Diffusion Processes and Dimensionality Reduction on Manifolds

Size: px
Start display at page:

Download "Diffusion Processes and Dimensionality Reduction on Manifolds"

Transcription

1 Faculty of Science Diffusion Processes and Dimensionality Reduction on Manifolds ESI, Vienna, Feb Stefan Sommer Department of Computer Science, University of Copenhagen February 23, 2015 Slide 1/22

2 Outline Dimensionality Reduction Diffusion PCA Development and Anisotropic Diffusions Examples Slide 2/22

3 Dimensionality Reduction in Non-Linear Manifolds dim. reduction and linearizations - mappings from non-linear manifolds to low dimensional Euclidean space that preserves structure of data Non-Euclidean generalizations of PCA: Principal Geodesic Analysis (PGA, Fletcher et al., 04) Geodesic PCA (GPCA, Huckeman et al., 10) Horizontal Component Analysis (HCA, Sommer, 13) Principal Nested Spheres ((C)PNS, Jung et al., 12) Barycentric Subspaces (BS, Pennec, 15) Slide 3/22

4 Dimensionality Reduction in Non-Linear Manifolds dim. reduction and linearizations - mappings from non-linear manifolds to low dimensional Euclidean space that preserves structure of data Non-Euclidean generalizations of PCA: PGA: analysis relative to Principal the Geodesic Analysis (PGA, Fletcher et al., 04) Geodesic PCA (GPCA, Huckeman et al., 10) data mean Horizontal Component Analysis (HCA, Sommer, 13) Principal Nested Spheres ((C)PNS, Jung et al., 12) Barycentric Subspaces (BS, Pennec, 15) Slide 3/22

5 Dimensionality Reduction in Non-Linear Manifolds dim. reduction and linearizations - mappings from non-linear manifolds to low dimensional Euclidean space that preserves structure of data Non-Euclidean generalizations of PCA: PGA: analysis Principal Geodesic Analysis (PGA, Fletcher et al., 04) relative to the Geodesic PCA (GPCA, Huckeman et al., 10) data mean Horizontal Component Analysis (HCA, Sommer, 13) Principal Nested Spheres ((C)PNS, Jung et al., 12) Barycentric Subspaces (BS, Pennec, 15) data points on non-linear manifold Slide 3/22

6 Dimensionality Reduction in Non-Linear Manifolds dim. reduction and linearizations - mappings from non-linear manifolds to low dimensional Euclidean space that preserves structure of data Non-Euclidean generalizations of PCA: PGA: analysis Principal Geodesic Analysis (PGA, Fletcher et al., 04) relative to Geodesic the PCA (GPCA, Huckeman et al., 10) data mean Horizontal Component Analysis (HCA, Sommer, 13) Principal Nested Spheres ((C)PNS, Jung et al., 12) Barycentric Subspaces (BS, Pennec, 15) intrinsic mean µ Slide 3/22

7 Dimensionality Reduction in Non-Linear Manifolds dim. reduction and linearizations - mappings from non-linear manifolds to low dimensional Euclidean space that preserves structure of data Non-Euclidean generalizations of PCA: PGA: analysis relative to Principal the Geodesic Analysis (PGA, Fletcher et al., 04) Geodesic PCA (GPCA, Huckeman et al., 10) data mean Horizontal Component Analysis (HCA, Sommer, 13) Principal Nested Spheres ((C)PNS, Jung et al., 12) Barycentric Subspaces (BS, Pennec, 15) tangent space T µ M Slide 3/22

8 Dimensionality Reduction in Non-Linear Manifolds dim. reduction and linearizations - mappings from non-linear manifolds to low dimensional Euclidean space that preserves structure of data Non-Euclidean generalizations of PCA: PGA: analysis relative to Principal the Geodesic Analysis (PGA, Fletcher et al., 04) Geodesic PCA (GPCA, Huckeman et al., 10) data mean Horizontal Component Analysis (HCA, Sommer, 13) Principal Nested Spheres ((C)PNS, Jung et al., 12) Barycentric Subspaces (BS, Pennec, 15) projection of data point to T µ M Slide 3/22

9 Dimensionality Reduction in Non-Linear Manifolds dim. reduction and linearizations - mappings from non-linear manifolds to low dimensional Euclidean space that preserves structure of data PGA: analysis Non-Euclidean generalizations of PCA: relative to the Principal Geodesic Analysis (PGA, Fletcher et al., 04) data mean Geodesic PCA (GPCA, Huckeman et al., 10) Horizontal Component Analysis (HCA, Sommer, 13) Principal Nested Spheres ((C)PNS, Jung et al., 12) Barycentric Subspaces Euclidean (BS, PCA Pennec, in tangent 15) space Slide 3/22

10 Dimensionality Reduction in Non-Linear Manifolds dim. reduction and linearizations - mappings from non-linear manifolds to low dimensional Euclidean space that preserves structure of data Non-Euclidean generalizations of PCA: Principal Geodesic Analysis (PGA, Fletcher et al., 04) Geodesic PCA (GPCA, Huckeman et al., 10) Horizontal Component Analysis (HCA, Sommer, 13) Principal Nested Spheres ((C)PNS, Jung et al., 12) Barycentric Subspaces (BS, Pennec, 15) Slide 3/22

11 PGA: Dimensionality Reduction in Non-Linear Manifolds dim. reduction and linearizations - mappings from non-linear manifolds to low dimensional Euclidean GPCA: space that preserves structure of data HCA: Non-Euclidean generalizations of PCA: Principal Geodesic Analysis (PGA, Fletcher et al., 04) Geodesic PCA (GPCA, Huckeman et al., 10) Horizontal Component Analysis (HCA, Sommer, 13) Principal Nested Spheres ((C)PNS, Jung et al., 12) Barycentric Subspaces (BS, Pennec, 15) Slide 3/22

12 PGA, GPCA, HCA, PNS,... search for explicitly constructed parametric subspaces: geodesic sprays, geodesics, iterated development,... in general manifolds, these subspaces are not totally geodesic projections to subspaces are problematic: geodesics may be dense on tori Slide 4/22

13 Generalizing Linear Statistics Euclidean Riemannian norm x y distances d(x,y) vectors v 0 for geodesics linear subspaces geodesic sprays why are geodesics fundamental when estimating covariance? Euclidean space analogies can lead to non-local constructions to goal of this talk is to get closer to constructions defined infinitesimally Slide 5/22

14 Euclidean PCA Usual formulation: eigendecomposition (u 1,λ 1 ),...,(u d,λ d ) of sample covar. matrix C principal components: x n = U T (y n µ) Probabilistic interpretation (Tipping, Bishop, 99): latent variable model y = Wx + µ + ε, x N (0,I), ε N (0,σ 2 I) marginal distribution y N (µ,c σ ), C σ = WW T + σ 2 I MLE of W : W ML = U(Λ σ 2 I) 1/2 + rotation Λ = diag(λ 1,...,λ d ) principal components: E[x n y n ] = (W T W + σ 2 I) 1 W T ML (y n µ) Slide 6/22

15 Diffusion PCA probabilistic PCA does not explicitly use subspaces on Riemannian manifolds, the Eells-Elworthy-Malliavin construction gives a map : FM Dens(M) Diff Γ Dens(M): the image Diff (FM), the set of (normalized) densities resulting from diffusions in FM µ Γ anisotropic normal distribution with µ = Diff (x,x α) = p µ µ 0, define the log-likelihood lnl(x,x α ) = lnl(µ) = N i=1 lnp µ (y i ) Diffusion PCA: maxim. lnl(x,x α ) for (x,x α ) FM MLE of data y i under the assumption y µ Γ Slide 7/22

16 Diffusion PCA all geometric complexities are hidden in the diffusion processes no parametric subspaces, no projections to dense geodesics principal components: mean sample paths reaching y i ˆx i (t) = E[x(t) x(1) = y i ] path dependency can be integrated out x i = 1 0 d dt ˆx i(t)dt = ˆx i (1) x i provides a linear view of the data Slide 8/22

17 Diffusion PCA all geometric complexities are hidden in the diffusion processes no parametric subspaces, no projections to dense geodesics principal components: mean sample paths reaching y i ˆx i (t) = E[x(t) x(1) = y i ] path dependency can be integrated out x i = 1 0 d dt ˆx i(t)dt = ˆx i (1) x i provides a linear view of the data Slide 8/22

18 Statistical Manifold: Geometry of Γ Densities Dens(M) = {µ Ω n (M) : Fisher-Rao metric: Gµ FR (α,β) = α β M µ µ µ Γ finite dim. subset of Dens(M) properties of Diff M : FM Dens(M) µ = 1,µ > 0} naturally defined on bundle of symmetric positive T 0 2 tensors Slide 9/22

19 SDEs On Manifolds stationary driftless diffusion SDE in R n : dx t = σdw t, σ M n d diffusion field σ, infinitesimal generator σσ T curvature: stationary field/generator cannot be defined due to holonomy Slide 10/22

20 SDEs On Manifolds stationary driftless diffusion SDE in R n : dx t = σdw t, σ M n d diffusion field σ, infinitesimal generator σσ T curvature: stationary field/generator cannot be defined due to holonomy Slide 10/22

21 The Frame Bundle the manifold and frames (bases) for the tangent spaces T p M F(M) consists of pairs u = (x,x α ), x M, X α frame for T x M curves in the horizontal part of F(M) correspond to curves in M and parallel transport of frames Slide 11/22

22 The Frame Bundle the manifold and frames (bases) for the tangent spaces T p M F(M) consists of pairs u = (x,x α ), x M, X α frame for T x M curves in the horizontal part of F(M) correspond to curves in M and parallel transport of frames Slide 11/22

23 The Frame Bundle the manifold and frames (bases) for the tangent spaces T p M F(M) consists of pairs u = (x,x α ), x M, X α frame for T x M curves in the horizontal part of F(M) correspond to curves in M and parallel transport of frames Slide 11/22

24 SDEs On Manifolds: Eells-Elworthy-Malliavin construction H i, i = 1...,n horizontal vector fields on F(M): SDE in R n : SDE in F(M): H i (u) = π 1 (u i ) dx t = σ(x t )dw t, σ(x t ) M n n du t = H i (U t ) dx i t = H i (U t ) σ(x t ) i j dw j t SDE on M: π F(M) (U t ) Slide 12/22

25 SDEs On Manifolds: Eells-Elworthy-Malliavin construction H i, i = 1...,n horizontal vector fields on F(M): SDE in R n : SDE in F(M): H i (u) = π 1 (u i ) dx t = σ(x t )dw t, σ(x t ) M n n du t = H i (U t ) dx i t = H i (U t ) σ(x t ) i j dw j t SDE on M: π F(M) (U t ) Slide 12/22

26 Anisotropic Diffusions Diff (x,x α ) = π F(M) (U 1 ) stochastic development / rolling without slipping in R n, sample path increments x ti = x ti+1 x ti are normally distributed N (0,( t) 1 Σ) with log-probability ln p Σ (x t ) 1 t Formally, we can set ln p Σ (x t ) N 1 xt T i Σ 1 x ti + c i=1 1 0 ẋ t 2 Σ dt + c Slide 13/22

27 Most Probable Paths Let φ be path development and U t a heat diffusion. As N, the finite path measure pullback φ v of Wiener measure v on W([0,1],M) (cont. paths from x) (Andersson, Driver, 99) geodesics and be formally viewed as most probable paths (MPPs) for the pull-back path density, i.e. MPPs for the Euclidean R n diffusion Anisotropic case: (x t,x α,t ) path in FM, X α,t represents Σ 1/2 and defines invertible map R n T xt M. Inner product on T xt M v,w Xα,t = X 1 α,t v,x 1 α,t w R n Slide 14/22

28 Sub-Riemannian Geometry optimal control problem with nonholonomic constraints let x t = arg min c t,c 0 =x,c 1 =y 1 0 ċ t 2 X α,t dt ṽ, w HFM = X 1 α,t π (ṽ),x 1 α,t π ( w) R n on H (xt,x α,t )FM. This defines a sub-riemannian metric G on TFM and equivalent problem (x t,x α,t ) = arg min (c t,c α,t ),c 0 =x,c 1 =y 1 with constraints (ċ t,ċ α,t ) H (ct,c α,t )FM 0 (ċ t,ċ α,t ) 2 HFM dt (1) Slide 15/22

29 Evolution Equations geodesics satisfy the Hamilton-Jacobi equations ẏ k t = G kj (y t )ξ t,j, ξt,k = 1 2 G pq y k ξ t,pξ t,q in coordinates (x i ) for M, X i α for X α, and W encoding the inner product W kl = δ αβ X k α X l β : ẋ i = W ij ξ j W ih Γ j β h ξ j β, Ẋα i = Γ i α h W hj ξ j + Γ i α k W kh Γ j β ξ i = W hl Γ k δ l,i ξ hξ kδ 1 ( ) Γ h γ k,i 2 W kh Γ k δ h + Γh γ k W kh Γ k δ h,i ξ hγ ξ kδ ( ξ i α = Γ h γ k,i α W kh Γ k δ h ξ h γ ξ kδ W hl,i α Γ k δ l + W hl Γ k δ l,i α )ξ h ξ kδ 1 ( ) W hk,i 2 α ξ h ξ k + Γ h γ k W kh,i α Γ k δ h ξ h γ ξ kδ h ξ j β Slide 16/22

30 S 2 (a) cov. diag(1, 1) (b) cov. diag(2,.5) (c) cov. diag(4,.25) Slide 17/22

31 S2 (d) cov. diag(1, 1) (e) cov. diag(2,.5) (f) cov. diag(4,.25) Slide 17/22

32 Landmark LDDMM Christoffel symbols (Michelli et al. 08) Γ k ij = 1 2 g ir ( ) g kl g rs,l g sl g rk,l g rl g ks,l gsj mix of transported frame and cometric: F d M bundle of rank d linear maps R d T x M, ξ, ξ T F d M, cometric ξ, ξ = δ αβ (ξ π 1 X α)( ξ π 1 X β) + λ the whole frame need not be transported ξ, ξ g R Slide 18/22

33 Landmark LDDMM Examples Slide 19/22

34 Summary Diffusion PCA: dim. reduction or linearization by MLE of densities generated by diffusion processes infinitesimal definition, no subspaces, no projections diffusion map Diff : FM Dens(M) from Eells-Elworthy-Malliavin construction of Brownian motion MPPs for anisotropic diffusions generalize geodesics 1 Sommer: Diffusion Processes and PCA on Manifolds, Oberwolfach extended abstract, Sommer: Anisotropic Distributions on Manifolds: Template Estimation and Most Probable Paths, in preparation. Slide 20/22

Brownian Motion on Manifold

Brownian Motion on Manifold Brownian Motion on Manifold QI FENG Purdue University feng71@purdue.edu August 31, 2014 QI FENG (Purdue University) Brownian Motion on Manifold August 31, 2014 1 / 26 Overview 1 Extrinsic construction

More information

1. Geometry of the unit tangent bundle

1. 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 information

arxiv: v1 [cs.cv] 19 May 2018

arxiv: v1 [cs.cv] 19 May 2018 Latent Space Non-Linear Statistics arxiv:85.763v [cs.cv] 9 May 8 Line Kühnel University of Copenhagen Denmark kuhnel@di.ku.dk Tom Fletcher University of Utah USA fletcher@sci.utah.edu Stefan Sommer University

More information

CALCULUS ON MANIFOLDS. 1. Riemannian manifolds Recall that for any smooth manifold M, dim M = n, the union T M =

CALCULUS ON MANIFOLDS. 1. Riemannian manifolds Recall that for any smooth manifold M, dim M = n, the union T M = CALCULUS ON MANIFOLDS 1. Riemannian manifolds Recall that for any smooth manifold M, dim M = n, the union T M = a M T am, called the tangent bundle, is itself a smooth manifold, dim T M = 2n. Example 1.

More information

Welcome to Copenhagen!

Welcome to Copenhagen! Welcome to Copenhagen! Schedule: Monday Tuesday Wednesday Thursday Friday 8 Registration and welcome 9 Crash course on Crash course on Introduction to Differential and Differential and Information Geometry

More information

THEODORE VORONOV DIFFERENTIAL GEOMETRY. Spring 2009

THEODORE VORONOV DIFFERENTIAL GEOMETRY. Spring 2009 [under construction] 8 Parallel transport 8.1 Equation of parallel transport Consider a vector bundle E B. We would like to compare vectors belonging to fibers over different points. Recall that this was

More information

Harmonic Morphisms - Basics

Harmonic Morphisms - Basics Department of Mathematics Faculty of Science Lund University Sigmundur.Gudmundsson@math.lu.se March 11, 2014 Outline Harmonic Maps in Gaussian Geometry 1 Harmonic Maps in Gaussian Geometry Holomorphic

More information

Data Analysis and Manifold Learning Lecture 6: Probabilistic PCA and Factor Analysis

Data Analysis and Manifold Learning Lecture 6: Probabilistic PCA and Factor Analysis Data Analysis and Manifold Learning Lecture 6: Probabilistic PCA and Factor Analysis Radu Horaud INRIA Grenoble Rhone-Alpes, France Radu.Horaud@inrialpes.fr http://perception.inrialpes.fr/ Outline of Lecture

More information

Riemannian geometry of surfaces

Riemannian geometry of surfaces Riemannian geometry of surfaces In this note, we will learn how to make sense of the concepts of differential geometry on a surface M, which is not necessarily situated in R 3. This intrinsic approach

More information

INTRO TO SUBRIEMANNIAN GEOMETRY

INTRO TO SUBRIEMANNIAN GEOMETRY INTRO TO SUBRIEMANNIAN GEOMETRY 1. Introduction to subriemannian geometry A lot of this tal is inspired by the paper by Ines Kath and Oliver Ungermann on the arxiv, see [3] as well as [1]. Let M be a smooth

More information

Manifold Learning for Signal and Visual Processing Lecture 9: Probabilistic PCA (PPCA), Factor Analysis, Mixtures of PPCA

Manifold Learning for Signal and Visual Processing Lecture 9: Probabilistic PCA (PPCA), Factor Analysis, Mixtures of PPCA Manifold Learning for Signal and Visual Processing Lecture 9: Probabilistic PCA (PPCA), Factor Analysis, Mixtures of PPCA Radu Horaud INRIA Grenoble Rhone-Alpes, France Radu.Horaud@inria.fr http://perception.inrialpes.fr/

More information

CS 468. Differential Geometry for Computer Science. Lecture 17 Surface Deformation

CS 468. Differential Geometry for Computer Science. Lecture 17 Surface Deformation CS 468 Differential Geometry for Computer Science Lecture 17 Surface Deformation Outline Fundamental theorem of surface geometry. Some terminology: embeddings, isometries, deformations. Curvature flows

More information

The Uses of Differential Geometry in Finance

The Uses of Differential Geometry in Finance The Uses of Differential Geometry in Finance Andrew Lesniewski Bloomberg, November 21 2005 The Uses of Differential Geometry in Finance p. 1 Overview Joint with P. Hagan and D. Woodward Motivation: Varadhan

More information

LECTURE 10: THE PARALLEL TRANSPORT

LECTURE 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 information

Chapter 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 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 information

Math 6455 Nov 1, Differential Geometry I Fall 2006, Georgia Tech

Math 6455 Nov 1, Differential Geometry I Fall 2006, Georgia Tech Math 6455 Nov 1, 26 1 Differential Geometry I Fall 26, Georgia Tech Lecture Notes 14 Connections Suppose that we have a vector field X on a Riemannian manifold M. How can we measure how much X is changing

More information

RESEARCH REPORT. Anisotropically Weighted and Nonholonomically Constrained Evolutions on Manifolds. Stefan Sommer. No. 12, September 2016

RESEARCH REPORT. Anisotropically Weighted and Nonholonomically Constrained Evolutions on Manifolds.   Stefan Sommer. No. 12, September 2016 CENTRE FOR STOCHASTIC GEOMETRY AND ADVANCED BIOIMAGING www.csgb.dk RESEARCH REPORT 6 Stefan Sommer Anisotropicall Weighted and Nonholonomicall Constrained Evolutions on Manifolds No., September 6 Anisotropicall

More information

Divergence theorems in path space II: degenerate diffusions

Divergence theorems in path space II: degenerate diffusions Divergence theorems in path space II: degenerate diffusions Denis Bell 1 Department of Mathematics, University of North Florida, 4567 St. Johns Bluff Road South, Jacksonville, FL 32224, U. S. A. email:

More information

DIFFERENTIAL GEOMETRY CLASS NOTES INSTRUCTOR: F. MARQUES. September 25, 2015

DIFFERENTIAL GEOMETRY CLASS NOTES INSTRUCTOR: F. MARQUES. September 25, 2015 DIFFERENTIAL GEOMETRY CLASS NOTES INSTRUCTOR: F. MARQUES MAGGIE MILLER September 25, 2015 1. 09/16/2015 1.1. Textbooks. Textbooks relevant to this class are Riemannian Geometry by do Carmo Riemannian Geometry

More information

Manifold Monte Carlo Methods

Manifold Monte Carlo Methods Manifold Monte Carlo Methods Mark Girolami Department of Statistical Science University College London Joint work with Ben Calderhead Research Section Ordinary Meeting The Royal Statistical Society October

More information

Tools of stochastic calculus

Tools of stochastic calculus slides for the course Interest rate theory, University of Ljubljana, 212-13/I, part III József Gáll University of Debrecen Nov. 212 Jan. 213, Ljubljana Itô integral, summary of main facts Notations, basic

More information

Hodge de Rham decomposition for an L 2 space of differfential 2-forms on path spaces

Hodge de Rham decomposition for an L 2 space of differfential 2-forms on path spaces Hodge de Rham decomposition for an L 2 space of differfential 2-forms on path spaces K. D. Elworthy and Xue-Mei Li For a compact Riemannian manifold the space L 2 A of L 2 differential forms decomposes

More information

Divergence Theorems in Path Space. Denis Bell University of North Florida

Divergence Theorems in Path Space. Denis Bell University of North Florida Divergence Theorems in Path Space Denis Bell University of North Florida Motivation Divergence theorem in Riemannian geometry Theorem. Let M be a closed d-dimensional Riemannian manifold. Then for any

More information

Interest Rate Models:

Interest Rate Models: 1/17 Interest Rate Models: from Parametric Statistics to Infinite Dimensional Stochastic Analysis René Carmona Bendheim Center for Finance ORFE & PACM, Princeton University email: rcarmna@princeton.edu

More information

5.2 The Levi-Civita Connection on Surfaces. 1 Parallel transport of vector fields on a surface M

5.2 The Levi-Civita Connection on Surfaces. 1 Parallel transport of vector fields on a surface M 5.2 The Levi-Civita Connection on Surfaces In this section, we define the parallel transport of vector fields on a surface M, and then we introduce the concept of the Levi-Civita connection, which is also

More information

Hamiltonian Systems of Negative Curvature are Hyperbolic

Hamiltonian Systems of Negative Curvature are Hyperbolic Hamiltonian Systems of Negative Curvature are Hyperbolic A. A. Agrachev N. N. Chtcherbakova Abstract The curvature and the reduced curvature are basic differential invariants of the pair: Hamiltonian system,

More information

Comparison for infinitesimal automorphisms. of parabolic geometries

Comparison for infinitesimal automorphisms. of parabolic geometries Comparison techniques for infinitesimal automorphisms of parabolic geometries University of Vienna Faculty of Mathematics Srni, January 2012 This talk reports on joint work in progress with Karin Melnick

More information

Bulletin of the Transilvania University of Braşov Vol 7(56), No Series III: Mathematics, Informatics, Physics, 1-12

Bulletin of the Transilvania University of Braşov Vol 7(56), No Series III: Mathematics, Informatics, Physics, 1-12 Bulletin of the Transilvania University of Braşov Vol 7(56), No. 1-2014 Series III: Mathematics, Informatics, Physics, 1-12 EQUATION GEODESIC IN A TWO-DIMENSIONAL FINSLER SPACE WITH SPECIAL (α, β)-metric

More information

Information geometry for bivariate distribution control

Information geometry for bivariate distribution control Information geometry for bivariate distribution control C.T.J.Dodson + Hong Wang Mathematics + Control Systems Centre, University of Manchester Institute of Science and Technology Optimal control of stochastic

More information

Distances between spectral densities. V x. V y. Genesis of this talk. The key point : the value of chordal distances. Spectrum approximation problem

Distances between spectral densities. V x. V y. Genesis of this talk. The key point : the value of chordal distances. Spectrum approximation problem Genesis of this talk Distances between spectral densities The shortest distance between two points is always under construction. (R. McClanahan) R. Sepulchre -- University of Cambridge Celebrating Anders

More information

Stochastic Mechanics of Particles and Fields

Stochastic Mechanics of Particles and Fields Stochastic Mechanics of Particles and Fields Edward Nelson Department of Mathematics, Princeton University These slides are posted at http://math.princeton.edu/ nelson/papers/xsmpf.pdf A preliminary draft

More information

Sub-Riemannian geometry in groups of diffeomorphisms and shape spaces

Sub-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 information

Fundamental Materials of Riemannian Geometry

Fundamental Materials of Riemannian Geometry Chapter 1 Fundamental aterials of Riemannian Geometry 1.1 Introduction In this chapter, we give fundamental materials in Riemannian geometry. In this book, we assume basic materials on manifolds. We give,

More information

UNIVERSITY OF DUBLIN

UNIVERSITY OF DUBLIN UNIVERSITY OF DUBLIN TRINITY COLLEGE JS & SS Mathematics SS Theoretical Physics SS TSM Mathematics Faculty of Engineering, Mathematics and Science school of mathematics Trinity Term 2015 Module MA3429

More information

On the exponential map on Riemannian polyhedra by Monica Alice Aprodu. Abstract

On the exponential map on Riemannian polyhedra by Monica Alice Aprodu. Abstract Bull. Math. Soc. Sci. Math. Roumanie Tome 60 (108) No. 3, 2017, 233 238 On the exponential map on Riemannian polyhedra by Monica Alice Aprodu Abstract We prove that Riemannian polyhedra admit explicit

More information

Brownian Motion and lorentzian manifolds

Brownian Motion and lorentzian manifolds The case of Jürgen Angst Institut de Recherche Mathématique Avancée Université Louis Pasteur, Strasbourg École d été de Probabilités de Saint-Flour juillet 2008, Saint-Flour 1 Construction of the diffusion,

More information

4.7 The Levi-Civita connection and parallel transport

4.7 The Levi-Civita connection and parallel transport Classnotes: Geometry & Control of Dynamical Systems, M. Kawski. April 21, 2009 138 4.7 The Levi-Civita connection and parallel transport In the earlier investigation, characterizing the shortest curves

More information

Some topics in sub-riemannian geometry

Some topics in sub-riemannian geometry Some topics in sub-riemannian geometry Luca Rizzi CNRS, Institut Fourier Mathematical Colloquium Universität Bern - December 19 2016 Sub-Riemannian geometry Known under many names: Carnot-Carathéodory

More information

Course Summary Math 211

Course Summary Math 211 Course Summary Math 211 table of contents I. Functions of several variables. II. R n. III. Derivatives. IV. Taylor s Theorem. V. Differential Geometry. VI. Applications. 1. Best affine approximations.

More information

Chap. 1. Some Differential Geometric Tools

Chap. 1. Some Differential Geometric Tools Chap. 1. Some Differential Geometric Tools 1. Manifold, Diffeomorphism 1.1. The Implicit Function Theorem ϕ : U R n R n p (0 p < n), of class C k (k 1) x 0 U such that ϕ(x 0 ) = 0 rank Dϕ(x) = n p x U

More information

A brief introduction to Semi-Riemannian geometry and general relativity. Hans Ringström

A brief introduction to Semi-Riemannian geometry and general relativity. Hans Ringström A brief introduction to Semi-Riemannian geometry and general relativity Hans Ringström May 5, 2015 2 Contents 1 Scalar product spaces 1 1.1 Scalar products...................................... 1 1.2 Orthonormal

More information

COMS 4721: Machine Learning for Data Science Lecture 19, 4/6/2017

COMS 4721: Machine Learning for Data Science Lecture 19, 4/6/2017 COMS 4721: Machine Learning for Data Science Lecture 19, 4/6/2017 Prof. John Paisley Department of Electrical Engineering & Data Science Institute Columbia University PRINCIPAL COMPONENT ANALYSIS DIMENSIONALITY

More information

Hyperbolic Geometry on Geometric Surfaces

Hyperbolic 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 information

LAN property for sde s with additive fractional noise and continuous time observation

LAN property for sde s with additive fractional noise and continuous time observation LAN property for sde s with additive fractional noise and continuous time observation Eulalia Nualart (Universitat Pompeu Fabra, Barcelona) joint work with Samy Tindel (Purdue University) Vlad s 6th birthday,

More information

Two dimensional manifolds

Two dimensional manifolds Two dimensional manifolds We are given a real two-dimensional manifold, M. A point of M is denoted X and local coordinates are X (x, y) R 2. If we use different local coordinates, (x, y ), we have x f(x,

More information

Quasi-invariant measures on the path space of a diffusion

Quasi-invariant measures on the path space of a diffusion Quasi-invariant measures on the path space of a diffusion Denis Bell 1 Department of Mathematics, University of North Florida, 4567 St. Johns Bluff Road South, Jacksonville, FL 32224, U. S. A. email: dbell@unf.edu,

More information

DIFFERENTIAL GEOMETRY. LECTURE 12-13,

DIFFERENTIAL GEOMETRY. LECTURE 12-13, DIFFERENTIAL GEOMETRY. LECTURE 12-13, 3.07.08 5. Riemannian metrics. Examples. Connections 5.1. Length of a curve. Let γ : [a, b] R n be a parametried curve. Its length can be calculated as the limit of

More information

Exercises in Geometry II University of Bonn, Summer semester 2015 Professor: Prof. Christian Blohmann Assistant: Saskia Voss Sheet 1

Exercises 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 information

On the existence of isoperimetric extremals of rotation and the fundamental equations of rotary diffeomorphisms

On the existence of isoperimetric extremals of rotation and the fundamental equations of rotary diffeomorphisms XV II th International Conference of Geometry, Integrability and Quantization Sveti Konstantin i Elena 2016 On the existence of isoperimetric extremals of rotation and the fundamental equations of rotary

More information

Curved Spacetime I. Dr. Naylor

Curved Spacetime I. Dr. Naylor Curved Spacetime I Dr. Naylor Last Week Einstein's principle of equivalence We discussed how in the frame of reference of a freely falling object we can construct a locally inertial frame (LIF) Space tells

More information

Generalized Gaussian Bridges of Prediction-Invertible Processes

Generalized Gaussian Bridges of Prediction-Invertible Processes Generalized Gaussian Bridges of Prediction-Invertible Processes Tommi Sottinen 1 and Adil Yazigi University of Vaasa, Finland Modern Stochastics: Theory and Applications III September 1, 212, Kyiv, Ukraine

More information

Continuous dependence estimates for the ergodic problem with an application to homogenization

Continuous dependence estimates for the ergodic problem with an application to homogenization Continuous dependence estimates for the ergodic problem with an application to homogenization Claudio Marchi Bayreuth, September 12 th, 2013 C. Marchi (Università di Padova) Continuous dependence Bayreuth,

More information

Lagrangian Formulation of Elastic Wave Equation on Riemannian Manifolds

Lagrangian Formulation of Elastic Wave Equation on Riemannian Manifolds RWE-C3-EAFIT Lagrangian Formulation of Elastic Wave Equation on Riemannian Manifolds Hector Roman Quiceno E. Advisors Ph.D Jairo Alberto Villegas G. Ph.D Diego Alberto Gutierrez I. Centro de Ciencias de

More information

Gaussian processes for inference in stochastic differential equations

Gaussian processes for inference in stochastic differential equations Gaussian processes for inference in stochastic differential equations Manfred Opper, AI group, TU Berlin November 6, 2017 Manfred Opper, AI group, TU Berlin (TU Berlin) inference in SDE November 6, 2017

More information

Invariant Nonholonomic Riemannian Structures on Three-Dimensional Lie Groups

Invariant Nonholonomic Riemannian Structures on Three-Dimensional Lie Groups Invariant Nonholonomic Riemannian Structures on Three-Dimensional Lie Groups Dennis I. Barrett Geometry, Graphs and Control (GGC) Research Group Department of Mathematics, Rhodes University Grahamstown,

More information

1 First and second variational formulas for area

1 First and second variational formulas for area 1 First and second variational formulas for area In this chapter, we will derive the first and second variational formulas for the area of a submanifold. This will be useful in our later discussion on

More information

Geodesic flows in manifolds of nonpositive curvature Patrick Eberlein. Table of Contents

Geodesic flows in manifolds of nonpositive curvature Patrick Eberlein. Table of Contents 1 Geodesic flows in manifolds of nonpositive curvature Patrick Eberlein Table of Contents I. Introduction - a quick historical survey of geodesic flows on negatively curved spaces. II. Preliminaries on

More information

Tangent bundles, vector fields

Tangent bundles, vector fields Location Department of Mathematical Sciences,, G5-109. Main Reference: [Lee]: J.M. Lee, Introduction to Smooth Manifolds, Graduate Texts in Mathematics 218, Springer-Verlag, 2002. Homepage for the book,

More information

Mathematical Relativity, Spring 2017/18 Instituto Superior Técnico

Mathematical Relativity, Spring 2017/18 Instituto Superior Técnico Mathematical Relativity, Spring 2017/18 Instituto Superior Técnico 1. Starting from R αβµν Z ν = 2 [α β] Z µ, deduce the components of the Riemann curvature tensor in terms of the Christoffel symbols.

More information

General tensors. Three definitions of the term V V. q times. A j 1...j p. k 1...k q

General tensors. Three definitions of the term V V. q times. A j 1...j p. k 1...k q General tensors Three definitions of the term Definition 1: A tensor of order (p,q) [hence of rank p+q] is a multilinear function A:V V }{{ V V R. }}{{} p times q times (Multilinear means linear in each

More information

Curvature homogeneity of type (1, 3) in pseudo-riemannian manifolds

Curvature homogeneity of type (1, 3) in pseudo-riemannian manifolds Curvature homogeneity of type (1, 3) in pseudo-riemannian manifolds Cullen McDonald August, 013 Abstract We construct two new families of pseudo-riemannian manifolds which are curvature homegeneous of

More information

Chapter 7 Curved Spacetime and General Covariance

Chapter 7 Curved Spacetime and General Covariance Chapter 7 Curved Spacetime and General Covariance In this chapter we generalize the discussion of preceding chapters to extend covariance to more general curved spacetimes. 145 146 CHAPTER 7. CURVED SPACETIME

More information

arxiv: v1 [math-ph] 2 Aug 2010

arxiv: v1 [math-ph] 2 Aug 2010 arxiv:1008.0360v1 [math-ph] 2 Aug 2010 Fractional Exact Solutions and Solitons in Gravity Dumitru Baleanu Department of Mathematics and Computer Sciences, Çankaya University, 06530, Ankara, Turkey Sergiu

More information

ECE 521. Lecture 11 (not on midterm material) 13 February K-means clustering, Dimensionality reduction

ECE 521. Lecture 11 (not on midterm material) 13 February K-means clustering, Dimensionality reduction ECE 521 Lecture 11 (not on midterm material) 13 February 2017 K-means clustering, Dimensionality reduction With thanks to Ruslan Salakhutdinov for an earlier version of the slides Overview K-means clustering

More information

Integration of non linear conservation laws?

Integration of non linear conservation laws? Integration of non linear conservation laws? Frédéric Hélein, Institut Mathématique de Jussieu, Paris 7 Advances in Surface Theory, Leicester, June 13, 2013 Harmonic maps Let (M, g) be an oriented Riemannian

More information

Data Analysis and Manifold Learning Lecture 9: Diffusion on Manifolds and on Graphs

Data Analysis and Manifold Learning Lecture 9: Diffusion on Manifolds and on Graphs Data Analysis and Manifold Learning Lecture 9: Diffusion on Manifolds and on Graphs Radu Horaud INRIA Grenoble Rhone-Alpes, France Radu.Horaud@inrialpes.fr http://perception.inrialpes.fr/ Outline of Lecture

More information

Mathematical Methods for Neurosciences. ENS - Master MVA Paris 6 - Master Maths-Bio ( )

Mathematical Methods for Neurosciences. ENS - Master MVA Paris 6 - Master Maths-Bio ( ) Mathematical Methods for Neurosciences. ENS - Master MVA Paris 6 - Master Maths-Bio (2014-2015) Etienne Tanré - Olivier Faugeras INRIA - Team Tosca November 26th, 2014 E. Tanré (INRIA - Team Tosca) Mathematical

More information

ON TOTALLY REAL SUBMANIFOLDS IN A NEARLY KÄHLER MANIFOLD *

ON TOTALLY REAL SUBMANIFOLDS IN A NEARLY KÄHLER MANIFOLD * PORTUGALIAE MATHEMATICA Vol. 58 Fasc. 2 2001 Nova Série ON TOTALLY REAL SUBMANIFOLDS IN A NEARLY KÄHLER MANIFOLD * Zhong Hua Hou Abstract: Let M m be a totally real submanifold of a nearly Kähler manifold

More information

Dyson series for the PDEs arising in Mathematical Finance I

Dyson series for the PDEs arising in Mathematical Finance I for the PDEs arising in Mathematical Finance I 1 1 Penn State University Mathematical Finance and Probability Seminar, Rutgers, April 12, 2011 www.math.psu.edu/nistor/ This work was supported in part by

More information

Relativistic Mechanics

Relativistic Mechanics Physics 411 Lecture 9 Relativistic Mechanics Lecture 9 Physics 411 Classical Mechanics II September 17th, 2007 We have developed some tensor language to describe familiar physics we reviewed orbital motion

More information

PCA and admixture models

PCA and admixture models PCA and admixture models CM226: Machine Learning for Bioinformatics. Fall 2016 Sriram Sankararaman Acknowledgments: Fei Sha, Ameet Talwalkar, Alkes Price PCA and admixture models 1 / 57 Announcements HW1

More information

Vectors. January 13, 2013

Vectors. January 13, 2013 Vectors January 13, 2013 The simplest tensors are scalars, which are the measurable quantities of a theory, left invariant by symmetry transformations. By far the most common non-scalars are the vectors,

More information

Lecture 5. vector bundles, gauge theory

Lecture 5. vector bundles, gauge theory Lecture 5 vector bundles, gauge theory tangent bundle In Lecture 2, we defined the tangent space T p M at each point p on M. Let us consider a collection of tangent bundles over every point on M, TM =

More information

The Algebra of Tensors; Tensors on a Vector Space Definition. Suppose V 1,,V k and W are vector spaces. A map. F : V 1 V k

The Algebra of Tensors; Tensors on a Vector Space Definition. Suppose V 1,,V k and W are vector spaces. A map. F : V 1 V k The Algebra of Tensors; Tensors on a Vector Space Definition. Suppose V 1,,V k and W are vector spaces. A map F : V 1 V k is said to be multilinear if it is linear as a function of each variable seperately:

More information

Riemannian and Sub-Riemannian Geodesic Flows

Riemannian and Sub-Riemannian Geodesic Flows Riemannian and Sub-Riemannian Geodesic Flows Mauricio Godoy Molina 1 Joint with E. Grong Universidad de La Frontera (Temuco, Chile) Potsdam, February 2017 1 Partially funded by grant Anillo ACT 1415 PIA

More information

General Relativity and Cosmology Mock exam

General Relativity and Cosmology Mock exam Physikalisches Institut Mock Exam Universität Bonn 29. June 2011 Theoretische Physik SS 2011 General Relativity and Cosmology Mock exam Priv. Doz. Dr. S. Förste Exercise 1: Overview Give short answers

More information

A Variational Model for Data Fitting on Manifolds by Minimizing the Acceleration of a Bézier Curve

A Variational Model for Data Fitting on Manifolds by Minimizing the Acceleration of a Bézier Curve A Variational Model for Data Fitting on Manifolds by Minimizing the Acceleration of a Bézier Curve Ronny Bergmann a Technische Universität Chemnitz Kolloquium, Department Mathematik, FAU Erlangen-Nürnberg,

More information

The concentration of a drug in blood. Exponential decay. Different realizations. Exponential decay with noise. dc(t) dt.

The concentration of a drug in blood. Exponential decay. Different realizations. Exponential decay with noise. dc(t) dt. The concentration of a drug in blood Exponential decay C12 concentration 2 4 6 8 1 C12 concentration 2 4 6 8 1 dc(t) dt = µc(t) C(t) = C()e µt 2 4 6 8 1 12 time in minutes 2 4 6 8 1 12 time in minutes

More information

arxiv: v1 [math-ph] 2 Aug 2010

arxiv: v1 [math-ph] 2 Aug 2010 arxiv:1008.0363v1 [math-ph] 2 Aug 2010 Fractional Analogous Models in Mechanics and Gravity Theories Dumitru Baleanu Department of Mathematics and Computer Sciences, Çankaya University, 06530, Ankara,

More information

Loos Symmetric Cones. Jimmie Lawson Louisiana State University. July, 2018

Loos Symmetric Cones. Jimmie Lawson Louisiana State University. July, 2018 Louisiana State University July, 2018 Dedication I would like to dedicate this talk to Joachim Hilgert, whose 60th birthday we celebrate at this conference and with whom I researched and wrote a big blue

More information

Geometry of Horizontal Bundles and Connections

Geometry of Horizontal Bundles and Connections Geometry of Horizontal Bundles and Connections Justin M. Ryan Wichita State University Department of Mathematics, Statistics, and Physics 23 April 2014 The Vertical Bundle Let fi : E æ M be a smooth fiber

More information

Elements of differential geometry

Elements of differential geometry Elements of differential geometry R.Beig (Univ. Vienna) ESI-EMS-IAMP School on Mathematical GR, 28.7. - 1.8. 2014 1. tensor algebra 2. manifolds, vector and covector fields 3. actions under diffeos and

More information

William P. Thurston. The Geometry and Topology of Three-Manifolds

William P. Thurston. The Geometry and Topology of Three-Manifolds William P. Thurston The Geometry and Topology of Three-Manifolds Electronic version 1.1 - March 00 http://www.msri.org/publications/books/gt3m/ This is an electronic edition of the 1980 notes distributed

More information

Method 1: Geometric Error Optimization

Method 1: Geometric Error Optimization Method 1: Geometric Error Optimization we need to encode the constraints ŷ i F ˆx i = 0, rank F = 2 idea: reconstruct 3D point via equivalent projection matrices and use reprojection error equivalent projection

More information

is constant [3]. In a recent work, T. IKAWA proved the following theorem for helices on a Lorentzian submanifold [1].

is constant [3]. In a recent work, T. IKAWA proved the following theorem for helices on a Lorentzian submanifold [1]. ANALELE ŞTIINŢIFICE ALE UNIVERSITĂŢII AL.I.CUZA IAŞI Tomul XLVI, s.i a, Matematică, 2000, f.2. ON GENERAL HELICES AND SUBMANIFOLDS OF AN INDEFINITE RIEMANNIAN MANIFOLD BY N. EKMEKCI Introduction. A regular

More information

Yuri Fedorov Multidimensional Integrable Generalizations of the nonholonomic Chaplygin sphere problem. November 29, 2006

Yuri Fedorov Multidimensional Integrable Generalizations of the nonholonomic Chaplygin sphere problem. November 29, 2006 Yuri Fedorov Multidimensional Integrable Generalizations of the nonholonomic Chaplygin sphere problem November 29, 2006 1 Preservation of Invariant Measure ẋ = v(x), x = (x 1,..., x n ) ( ) The volume

More information

Subriemannian geodesics - an introduction

Subriemannian geodesics - an introduction - an introduction Department of Theoretical Physics Comenius University Bratislava fecko@fmph.uniba.sk Student Colloqium and School on Mathematical Physics, Stará Lesná, Slovakia, August 24-30, 2009 -

More information

INTRODUCTION TO GEOMETRY

INTRODUCTION TO GEOMETRY INTRODUCTION TO GEOMETRY ERIKA DUNN-WEISS Abstract. This paper is an introduction to Riemannian and semi-riemannian manifolds of constant sectional curvature. We will introduce the concepts of moving frames,

More information

H-convex Riemannian submanifolds

H-convex Riemannian submanifolds H-convex Riemannian submanifolds Constantin Udrişte and Teodor Oprea Abstract. Having in mind the well known model of Euclidean convex hypersurfaces [4], [5] and the ideas in [1], many authors defined

More information

Index theory on manifolds with corners: Generalized Gauss-Bonnet formulas

Index theory on manifolds with corners: Generalized Gauss-Bonnet formulas Index theory on singular manifolds I p. 1/4 Index theory on singular manifolds I Index theory on manifolds with corners: Generalized Gauss-Bonnet formulas Paul Loya Index theory on singular manifolds I

More information

Sub-Riemannian Geometry: Basic Ideas and Examples

Sub-Riemannian Geometry: Basic Ideas and Examples Sub-Riemannian Geometry: Basic Ideas and Examples Abubakr Muhammad Dept. of Electrical Engineering Georgia Institute of Technology, Atlanta, GA, USA. abubakr@ece.gatech.edu Keywords Differential geometry;

More information

Math 550 / David Dumas / Fall Problems

Math 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 information

Stochastic gradient descent on Riemannian manifolds

Stochastic gradient descent on Riemannian manifolds Stochastic gradient descent on Riemannian manifolds Silvère Bonnabel 1 Centre de Robotique - Mathématiques et systèmes Mines ParisTech SMILE Seminar Mines ParisTech Novembre 14th, 2013 1 silvere.bonnabel@mines-paristech

More information

η = (e 1 (e 2 φ)) # = e 3

η = (e 1 (e 2 φ)) # = e 3 Research Statement My research interests lie in differential geometry and geometric analysis. My work has concentrated according to two themes. The first is the study of submanifolds of spaces with riemannian

More information

Statistical Analysis of Tensor Fields

Statistical Analysis of Tensor Fields Statistical Analysis of Tensor Fields Yuchen Xie Baba C. Vemuri Jeffrey Ho Department of Computer and Information Sciences and Engineering University of Florida Abstract. In this paper, we propose a Riemannian

More information

Machine Learning. B. Unsupervised Learning B.2 Dimensionality Reduction. Lars Schmidt-Thieme, Nicolas Schilling

Machine Learning. B. Unsupervised Learning B.2 Dimensionality Reduction. Lars Schmidt-Thieme, Nicolas Schilling Machine Learning B. Unsupervised Learning B.2 Dimensionality Reduction Lars Schmidt-Thieme, Nicolas Schilling Information Systems and Machine Learning Lab (ISMLL) Institute for Computer Science University

More information

Applications of Information Geometry to Hypothesis Testing and Signal Detection

Applications of Information Geometry to Hypothesis Testing and Signal Detection CMCAA 2016 Applications of Information Geometry to Hypothesis Testing and Signal Detection Yongqiang Cheng National University of Defense Technology July 2016 Outline 1. Principles of Information Geometry

More information

Second-Order Inference for Gaussian Random Curves

Second-Order Inference for Gaussian Random Curves Second-Order Inference for Gaussian Random Curves With Application to DNA Minicircles Victor Panaretos David Kraus John Maddocks Ecole Polytechnique Fédérale de Lausanne Panaretos, Kraus, Maddocks (EPFL)

More information

Tensor Analysis in Euclidean Space

Tensor Analysis in Euclidean Space Tensor Analysis in Euclidean Space James Emery Edited: 8/5/2016 Contents 1 Classical Tensor Notation 2 2 Multilinear Functionals 4 3 Operations With Tensors 5 4 The Directional Derivative 5 5 Curvilinear

More information