Decreasing limits • Taking maximum (u, v ∈ F (X ) ⇒ max{u, v } ∈ F (X )). • Taking upper envelopes: F ⊂ F (X ) locally bounded above, then ∗ v ≡ sup u u∈F Like subharmonic functions! Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 14 / 36 Examples. Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 15 / 36 Examples. F ≡ {trA ≥ 0}. Blaine Lawson F (X ) = {the subharmonic functions} A Geometric Perspective on Nonlinear PDE’s July 11, 2013 15 / 36 Examples. F ≡ {trA ≥ 0}. F ≡ {A ≥ 0}. Blaine Lawson F (X ) = {the subharmonic functions} F (X ) = {the convex functions} A Geometric Perspective on Nonlinear PDE’s July 11, 2013 15 / 36 Examples. F ≡ {trA ≥ 0}. F ≡ {A ≥ 0}. F ≡ {AC ≥ 0}. Blaine Lawson F (X ) = {the subharmonic functions} F (X ) = {the convex functions} F (X ) = {the plurisubharmonic functions} A Geometric Perspective on Nonlinear PDE’s July 11, 2013 15 / 36 Examples. F ≡ {trA ≥ 0}. F ≡ {A ≥ 0}. F ≡ {AC ≥ 0}. F ≡ {AH ≥ 0}. Blaine Lawson F (X ) = {the subharmonic functions} F (X ) = {the convex functions} F (X ) = {the plurisubharmonic functions} F (X ) = {the H-plurisubharmonic functions} A Geometric Perspective on Nonlinear PDE’s July 11, 2013 15 / 36 Examples. F ≡ {trA ≥ 0}. F ≡ {A ≥ 0}. F ≡ {AC ≥ 0}. F ≡ {AH ≥ 0}. F (X ) = {the subharmonic functions} F (X ) = {the convex functions} F (X ) = {the plurisubharmonic functions} F (X ) = {the H-plurisubharmonic functions} Fk ≡ {λk (A) ≥ 0}. Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 15 / 36 Analogues in Calibrated Geometry. Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 16 / 36 Analogues in Calibrated Geometry. Let φ ∈ Λ p Rn be a calibration, Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 16 / 36 Analogues in Calibrated Geometry. Let φ ∈ Λ p Rn be a calibration, i.e. φP ≤ volP for all oriented p-planes P. Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 16 / 36 Analogues in Calibrated Geometry. Let φ ∈ Λ p Rn be a calibration, i.e. φP ≤ volP for all oriented p-planes P. Let G l (φ) ≡ Blaine Lawson P : φP = volP A Geometric Perspective on Nonlinear PDE’s July 11, 2013 16 / 36 Analogues in Calibrated Geometry. Let φ ∈ Λ p Rn be a calibration, i.e. φP ≤ volP for all oriented p-planes P. Let G l (φ) ≡ P : φP = volP Definition. F (φ) ≡ Blaine Lawson A : tr AP ≥ 0 ∀ P ∈ G l (φ) A Geometric Perspective on Nonlinear PDE’s July 11, 2013 16 / 36 Analogues in Calibrated Geometry. Let φ ∈ Λ p Rn be a calibration, i.e. φP ≤ volP for all oriented p-planes P. Let G l (φ) ≡ P : φP = volP Definition. F (φ) ≡ A : tr AP ≥ 0 ∀ P ∈ G l (φ) φ-subharmonic functions. Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 16 / 36 What about Supersolutions? Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 17 / 36 What about Supersolutions? A PARALLEL STORY UNDER DUALITY Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 17 / 36 Duality Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 18 / 36 Duality Given a subequation F , define the dual of F by e ≡ ∼ (−IntF) = −(∼ IntF) F Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 18 / 36 Duality Given a subequation F , define the dual of F by e ≡ ∼ (−IntF) = −(∼ IntF) F We make mild topological assumptions on F : Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 18 / 36 Duality Given a subequation F , define the dual of F by e ≡ ∼ (−IntF) = −(∼ IntF) F We make mild topological assumptions on F : (i) F = IntF Blaine Lawson (ii) Fx = Intx Fx (iii) Intx F = (IntF ) ∩ Fx A Geometric Perspective on Nonlinear PDE’s July 11, 2013 18 / 36 Duality Given a subequation F , define the dual of F by e ≡ ∼ (−IntF) = −(∼ IntF) F We make mild topological assumptions on F : (i) F = IntF (ii) Fx = Intx Fx (iii) Intx F = (IntF ) ∩ Fx Then e is a subequation F Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 18 / 36 Duality Given a subequation F , define the dual of F by e ≡ ∼ (−IntF) = −(∼ IntF) F We make mild topological assumptions on F : (i) F = IntF (ii) Fx = Intx Fx (iii) Intx F = (IntF ) ∩ Fx Then e is a subequation F e e = F F Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 18 / 36 Examples. Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 19 / 36 Solutions – F -harmonicity Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 20 / 36 Solutions – F -harmonicity Definition. A function u ∈ C(X ) is called F -harmonic if u is F -subharmonic Blaine Lawson and e -subharmonic −u is F A Geometric Perspective on Nonlinear PDE’s July 11, 2013 20 / 36 The Dirichlet Problem – F -Boundary Convexity Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 21 / 36 The Dirichlet Problem – F -Boundary Convexity Consider a domain Ω ⊂⊂ Rn with smooth boundary ∂Ω. Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 21 / 36 The Dirichlet Problem – F -Boundary Convexity Consider a domain Ω ⊂⊂ Rn with smooth boundary ∂Ω. Let F ⊂ Sym2 (Rn ) be a constant coefficient pure second-order subequation Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 21 / 36 The Dirichlet Problem – F -Boundary Convexity Consider a domain Ω ⊂⊂ Rn with smooth boundary ∂Ω. Let F ⊂ Sym2 (Rn ) be a constant coefficient pure second-order subequation Definition. ∂Ω is strictly F -convex at a point x ∈ ∂Ω if its second fundamental form IIx (w.r.t. the interior normal) satisfes Blaine Lawson t 0 0 IIx ∈ IntF for all t ≥ some t0 . A Geometric Perspective on Nonlinear PDE’s July 11, 2013 21 / 36 The Dirichlet Problem Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 22 / 36 The Dirichlet Problem THEOREM. (Harvey+L.) Let F be a constant coefficient subequation and e -convex. suppose ∂Ω ⊂⊂ Rn is both strictly F and F Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 22 / 36 The Dirichlet Problem THEOREM. (Harvey+L.) Let F be a constant coefficient subequation and e -convex. Then for every suppose ∂Ω ⊂⊂ Rn is both strictly F and F ϕ ∈ C(∂Ω), there exists u ∈ C(Ω) such that u Ω is F -harmonic, and u ∂Ω = ϕ. Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 22 / 36 The Dirichlet Problem THEOREM. (Harvey+L.) Let F be a constant coefficient subequation and e -convex. Then for every suppose ∂Ω ⊂⊂ Rn is both strictly F and F ϕ ∈ C(∂Ω), there exists u ∈ C(Ω) such that u Ω is F -harmonic, and u ∂Ω = ϕ. Uniqueness holds if F is pure second-order, Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 22 / 36 The Dirichlet Problem THEOREM. (Harvey+L.) Let F be a constant coefficient subequation and e -convex. Then for every suppose ∂Ω ⊂⊂ Rn is both strictly F and F ϕ ∈ C(∂Ω), there exists u ∈ C(Ω) such that u Ω is F -harmonic, and u ∂Ω = ϕ. Uniqueness holds if F is pure second-order, (or, more generally, gradient-independent). Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 22 / 36 Removable Singularities – Monotonicity Cones Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 23 / 36 Removable Singularities – Monotonicity Cones Definition. A monotonicity cone for a subequation F is a convex cone subequation M such that F + M ⊂ F. Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 23 / 36 Removable Singularities – Monotonicity Cones Definition. A monotonicity cone for a subequation F is a convex cone subequation M such that F + M ⊂ F. Definition. A closed subset E ⊂ X open ⊂ Rn is C ∞ M-polar if E ⊂ {x : ψ(x) = −∞} for some ψ ∈ M(X ) which is smooth outside of E. Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 23 / 36 Removable Singularities – Monotonicity Cones Definition. A monotonicity cone for a subequation F is a convex cone subequation M such that F + M ⊂ F. Definition. A closed subset E ⊂ X open ⊂ Rn is C ∞ M-polar if E ⊂ {x : ψ(x) = −∞} for some ψ ∈ M(X ) which is smooth outside of E. THEOREM. (Harvey+L.) Let F be a subequation with monotonicity cone M. Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 23 / 36 Removable Singularities – Monotonicity Cones Definition. A monotonicity cone for a subequation F is a convex cone subequation M such that F + M ⊂ F. Definition. A closed subset E ⊂ X open ⊂ Rn is C ∞ M-polar if E ⊂ {x : ψ(x) = −∞} for some ψ ∈ M(X ) which is smooth outside of E. THEOREM. (Harvey+L.) Let F be a subequation with monotonicity cone M. Let E ⊂ X be a closed subset (with no interior) which is C ∞ M-polar. Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 23 / 36 Removable Singularities – Monotonicity Cones Definition. A monotonicity cone for a subequation F is a convex cone subequation M such that F + M ⊂ F. Definition. A closed subset E ⊂ X open ⊂ Rn is C ∞ M-polar if E ⊂ {x : ψ(x) = −∞} for some ψ ∈ M(X ) which is smooth outside of E. THEOREM. (Harvey+L.) Let F be a subequation with monotonicity cone M. Let E ⊂ X be a closed subset (with no interior) which is C ∞ M-polar. Then any u ∈ F (X − E) which is locally bounded across E e ∈ F (X ). extends canonically to a function u Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 23 / 36 Removable Singularities – Monotonicity Cones Definition. A monotonicity cone for a subequation F is a convex cone subequation M such that F + M ⊂ F. Definition. A closed subset E ⊂ X open ⊂ Rn is C ∞ M-polar if E ⊂ {x : ψ(x) = −∞} for some ψ ∈ M(X ) which is smooth outside of E. THEOREM. (Harvey+L.) Let F be a subequation with monotonicity cone M. Let E ⊂ X be a closed subset (with no interior) which is C ∞ M-polar. Then any u ∈ F (X − E) which is locally bounded across E e ∈ F (X ). extends canonically to a function u THEOREM. (Harvey+L.) Let F and M be as above. Then for u ∈ C(X ) u is F harmonic on X − E Blaine Lawson ⇒ u is F harmonic on X . A Geometric Perspective on Nonlinear PDE’s July 11, 2013 23 / 36 Removable Singularities – Example Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 24 / 36 Removable Singularities – Example Fix 1 ≤ p ≤ n and let Pp ≡ {A : λ1 (A) + · · · + λ[p] (A) + (p − [p])λp+1 (A) ≥ 0} Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 24 / 36 Removable Singularities – Example Fix 1 ≤ p ≤ n and let Pp ≡ {A : λ1 (A) + · · · + λ[p] (A) + (p − [p])λp+1 (A) ≥ 0} The p-convexity subequation Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 24 / 36 Removable Singularities – Example Fix 1 ≤ p ≤ n and let Pp ≡ {A : λ1 (A) + · · · + λ[p] (A) + (p − [p])λp+1 (A) ≥ 0} The p-convexity subequation THEOREM. (Harvey+L.) Let F be a subequation with monotonicity cone Pp . Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 24 / 36 Removable Singularities – Example Fix 1 ≤ p ≤ n and let Pp ≡ {A : λ1 (A) + · · · + λ[p] (A) + (p − [p])λp+1 (A) ≥ 0} The p-convexity subequation THEOREM. (Harvey+L.) Let F be a subequation with monotonicity cone Pp . Then every closed set E ⊂ X of locally finite Hausdorff (p − 2)-measure is removable for F -subharmonics and F -harmonics as above. Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 24 / 36 Removable Singularities – Example Fix 1 ≤ p ≤ n and let Pp ≡ {A : λ1 (A) + · · · + λ[p] (A) + (p − [p])λp+1 (A) ≥ 0} The p-convexity subequation THEOREM. (Harvey+L.) Let F be a subequation with monotonicity cone Pp . Then every closed set E ⊂ X of locally finite Hausdorff (p − 2)-measure is removable for F -subharmonics and F -harmonics as above. The proof uses Riesz potentials ψ ≡ Kp ∗ µ Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 24 / 36 Removable Singularities – Example Fix 1 ≤ p ≤ n and let Pp ≡ {A : λ1 (A) + · · · + λ[p] (A) + (p − [p])λp+1 (A) ≥ 0} The p-convexity subequation THEOREM. (Harvey+L.) Let F be a subequation with monotonicity cone Pp . Then every closed set E ⊂ X of locally finite Hausdorff (p − 2)-measure is removable for F -subharmonics and F -harmonics as above. The proof uses Riesz potentials ψ ≡ Kp ∗ µ 1 − |x|p−2 2 < p ≤ n − 1 Kp (x) ≡ log|x| p=2 2−p |x| 1≤p<2 Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 24 / 36 Removable Singularities – Riesz Characteristic Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 25 / 36 Removable Singularities – Riesz Characteristic Let F ⊂ Sym2 (Rn ) be a pure second-order c.c. subequation. Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 25 / 36 Removable Singularities – Riesz Characteristic Let F ⊂ Sym2 (Rn ) be a pure second-order c.c. subequation. Definition. The Riesz characteristic of F is the number αF ≡ sup{α : I − αPe ∈ F ∀ e} where Pe is orthogonal projection onto the e-line. Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 25 / 36 Removable Singularities – Riesz Characteristic Let F ⊂ Sym2 (Rn ) be a pure second-order c.c. subequation. Definition. The Riesz characteristic of F is the number αF ≡ sup{α : I − αPe ∈ F ∀ e} where Pe is orthogonal projection onto the e-line. THEOREM. (Harvey+L.) Suppose F ⊂ Sym2 (Rn ) has Riesz characteristic α. Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 25 / 36 Removable Singularities – Riesz Characteristic Let F ⊂ Sym2 (Rn ) be a pure second-order c.c. subequation. Definition. The Riesz characteristic of F is the number αF ≡ sup{α : I − αPe ∈ F ∀ e} where Pe is orthogonal projection onto the e-line. THEOREM. (Harvey+L.) Suppose F ⊂ Sym2 (Rn ) has Riesz characteristic α. Then every closed set E ⊂ X of locally finite Hausdorff (α − 2)-measure is removable for F -subharmonics and F -harmonics as above. Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 25 / 36 Example – Elementary Symmetric Functions of D 2 u Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 26 / 36 Example – Elementary Symmetric Functions of D 2 u Let Fk ≡ {A ∈ Sym2 (Rn ) : σ1 (A) ≥ 0, ..., σk (A) ≥ 0} Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 26 / 36 Example – Elementary Symmetric Functions of D 2 u Let Fk ≡ {A ∈ Sym2 (Rn ) : σ1 (A) ≥ 0, ..., σk (A) ≥ 0} Fk has Riesz characteristic Blaine Lawson A Geometric Perspective on Nonlinear PDE’s n k July 11, 2013 26 / 36 Example – Elementary Symmetric Functions of D 2 u Let Fk ≡ {A ∈ Sym2 (Rn ) : σ1 (A) ≥ 0, ..., σk (A) ≥ 0} Fk has Riesz characteristic n k THEOREM. Every closed E ⊂ X of locally finite Hausdorff ( kn − 2)-measure is removable for Fk -subharmonics and Fk -harmonics. Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 26 / 36 Another Example – Branches of the Complex Monge-Ampère Equation Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 27 / 36 Another Example – Branches of the Complex Monge-Ampère Equation In Cn PkC ≡ {A ∈ Sym2 (R2n ) : λk (AC ) ≥ 0} Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 27 / 36 Another Example – Branches of the Complex Monge-Ampère Equation In Cn PkC ≡ {A ∈ Sym2 (R2n ) : λk (AC ) ≥ 0} A set E ⊂ Cn is pluripolar if E = {u = −∞} for some plurisubharmonic function u. Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 27 / 36 Another Example – Branches of the Complex Monge-Ampère Equation In Cn PkC ≡ {A ∈ Sym2 (R2n ) : λk (AC ) ≥ 0} A set E ⊂ Cn is pluripolar if E = {u = −∞} for some plurisubharmonic function u. THEOREM. Any pluirpolar set is removable for all branches of the complex Monge-Ampère equation. Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 27 / 36 Differential Equations on Manifolds. Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 28 / 36 Differential Equations on Manifolds. Let X be any manifold. Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 28 / 36 Differential Equations on Manifolds. Let X be any manifold. Definition. The 2-jet bundle of X is the vector bundle J 2 (X ) −→ X whose fibre at x ∈ X is ∞ Jx2 (X ) ≡ Cx∞ /Cx,3 ∞ = germs of functions which vanish to order 3 at x. where Cx,3 Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 28 / 36 Differential Equations on Manifolds. Let X be any manifold. Definition. The 2-jet bundle of X is the vector bundle J 2 (X ) −→ X whose fibre at x ∈ X is ∞ Jx2 (X ) ≡ Cx∞ /Cx,3 ∞ = germs of functions which vanish to order 3 at x. where Cx,3 There is a short exact sequence 0 → Sym2 (T∗ X) → J2 (X) → J1 (X) → 0. Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 28 / 36 All of our Definitions Transfer to this Setting. Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 29 / 36 All of our Definitions Transfer to this Setting. Definition. A second-order subequation on X is a closed subset F ⊂ J 2 (X ) Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 29 / 36 All of our Definitions Transfer to this Setting. Definition. A second-order subequation on X is a closed subset F ⊂ J 2 (X ) satisfying the positivity, negativity and topological conditions above: F +P ⊂ F Blaine Lawson F +N ⊂ F A Geometric Perspective on Nonlinear PDE’s July 11, 2013 29 / 36 All of our Definitions Transfer to this Setting. Definition. A second-order subequation on X is a closed subset F ⊂ J 2 (X ) satisfying the positivity, negativity and topological conditions above: F +P ⊂ F F +N ⊂ F and (i) F = IntF Blaine Lawson (ii) Fx = Intx Fx (iii) Intx F = (IntF ) ∩ Fx A Geometric Perspective on Nonlinear PDE’s July 11, 2013 29 / 36 All of our Definitions Transfer to this Setting. Definition. A second-order subequation on X is a closed subset F ⊂ J 2 (X ) satisfying the positivity, negativity and topological conditions above: F +P ⊂ F F +N ⊂ F and (i) F = IntF (ii) Fx = Intx Fx (iii) Intx F = (IntF ) ∩ Fx e , F -harmonic, etc. as before. Concepts of F -subharmonic, F Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 29 / 36 The Riemannian Hessian gives a Splitting of the Sequence 0 → Sym2 (T ∗ X ) → J 2 (X ) → J 1 (X ) → 0. Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 30 / 36 The Riemannian Hessian gives a Splitting of the Sequence 0 → Sym2 (T ∗ X ) → J 2 (X ) → J 1 (X ) → 0. so that J 2 (X ) = R ⊕ T ∗ X ⊕ Sym2 (T ∗ X ) Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 30 / 36 Jet-Equivalence. Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 31 / 36 Jet-Equivalence. Definition. An automorphism of J 2 (X ) is a bundle isomorphism Φ : J 2 (X ) −→ J 2 (X ) which respects the exact sequence 0 → Sym2 (T ∗ X ) → J 2 (X ) → R ⊕ T ∗ X → 0. Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 31 / 36 Jet-Equivalence. Definition. An automorphism of J 2 (X ) is a bundle isomorphism Φ : J 2 (X ) −→ J 2 (X ) which respects the exact sequence 0 → Sym2 (T ∗ X ) → J 2 (X ) → R ⊕ T ∗ X → 0. and there are bundle isomorphisms g : T ∗X → T ∗X Blaine Lawson and h : T ∗X → T ∗X A Geometric Perspective on Nonlinear PDE’s July 11, 2013 31 / 36 Jet-Equivalence. Definition. An automorphism of J 2 (X ) is a bundle isomorphism Φ : J 2 (X ) −→ J 2 (X ) which respects the exact sequence 0 → Sym2 (T ∗ X ) → J 2 (X ) → R ⊕ T ∗ X → 0. and there are bundle isomorphisms g : T ∗X → T ∗X and h : T ∗X → T ∗X so the restriction of Φ to Sym2 (T ∗ X ) is Φ(A) = gAg t Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 31 / 36 Jet-Equivalence. Definition. An automorphism of J 2 (X ) is a bundle isomorphism Φ : J 2 (X ) −→ J 2 (X ) which respects the exact sequence 0 → Sym2 (T ∗ X ) → J 2 (X ) → R ⊕ T ∗ X → 0. and there are bundle isomorphisms g : T ∗X → T ∗X and h : T ∗X → T ∗X so the restriction of Φ to Sym2 (T ∗ X ) is Φ(A) = gAg t and the induced map on R ⊕ T ∗ X is Id⊕h. Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 31 / 36 Jet-Equivalence. Definition. An automorphism of J 2 (X ) is a bundle isomorphism Φ : J 2 (X ) −→ J 2 (X ) which respects the exact sequence 0 → Sym2 (T ∗ X ) → J 2 (X ) → R ⊕ T ∗ X → 0. and there are bundle isomorphisms g : T ∗X → T ∗X and h : T ∗X → T ∗X so the restriction of Φ to Sym2 (T ∗ X ) is Φ(A) = gAg t and the induced map on R ⊕ T ∗ X is Id⊕h. An affine automorphism is Ψ = J + Φ where Φ is an automorphism and J is a section of J 2 (X ). Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 31 / 36 Jet-Equivalence. Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 32 / 36 Jet-Equivalence. Definition. Two subequations F , F 0 are affinely jet-equivalent if there exists an affine automorphism Ψ : J 2 (X ) → J 2 (X ) such that Φ(F ) = F 0 Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 32 / 36 The Dirichlet Problem. Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 33 / 36 The Dirichlet Problem. Let F be a subequation on a manifold X . Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 33 / 36 The Dirichlet Problem. Let F be a subequation on a manifold X . Suppose there exists some strictly M-subharmonic function on X where M is a monotonicity cone for F . Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 33 / 36 The Dirichlet Problem. Let F be a subequation on a manifold X . Suppose there exists some strictly M-subharmonic function on X where M is a monotonicity cone for F . THEOREM. (Harvey+L.) Suppose F is locally affinely jet-equivalent to a constant coefficient subequation. Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 33 / 36 The Dirichlet Problem. Let F be a subequation on a manifold X . Suppose there exists some strictly M-subharmonic function on X where M is a monotonicity cone for F . THEOREM. (Harvey+L.) Suppose F is locally affinely jet-equivalent to a constant coefficient subequation. Then for every domain Ω ⊂⊂ X whose e convex, boundary is strictly F and F Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 33 / 36 The Dirichlet Problem. Let F be a subequation on a manifold X . Suppose there exists some strictly M-subharmonic function on X where M is a monotonicity cone for F . THEOREM. (Harvey+L.) Suppose F is locally affinely jet-equivalent to a constant coefficient subequation. Then for every domain Ω ⊂⊂ X whose e convex, the Dirichlet problem for F -harmonic boundary is strictly F and F functions is uniquely solvable for all continuous boundary data ϕ ∈ ∂Ω. Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 33 / 36 Example: Manifolds with G-Structure. Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 34 / 36 Example: Manifolds with G-Structure. Fix a subgroup G ⊂ O(n) Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 34 / 36 Example: Manifolds with G-Structure. Fix a subgroup G ⊂ O(n) and let F ⊂ J2 = R × Rn × Sym2 (Rn ) be a constant coefficient subequation which is G-invariant. Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 34 / 36 Example: Manifolds with G-Structure. Fix a subgroup G ⊂ O(n) and let F ⊂ J2 = R × Rn × Sym2 (Rn ) be a constant coefficient subequation which is G-invariant. Then F determines a subequation F on every riemannian manifold with G-structure (not necessarily integrable). Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 34 / 36 Example: Manifolds with G-Structure. Fix a subgroup G ⊂ O(n) and let F ⊂ J2 = R × Rn × Sym2 (Rn ) be a constant coefficient subequation which is G-invariant. Then F determines a subequation F on every riemannian manifold with G-structure (not necessarily integrable). Furthermore, each such F is locally jet-equivalent to the constant coefficient equation F. Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 34 / 36 Additional Applications. Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 35 / 36 Additional Applications. Jet Equivalence allows one to: • Solve the Dirichlet Problem for inhomogeneous equations, e.g., λk (Hess u) = f (x). Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 35 / 36 Additional Applications. Jet Equivalence allows one to: • Solve the Dirichlet Problem for inhomogeneous equations, e.g., λk (Hess u) = f (x). • Solve the Dirichlet Problem for parabolic equations. Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 35 / 36 Additional Applications. Jet Equivalence allows one to: • Solve the Dirichlet Problem for inhomogeneous equations, e.g., λk (Hess u) = f (x). • Solve the Dirichlet Problem for parabolic equations. • Solve the Dirichlet Problem for equations with obstacles. Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 35 / 36 HAPPY BIRTHDAY ROBERT!! Blaine Lawson A Geometric Perspective on Nonlinear PDE’s July 11, 2013 36 / 36