Semidefinite Programming and Harmonic Analysis

Size: px
Start display at page:

Download "Semidefinite Programming and Harmonic Analysis"

Transcription

1 1 / 74 Semidefinite Programming and Harmonic Analysis Cristóbal Guzmán CS Discrete Fourier Analysis and Applications March 7, 2012

2 2 / 74 Motivation SDP gives best relaxations known for some combinatorial optimization problems. Medium/large scale SDP are beyond the grasp of computational tractability (although polynomial). Many combinatorial applications are invariant under a group action. These symmetries lead to nice block decompositions of the SDP, making them computationally tractable.

3 3 / 74 Objectives Show the modeling capabilities of SDP. Show through examples how to exploit symmetries for simplifying SDP formulations. Generalize SDP models to infinite dimensions, through combinatorial applications.

4 4 / 74 Outline 1 Introduction Semidefinite Programming Lovász Theta Function Distance Graphs Infinite-Dimensional Lovász Relaxation 2 Exploiting Symmetry The Automorphism Group Symmetry and Optimization 3 Harmonic Analysis Rudiments of Representation Theory Application to Binary Codes Application to Kissing Numbers

5 5 / 74 Outline 1 Introduction Semidefinite Programming Lovász Theta Function Distance Graphs Infinite-Dimensional Lovász Relaxation 2 Exploiting Symmetry The Automorphism Group Symmetry and Optimization 3 Harmonic Analysis Rudiments of Representation Theory Application to Binary Codes Application to Kissing Numbers

6 6 / 74 Outline 1 Introduction Semidefinite Programming Lovász Theta Function Distance Graphs Infinite-Dimensional Lovász Relaxation 2 Exploiting Symmetry The Automorphism Group Symmetry and Optimization 3 Harmonic Analysis Rudiments of Representation Theory Application to Binary Codes Application to Kissing Numbers

7 Semidefinite Programming SDP: optimization of a linear function over the intersection of the cone of p.s.d. matrices with an affine subspace Figure: Spectrahedron given by 1 x 1 x 2 x 1 1 x 3 x 2 x / 74

8 8 / 74 Semidefinite Programming SDP: optimization of a linear function over the intersection of the cone of p.s.d. matrices with an affine subspace Figure: Spectrahedron given by 1 x 1 x 2 x 1 1 x 3 x 2 x 3 1 Primal problem (in standard form) m (P) min{c T x : A i x i B 0}. i=1 0.

9 9 / 74 Semidefinite Programming SDP: optimization of a linear function over the intersection of the cone of p.s.d. matrices with an affine subspace Figure: Spectrahedron given by 1 x 1 x 2 x 1 1 x 3 x 2 x 3 1 Primal problem (in standard form) m (P) min{c T x : A i x i B 0}. Dual problem i=1 0. (D) max{tr(bλ) : Tr(A i Λ) = c i, Λ 0}.

10 10 / 74 Remarks SDP is a matrix analogue of Linear Programming (A i, B diagonal) Expresive power: stability of linear systems, eigenvalue optimization, polynomial optimization, combinatorial relaxations, etc. Computationally Tractable : polynomial time interior point methods, and fast first-order methods available. Under strict primal (or dual) feasibility, we have strong duality OPT (P) = OPT (D) Note that these values are not necessarily achieved!

11 11 / 74 Outline 1 Introduction Semidefinite Programming Lovász Theta Function Distance Graphs Infinite-Dimensional Lovász Relaxation 2 Exploiting Symmetry The Automorphism Group Symmetry and Optimization 3 Harmonic Analysis Rudiments of Representation Theory Application to Binary Codes Application to Kissing Numbers

12 12 / 74 Independent sets in graphs G = (V, E) graph, V vertex set, E V 2 edge set. C V is independent if all pairs in C are not adjacent, ie x, y C : {x, y} / E. Figure: Blue vertices form an independent set.

13 13 / 74 Independent sets in graphs G = (V, E) graph, V vertex set, E V 2 edge set. C V is independent if all pairs in C are not adjacent, ie x, y C : {x, y} / E. Figure: Blue vertices form an independent set. For a finite graph G define the independence number α(g) = max{ C : C independent set}. Finding (or even approximating) α is NP-hard

14 14 / 74 Lovász Theta Function (1979) Theorem α(g) ϑ 1 (G). max x V y V K (x, y) s.t. Tr(K ) = 1, ϑ 1 (G) = K (x, y) = 0 if xy E K 0 Proof. Let C V be an independent set, and 1 C {0, 1} V its characteristic vector. Then K = 1 C 1 C1 T C is feasible for the SDP and x,y V K (x, y) = C. Thus, C ϑ 1 (G).

15 15 / 74 Theta Function: Dual Form min λ s.t. K (x, x) = λ 1 x V ϑ 1 (G) = K (x, y) = 1 xy / E K 0

16 16 / 74 Remarks: Theta Function: Dual Form min λ s.t. K (x, x) = λ 1 x V ϑ 1 (G) = K (x, y) = 1 xy / E K 0 Lovász relaxation is stronger than LP relaxation. Lovász upper bound is weak, there exists an infinite family of graphs with ϑ α > n 2 c logn This makes sense: the stability number is hard to approximate up to a factor n 1 ɛ For symmetric graphs, Lovász relaxation might give tighter bounds

17 17 / 74 Outline 1 Introduction Semidefinite Programming Lovász Theta Function Distance Graphs Infinite-Dimensional Lovász Relaxation 2 Exploiting Symmetry The Automorphism Group Symmetry and Optimization 3 Harmonic Analysis Rudiments of Representation Theory Application to Binary Codes Application to Kissing Numbers

18 18 / 74 Distance Graphs Let V be a (separable) compact metric space, with distance d : V V R +. µ is a finite (regular Borel) measure. Adjacency in the distance graph is determined by the distance d Definition Let I R +. The distance graph G(V, I) = (V, E) is a graph with vertex set V and edge set E = {xy : d(x, y) I}.

19 19 / 74 A Familiar Example: Finite Connected Graphs Let G = (V, E) be a finite graph Let d : V V N {+ } given by { length of shortest xy-path, xy-path d(x, y) = +, xy-path If G is connected, then d is a metric on V Let µ(a) = A, so µ is the uniform measure In this case, E can be defined only in terms of the metric, namely xy E iff d(x, y) = 1

20 20 / 74 Example: Kissing Numbers τ n = maximum number of unit spheres which can touch a central unit sphere without pairwise overlapping Figure: Optimal kissing configurations for dimensions 2 and 3

21 21 / 74 Example: Kissing Numbers τ n = maximum number of unit spheres which can touch a central unit sphere without pairwise overlapping Figure: Optimal kissing configurations for dimensions 2 and : τ 3 =? 12 (Isaac Newton) or 13 (David Gregory)?

22 22 / 74 Kissing Numbers: Some History τ 1 = 2, τ 2 = 6 are trivial τ 3 = 12, Shütte, van der Waerden (1953), by elementary geometry (difficult) τ 8 = 240, τ 24 = , Odlyzko, Sloane and Levenshtein (1979), by LP (relatively easy, and elegant). τ 4 = 24, Musin (2003), by LP and elementary geometry (difficult) Unified proof (and more) with SDP!

23 23 / 74 Outline 1 Introduction Semidefinite Programming Lovász Theta Function Distance Graphs Infinite-Dimensional Lovász Relaxation 2 Exploiting Symmetry The Automorphism Group Symmetry and Optimization 3 Harmonic Analysis Rudiments of Representation Theory Application to Binary Codes Application to Kissing Numbers

24 24 / 74 Kissing Numbers and Distance Graphs V = S n 1 := {x R n : x 2 2 = 1}, the unit sphere. d(x, y) = arccos(x y) is a metric on V. e.g. d(x, x) = π. ω, the surface area induced by the Lebesgue measure is a measure on V. ω(s n 1 ) = 2π n/2 /Γ(n/2).

25 25 / 74 Kissing Numbers and Distance Graphs V = S n 1 := {x R n : x 2 2 = 1}, the unit sphere. d(x, y) = arccos(x y) is a metric on V. e.g. d(x, x) = π. ω, the surface area induced by the Lebesgue measure is a measure on V. ω(s n 1 ) = 2π n/2 /Γ(n/2). G(V, (0, θ)) is a distance graph. Stable sets in this graph are called spherical codes with minimal angular distance θ. A(n, θ) = α( G(S n 1, (0, θ)) ) Note: n-th kissing number is equivalent to A(n, π/3).

26 26 / 74 Positive Hilbert-Schmidt Kernels For Lovász SDP we need p.s.d infinite matrices C(S n 1 ) = {f : S n 1 R f continuous} Hilbert-Schmidt kernel: elements of C(S n 1 S n 1 ) Symmetric (or Hermitian) HS kernel: K (x, y) = K (y, x), for all x, y S n 1 Definition A symmetric HS kernel K is called positive if for all f C(S n 1 S n 1 ): K (x, y)f (x)f (y) dω(x) dω(y) 0 S n 1 S n 1 As in the matrix case, we will use the shorthand notation K 0 for positive kernels K.

27 27 / 74 Lovász Relaxation For a distance graph G(V, I), we get an infinite-dimensional SDP relaxation min λ s.t. K (x, x) = λ 1, x V ϑ(g(v, I)) = K (x, y) = 1, xy / E K 0

28 28 / 74 Lovász Relaxation For Kissing numbers min λ ϑ(s n 1 s.t. K (x, x) = λ 1, x S, (0, π/3)) = n 1 K (x, y) = 1, x y π/3 K 0

29 29 / 74 Theorem α(g(v, I)) ϑ(g(v, I)). Proof. Let C be a stable set in G(V, I) and let K be any feasible kernel for the relaxation. We will prove C λ. Consider the matrix (K (c, c )) c,c C2. It is p.s.d., thus 1 T K 1 = c,c K (c, c ) 0. On the other hand c,c K (c, c ) = K (c, c) + K (c, c ) c c c C K (c, c) C ( C 1) Therefore, C 1 K (c, c) = λ 1.

30 30 / 74 Error-Correcting (Binary) Codes Model for robust communication accross noisy channels, and store/retrieve information on media Idea: send redundant messages, so receiver can detect and correct errors, under some assumptions.

31 31 / 74 Error-Correcting (Binary) Codes Model for robust communication accross noisy channels, and store/retrieve information on media Idea: send redundant messages, so receiver can detect and correct errors, under some assumptions. V = {0, 1} n, the Hamming cube. δ(x, y) = {i [n] : x i y i } is a metric on V

32 32 / 74 Error-Correcting (Binary) Codes Model for robust communication accross noisy channels, and store/retrieve information on media Idea: send redundant messages, so receiver can detect and correct errors, under some assumptions. V = {0, 1} n, the Hamming cube. δ(x, y) = {i [n] : x i y i } is a metric on V A(n, d)= maximal cardinality of C V such that every two points in C have Hamming distance at least d. Thus, A(n, d) = α({0, 1} n, {1,..., d 1}) A stable set in this graph can correct (d 1)/2 errors A(n, d) is unknown to large extend

33 33 / 74 Outline 1 Introduction Semidefinite Programming Lovász Theta Function Distance Graphs Infinite-Dimensional Lovász Relaxation 2 Exploiting Symmetry The Automorphism Group Symmetry and Optimization 3 Harmonic Analysis Rudiments of Representation Theory Application to Binary Codes Application to Kissing Numbers

34 34 / 74 Outline 1 Introduction Semidefinite Programming Lovász Theta Function Distance Graphs Infinite-Dimensional Lovász Relaxation 2 Exploiting Symmetry The Automorphism Group Symmetry and Optimization 3 Harmonic Analysis Rudiments of Representation Theory Application to Binary Codes Application to Kissing Numbers

35 35 / 74 Automorphism Group Definition Given a distance graph G(V, I), we define its automorphism group as Aut(G) = {u : V V : u bijective, d(x, y) = d(ux, uy) x, y V }

36 36 / 74 Automorphism Group Definition Given a distance graph G(V, I), we define its automorphism group as Aut(G) = {u : V V : u bijective, d(x, y) = d(ux, uy) x, y V } Examples: For the distance graph defined by shortest paths in G = (V, E), the automorphism group is Aut(G) = {u : V V : u bijective, (x, y) E iff (ux, uy) E} Deciding whether this group is trivial is as hard as the Graph Isomorphism Problem

37 37 / 74 Automorphism Group Aut(G) = {u : V V : u bij., d(x, y) = d(ux, uy) x, y V } Examples: For the Hamming cube, Aut(G) has order 2 n n!. It is generated by all n! permutations and all 2 n switches 0 1. For S n 1, the automorphism group is the orthogonal group Aut(S n 1 ) = O(R n ) := {u M n (R) : u T u = I} This group not only preserves distance, but also is invariant for, x, y = ux, uy u O(R n )

38 38 / 74 Outline 1 Introduction Semidefinite Programming Lovász Theta Function Distance Graphs Infinite-Dimensional Lovász Relaxation 2 Exploiting Symmetry The Automorphism Group Symmetry and Optimization 3 Harmonic Analysis Rudiments of Representation Theory Application to Binary Codes Application to Kissing Numbers

39 39 / 74 Symmetry and Optimization Previous examples are invariant by Aut(G) (e.g. flip-permute in {0, 1} n, rotate kissing configurations). Lovász SDP is reduced by exploting these symmetries. Definition A positive kernel K is Aut(V )-invariant if for all u Aut(V ) K (ux, uy) = K (x, y)

40 Symmetry and Optimization Previous examples are invariant by Aut(G) (e.g. flip-permute in {0, 1} n, rotate kissing configurations). Lovász SDP is reduced by exploting these symmetries. Definition A positive kernel K is Aut(V )-invariant if for all u Aut(V ) K (ux, uy) = K (x, y) Remark: If K is feasible for the Lovász SDP, then so it is its Aut(V )-invariant group average: K (x, y) = 1 Aut(V ) u Aut(V ) K (ux, uy) Change of variables by u preserves feasibility and value. 40 / 74

41 41 / 74 Lovász SDP for A(n, d): Example: Binary codes SDP min λ s.t. K (x, x) = λ 1 x {0, 1} n ϑ = K (x, y) 1 δ(x, y) {d,..., n} K 0 K is Aut({0, 1} n )-inv.

42 42 / 74 Lovász SDP for A(n, d): Example: Binary codes SDP min λ s.t. K (x, x) = λ 1 x {0, 1} n ϑ = K (x, y) 1 δ(x, y) {d,..., n} K 0 K is Aut({0, 1} n )-inv. We will show that the two last constraints can be re-written as K (x, y) = n f k Kk n (δ(x, y)) k=1 where f k 0 are constants, and Kk n(t) = n are the Krawtchouk polynomials. t=0 ( t i )( n t k i) ( 1) i,

43 43 / 74 Example: Binary codes LP By this last theorem, we have the following LP in f 1,..., f n min 1 + n ϑ k=0 f kkk n(0) = s.t. n k=0 f kkk n (t) 1 t = d,..., n f k 0 t = 1,..., n The proof of the representation theorem requires tools from Harmonic Analysis.

44 44 / 74 Outline 1 Introduction Semidefinite Programming Lovász Theta Function Distance Graphs Infinite-Dimensional Lovász Relaxation 2 Exploiting Symmetry The Automorphism Group Symmetry and Optimization 3 Harmonic Analysis Rudiments of Representation Theory Application to Binary Codes Application to Kissing Numbers

45 45 / 74 Outline 1 Introduction Semidefinite Programming Lovász Theta Function Distance Graphs Infinite-Dimensional Lovász Relaxation 2 Exploiting Symmetry The Automorphism Group Symmetry and Optimization 3 Harmonic Analysis Rudiments of Representation Theory Application to Binary Codes Application to Kissing Numbers

46 46 / 74 Aut(V ) as a Topological Group Γ =Aut(V ) is a compact topological group (product and inversion are continuous) There always exists the Haar (a.k.a left-invariant) measure f (vu) = f (u) v Γ, f C(Γ) Γ Γ Examples: uniform distribution, Lebesgue measure. Given a positive kernel K, we define its group average as K (x, y) = K (ux, uy)du Γ

47 47 / 74 Γ =Aut(V ) acts on C(V ) by uf (x) = f (u 1 x) This action is linear u(αf + βg) = αuf + βug (f, g) := fg is an invariant inner product for C(V ) Γ (f, g) = (uf, ug) f, g C(V ), u Γ

48 Γ =Aut(V ) acts on C(V ) by uf (x) = f (u 1 x) This action is linear u(αf + βg) = αuf + βug (f, g) := fg is an invariant inner product for C(V ) Γ (f, g) = (uf, ug) f, g C(V ), u Γ A subspace S C(V ) is Γ-invariant if us = S, u Γ. S is Γ-irreducible if {0} and S are the only Γ-invariant subspaces of S If V is Γ-invariant, then V is Γ-invariant too Irreducible spaces are either equivalent ( linear map between orthonormal bases) or orthogonal! 48 / 74

49 49 / 74 Peter-Weyl Theorem Theorem (Peter & Weyl (1927)) All irreducible subspaces of C(V ) are of finite dimension The space C(V ) decomposes orthogonally as C(V ) = H k, k N Each space H k decomposes orthogonally in irreducible spaces m k H k = H k,i, where H k,i = Hk,i iff k = k The dimension h k of H k,i is finite, but the multiplicity m k might be infinite. i=1

50 50 / 74 Bochner s Representation Theorem Theorem (Bochner (1941)) Let e k,i,l be an complete orthonormal system for C(V ) (by Peter-Weyl). Every Aut(V )-invariant, positive K is s.t. K (x, y) = k N or more economically as h k f k,ij i,j {1,...,m k } l=1 K (x, y) = F k, Z (x,y) k, k N e k,i,l (x)e k,j,l (y), with (F k ) ij = f k,ij and (Z (x,y) k ) ij = h k l=1 e k,i,l(x)e k,j,l (y). Each F k is Hermitian and positive. The series converges absolutely and uniformly.

51 51 / 74 About Bochner s Theorem Recall the spectral theorem for symmetric matrices n K = PDP T i.e. K (x, y) = λ k e k (x)e k (y). k=1

52 52 / 74 About Bochner s Theorem Recall the spectral theorem for symmetric matrices n K = PDP T i.e. K (x, y) = λ k e k (x)e k (y). k=1 Peter-Weyl theorem is nothing but R n = e 1... e n. Matrix diagonalization contains one block of 1-dimensional (equivalent) subspaces (multiplicity n)

53 53 / 74 About Bochner s Theorem Recall the spectral theorem for symmetric matrices n K = PDP T i.e. K (x, y) = λ k e k (x)e k (y). k=1 Peter-Weyl theorem is nothing but R n = e 1... e n. Matrix diagonalization contains one block of 1-dimensional (equivalent) subspaces (multiplicity n) These are the irreducible subspaces for Γ = {1}. Not so interesting after all, since we only imposed K invariant for the identity.

54 54 / 74 Outline 1 Introduction Semidefinite Programming Lovász Theta Function Distance Graphs Infinite-Dimensional Lovász Relaxation 2 Exploiting Symmetry The Automorphism Group Symmetry and Optimization 3 Harmonic Analysis Rudiments of Representation Theory Application to Binary Codes Application to Kissing Numbers

55 55 / 74 Particular cases: Boolean Harmonics Theorem (Peter-Weyl for the Hamming Cube) C({0, 1} n ) = H 0 H 1... H n. H k is spanned by the characters χ y (x) = ( 1) y x of deg. k H k is irreducible, with dimension h k = ( ) n k

56 56 / 74 Particular cases: Boolean Harmonics Theorem (Peter-Weyl for the Hamming Cube) C({0, 1} n ) = H 0 H 1... H n. H k is spanned by the characters χ y (x) = ( 1) y x of deg. k H k is irreducible, with dimension h k = ( ) n k Remarks: Since the multiplicities are 1, positive kernels only involve psd matrices of size 1 1, ie, nonnegative scalars. Aut({0, 1} n ) acts distance-transitively on pairs δ(x, y) = δ(x, y ) u : (ux, uy) = (x, y ). This implies, invariant kernels only depend on the distance.

57 57 / 74 Boolean Harmonics Representation: Proof Idea By Fourier analysis, (χ y ) y {0,1} n is an orthonormal basis for C({0, 1} n ). dim χ y =1, and is invariant under ({0, 1} n, + mod 2 ) (switches) uχ y (x) = χ y (x u) = ( 1) y ( u) χ y (x) χ y. Under the complete group Aut({0, 1} n ) (i.e., also permutations), these one dimensional subspaces are grouped together, according to their degree H k = {χ y : deg(y) = k} are Aut({0, 1} n )-invariant

58 58 / 74 Boolean Harmonics Representation: Proof Idea Claim: H k is irreducible, for k = 0,..., n. But first Definition f : {0, 1} n C is a zonal spherical function with pole e {0, 1} n if u Aut({0, 1} n ) & ue = e uf = f. Lemma Let U C({0, 1} n ) be an invariant subspace and let e {0, 1}. If the subspace of zonal spherical functions with pole e that lie in U has dimension one, then U is irreducible.

59 59 / 74 Proof of the lemma Proof. If U is not irreducible, then U = V V, both invariant. Let e 1,..., e n and f 1,..., f m orthonormal bases for V and V, resp. Let z V (x) = n e i (e)e i (x), z V (x) = i=1 m f j (e)f j (x) j=1 These are two l.i. zonal spherical functions with pole e, a contradiction.

60 60 / 74 Boolean Harmonics: Proof Idea Apply the lemma to H k, for pole e = (0,..., 0). Let f H k be a zonal spherical function with pole 0

61 61 / 74 Boolean Harmonics: Proof Idea Apply the lemma to H k, for pole e = (0,..., 0). Let f H k be a zonal spherical function with pole 0 Group stabilizing 0 = all permutations (no switching!).

62 Boolean Harmonics: Proof Idea Apply the lemma to H k, for pole e = (0,..., 0). Let f H k be a zonal spherical function with pole 0 Group stabilizing 0 = all permutations (no switching!). Thus, for any u σ n, uf = f. Take Fourier transforms u σ n ˆf (y)χuy (x) = ˆf (y)χy (x). y {0,1} n deg(y)=k y {0,1} n deg(y)=k 62 / 74

63 Boolean Harmonics: Proof Idea Apply the lemma to H k, for pole e = (0,..., 0). Let f H k be a zonal spherical function with pole 0 Group stabilizing 0 = all permutations (no switching!). Thus, for any u σ n, uf = f. Take Fourier transforms u σ n ˆf (y)χuy (x) = ˆf (y)χy (x). y {0,1} n deg(y)=k y {0,1} n deg(y)=k Therefore, all Fourier coefficients have to coincide! i.e., subspace has dimension / 74

64 64 / 74 Proof. Bochner s Representation Theorem H k s are not equivalent, so they are in different blocks. Let us compute Bochner s representation functions Z k (x, x ) = χ y (x)χ y (x ). deg(y)=k Recall that Z k (x, x ) only depends on t = δ(x, x ). Thus, Z k (x, x ) = Z k ((0,..., 0), (1,..., 1, 0,..., 0) }{{}}{{} t n t

65 65 / 74 Proof. Bochner s Representation Theorem H k s are not equivalent, so they are in different blocks. Let us compute Bochner s representation functions Z k (x, x ) = χ y (x)χ y (x ). deg(y)=k Recall that Z k (x, x ) only depends on t = δ(x, x ). Thus, Z k (x, x ) = Z k ((0,..., 0), (1,..., 1, 0,..., 0) }{{}}{{} t n t k ( )( ) t n t = ( 1) i i k i i=1

66 66 / 74 Proof. Bochner s Representation Theorem H k s are not equivalent, so they are in different blocks. Let us compute Bochner s representation functions Z k (x, x ) = χ y (x)χ y (x ). deg(y)=k Recall that Z k (x, x ) only depends on t = δ(x, x ). Thus, Z k (x, x ) = Z k ((0,..., 0), (1,..., 1, 0,..., 0) }{{}}{{} t n t k ( )( ) t n t = ( 1) i i k i i=1 = K n k (t)

67 67 / 74 Lovász SDP for A(n, d): Finally: Binary Codes Invariant Relaxation min λ s.t. K (x, x) = λ 1 x {0, 1} n ϑ = K (x, y) 1 δ(x, y) {d,..., n} K 0 K is Aut({0, 1} n )-inv. By Bochner s theorem for Boolean harmonics, we get an LP in f 1,..., f n min 1 + n ϑ k=0 f kkk n(0) = s.t. n k=0 f kkk n (t) 1 t = d,..., n f k 0 t = 1,..., n

68 68 / 74 Outline 1 Introduction Semidefinite Programming Lovász Theta Function Distance Graphs Infinite-Dimensional Lovász Relaxation 2 Exploiting Symmetry The Automorphism Group Symmetry and Optimization 3 Harmonic Analysis Rudiments of Representation Theory Application to Binary Codes Application to Kissing Numbers

69 69 / 74 Spherical Harmonics Theorem (Peter-Weyl for Spherical Harmonics) The space of complex-valued continuous functions on the unit sphere decomposes orthogonally into pairwise O(R n )-irreducible spaces C(S n 1 ) = k N H k, where H k is the space of homogeneous polynomials f of degree k which vanish under the Laplace operator f (x) = 0, x S n 1. H k has dimension h k = ( ) ( n+k 1 n 1 n+k 3 ) n 1.

70 70 / 74 Spherical Harmonics (Bochner theorem) Theorem (Schoenberg (1940)) Every positive, O(R n )-invariant kernel K C(S n 1 S n 1 ) can be written as where K (x, y) = k=0 f k Pk n (x y), The RHS converges absolutely and uniformly f k 0 for all k, and k f k < + (P n k ) k N is an orthonormal basis for L 2 ([ 1, 1], (1 t 2 ) (n 3)/2 dt) (Jacobi polynomials).

71 71 / 74 Kissing Numbers Invariant Relaxation min λ s.t. K (x, x) = λ 1, x S n 1 τ n = K (x, y) = 1, x y π/3 K 0 K is O(R n ) invariant By Schoenberg s characterization, we get an infinite-dimensional LP min λ s.t. τ n = f k 0, k N k N f kpk n(1) = λ 1 k N f kpk n (t) 1, t [ 1, 1/2]

72 72 / 74 Kissing Numbers Invariant relaxation min τ n = λ s.t. f k 0, k N k N f kpk n(1) = λ 1 k N f kpk n (t) 1, t [ 1, 1/2] We can truncate the series, and use sums-of-squares relaxations for the last constraint (recall Hilbert s seventeenth problem). This still provides upper bounds. This gives new proofs for τ 8 = 240, and τ 24 =

73 73 / 74 Other Discrete Geometric Problems Problem Vertex set Harmonic New results analysis Kissing S n 1 Compact, Unified numbers non-abelian proofs Coloring R n Non-compact, Bounds for abelian n = 4,..., 24, new asymptotics Sphere R n Non-compact, Bounds for packing abelian n = 4,..., 36 Body R n O(R n ) Non-compact, In progress Packing non-abelian

74 74 / 74 Bibliography F. Vallentin. Lecture notes: Semidefinite programing and harmonic analysis. Tutorial for HPOPT 2008, in Tilburg Univ., 2008 C. Bachoc, D.C. Gijswijt, A. Schrijver, F. Vallentin. Invariant semidefinite programs. In Handbook on Semidefinite, Conic and Polynomial Optimization (M.F. Anjos, J.B. Lasserre (ed.)). Springer, H. Cohn, N. Elkies. New upper bounds on sphere packings, I & II. Annals of Mathematics 157, W. Rudin. Fourier analysis on groups. Wiley classics library, 1990

Invariant Semidefinite Programs

Invariant Semidefinite Programs Invariant Semidefinite Programs Christine Bachoc Université Bordeaux I, IMB Modern trends in Optimization and its Application, IPAM Optimization Tutorials, september 14-17, 2010 Outline of Part I Invariant

More information

Applications of semidefinite programming in Algebraic Combinatorics

Applications of semidefinite programming in Algebraic Combinatorics Applications of semidefinite programming in Algebraic Combinatorics Tohoku University The 23rd RAMP Symposium October 24, 2011 We often want to 1 Bound the value of a numerical parameter of certain combinatorial

More information

SDP Relaxations for MAXCUT

SDP Relaxations for MAXCUT SDP Relaxations for MAXCUT from Random Hyperplanes to Sum-of-Squares Certificates CATS @ UMD March 3, 2017 Ahmed Abdelkader MAXCUT SDP SOS March 3, 2017 1 / 27 Overview 1 MAXCUT, Hardness and UGC 2 LP

More information

Packing, coding, and ground states From information theory to physics. Lecture III. Packing and energy minimization bounds in compact spaces

Packing, coding, and ground states From information theory to physics. Lecture III. Packing and energy minimization bounds in compact spaces Packing, coding, and ground states From information theory to physics Lecture III. Packing and energy minimization bounds in compact spaces Henry Cohn Microsoft Research New England Pair correlations For

More information

REPRESENTATION THEORY WEEK 7

REPRESENTATION THEORY WEEK 7 REPRESENTATION THEORY WEEK 7 1. Characters of L k and S n A character of an irreducible representation of L k is a polynomial function constant on every conjugacy class. Since the set of diagonalizable

More information

MIT Algebraic techniques and semidefinite optimization May 9, Lecture 21. Lecturer: Pablo A. Parrilo Scribe:???

MIT Algebraic techniques and semidefinite optimization May 9, Lecture 21. Lecturer: Pablo A. Parrilo Scribe:??? MIT 6.972 Algebraic techniques and semidefinite optimization May 9, 2006 Lecture 2 Lecturer: Pablo A. Parrilo Scribe:??? In this lecture we study techniques to exploit the symmetry that can be present

More information

Semidefinite and Second Order Cone Programming Seminar Fall 2001 Lecture 5

Semidefinite and Second Order Cone Programming Seminar Fall 2001 Lecture 5 Semidefinite and Second Order Cone Programming Seminar Fall 2001 Lecture 5 Instructor: Farid Alizadeh Scribe: Anton Riabov 10/08/2001 1 Overview We continue studying the maximum eigenvalue SDP, and generalize

More information

Applications of Semidefinite Programming to Coding Theory

Applications of Semidefinite Programming to Coding Theory Applications of Semidefinite Programming to Coding Theory Christine Bachoc Institut de Mathématiques de Bordeaux Université Bordeaux 1 351, cours de la Libération, 33405 Talence, France Email: {christine.bachoc}@math.u-bordeaux1.fr

More information

Introduction to Semidefinite Programming I: Basic properties a

Introduction to Semidefinite Programming I: Basic properties a Introduction to Semidefinite Programming I: Basic properties and variations on the Goemans-Williamson approximation algorithm for max-cut MFO seminar on Semidefinite Programming May 30, 2010 Semidefinite

More information

Representation Theory

Representation Theory Representation Theory Representations Let G be a group and V a vector space over a field k. A representation of G on V is a group homomorphism ρ : G Aut(V ). The degree (or dimension) of ρ is just dim

More information

Spherical Euclidean Distance Embedding of a Graph

Spherical Euclidean Distance Embedding of a Graph Spherical Euclidean Distance Embedding of a Graph Hou-Duo Qi University of Southampton Presented at Isaac Newton Institute Polynomial Optimization August 9, 2013 Spherical Embedding Problem The Problem:

More information

Semidefinite programming bounds for spherical codes

Semidefinite programming bounds for spherical codes Semidefinite programming bounds for spherical codes Christine Bachoc IMB, Université Bordeaux I 351, cours de la Libération 33405 Talence France bachoc@mathu-bordeaux1fr Fran Vallentin Centrum voor Wisunde

More information

Invariant semidefinite programs

Invariant semidefinite programs Invariant semidefinite programs Christine Bachoc 1, Dion C. Gijswijt 2, Alexander Schrijver 3, and Frank Vallentin 4 1 Laboratoire A2X, Université Bordeaux I, 351, cours de la Libération, 33405 Talence,

More information

CSC Linear Programming and Combinatorial Optimization Lecture 10: Semidefinite Programming

CSC Linear Programming and Combinatorial Optimization Lecture 10: Semidefinite Programming CSC2411 - Linear Programming and Combinatorial Optimization Lecture 10: Semidefinite Programming Notes taken by Mike Jamieson March 28, 2005 Summary: In this lecture, we introduce semidefinite programming

More information

Since G is a compact Lie group, we can apply Schur orthogonality to see that G χ π (g) 2 dg =

Since G is a compact Lie group, we can apply Schur orthogonality to see that G χ π (g) 2 dg = Problem 1 Show that if π is an irreducible representation of a compact lie group G then π is also irreducible. Give an example of a G and π such that π = π, and another for which π π. Is this true for

More information

Elementary linear algebra

Elementary linear algebra Chapter 1 Elementary linear algebra 1.1 Vector spaces Vector spaces owe their importance to the fact that so many models arising in the solutions of specific problems turn out to be vector spaces. The

More information

Dimension reduction for semidefinite programming

Dimension reduction for semidefinite programming 1 / 22 Dimension reduction for semidefinite programming Pablo A. Parrilo Laboratory for Information and Decision Systems Electrical Engineering and Computer Science Massachusetts Institute of Technology

More information

6-1 The Positivstellensatz P. Parrilo and S. Lall, ECC

6-1 The Positivstellensatz P. Parrilo and S. Lall, ECC 6-1 The Positivstellensatz P. Parrilo and S. Lall, ECC 2003 2003.09.02.10 6. The Positivstellensatz Basic semialgebraic sets Semialgebraic sets Tarski-Seidenberg and quantifier elimination Feasibility

More information

CHAPTER 6. Representations of compact groups

CHAPTER 6. Representations of compact groups CHAPTER 6 Representations of compact groups Throughout this chapter, denotes a compact group. 6.1. Examples of compact groups A standard theorem in elementary analysis says that a subset of C m (m a positive

More information

MAT 445/ INTRODUCTION TO REPRESENTATION THEORY

MAT 445/ INTRODUCTION TO REPRESENTATION THEORY MAT 445/1196 - INTRODUCTION TO REPRESENTATION THEORY CHAPTER 1 Representation Theory of Groups - Algebraic Foundations 1.1 Basic definitions, Schur s Lemma 1.2 Tensor products 1.3 Unitary representations

More information

LOWER BOUNDS FOR MEASURABLE CHROMATIC NUMBERS

LOWER BOUNDS FOR MEASURABLE CHROMATIC NUMBERS LOWER BOUNDS FOR MEASURABLE CHROMATIC NUMBERS CHRISTINE BACHOC, GABRIELE NEBE, FERNANDO MÁRIO DE OLIVEIRA FILHO, AND FRANK VALLENTIN ABSTRACT. The Lovász theta function provides a lower bound for the chromatic

More information

Approximation Algorithms

Approximation Algorithms Approximation Algorithms Chapter 26 Semidefinite Programming Zacharias Pitouras 1 Introduction LP place a good lower bound on OPT for NP-hard problems Are there other ways of doing this? Vector programs

More information

Lecture Semidefinite Programming and Graph Partitioning

Lecture Semidefinite Programming and Graph Partitioning Approximation Algorithms and Hardness of Approximation April 16, 013 Lecture 14 Lecturer: Alantha Newman Scribes: Marwa El Halabi 1 Semidefinite Programming and Graph Partitioning In previous lectures,

More information

BOUNDS FOR SOLID ANGLES OF LATTICES OF RANK THREE

BOUNDS FOR SOLID ANGLES OF LATTICES OF RANK THREE BOUNDS FOR SOLID ANGLES OF LATTICES OF RANK THREE LENNY FUKSHANSKY AND SINAI ROBINS Abstract. We find sharp absolute consts C and C with the following property: every well-rounded lattice of rank 3 in

More information

Spectral Measures, the Spectral Theorem, and Ergodic Theory

Spectral Measures, the Spectral Theorem, and Ergodic Theory Spectral Measures, the Spectral Theorem, and Ergodic Theory Sam Ziegler The spectral theorem for unitary operators The presentation given here largely follows [4]. will refer to the unit circle throughout.

More information

Semidefinite Programming

Semidefinite Programming Semidefinite Programming Notes by Bernd Sturmfels for the lecture on June 26, 208, in the IMPRS Ringvorlesung Introduction to Nonlinear Algebra The transition from linear algebra to nonlinear algebra has

More information

arxiv: v1 [math.mg] 12 Jun 2012

arxiv: v1 [math.mg] 12 Jun 2012 UPPER BOUNDS FOR PACKINGS OF SPHERES OF SEVERAL RADII DAVID DE LAAT, FERNANDO MÁRIO DE OLIVEIRA FILHO, AND FRANK VALLENTIN arxiv:1206.2608v1 [math.mg] 12 Jun 2012 Abstract. We give theorems that can be

More information

REPRESENTATION THEORY OF S n

REPRESENTATION THEORY OF S n REPRESENTATION THEORY OF S n EVAN JENKINS Abstract. These are notes from three lectures given in MATH 26700, Introduction to Representation Theory of Finite Groups, at the University of Chicago in November

More information

1 The independent set problem

1 The independent set problem ORF 523 Lecture 11 Spring 2016, Princeton University Instructor: A.A. Ahmadi Scribe: G. Hall Tuesday, March 29, 2016 When in doubt on the accuracy of these notes, please cross chec with the instructor

More information

The Hamming Codes and Delsarte s Linear Programming Bound

The Hamming Codes and Delsarte s Linear Programming Bound The Hamming Codes and Delsarte s Linear Programming Bound by Sky McKinley Under the Astute Tutelage of Professor John S. Caughman, IV A thesis submitted in partial fulfillment of the requirements for the

More information

Chapter 7: Bounded Operators in Hilbert Spaces

Chapter 7: Bounded Operators in Hilbert Spaces Chapter 7: Bounded Operators in Hilbert Spaces I-Liang Chern Department of Applied Mathematics National Chiao Tung University and Department of Mathematics National Taiwan University Fall, 2013 1 / 84

More information

The 123 Theorem and its extensions

The 123 Theorem and its extensions The 123 Theorem and its extensions Noga Alon and Raphael Yuster Department of Mathematics Raymond and Beverly Sackler Faculty of Exact Sciences Tel Aviv University, Tel Aviv, Israel Abstract It is shown

More information

The maximal stable set problem : Copositive programming and Semidefinite Relaxations

The maximal stable set problem : Copositive programming and Semidefinite Relaxations The maximal stable set problem : Copositive programming and Semidefinite Relaxations Kartik Krishnan Department of Mathematical Sciences Rensselaer Polytechnic Institute Troy, NY 12180 USA kartis@rpi.edu

More information

CONSTRAINED PERCOLATION ON Z 2

CONSTRAINED PERCOLATION ON Z 2 CONSTRAINED PERCOLATION ON Z 2 ZHONGYANG LI Abstract. We study a constrained percolation process on Z 2, and prove the almost sure nonexistence of infinite clusters and contours for a large class of probability

More information

Chapter 2 Linear Transformations

Chapter 2 Linear Transformations Chapter 2 Linear Transformations Linear Transformations Loosely speaking, a linear transformation is a function from one vector space to another that preserves the vector space operations. Let us be more

More information

Topics in Representation Theory: Fourier Analysis and the Peter Weyl Theorem

Topics in Representation Theory: Fourier Analysis and the Peter Weyl Theorem Topics in Representation Theory: Fourier Analysis and the Peter Weyl Theorem 1 Fourier Analysis, a review We ll begin with a short review of simple facts about Fourier analysis, before going on to interpret

More information

Recall that any inner product space V has an associated norm defined by

Recall that any inner product space V has an associated norm defined by Hilbert Spaces Recall that any inner product space V has an associated norm defined by v = v v. Thus an inner product space can be viewed as a special kind of normed vector space. In particular every inner

More information

Lecture Note 5: Semidefinite Programming for Stability Analysis

Lecture Note 5: Semidefinite Programming for Stability Analysis ECE7850: Hybrid Systems:Theory and Applications Lecture Note 5: Semidefinite Programming for Stability Analysis Wei Zhang Assistant Professor Department of Electrical and Computer Engineering Ohio State

More information

MOMENT METHODS IN ENERGY MINIMIZATION: NEW BOUNDS FOR RIESZ MINIMAL ENERGY PROBLEMS

MOMENT METHODS IN ENERGY MINIMIZATION: NEW BOUNDS FOR RIESZ MINIMAL ENERGY PROBLEMS MOMENT METHODS IN ENERGY MINIMIZATION: NEW BOUNDS FOR RIESZ MINIMAL ENERGY PROBLEMS DAVID DE LAAT Abstract. We use moment methods to construct a converging hierarchy of optimization problems to lower bound

More information

Copositive Programming and Combinatorial Optimization

Copositive Programming and Combinatorial Optimization Copositive Programming and Combinatorial Optimization Franz Rendl http://www.math.uni-klu.ac.at Alpen-Adria-Universität Klagenfurt Austria joint work with M. Bomze (Wien) and F. Jarre (Düsseldorf) and

More information

U.C. Berkeley CS294: Beyond Worst-Case Analysis Handout 12 Luca Trevisan October 3, 2017

U.C. Berkeley CS294: Beyond Worst-Case Analysis Handout 12 Luca Trevisan October 3, 2017 U.C. Berkeley CS94: Beyond Worst-Case Analysis Handout 1 Luca Trevisan October 3, 017 Scribed by Maxim Rabinovich Lecture 1 In which we begin to prove that the SDP relaxation exactly recovers communities

More information

Lift-and-Project Techniques and SDP Hierarchies

Lift-and-Project Techniques and SDP Hierarchies MFO seminar on Semidefinite Programming May 30, 2010 Typical combinatorial optimization problem: max c T x s.t. Ax b, x {0, 1} n P := {x R n Ax b} P I := conv(k {0, 1} n ) LP relaxation Integral polytope

More information

Highest-weight Theory: Verma Modules

Highest-weight Theory: Verma Modules Highest-weight Theory: Verma Modules Math G4344, Spring 2012 We will now turn to the problem of classifying and constructing all finitedimensional representations of a complex semi-simple Lie algebra (or,

More information

Lectures 15: Cayley Graphs of Abelian Groups

Lectures 15: Cayley Graphs of Abelian Groups U.C. Berkeley CS294: Spectral Methods and Expanders Handout 15 Luca Trevisan March 14, 2016 Lectures 15: Cayley Graphs of Abelian Groups In which we show how to find the eigenvalues and eigenvectors of

More information

REAL AND COMPLEX ANALYSIS

REAL AND COMPLEX ANALYSIS REAL AND COMPLE ANALYSIS Third Edition Walter Rudin Professor of Mathematics University of Wisconsin, Madison Version 1.1 No rights reserved. Any part of this work can be reproduced or transmitted in any

More information

CS675: Convex and Combinatorial Optimization Fall 2016 Combinatorial Problems as Linear and Convex Programs. Instructor: Shaddin Dughmi

CS675: Convex and Combinatorial Optimization Fall 2016 Combinatorial Problems as Linear and Convex Programs. Instructor: Shaddin Dughmi CS675: Convex and Combinatorial Optimization Fall 2016 Combinatorial Problems as Linear and Convex Programs Instructor: Shaddin Dughmi Outline 1 Introduction 2 Shortest Path 3 Algorithms for Single-Source

More information

Topics in Harmonic Analysis Lecture 1: The Fourier transform

Topics in Harmonic Analysis Lecture 1: The Fourier transform Topics in Harmonic Analysis Lecture 1: The Fourier transform Po-Lam Yung The Chinese University of Hong Kong Outline Fourier series on T: L 2 theory Convolutions The Dirichlet and Fejer kernels Pointwise

More information

Hypercube Coloring and the Structure of Binary Codes

Hypercube Coloring and the Structure of Binary Codes Hypercube Coloring and the Structure of Binary Codes by James Gregory Rix A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE in The College of Graduate Studies

More information

THE EULER CHARACTERISTIC OF A LIE GROUP

THE EULER CHARACTERISTIC OF A LIE GROUP THE EULER CHARACTERISTIC OF A LIE GROUP JAY TAYLOR 1 Examples of Lie Groups The following is adapted from [2] We begin with the basic definition and some core examples Definition A Lie group is a smooth

More information

Permutation groups/1. 1 Automorphism groups, permutation groups, abstract

Permutation groups/1. 1 Automorphism groups, permutation groups, abstract Permutation groups Whatever you have to do with a structure-endowed entity Σ try to determine its group of automorphisms... You can expect to gain a deep insight into the constitution of Σ in this way.

More information

Representations of algebraic groups and their Lie algebras Jens Carsten Jantzen Lecture III

Representations of algebraic groups and their Lie algebras Jens Carsten Jantzen Lecture III Representations of algebraic groups and their Lie algebras Jens Carsten Jantzen Lecture III Lie algebras. Let K be again an algebraically closed field. For the moment let G be an arbitrary algebraic group

More information

Tomato Packing and Lettuce-Based Crypto

Tomato Packing and Lettuce-Based Crypto Tomato Packing and Lettuce-Based Crypto A. S. Mosunov and L. A. Ruiz-Lopez University of Waterloo Joint PM & C&O Colloquium March 30th, 2017 Spring in Waterloo Love is everywhere... Picture from https://avatanplus.com/files/resources/mid/

More information

12. Hilbert Polynomials and Bézout s Theorem

12. Hilbert Polynomials and Bézout s Theorem 12. Hilbert Polynomials and Bézout s Theorem 95 12. Hilbert Polynomials and Bézout s Theorem After our study of smooth cubic surfaces in the last chapter, let us now come back to the general theory of

More information

Sphere packing, lattice packing, and related problems

Sphere packing, lattice packing, and related problems Sphere packing, lattice packing, and related problems Abhinav Kumar Stony Brook April 25, 2018 Sphere packings Definition A sphere packing in R n is a collection of spheres/balls of equal size which do

More information

LECTURE 16: LIE GROUPS AND THEIR LIE ALGEBRAS. 1. Lie groups

LECTURE 16: LIE GROUPS AND THEIR LIE ALGEBRAS. 1. Lie groups LECTURE 16: LIE GROUPS AND THEIR LIE ALGEBRAS 1. Lie groups A Lie group is a special smooth manifold on which there is a group structure, and moreover, the two structures are compatible. Lie groups are

More information

Robust and Optimal Control, Spring 2015

Robust and Optimal Control, Spring 2015 Robust and Optimal Control, Spring 2015 Instructor: Prof. Masayuki Fujita (S5-303B) G. Sum of Squares (SOS) G.1 SOS Program: SOS/PSD and SDP G.2 Duality, valid ineqalities and Cone G.3 Feasibility/Optimization

More information

LIFTS OF CONVEX SETS AND CONE FACTORIZATIONS JOÃO GOUVEIA, PABLO A. PARRILO, AND REKHA THOMAS

LIFTS OF CONVEX SETS AND CONE FACTORIZATIONS JOÃO GOUVEIA, PABLO A. PARRILO, AND REKHA THOMAS LIFTS OF CONVEX SETS AND CONE FACTORIZATIONS Abstract. In this paper we address the basic geometric question of when a given convex set is the image under a linear map of an affine slice of a given closed

More information

TOPICS IN HARMONIC ANALYSIS WITH APPLICATIONS TO RADAR AND SONAR. Willard Miller

TOPICS IN HARMONIC ANALYSIS WITH APPLICATIONS TO RADAR AND SONAR. Willard Miller TOPICS IN HARMONIC ANALYSIS WITH APPLICATIONS TO RADAR AND SONAR Willard Miller October 23 2002 These notes are an introduction to basic concepts and tools in group representation theory both commutative

More information

Algebraic Varieties. Notes by Mateusz Micha lek for the lecture on April 17, 2018, in the IMPRS Ringvorlesung Introduction to Nonlinear Algebra

Algebraic Varieties. Notes by Mateusz Micha lek for the lecture on April 17, 2018, in the IMPRS Ringvorlesung Introduction to Nonlinear Algebra Algebraic Varieties Notes by Mateusz Micha lek for the lecture on April 17, 2018, in the IMPRS Ringvorlesung Introduction to Nonlinear Algebra Algebraic varieties represent solutions of a system of polynomial

More information

A ten page introduction to conic optimization

A ten page introduction to conic optimization CHAPTER 1 A ten page introduction to conic optimization This background chapter gives an introduction to conic optimization. We do not give proofs, but focus on important (for this thesis) tools and concepts.

More information

1 Fields and vector spaces

1 Fields and vector spaces 1 Fields and vector spaces In this section we revise some algebraic preliminaries and establish notation. 1.1 Division rings and fields A division ring, or skew field, is a structure F with two binary

More information

Spectral Theorem for Self-adjoint Linear Operators

Spectral Theorem for Self-adjoint Linear Operators Notes for the undergraduate lecture by David Adams. (These are the notes I would write if I was teaching a course on this topic. I have included more material than I will cover in the 45 minute lecture;

More information

Semidefinite Programming Basics and Applications

Semidefinite Programming Basics and Applications Semidefinite Programming Basics and Applications Ray Pörn, principal lecturer Åbo Akademi University Novia University of Applied Sciences Content What is semidefinite programming (SDP)? How to represent

More information

Chapter 1. Preliminaries

Chapter 1. Preliminaries Introduction This dissertation is a reading of chapter 4 in part I of the book : Integer and Combinatorial Optimization by George L. Nemhauser & Laurence A. Wolsey. The chapter elaborates links between

More information

OPERATOR THEORY ON HILBERT SPACE. Class notes. John Petrovic

OPERATOR THEORY ON HILBERT SPACE. Class notes. John Petrovic OPERATOR THEORY ON HILBERT SPACE Class notes John Petrovic Contents Chapter 1. Hilbert space 1 1.1. Definition and Properties 1 1.2. Orthogonality 3 1.3. Subspaces 7 1.4. Weak topology 9 Chapter 2. Operators

More information

Qualifying Exams I, Jan where µ is the Lebesgue measure on [0,1]. In this problems, all functions are assumed to be in L 1 [0,1].

Qualifying Exams I, Jan where µ is the Lebesgue measure on [0,1]. In this problems, all functions are assumed to be in L 1 [0,1]. Qualifying Exams I, Jan. 213 1. (Real Analysis) Suppose f j,j = 1,2,... and f are real functions on [,1]. Define f j f in measure if and only if for any ε > we have lim µ{x [,1] : f j(x) f(x) > ε} = j

More information

Semidefinite Programming

Semidefinite Programming Semidefinite Programming Basics and SOS Fernando Mário de Oliveira Filho Campos do Jordão, 2 November 23 Available at: www.ime.usp.br/~fmario under talks Conic programming V is a real vector space h, i

More information

BRST and Dirac Cohomology

BRST and Dirac Cohomology BRST and Dirac Cohomology Peter Woit Columbia University Dartmouth Math Dept., October 23, 2008 Peter Woit (Columbia University) BRST and Dirac Cohomology October 2008 1 / 23 Outline 1 Introduction 2 Representation

More information

Lecture 1. 1 Conic programming. MA 796S: Convex Optimization and Interior Point Methods October 8, Consider the conic program. min.

Lecture 1. 1 Conic programming. MA 796S: Convex Optimization and Interior Point Methods October 8, Consider the conic program. min. MA 796S: Convex Optimization and Interior Point Methods October 8, 2007 Lecture 1 Lecturer: Kartik Sivaramakrishnan Scribe: Kartik Sivaramakrishnan 1 Conic programming Consider the conic program min s.t.

More information

L(C G (x) 0 ) c g (x). Proof. Recall C G (x) = {g G xgx 1 = g} and c g (x) = {X g Ad xx = X}. In general, it is obvious that

L(C G (x) 0 ) c g (x). Proof. Recall C G (x) = {g G xgx 1 = g} and c g (x) = {X g Ad xx = X}. In general, it is obvious that ALGEBRAIC GROUPS 61 5. Root systems and semisimple Lie algebras 5.1. Characteristic 0 theory. Assume in this subsection that chark = 0. Let me recall a couple of definitions made earlier: G is called reductive

More information

Boolean Inner-Product Spaces and Boolean Matrices

Boolean Inner-Product Spaces and Boolean Matrices Boolean Inner-Product Spaces and Boolean Matrices Stan Gudder Department of Mathematics, University of Denver, Denver CO 80208 Frédéric Latrémolière Department of Mathematics, University of Denver, Denver

More information

LINEAR ALGEBRA BOOT CAMP WEEK 4: THE SPECTRAL THEOREM

LINEAR ALGEBRA BOOT CAMP WEEK 4: THE SPECTRAL THEOREM LINEAR ALGEBRA BOOT CAMP WEEK 4: THE SPECTRAL THEOREM Unless otherwise stated, all vector spaces in this worksheet are finite dimensional and the scalar field F is R or C. Definition 1. A linear operator

More information

SPHERICAL UNITARY REPRESENTATIONS FOR REDUCTIVE GROUPS

SPHERICAL UNITARY REPRESENTATIONS FOR REDUCTIVE GROUPS SPHERICAL UNITARY REPRESENTATIONS FOR REDUCTIVE GROUPS DAN CIUBOTARU 1. Classical motivation: spherical functions 1.1. Spherical harmonics. Let S n 1 R n be the (n 1)-dimensional sphere, C (S n 1 ) the

More information

Real symmetric matrices/1. 1 Eigenvalues and eigenvectors

Real symmetric matrices/1. 1 Eigenvalues and eigenvectors Real symmetric matrices 1 Eigenvalues and eigenvectors We use the convention that vectors are row vectors and matrices act on the right. Let A be a square matrix with entries in a field F; suppose that

More information

Short note on compact operators - Monday 24 th March, Sylvester Eriksson-Bique

Short note on compact operators - Monday 24 th March, Sylvester Eriksson-Bique Short note on compact operators - Monday 24 th March, 2014 Sylvester Eriksson-Bique 1 Introduction In this note I will give a short outline about the structure theory of compact operators. I restrict attention

More information

LECTURE 4: REPRESENTATION THEORY OF SL 2 (F) AND sl 2 (F)

LECTURE 4: REPRESENTATION THEORY OF SL 2 (F) AND sl 2 (F) LECTURE 4: REPRESENTATION THEORY OF SL 2 (F) AND sl 2 (F) IVAN LOSEV In this lecture we will discuss the representation theory of the algebraic group SL 2 (F) and of the Lie algebra sl 2 (F), where F is

More information

MATH32062 Notes. 1 Affine algebraic varieties. 1.1 Definition of affine algebraic varieties

MATH32062 Notes. 1 Affine algebraic varieties. 1.1 Definition of affine algebraic varieties MATH32062 Notes 1 Affine algebraic varieties 1.1 Definition of affine algebraic varieties We want to define an algebraic variety as the solution set of a collection of polynomial equations, or equivalently,

More information

Codes and Rings: Theory and Practice

Codes and Rings: Theory and Practice Codes and Rings: Theory and Practice Patrick Solé CNRS/LAGA Paris, France, January 2017 Geometry of codes : the music of spheres R = a finite ring with identity. A linear code of length n over a ring R

More information

MIT Algebraic techniques and semidefinite optimization February 14, Lecture 3

MIT Algebraic techniques and semidefinite optimization February 14, Lecture 3 MI 6.97 Algebraic techniques and semidefinite optimization February 4, 6 Lecture 3 Lecturer: Pablo A. Parrilo Scribe: Pablo A. Parrilo In this lecture, we will discuss one of the most important applications

More information

Copositive Programming and Combinatorial Optimization

Copositive Programming and Combinatorial Optimization Copositive Programming and Combinatorial Optimization Franz Rendl http://www.math.uni-klu.ac.at Alpen-Adria-Universität Klagenfurt Austria joint work with I.M. Bomze (Wien) and F. Jarre (Düsseldorf) IMA

More information

LECTURE 25-26: CARTAN S THEOREM OF MAXIMAL TORI. 1. Maximal Tori

LECTURE 25-26: CARTAN S THEOREM OF MAXIMAL TORI. 1. Maximal Tori LECTURE 25-26: CARTAN S THEOREM OF MAXIMAL TORI 1. Maximal Tori By a torus we mean a compact connected abelian Lie group, so a torus is a Lie group that is isomorphic to T n = R n /Z n. Definition 1.1.

More information

Finite-dimensional spaces. C n is the space of n-tuples x = (x 1,..., x n ) of complex numbers. It is a Hilbert space with the inner product

Finite-dimensional spaces. C n is the space of n-tuples x = (x 1,..., x n ) of complex numbers. It is a Hilbert space with the inner product Chapter 4 Hilbert Spaces 4.1 Inner Product Spaces Inner Product Space. A complex vector space E is called an inner product space (or a pre-hilbert space, or a unitary space) if there is a mapping (, )

More information

Math Linear Algebra II. 1. Inner Products and Norms

Math Linear Algebra II. 1. Inner Products and Norms Math 342 - Linear Algebra II Notes 1. Inner Products and Norms One knows from a basic introduction to vectors in R n Math 254 at OSU) that the length of a vector x = x 1 x 2... x n ) T R n, denoted x,

More information

16.1 L.P. Duality Applied to the Minimax Theorem

16.1 L.P. Duality Applied to the Minimax Theorem CS787: Advanced Algorithms Scribe: David Malec and Xiaoyong Chai Lecturer: Shuchi Chawla Topic: Minimax Theorem and Semi-Definite Programming Date: October 22 2007 In this lecture, we first conclude our

More information

MATHEMATICS. Course Syllabus. Section A: Linear Algebra. Subject Code: MA. Course Structure. Ordinary Differential Equations

MATHEMATICS. Course Syllabus. Section A: Linear Algebra. Subject Code: MA. Course Structure. Ordinary Differential Equations MATHEMATICS Subject Code: MA Course Structure Sections/Units Section A Section B Section C Linear Algebra Complex Analysis Real Analysis Topics Section D Section E Section F Section G Section H Section

More information

CSC Linear Programming and Combinatorial Optimization Lecture 12: The Lift and Project Method

CSC Linear Programming and Combinatorial Optimization Lecture 12: The Lift and Project Method CSC2411 - Linear Programming and Combinatorial Optimization Lecture 12: The Lift and Project Method Notes taken by Stefan Mathe April 28, 2007 Summary: Throughout the course, we have seen the importance

More information

The following definition is fundamental.

The following definition is fundamental. 1. Some Basics from Linear Algebra With these notes, I will try and clarify certain topics that I only quickly mention in class. First and foremost, I will assume that you are familiar with many basic

More information

ELE539A: Optimization of Communication Systems Lecture 15: Semidefinite Programming, Detection and Estimation Applications

ELE539A: Optimization of Communication Systems Lecture 15: Semidefinite Programming, Detection and Estimation Applications ELE539A: Optimization of Communication Systems Lecture 15: Semidefinite Programming, Detection and Estimation Applications Professor M. Chiang Electrical Engineering Department, Princeton University March

More information

Chapter 3. Some Applications. 3.1 The Cone of Positive Semidefinite Matrices

Chapter 3. Some Applications. 3.1 The Cone of Positive Semidefinite Matrices Chapter 3 Some Applications Having developed the basic theory of cone programming, it is time to apply it to our actual subject, namely that of semidefinite programming. Indeed, any semidefinite program

More information

Lower bounds on the size of semidefinite relaxations. David Steurer Cornell

Lower bounds on the size of semidefinite relaxations. David Steurer Cornell Lower bounds on the size of semidefinite relaxations David Steurer Cornell James R. Lee Washington Prasad Raghavendra Berkeley Institute for Advanced Study, November 2015 overview of results unconditional

More information

here, this space is in fact infinite-dimensional, so t σ ess. Exercise Let T B(H) be a self-adjoint operator on an infinitedimensional

here, this space is in fact infinite-dimensional, so t σ ess. Exercise Let T B(H) be a self-adjoint operator on an infinitedimensional 15. Perturbations by compact operators In this chapter, we study the stability (or lack thereof) of various spectral properties under small perturbations. Here s the type of situation we have in mind:

More information

Representation Theory. Ricky Roy Math 434 University of Puget Sound

Representation Theory. Ricky Roy Math 434 University of Puget Sound Representation Theory Ricky Roy Math 434 University of Puget Sound May 2, 2010 Introduction In our study of group theory, we set out to classify all distinct groups of a given order up to isomorphism.

More information

Recall the convention that, for us, all vectors are column vectors.

Recall the convention that, for us, all vectors are column vectors. Some linear algebra Recall the convention that, for us, all vectors are column vectors. 1. Symmetric matrices Let A be a real matrix. Recall that a complex number λ is an eigenvalue of A if there exists

More information

1 Math 241A-B Homework Problem List for F2015 and W2016

1 Math 241A-B Homework Problem List for F2015 and W2016 1 Math 241A-B Homework Problem List for F2015 W2016 1.1 Homework 1. Due Wednesday, October 7, 2015 Notation 1.1 Let U be any set, g be a positive function on U, Y be a normed space. For any f : U Y let

More information

CHAPTER VIII HILBERT SPACES

CHAPTER VIII HILBERT SPACES CHAPTER VIII HILBERT SPACES DEFINITION Let X and Y be two complex vector spaces. A map T : X Y is called a conjugate-linear transformation if it is a reallinear transformation from X into Y, and if T (λx)

More information

8 Approximation Algorithms and Max-Cut

8 Approximation Algorithms and Max-Cut 8 Approximation Algorithms and Max-Cut 8. The Max-Cut problem Unless the widely believed P N P conjecture is false, there is no polynomial algorithm that can solve all instances of an NP-hard problem.

More information

x i e i ) + Q(x n e n ) + ( i<n c ij x i x j

x i e i ) + Q(x n e n ) + ( i<n c ij x i x j Math 210A. Quadratic spaces over R 1. Algebraic preliminaries Let V be a finite free module over a nonzero commutative ring F. Recall that a quadratic form on V is a map Q : V F such that Q(cv) = c 2 Q(v)

More information

Relaxations of combinatorial problems via association schemes

Relaxations of combinatorial problems via association schemes 1 Relaxations of combinatorial problems via association schemes Etienne de Klerk, Fernando M. de Oliveira Filho, and Dmitrii V. Pasechnik Tilburg University, The Netherlands; Nanyang Technological University,

More information

Modeling with semidefinite and copositive matrices

Modeling with semidefinite and copositive matrices Modeling with semidefinite and copositive matrices Franz Rendl http://www.math.uni-klu.ac.at Alpen-Adria-Universität Klagenfurt Austria F. Rendl, Singapore workshop 2006 p.1/24 Overview Node and Edge relaxations

More information

Spectral Properties of Elliptic Operators In previous work we have replaced the strong version of an elliptic boundary value problem

Spectral Properties of Elliptic Operators In previous work we have replaced the strong version of an elliptic boundary value problem Spectral Properties of Elliptic Operators In previous work we have replaced the strong version of an elliptic boundary value problem L u x f x BC u x g x with the weak problem find u V such that B u,v

More information