Metamorphoses for Pattern Matching
|
|
- Tiffany Cole
- 5 years ago
- Views:
Transcription
1 Metamorphoses for Pattern Matching Laurent Younes Joint work with Darryl Holm and Alain Trouvé Johns Hopkins University July Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
2 Pattern Matching The issue addressed in Diffeomorphic Metric Matching is the comparison of deformable structures via the action of diffeomorphisms. Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
3 Pattern Matching The issue addressed in Diffeomorphic Metric Matching is the comparison of deformable structures via the action of diffeomorphisms. In particular, important problems are: Finding correspondences between images, surfaces, curves... Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
4 Pattern Matching The issue addressed in Diffeomorphic Metric Matching is the comparison of deformable structures via the action of diffeomorphisms. In particular, important problems are: Finding correspondences between images, surfaces, curves... Comparing deformable structures (with distances). Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
5 Pattern Matching The issue addressed in Diffeomorphic Metric Matching is the comparison of deformable structures via the action of diffeomorphisms. In particular, important problems are: Finding correspondences between images, surfaces, curves... Comparing deformable structures (with distances). Building Riemannian manifolds on shapes. Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
6 Pattern Matching The issue addressed in Diffeomorphic Metric Matching is the comparison of deformable structures via the action of diffeomorphisms. In particular, important problems are: Finding correspondences between images, surfaces, curves... Comparing deformable structures (with distances). Building Riemannian manifolds on shapes. Important applications in medical imaging (computational anatomy). Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
7 Deformable templates The deformable template paradigm (U. Grenander) can be transcribed in out context as follows: Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
8 Deformable templates The deformable template paradigm (U. Grenander) can be transcribed in out context as follows: 1. Equip the group of diffeomorphisms with a right-invariant Riemannian metric. Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
9 Deformable templates The deformable template paradigm (U. Grenander) can be transcribed in out context as follows: 1. Equip the group of diffeomorphisms with a right-invariant Riemannian metric. 2. Assume that an object n temp is given (the template)and consider the orbit N = Diff.n temp. Equip the orbit with the Riemannian projection of the metric on Diff. Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
10 Optimal matching with deformable templates Geodesics provide optimal correspondences: with fixed n 0 and n 1, minimize 1 u t 2 dt with constraints ṅ t u t.n t = 0. 0 Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
11 Optimal matching with deformable templates Geodesics provide optimal correspondences: with fixed n 0 and n 1, minimize 1 u t 2 dt with constraints ṅ t u t.n t = 0. Problem: restricted to an orbit. 0 Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
12 Optimal matching with deformable templates Geodesics provide optimal correspondences: with fixed n 0 and n 1, minimize 1 u t 2 dt with constraints ṅ t u t.n t = 0. Problem: restricted to an orbit. Inexact matching: minimize with ġ t u t g t = u t 2 dt + Error(n 1, g 1.n 0 ) Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
13 Optimal matching with deformable templates Geodesics provide optimal correspondences: with fixed n 0 and n 1, minimize 1 u t 2 dt with constraints ṅ t u t.n t = 0. Problem: restricted to an orbit. Inexact matching: minimize with ġ t u t g t = 0. 1 But the metric aspect is lost. 0 0 u t 2 dt + Error(n 1, g 1.n 0 ) Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
14 Metamorphoses Definition Let N be a manifold acted upon by a Lie group G. A metamorphosis is a pair of curves (g t, η t ) G N parameterized by t, with g 0 = id G. Its image is the curve n t N defined by the action n t = g t.η t. Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
15 Invariant Lagrangian minimization Given n 0 and n 1, we will optimize, over metamorphoses (g t, η t ), the integrated Lagrangian 1 0 L(g t, ġ t, η t, η t )dt with end-point conditions g 0 = id G, η 0 = n 0 and g 1 η 1 = n 1. Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
16 Let g denote the Lie algebra of G. We assume that L is invariant by the right action of G on G N defined by (g, η)h = (gh, h 1 η). Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
17 Let g denote the Lie algebra of G. We assume that L is invariant by the right action of G on G N defined by (g, η)h = (gh, h 1 η). Thus, there exists a function l defined on g TN such that L(g, U g, η, ξ η ) = l(u g g 1, gη, gξ η ). Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
18 Let g denote the Lie algebra of G. We assume that L is invariant by the right action of G on G N defined by (g, η)h = (gh, h 1 η). Thus, there exists a function l defined on g TN such that L(g, U g, η, ξ η ) = l(u g g 1, gη, gξ η ). For a metamorphosis (g t, η t ), let u t = ġ t gt 1, n t = g t η t and ν t = g t η t, We have L(g t, ġ t, η t, η t ) = l(u t, n t, ν t ). Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
19 Notation The pairing between a linear form p and a vector u is denoted ( p u ). Duality with respect to this pairing is denoted with a exponent: ( p Au ) = ( A p u ). When G acts on a manifold Ñ, define the operator on T Ñ Ñ, taking values in g by ( δ ñ u ) = ( δ uñ ). Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
20 Euler Equations Compute the first variation for 1 0 l(u t, n t, ν t )dt with fixed boundary conditions n 0 and n 1. Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
21 Euler Equations Compute the first variation for 1 0 l(u t, n t, ν t )dt with fixed boundary conditions n 0 and n 1. Consider a variation δu and ω = δn. From n = gη and ν = g η we get ṅ = ν + un and ω = δν + uω + δun. (Derivatives are assumed to be taken in a chart.) Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
22 Euler Equations Compute the first variation for 1 0 l(u t, n t, ν t )dt with fixed boundary conditions n 0 and n 1. Consider a variation δu and ω = δn. From n = gη and ν = g η we get ṅ = ν + un and ω = δν + uω + δun. (Derivatives are assumed to be taken in a chart.) Write 1 0 ( ( δl ) ( δl ) ( δl )) δu δu + δn ω + ω uω δu n dt = 0. δν Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
23 From the δu term δl δu + δl δν n = 0. Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
24 From the δu term From the ω term: with δl δu + δl δν n = 0. δl t δν + u δl δν δl δn = 0 ( δl δν uω) = ( u δl δν ω). Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
25 We therefore obtain the system of equations δl δu + δl δν n = 0 δl t δν + u δl δν = δl δn ṅ = ν + un Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
26 We therefore obtain the system of equations δl δu + δl δν n = 0 δl t δν + u δl δν = δl δn ṅ = ν + un Note that δl δν n associated to the invariance of the Lagrangian. The special form of the boundary conditions ensure that this momenum is zero. δu + δl Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
27 Euler-Poincaré reduction Make the variations ξ = δgg 1 and ϖ = gδη. Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
28 Euler-Poincaré reduction Make the variations ξ = δgg 1 and ϖ = gδη. We have δu = ξ + [ξ, u] (EP reduction on the group, cf. Holm, Marsden, Ratiu). Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
29 Euler-Poincaré reduction Make the variations ξ = δgg 1 and ϖ = gδη. We have δu = ξ + [ξ, u] (EP reduction on the group, cf. Holm, Marsden, Ratiu). We also have δn = δ(gη) = ϖ + ξn. Using ν = g η, yielding δν = gδ η + ξν and ϖ = gδη, yielding ϖ = uϖ + g η. we get δν = ϖ + ξν uϖ. Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
30 Euler-Poincaré reduction Make the variations ξ = δgg 1 and ϖ = gδη. We have δu = ξ + [ξ, u] (EP reduction on the group, cf. Holm, Marsden, Ratiu). We also have δn = δ(gη) = ϖ + ξn. Using ν = g η, yielding δν = gδ η + ξν and ϖ = gδη, yielding ϖ = uϖ + g η. we get δν = ϖ + ξν uϖ. At t = 0, g 0 = id and n 0 = g 0 η 0 = cst so that ξ 0 = 0 and ϖ 0 = 0. Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
31 Euler-Poincaré reduction Make the variations ξ = δgg 1 and ϖ = gδη. We have δu = ξ + [ξ, u] (EP reduction on the group, cf. Holm, Marsden, Ratiu). We also have δn = δ(gη) = ϖ + ξn. Using ν = g η, yielding δν = gδ η + ξν and ϖ = gδη, yielding ϖ = uϖ + g η. we get δν = ϖ + ξν uϖ. At t = 0, g 0 = id and n 0 = g 0 η 0 = cst so that ξ 0 = 0 and ϖ 0 = 0. At t = 1, g 1 η 1 = cst so that ξ 1 n 1 + ω 1 = 0. Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
32 Write 1 ( ( δl δu ξ ) ( δl ) ( δl )) ad u ξ + δn ϖ + ξn + ϖ + ξν uϖ dt = 0 δν 0 Using integration by parts we get the boundary equation ( ) ( ) δl δl + n 1 = 0. δu δν 1 1 Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
33 Evolution from ξ: δl t δu + δl ad u δu + δl δn n + δl δν ν = 0 Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
34 Evolution from ξ: δl t δu + δl ad u δu + δl δn n + δl δν ν = 0 Evolution from ϖ: δl t δν + u δl δν δl δn = 0. Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
35 EPMorph We therefore obtain the system δl t δu + δl ad u δu + δl δn n + δl δν ν = 0 δl t δν + u δl δν δl δn = 0 ) ) ( δl δu 1 + ṅ = ν + un ( δl δν 1 n 1 = 0 Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
36 Special case 1: Riemannian metric Define an invariant Riemannian metric on G N, yielding l(u, n, ν) = (u, ν) 2 n. Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
37 Special case 1: Riemannian metric Define an invariant Riemannian metric on G N, yielding l(u, n, ν) = (u, ν) 2 n. For example l(u, n, ν) = u 2 g + ν 2 n, for a given norm,. g, on g and a pre-existing Riemannian structure on N (Trouvé, Y.). Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
38 Special case 1: Riemannian metric Define an invariant Riemannian metric on G N, yielding l(u, n, ν) = (u, ν) 2 n. For example l(u, n, ν) = u 2 g + ν 2 n, for a given norm,. g, on g and a pre-existing Riemannian structure on N (Trouvé, Y.). This in turn projects into a Riemannian metric on N. Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
39 Special Case 2: Semi-direct product Assume that N is also a group and that g(nñ) = (gn)(gñ) (e.g., N is a vector space and the action of G is linear). Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
40 Special Case 2: Semi-direct product Assume that N is also a group and that g(nñ) = (gn)(gñ) (e.g., N is a vector space and the action of G is linear). Consider the semi-direct product G SN with (g, n)( g, ñ) = (g g, (gñ)n) with a right-invariant metric specified by (idg,id N ) at the identity. Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
41 Special Case 2: Semi-direct product Assume that N is also a group and that g(nñ) = (gn)(gñ) (e.g., N is a vector space and the action of G is linear). Consider the semi-direct product G SN with (g, n)( g, ñ) = (g g, (gñ)n) with a right-invariant metric specified by (idg,id N ) at the identity. Minimize the geodesic energy between (id G, n 0 ) and (g 1, n 1 ) with fixed n 0 and n 1. Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
42 Special Case 2: Semi-direct product Assume that N is also a group and that g(nñ) = (gn)(gñ) (e.g., N is a vector space and the action of G is linear). Consider the semi-direct product G SN with (g, n)( g, ñ) = (g g, (gñ)n) with a right-invariant metric specified by (idg,id N ) at the identity. Minimize the geodesic energy between (id G, n 0 ) and (g 1, n 1 ) with fixed n 0 and n 1. Proposition: This is metamorphosis with reduced Lagrangian l(u, n, ν) = (u, n 1 ν) 2 (id G,id N ). Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
43 Note that the evolution equations in this case correspond to the full conservation of momentum on the semi-direct product: ( δl δl ) (t), δu δν (t) ( δl = Ad(g,n) δl ) (t), 1 δu δν (t) Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
44 Note that the evolution equations in this case correspond to the full conservation of momentum on the semi-direct product: ( δl δl ) (t), δu δν (t) ( δl = Ad(g,n) δl ) (t), 1 δu δν (t) When N is a vector space, the Lagrangian does ot depend on n and the equations are δl δu + δl δν n = 0 δl t δν + u δl δν = 0 ṅ = ν + un Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
45 From now on, G is a group of diffeomorphisms over some open subset Ω R d. g is equipped with a Hilbert norm u t g with a continuous inclusion in C p 0 (Ω) (C p vector fields vanishing at infinity) with the supremum norm and p 1. Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
46 From now on, G is a group of diffeomorphisms over some open subset Ω R d. g is equipped with a Hilbert norm u t g with a continuous inclusion in C p 0 (Ω) (C p vector fields vanishing at infinity) with the supremum norm and p 1. Under this assumption, the flows associated to time-dependent v.f. that satisfy 1 0 u t 2 gdt < for a subgroup of Diff(Ω) (Trouvé; Dupuis et al.). Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
47 Notation We will write the inner product in g under the form u, v m g = ( L g u v ) where L g is the duality operator from g to g. Its inverse, a kernel operator, is denoted K g. Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
48 Landmarks and Peakons Take the space of configurations of Q landmarks N = Ω Q. Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
49 Landmarks and Peakons Take the space of configurations of Q landmarks N = Ω Q. Elements η, ν N are Q-tuples of points in Ω, with tangent vectors being Q-tuples of d-dimensional vectors. Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
50 Landmarks and Peakons Take the space of configurations of Q landmarks N = Ω Q. Elements η, ν N are Q-tuples of points in Ω, with tangent vectors being Q-tuples of d-dimensional vectors. Take l(u, n, ν) = u 2 g + 1 σ 2 Q k=1 ν k 2. Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
51 Landmarks and Peakons Take the space of configurations of Q landmarks N = Ω Q. Elements η, ν N are Q-tuples of points in Ω, with tangent vectors being Q-tuples of d-dimensional vectors. Take l(u, n, ν) = u 2 g + 1 σ 2 Q k=1 ν k 2. We have δl/δu = 2L g u, and (δl/δν) = (2/σ 2 )(ν 1,..., ν N ). Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
52 Let n = (q 1,..., q N ). We have δl Q δν n = 2 ν k σ 2 δ q k. k=1 Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
53 Let n = (q 1,..., q N ). We have δl Q δν n = 2 ν k σ 2 δ q k. k=1 Notation: if f is a vector field on R d and µ a measure on R d, ( f µ w ) = R d f (x) T w(x)dµ. (1) Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
54 Euler equations for landmark metamorphosis (with p k = ν k /σ 2 ) Q L g u = p k δ qk. k=1 dp k dt + Du(q k) T p k = 0, q k = u(q k ) + σ 2 p k, k = 1,..., Q k = 1,..., Q The case σ 2 = 0 gives the singular solutions to EPDiff. Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
55 EPMorph/EPDiff Comparison with two 1-D peakons p= p= p= p= p= p= p= p= p= p= Figure: Head-on collision between two peakons. The plots show the evolution of q = q 2 q 1 over time for several values of p = p 2 p 1, for σ 2 = 0 (left) and σ 2 = 10 4 (right) Rem: Many simulations for peakons with EPdiff, e.g.: Holm and Staley, McLachlan and Marsland... Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
56 Head-on collisions Head-on collision of two peakons; left: EPDIFF; right: metamorphoses Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
57 p= p= p= p= p= p= p= p= p= p= Figure: One peakon overtaking another. The plots show the evolution of q = q 2 q 1 over time for several values of p = p 2 p 1, for σ 2 = 0 (left) and σ 2 = 0.05 (right) Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
58 Back Collision Left: EPDIFF; center: metamorphoses, low difference of energy; right: metamorphoses, high difference of energy. Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
59 Image Metamorphosis Let H be a space of smooth functions from Ω to R, with g.n = n g 1. Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
60 Image Metamorphosis Let H be a space of smooth functions from Ω to R, with g.n = n g 1. Take l(u, ν) = u 2 g + 1 σ 2 ν 2 L 2. Denote z = ν/σ 2. The Euler equations are L g u = z n ż + div(zu) = 0 ṅ + n T u = σ 2 z Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
61 Remark: H 1 α in 1-D Let m = L g u = (1 2 x )u, and ρ = σz: t m + u x m + 2m x u = ρ x ρ with t ρ + x (ρu) = 0 This is equivalent to the compatibility for dλ/dt = 0 of ( x 2 ψ mλ + ρ2 λ 2) ψ = 0 ( 1 ) t ψ = 2λ + u x ψ ψ xu Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
62 This model of image metamorphoses is a semi-direct product model. The conserved momentum is (L g u, z), yielding the equations L g u t + z t n t = cst and z t = det(dgt 1 )z 0 gt 1. Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
63 Examples Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
64 Examples Interpolation between two face images (different persons). Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
65 Densities Use the action g.n = det D(g 1 ) n g 1. Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
66 Densities Use the action g.n = det D(g 1 ) n g 1. Reduced Lagrangian: l(u, ν) = 1 2 u 2 g + 1 2σ 2 ν 2 L 2. Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
67 Densities Use the action g.n = det D(g 1 ) n g 1. Reduced Lagrangian: l(u, ν) = 1 2 u 2 g + 1 2σ 2 ν 2 L 2. The equations are L g u = n z ż + z T u = 0 ṅ + div(nu) = σ 2 z Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
68 The conservation of momentum gives L g u + n z = cst and z = z 0 g 1. The constant in the first equation vanishes for horizontal geodesics in GSN/G. Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
69 Comparing measures In many instances, deformable objects are singular and should be compared as such. Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
70 Comparing measures In many instances, deformable objects are singular and should be compared as such. The simplest example is the case of unlabelled discrete point sets. Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
71 Comparing measures In many instances, deformable objects are singular and should be compared as such. The simplest example is the case of unlabelled discrete point sets. Such structures are conveniently represented as measures, e.g., (x i ) i I i I δ xi. Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
72 Comparing measures In many instances, deformable objects are singular and should be compared as such. The simplest example is the case of unlabelled discrete point sets. Such structures are conveniently represented as measures, e.g., (x i ) i I i I δ xi. To apply (Riemannian) metamorphoses to measures, one needs to embed them in a Hilbert space. Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
73 Measure Metamorphoses Let H be a Hilbert space of functions over R d and N = H. Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
74 Measure Metamorphoses Let H be a Hilbert space of functions over R d and N = H. We assume that H C 0 so that for all x R d, K x H such that, for all f H, Kx, f H = f (x). Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
75 Measure Metamorphoses Let H be a Hilbert space of functions over R d and N = H. We assume that H C 0 so that for all x R d, K x H such that, for all f H, Kx, f H = f (x). The kernel K H (x, y) := K x (y) satisfies the equation Kx, K y H = K H(x, y). Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
76 K H provides an isometry between N and H via the relation η K H η with K H η, f H = ( η f ). In particular, the dual inner product on N is given by η, η N = ( η K H η ). Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
77 Define the action of G Diff on N by ( gη f ) = ( η f g ), with g G, η N and f H. Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
78 Define the action of G Diff on N by ( gη f ) = ( η f g ), with g G, η N and f H. For η N and w g, and for all f H, ( wη f ) = ( η f T w ). Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
79 Define the action of G Diff on N by ( gη f ) = ( η f g ), with g G, η N and f H. For η N and w g, and for all f H, ( wη f ) = ( η f T w ). We use the semi-direct product model and define the Lagrangian l(u, ν) = 1 2 u 2 g + 1 2σ 2 ν 2 N. Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
80 E-L Equations Let f = δl/δν = (1/σ 2 )K H ν The E-L Equations are: L g u = f u ḟ t + f T u = 0 ṅ u.n = σ 2 L H f Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
81 Integrated form In integrated form, this yields u = K g ( f n) with ġ t = u t g t. n t = g t n 0 + σ 2 g t t 0 g 1 s K 1 H (f 0 g 1 )ds s Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
82 Initial Value Problem Theorem Under some conditions on H, f 0, n 0, for all T > 0, there exists a unique solution to the previous system over [0, T ] with initial conditions n 0 and f 0. (The conditions essentially imply the loss of one derivative for n 0 and f 0 : if H H p, then f 0 H p+1 and n 0 H p+1.) Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
83 Boundary Value Problem The BVP requires to minimize, with fixed n 0 and n 1 E(u, n) := 1 0 u t 2 gdt σ 2 ṅ t u t n t 2 N dt. Theorem Under some conditions on H, for any given n 0, n 1 H, there exists a minimizer for the BVP. 0 Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
84 Open issues Numerical schemes for measure metamorphoses. How smooth are geodesics? Extension of VPM methods to EPMorph? Happy Birthday Darryl ( ) Metamorphoses for Pattern Matching July / 43
SINGULAR SOLITONS, MOMENTUM MAPS & COMPUTATIONAL ANATOMY
SINGULAR SOLITONS, MOMENTUM MAPS & COMPUTATIONAL ANATOMY Darryl D. Holm CCS-Division @ Los Alamos National Laboratory Mathematics Department @ Imperial College London dholm@lanl.gov, d.holm@imperial.ac.uk
More informationThe momentum map representation of images
The momentum map representation of images M. Bruveris 1, F. Gay-Balmaz, D. D. Holm 1, and T. S. Ratiu 3 Revised version August 1, 1 arxiv:91.99v [nlin.cd] 8 Apr 15 Abstract This paper discusses the mathematical
More informationarxiv: v1 [cs.cv] 20 Nov 2017
STOCHASTIC METAMORPHOSIS WITH TEMPLATE UNCERTAINTIES ALEXIS ARNAUDON, DARRYL D HOLM, AND STEFAN SOMMER arxiv:1711.7231v1 [cs.cv] 2 Nov 217 Abstract. In this paper, we investigate two stochastic perturbations
More informationRESEARCH REPORT. Stochastic metamorphosis with template uncertainties. Alexis Arnaudon, Darryl D. Holm and Stefan Sommer
CENTRE FOR STOCHASTIC GEOMETRY AND ADVANCED BIOIMAGING 217 www.csgb.dk RESEARCH REPORT Alexis Arnaudon, Darryl D. Holm and Stefan Sommer Stochastic metamorphosis with template uncertainties No. 13, November
More informationSub-Riemannian geometry in groups of diffeomorphisms and shape spaces
Sub-Riemannian geometry in groups of diffeomorphisms and shape spaces Sylvain Arguillère, Emmanuel Trélat (Paris 6), Alain Trouvé (ENS Cachan), Laurent May 2013 Plan 1 Sub-Riemannian geometry 2 Right-invariant
More informationGeodesic shooting on shape spaces
Geodesic shooting on shape spaces Alain Trouvé CMLA, Ecole Normale Supérieure de Cachan GDR MIA Paris, November 21 2014 Outline Riemannian manifolds (finite dimensional) Spaces spaces (intrinsic metrics)
More informationHAPPY BIRTHDAY TONY! Chicago Blochfest June 30, 2105
HAPPY BIRTHDAY TONY! 1 CLEBSCH OPTIMAL CONTROL Tudor S. Ratiu Section de Mathématiques Ecole Polytechnique Fédérale de Lausanne, Switzerland tudor.ratiu@epfl.ch 2 PLAN OF THE PRESENTATION Symmetric representation
More informationThe Euler-Lagrange Equation for Interpolating Sequence of Landmark Datasets
The Euler-Lagrange Equation for Interpolating Sequence of Landmark Datasets Mirza Faisal Beg 1, Michael Miller 1, Alain Trouvé 2, and Laurent Younes 3 1 Center for Imaging Science & Department of Biomedical
More informationA CRASH COURSE IN EULER-POINCARÉ REDUCTION
A CRASH COURSE IN EULER-POINCARÉ REDUCTION HENRY O. JACOBS Abstract. The following are lecture notes from lectures given at the Fields Institute during the Legacy of Jerrold E. Marsden workshop. In these
More informationIntegrable evolution equations on spaces of tensor densities
Integrable evolution equations on spaces of tensor densities J. Lenells, G. Misiolek and F. Tiglay* April 11, 21 Lenells, Misiolek, Tiglay* AMS Meeting, Minnesota, April 11, 21 slide 1/22 A family of equations
More informationarxiv: v3 [cs.cv] 20 Oct 2018
STRING METHODS FOR STOCHASTIC IMAGE AND SHAPE MATCHING ALEXIS ARNAUDON, DARRYL D HOLM, AND STEFAN SOMMER arxiv:1805.06038v3 [cs.cv] 20 Oct 2018 Abstract. Matching of images and analysis of shape differences
More informationQuantising noncompact Spin c -manifolds
Quantising noncompact Spin c -manifolds Peter Hochs University of Adelaide Workshop on Positive Curvature and Index Theory National University of Singapore, 20 November 2014 Peter Hochs (UoA) Noncompact
More informationBACKGROUND IN SYMPLECTIC GEOMETRY
BACKGROUND IN SYMPLECTIC GEOMETRY NILAY KUMAR Today I want to introduce some of the symplectic structure underlying classical mechanics. The key idea is actually quite old and in its various formulations
More informationLECTURE 28: VECTOR BUNDLES AND FIBER BUNDLES
LECTURE 28: VECTOR BUNDLES AND FIBER BUNDLES 1. Vector Bundles In general, smooth manifolds are very non-linear. However, there exist many smooth manifolds which admit very nice partial linear structures.
More informationSome aspects of the Geodesic flow
Some aspects of the Geodesic flow Pablo C. Abstract This is a presentation for Yael s course on Symplectic Geometry. We discuss here the context in which the geodesic flow can be understood using techniques
More informationConstrained shape spaces of infinite dimension
Constrained shape spaces of infinite dimension Sylvain Arguillère, Emmanuel Trélat (Paris 6), Alain Trouvé (ENS Cachan), Laurent May 2013 Goal: Adding constraints to shapes in order to better fit the modeled
More informationPart III Symmetries, Fields and Particles
Part III Symmetries, Fields and Particles Theorems Based on lectures by N. Dorey Notes taken by Dexter Chua Michaelmas 2016 These notes are not endorsed by the lecturers, and I have modified them (often
More informationTravel Time Tomography and Tensor Tomography, I
Travel Time Tomography and Tensor Tomography, I Plamen Stefanov Purdue University Mini Course, MSRI 2009 Plamen Stefanov (Purdue University ) Travel Time Tomography and Tensor Tomography, I 1 / 17 Alternative
More informationA new class of pseudodifferential operators with mixed homogenities
A new class of pseudodifferential operators with mixed homogenities Po-Lam Yung University of Oxford Jan 20, 2014 Introduction Given a smooth distribution of hyperplanes on R N (or more generally on a
More informationEuler-Poincaré reduction in principal bundles by a subgroup of the structure group
Euler-Poincaré reduction in principal bundles by a subgroup of the structure group M. Castrillón López ICMAT(CSIC-UAM-UC3M-UCM) Universidad Complutense de Madrid (Spain) Joint work with Pedro L. García,
More informationThe Averaged Fluid and EPDiff Equations: (A Deja Vu Lecture) Jerrold E. Marsden
C A L T E C H Control & Dynamical Systems The Averaged Fluid and EPDiff Equations: (A Deja Vu Lecture) Jerrold E. Marsden Control and Dynamical Systems, Caltech http://www.cds.caltech.edu/ marsden/ Boulder,
More informationRegistration of anatomical images using geodesic paths of diffeomorphisms parameterized with stationary vector fields
Registration of anatomical images using geodesic paths of diffeomorphisms parameterized with stationary vector fields Monica Hernandez, Matias N. Bossa, and Salvador Olmos Communication Technologies Group
More informationHyperbolic Geometry on Geometric Surfaces
Mathematics Seminar, 15 September 2010 Outline Introduction Hyperbolic geometry Abstract surfaces The hemisphere model as a geometric surface The Poincaré disk model as a geometric surface Conclusion Introduction
More informationDynamics of symplectic fluids and point vortices
Dynamics of symplectic fluids and point vortices Boris Khesin June 2011 In memory of Vladimir Igorevich Arnold Abstract We present the Hamiltonian formalism for the Euler equation of symplectic fluids,
More informationTemplate estimation form unlabeled point set data and surfaces for Computational Anatomy
Template estimation form unlabeled point set data and surfaces for Computational Anatomy Joan Glaunès 1 and Sarang Joshi 1 Center for Imaging Science, Johns Hopkins University, joan@cis.jhu.edu SCI, University
More informationGlobal aspects of Lorentzian manifolds with special holonomy
1/13 Global aspects of Lorentzian manifolds with special holonomy Thomas Leistner International Fall Workshop on Geometry and Physics Évora, September 2 5, 2013 Outline 2/13 1 Lorentzian holonomy Holonomy
More informationVirasoro hair on locally AdS 3 geometries
Virasoro hair on locally AdS 3 geometries Kavli Institute for Theoretical Physics China Institute of Theoretical Physics ICTS (USTC) arxiv: 1603.05272, M. M. Sheikh-Jabbari and H. Y Motivation Introduction
More informationPseudo-Poincaré Inequalities and Applications to Sobolev Inequalities
Pseudo-Poincaré Inequalities and Applications to Sobolev Inequalities Laurent Saloff-Coste Abstract Most smoothing procedures are via averaging. Pseudo-Poincaré inequalities give a basic L p -norm control
More informationDouble Layer Potentials on Polygons and Pseudodifferential Operators on Lie Groupoids
Double Layer Potentials on Polygons and Pseudodifferential Operators on Lie Groupoids joint work with Hengguang Li Yu Qiao School of Mathematics and Information Science Shaanxi Normal University Xi an,
More informationCocycles and stream functions in quasigeostrophic motion
Journal of Nonlinear Mathematical Physics Volume 15, Number 2 (2008), 140 146 Letter Cocycles and stream functions in quasigeostrophic motion Cornelia VIZMAN West University of Timişoara, Romania E-mail:
More informationarxiv: v1 [math.oc] 19 Mar 2010
SHAPE SPLINES AND STOCHASTIC SHAPE EVOLUTIONS: A SECOND ORDER POINT OF VIEW ALAIN TROUVÉ AND FRANÇOIS-XAVIER VIALARD. arxiv:13.3895v1 [math.oc] 19 Mar 21 Contents 1. Introduction 1 2. Hamiltonian equations
More informationQUALIFYING EXAMINATION Harvard University Department of Mathematics Tuesday September 21, 2004 (Day 1)
QUALIFYING EXAMINATION Harvard University Department of Mathematics Tuesday September 21, 2004 (Day 1) Each of the six questions is worth 10 points. 1) Let H be a (real or complex) Hilbert space. We say
More informationSYMPLECTIC GEOMETRY: LECTURE 5
SYMPLECTIC GEOMETRY: LECTURE 5 LIAT KESSLER Let (M, ω) be a connected compact symplectic manifold, T a torus, T M M a Hamiltonian action of T on M, and Φ: M t the assoaciated moment map. Theorem 0.1 (The
More informationGeodesic Warps by Conformal Mappings
DOI 1.17/s11263-12-584-x Geodesic Warps by Conformal Mappings Stephen Marsland Robert I. McLachlan Klas Modin Matthew Perlmutter Received: 2 December 211 / Accepted: 2 October 212 Springer Science+Business
More informationWITTEN HELLFER SJÖSTRAND THEORY. 1.DeRham Hodge Theorem. 2.WHS-Theorem. 3.Mathematics behind WHS-Theorem. 4.WHS-Theorem in the presence of.
WITTEN HELLFER SJÖSTRAND THEORY 1.DeRham Hodge Theorem 2.WHS-Theorem 3.Mathematics behind WHS-Theorem 4.WHS-Theorem in the presence of symmetry 1 2 Notations M closed smooth manifold, (V, V ) euclidean
More informationGeodesic Warps by Conformal Mappings
!"#$%&'()*!"#$%&'()(&*+&,+-."+/,&0/.1$)#*&$+.+)/"*(&!"#$%&'()(&*+&,#(-&"#.%(/&0(1()(.$( 1 1 1 1 1 1 0 1 0 1 0 1 Geodesic Warps by Conformal Mappings Stephen Marsland 1, Robert I McLachlan, Klas Modin,
More informationPDiff: Irrotational Diffeomorphisms for Computational Anatomy
PDiff: Irrotational Diffeomorphisms for Computational Anatomy Jacob Hinkle and Sarang Joshi Scientific Computing and Imaging Institute, Department of Bioengineering, University of Utah, Salt Lake City,
More information1 Hermitian symmetric spaces: examples and basic properties
Contents 1 Hermitian symmetric spaces: examples and basic properties 1 1.1 Almost complex manifolds............................................ 1 1.2 Hermitian manifolds................................................
More informationExercises in Geometry II University of Bonn, Summer semester 2015 Professor: Prof. Christian Blohmann Assistant: Saskia Voss Sheet 1
Assistant: Saskia Voss Sheet 1 1. Conformal change of Riemannian metrics [3 points] Let (M, g) be a Riemannian manifold. A conformal change is a nonnegative function λ : M (0, ). Such a function defines
More informationDifferential Geometry Exercises
Differential Geometry Exercises Isaac Chavel Spring 2006 Jordan curve theorem We think of a regular C 2 simply closed path in the plane as a C 2 imbedding of the circle ω : S 1 R 2. Theorem. Given the
More informationSTRUCTURE OF GEODESICS IN A 13-DIMENSIONAL GROUP OF HEISENBERG TYPE
Steps in Differential Geometry, Proceedings of the Colloquium on Differential Geometry, 25 30 July, 2000, Debrecen, Hungary STRUCTURE OF GEODESICS IN A 13-DIMENSIONAL GROUP OF HEISENBERG TYPE Abstract.
More informationOptimal Collision Avoidance and Formation Switching on Riemannian Manifolds 1
Optimal Collision Avoidance and Formation Switching on Riemannian Manifolds Jianghai Hu and Shankar Sastry {jianghai,sastry}@eecs.berkeley.edu Department of Electrical Eng. & Comp. Science University of
More informationLuminy Lecture 2: Spectral rigidity of the ellipse joint work with Hamid Hezari. Luminy Lecture April 12, 2015
Luminy Lecture 2: Spectral rigidity of the ellipse joint work with Hamid Hezari Luminy Lecture April 12, 2015 Isospectral rigidity of the ellipse The purpose of this lecture is to prove that ellipses are
More informationRANDOM FIELDS AND GEOMETRY. Robert Adler and Jonathan Taylor
RANDOM FIELDS AND GEOMETRY from the book of the same name by Robert Adler and Jonathan Taylor IE&M, Technion, Israel, Statistics, Stanford, US. ie.technion.ac.il/adler.phtml www-stat.stanford.edu/ jtaylor
More informationMath Tune-Up Louisiana State University August, Lectures on Partial Differential Equations and Hilbert Space
Math Tune-Up Louisiana State University August, 2008 Lectures on Partial Differential Equations and Hilbert Space 1. A linear partial differential equation of physics We begin by considering the simplest
More informationAbout Lie Groups. timothy e. goldberg. October 6, Lie Groups Beginning Details The Exponential Map and Useful Curves...
About Lie Groups timothy e. goldberg October 6, 2005 Contents 1 Lie Groups 1 1.1 Beginning Details................................. 1 1.2 The Exponential Map and Useful Curves.................... 3 2 The
More informationInvariant Lagrangian Systems on Lie Groups
Invariant Lagrangian Systems on Lie Groups Dennis Barrett Geometry and Geometric Control (GGC) Research Group Department of Mathematics (Pure and Applied) Rhodes University, Grahamstown 6140 Eastern Cape
More informationLECTURE: KOBORDISMENTHEORIE, WINTER TERM 2011/12; SUMMARY AND LITERATURE
LECTURE: KOBORDISMENTHEORIE, WINTER TERM 2011/12; SUMMARY AND LITERATURE JOHANNES EBERT 1.1. October 11th. 1. Recapitulation from differential topology Definition 1.1. Let M m, N n, be two smooth manifolds
More informationarxiv: v3 [math.na] 7 Jul 2009
The variational particle-mesh method for matching curves arxiv:7.v [math.na] 7 Jul 9 C J Cotter Department of Aeronautics, Imperial College London, SW7 AZ E-mail: colin.cotter@imperial.ac.u Abstract. Diffeomorphic
More informationOn the coupling between an ideal fluid and immersed particles
On the coupling between an ideal fluid and immersed particles Henry O. Jacobs d,, Tudor S. Ratiu e, Mathieu Desbrun f a Imperial College London, Mathematics Department, 18 Queen s Gate Rd., London SW72AZ,
More informationL19: Fredholm theory. where E u = u T X and J u = Formally, J-holomorphic curves are just 1
L19: Fredholm theory We want to understand how to make moduli spaces of J-holomorphic curves, and once we have them, how to make them smooth and compute their dimensions. Fix (X, ω, J) and, for definiteness,
More informationFast Geodesic Regression for Population-Based Image Analysis
Fast Geodesic Regression for Population-Based Image Analysis Yi Hong 1, Polina Golland 2, and Miaomiao Zhang 2 1 Computer Science Department, University of Georgia 2 Computer Science and Artificial Intelligence
More informationLorentzian elasticity arxiv:
Lorentzian elasticity arxiv:1805.01303 Matteo Capoferri and Dmitri Vassiliev University College London 14 July 2018 Abstract formulation of elasticity theory Consider a manifold M equipped with non-degenerate
More informationPractice Qualifying Exam Questions, Differentiable Manifolds, Fall, 2009.
Practice Qualifying Exam Questions, Differentiable Manifolds, Fall, 2009. Solutions (1) Let Γ be a discrete group acting on a manifold M. (a) Define what it means for Γ to act freely. Solution: Γ acts
More informationÜbungen zu RT2 SS (4) Show that (any) contraction of a (p, q) - tensor results in a (p 1, q 1) - tensor.
Übungen zu RT2 SS 2010 (1) Show that the tensor field g µν (x) = η µν is invariant under Poincaré transformations, i.e. x µ x µ = L µ νx ν + c µ, where L µ ν is a constant matrix subject to L µ ρl ν ση
More informationA little taste of symplectic geometry
A little taste of symplectic geometry Timothy Goldberg Thursday, October 4, 2007 Olivetti Club Talk Cornell University 1 2 What is symplectic geometry? Symplectic geometry is the study of the geometry
More information3.2 Frobenius Theorem
62 CHAPTER 3. POINCARÉ, INTEGRABILITY, DEGREE 3.2 Frobenius Theorem 3.2.1 Distributions Definition 3.2.1 Let M be a n-dimensional manifold. A k-dimensional distribution (or a tangent subbundle) Δ : M Δ
More informationConstant symplectic 2-groupoids
Constant symplectic 2-groupoids Rajan Mehta Smith College April 17, 2017 The integration problem for Courant algebroids The integration problem for Courant algebroids Poisson manifolds integrate to symplectic
More informationReduction of Homogeneous Riemannian structures
Geometric Structures in Mathematical Physics, 2011 Reduction of Homogeneous Riemannian structures M. Castrillón López 1 Ignacio Luján 2 1 ICMAT (CSIC-UAM-UC3M-UCM) Universidad Complutense de Madrid 2 Universidad
More informationINSTANTON MODULI AND COMPACTIFICATION MATTHEW MAHOWALD
INSTANTON MODULI AND COMPACTIFICATION MATTHEW MAHOWALD () Instanton (definition) (2) ADHM construction (3) Compactification. Instantons.. Notation. Throughout this talk, we will use the following notation:
More informationL 2 Geometry of the Symplectomorphism Group
University of Notre Dame Workshop on Innite Dimensional Geometry, Vienna 2015 Outline 1 The Exponential Map on D s ω(m) 2 Existence of Multiplicity of Outline 1 The Exponential Map on D s ω(m) 2 Existence
More informationDual pairs for matrix groups
Dual pairs for matrix groups Paul Skerritt 1 and Cornelia Vizman 2 arxiv:1805.01519v1 [math.sg] 3 May 2018 Abstract In this paper we present two dual pairs that can be seen as the linear analogues of the
More information1. Geometry of the unit tangent bundle
1 1. Geometry of the unit tangent bundle The main reference for this section is [8]. In the following, we consider (M, g) an n-dimensional smooth manifold endowed with a Riemannian metric g. 1.1. Notations
More informationPoisson Equation on Closed Manifolds
Poisson Equation on Closed anifolds Andrew acdougall December 15, 2011 1 Introduction The purpose of this project is to investigate the poisson equation φ = ρ on closed manifolds (compact manifolds without
More informationGEOMETRIC QUANTIZATION
GEOMETRIC QUANTIZATION 1. The basic idea The setting of the Hamiltonian version of classical (Newtonian) mechanics is the phase space (position and momentum), which is a symplectic manifold. The typical
More informationRuijsenaars type deformation of hyperbolic BC n Sutherland m
Ruijsenaars type deformation of hyperbolic BC n Sutherland model March 2015 History 1. Olshanetsky and Perelomov discovered the hyperbolic BC n Sutherland model by a reduction/projection procedure, but
More informationLECTURE 10: THE PARALLEL TRANSPORT
LECTURE 10: THE PARALLEL TRANSPORT 1. The parallel transport We shall start with the geometric meaning of linear connections. Suppose M is a smooth manifold with a linear connection. Let γ : [a, b] M be
More informationGeometric inequalities for black holes
Geometric inequalities for black holes Sergio Dain FaMAF-Universidad Nacional de Córdoba, CONICET, Argentina. 3 August, 2012 Einstein equations (vacuum) The spacetime is a four dimensional manifold M with
More informationLECTURE 10: THE ATIYAH-GUILLEMIN-STERNBERG CONVEXITY THEOREM
LECTURE 10: THE ATIYAH-GUILLEMIN-STERNBERG CONVEXITY THEOREM Contents 1. The Atiyah-Guillemin-Sternberg Convexity Theorem 1 2. Proof of the Atiyah-Guillemin-Sternberg Convexity theorem 3 3. Morse theory
More informationCOHOMOGENEITY ONE HYPERSURFACES OF EUCLIDEAN SPACES
COHOMOGENEITY ONE HYPERSURFACES OF EUCLIDEAN SPACES FRANCESCO MERCURI, FABIO PODESTÀ, JOSÉ A. P. SEIXAS AND RUY TOJEIRO Abstract. We study isometric immersions f : M n R n+1 into Euclidean space of dimension
More informationIsothermal parameters
MTH.B402; Sect. 3 (20180626) 26 Isothermal parameters A Review of Complex Analysis. Let C be the complex plane. A C 1 -function 7 f : C D z w = f(z) C defined on a domain D is said to be holomorphic if
More informationSymmetric Image Normalization in the Diffeomorphic Space
Symmetric Image Normalization in the Diffeomorphic Space Brian Avants, Charles Epstein, James Gee Penn Image Computing & Science Lab Departments of Radiology and Mathematics University of Pennsylvania
More informationMetamorphic Geodesic Regression
Metamorphic Geodesic Regression Yi Hong, Sarang Joshi 3, Mar Sanchez 4, Martin Styner, and Marc Niethammer,2 UNC-Chapel Hill/ 2 BRIC, 3 University of Utah, 4 Emory University Abstract. We propose a metamorphic
More informationThe Karcher Mean of Points on SO n
The Karcher Mean of Points on SO n Knut Hüper joint work with Jonathan Manton (Univ. Melbourne) Knut.Hueper@nicta.com.au National ICT Australia Ltd. CESAME LLN, 15/7/04 p.1/25 Contents Introduction CESAME
More informationSeptember 21, :43pm Holm Vol 2 WSPC/Book Trim Size for 9in by 6in
1 GALILEO Contents 1.1 Principle of Galilean relativity 2 1.2 Galilean transformations 3 1.2.1 Admissible force laws for an N-particle system 6 1.3 Subgroups of the Galilean transformations 8 1.3.1 Matrix
More informationChapter 14. Basics of The Differential Geometry of Surfaces. Introduction. Parameterized Surfaces. The First... Home Page. Title Page.
Chapter 14 Basics of The Differential Geometry of Surfaces Page 649 of 681 14.1. Almost all of the material presented in this chapter is based on lectures given by Eugenio Calabi in an upper undergraduate
More informationPoincaré Duality Angles on Riemannian Manifolds with Boundary
Poincaré Duality Angles on Riemannian Manifolds with Boundary Clayton Shonkwiler Department of Mathematics University of Pennsylvania June 5, 2009 Realizing cohomology groups as spaces of differential
More informationCritical metrics on connected sums of Einstein four-manifolds
Critical metrics on connected sums of Einstein four-manifolds University of Wisconsin, Madison March 22, 2013 Kyoto Einstein manifolds Einstein-Hilbert functional in dimension 4: R(g) = V ol(g) 1/2 R g
More informationSymplectic geometry of deformation spaces
July 15, 2010 Outline What is a symplectic structure? What is a symplectic structure? Denition A symplectic structure on a (smooth) manifold M is a closed nondegenerate 2-form ω. Examples Darboux coordinates
More informationDEVELOPMENT OF MORSE THEORY
DEVELOPMENT OF MORSE THEORY MATTHEW STEED Abstract. In this paper, we develop Morse theory, which allows us to determine topological information about manifolds using certain real-valued functions defined
More informationHamiltonian flows, cotangent lifts, and momentum maps
Hamiltonian flows, cotangent lifts, and momentum maps Jordan Bell jordan.bell@gmail.com Department of Mathematics, University of Toronto April 3, 2014 1 Symplectic manifolds Let (M, ω) and (N, η) be symplectic
More informationTIME DISCRETE GEODESIC PATHS IN THE SPACE OF IMAGES
TIME ISCRETE GEOESIC PATHS IN THE SPACE OF IMAGES B. BERELS, A. EFFLAN, M. RUMPF Abstract. In this paper the space of images is considered as a Riemannian manifold using the metamorphosis approach [28,
More informationTHE CANONICAL PENCILS ON HORIKAWA SURFACES
THE CANONICAL PENCILS ON HORIKAWA SURFACES DENIS AUROUX Abstract. We calculate the monodromies of the canonical Lefschetz pencils on a pair of homeomorphic Horikawa surfaces. We show in particular that
More informationChapter 3: Duality Toolbox
3.: GENEAL ASPECTS 3..: I/UV CONNECTION Chapter 3: Duality Toolbox MIT OpenCourseWare Lecture Notes Hong Liu, Fall 04 Lecture 8 As seen before, equipped with holographic principle, we can deduce N = 4
More informationMath 550 / David Dumas / Fall Problems
Math 550 / David Dumas / Fall 2014 Problems Please note: This list was last updated on November 30, 2014. Problems marked with * are challenge problems. Some problems are adapted from the course texts;
More informationChapter 3. Riemannian Manifolds - I. The subject of this thesis is to extend the combinatorial curve reconstruction approach to curves
Chapter 3 Riemannian Manifolds - I The subject of this thesis is to extend the combinatorial curve reconstruction approach to curves embedded in Riemannian manifolds. A Riemannian manifold is an abstraction
More informationQuantising proper actions on Spin c -manifolds
Quantising proper actions on Spin c -manifolds Peter Hochs University of Adelaide Differential geometry seminar Adelaide, 31 July 2015 Joint work with Mathai Varghese (symplectic case) Geometric quantization
More informationVariational Principle and Einstein s equations
Chapter 15 Variational Principle and Einstein s equations 15.1 An useful formula There exists an useful equation relating g µν, g µν and g = det(g µν ) : g x α = ggµν g µν x α. (15.1) The proof is the
More informationThe Symmetric Space for SL n (R)
The Symmetric Space for SL n (R) Rich Schwartz November 27, 2013 The purpose of these notes is to discuss the symmetric space X on which SL n (R) acts. Here, as usual, SL n (R) denotes the group of n n
More informationExotic nearly Kähler structures on S 6 and S 3 S 3
Exotic nearly Kähler structures on S 6 and S 3 S 3 Lorenzo Foscolo Stony Brook University joint with Mark Haskins, Imperial College London Friday Lunch Seminar, MSRI, April 22 2016 G 2 cones and nearly
More informationLECTURE 1: LINEAR SYMPLECTIC GEOMETRY
LECTURE 1: LINEAR SYMPLECTIC GEOMETRY Contents 1. Linear symplectic structure 3 2. Distinguished subspaces 5 3. Linear complex structure 7 4. The symplectic group 10 *********************************************************************************
More informationRiemannian Metrics on the Space of Solid Shapes
Riemannian Metrics on the Space of Solid Shapes P. Thomas Fletcher and Ross T. Whitaker School of Computing University of Utah fletcher@sci.utah.edu, whitaker@cs.utah.edu Abstract. We present a new framework
More informationEinstein-Hilbert action on Connes-Landi noncommutative manifolds
Einstein-Hilbert action on Connes-Landi noncommutative manifolds Yang Liu MPIM, Bonn Analysis, Noncommutative Geometry, Operator Algebras Workshop June 2017 Motivations and History Motivation: Explore
More informationGravitation: Tensor Calculus
An Introduction to General Relativity Center for Relativistic Astrophysics School of Physics Georgia Institute of Technology Notes based on textbook: Spacetime and Geometry by S.M. Carroll Spring 2013
More informationModuli spaces of graphs and homology operations on loop spaces of manifolds
Moduli spaces of graphs and homology operations on loop spaces of manifolds Ralph L. Cohen Stanford University July 2, 2005 String topology:= intersection theory in loop spaces (and spaces of paths) of
More informationAn Invitation to Geometric Quantization
An Invitation to Geometric Quantization Alex Fok Department of Mathematics, Cornell University April 2012 What is quantization? Quantization is a process of associating a classical mechanical system to
More informationShape deformation analysis from the optimal control viewpoint
Shape deformation analysis from the optimal control viewpoint Sylvain Arguillère Emmanuel Trélat Alain Trouvé Laurent Younes Abstract A crucial problem in shape deformation analysis is to determine a deformation
More informationShape analysis on Lie groups and Homogeneous Manifolds with applications in computer animation
Shape analysis on Lie groups and Homogeneous Manifolds with applications in computer animation Department of Mathematical Sciences, NTNU, Trondheim, Norway joint work with Markus Eslitzbichler and Alexander
More informationLecture I: Constrained Hamiltonian systems
Lecture I: Constrained Hamiltonian systems (Courses in canonical gravity) Yaser Tavakoli December 15, 2014 1 Introduction In canonical formulation of general relativity, geometry of space-time is given
More informationHard Lefschetz Theorem for Vaisman manifolds
Hard Lefschetz Theorem for Vaisman manifolds Antonio De Nicola CMUC, University of Coimbra, Portugal joint work with B. Cappelletti-Montano (Univ. Cagliari), J.C. Marrero (Univ. La Laguna) and I. Yudin
More information