Rank Minimization over Finite Fields

Size: px
Start display at page:

Download "Rank Minimization over Finite Fields"

Transcription

1 Rank Minimization over Finite Fields Vincent Y. F. Tan Laura Balzano, Stark C. Draper Department of Electrical and Computer Engineering, University of Wisconsin-Madison ISIT 2011 Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

2 Connection between Two Areas of Study Area of Matrix Completion Rank-Metric Codes Study Rank Minimization Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

3 Connection between Two Areas of Study Area of Matrix Completion Rank-Metric Codes Study Rank Minimization Applications Collaborative Filtering Crisscross Error Correction Minimal Realization Network Coding Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

4 Connection between Two Areas of Study Area of Matrix Completion Rank-Metric Codes Study Rank Minimization Applications Collaborative Filtering Crisscross Error Correction Minimal Realization Network Coding Field Real R, Complex C Finite Field F q Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

5 Connection between Two Areas of Study Area of Matrix Completion Rank-Metric Codes Study Rank Minimization Applications Collaborative Filtering Crisscross Error Correction Minimal Realization Network Coding Field Real R, Complex C Finite Field F q Techniques Functional Analysis Algebraic coding Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

6 Connection between Two Areas of Study Area of Matrix Completion Rank-Metric Codes Study Rank Minimization Applications Collaborative Filtering Crisscross Error Correction Minimal Realization Network Coding Field Real R, Complex C Finite Field F q Techniques Functional Analysis Algebraic coding Decoding Convex Optimization Berlekamp-Massey Nuclear Norm Error Trapping Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

7 Connection between Two Areas of Study Area of Matrix Completion Rank-Metric Codes Study Rank Minimization Applications Collaborative Filtering Crisscross Error Correction Minimal Realization Network Coding Field Real R, Complex C Finite Field F q Techniques Functional Analysis Algebraic coding Decoding Convex Optimization Berlekamp-Massey Nuclear Norm Error Trapping Can we draw analogies between the two areas of study? Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

8 Connection between Two Areas of Study Area of Matrix Completion Rank-Metric Codes Study Rank Minimization Applications Collaborative Filtering Crisscross Error Correction Minimal Realization Network Coding Field Real R, Complex C Finite Field F q Techniques Functional Analysis Algebraic coding Decoding Convex Optimization Berlekamp-Massey Nuclear Norm Error Trapping Can we draw analogies between the two areas of study? Rank Minimization over finite fields Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

9 Minimum rank distance decoding Transmit some codeword C C A rank-metric code C is a non-empty subset of F m n q the rank distance endowed with Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

10 Minimum rank distance decoding Transmit some codeword C C A rank-metric code C is a non-empty subset of F m n q the rank distance Receive some received matrix R F m n q endowed with Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

11 Minimum rank distance decoding Transmit some codeword C C A rank-metric code C is a non-empty subset of F m n q the rank distance Receive some received matrix R F m n q Decoding problem (under mild conditions) is Ĉ = arg min C C rank(r C) Minimum distance decoding since rank induces a metric endowed with Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

12 Minimum rank distance decoding Transmit some codeword C C A rank-metric code C is a non-empty subset of F m n q the rank distance Receive some received matrix R F m n q Decoding problem (under mild conditions) is Ĉ = arg min C C rank(r C) Minimum distance decoding since rank induces a metric endowed with X R C is known as the error matrix assumed low-rank Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

13 Minimum rank distance decoding Transmit some codeword C C A rank-metric code C is a non-empty subset of F m n q the rank distance Receive some received matrix R F m n q Decoding problem (under mild conditions) is Ĉ = arg min C C rank(r C) Minimum distance decoding since rank induces a metric endowed with X R C is known as the error matrix assumed low-rank Assume linear code. Rank minimization over finite field: ˆX = arg min X coset rank(x) Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

14 When do low-rank errors occur? Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

15 When do low-rank errors occur? Probabilistic crisscross error correction Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

16 When do Probabilistic low-rank errors Crisscross occur? Error Correction Probabilistic crisscross error correction Roth, IT Trans 1997 Ron M. Roth, Member, IEEE Abstract The crisscross error model in data arrays is considered, where the corrupted symbols are confined to a prescribed number of rows or columns (or both). Under the additional assumption that the corrupted entries are uniformly distributed over the channel alphabet, and by allowing a small decoding error probability, a coding scheme is presented where the redundancy can get close to one half the redundancy required in minimumdistance decoding of crisscross errors. Index Terms Crisscross errors, probabilistic coding, rank metric. I. INTRODUCTION ONSIDER an application where information symbols C(such as bits or bytes) are stored in arrays, with the possibility of some of the symbols recorded erroneously. The error patterns are such that all corrupted symbols are confined to a prescribed number of rows or columns (or both). We refer to such an error model as crisscross errors. A crisscross error pattern that is confined to two rows and three columns Fig. 1. Typical crisscross error pattern. is shown in Fig. 1. Crisscross Dataerrorstorage can be found applications: in various data storage Data stored in arrays applications; see, for instance, [3], [5], [6], [11], [16] [19]. arrays over is the cover weight of their difference. An Such errors may occur in memory chip arrays, where row array code over is a -dimensional linear or column failures occur due to the malfunctioning of row subspace of the vector space of all matrices over Error patterns confined to two rows and three columns above drivers or column amplifiers. Crisscross errors can also be such that is the smallest cover distance between any found in helical tapes, where the tracks are recorded in a two distinct elements of or, equivalently, the smallest cover direction which is (conceptually) perpendicular to the direction weight of any nonzero element of. The parameter is of the movement of the tape; misalignment of the reading head referred to as the minimum cover distance of and the term causes burst errors to occur along the track (and across the stands for the redundancy of. tape), whereas scratches on the tape usually occur along the The Singleton bound on the minimum cover distance states tape Vincent (and across Tan the (UW-Madison) tracks). Crisscross error-correcting Rank Minimization codes thatover the minimum Finite Fields cover distance and the redundancy ISITof 2011 any 4 / 20

17 When do Probabilistic low-rank errors Crisscross occur? Error Correction Probabilistic crisscross error correction Roth, IT Trans 1997 Ron M. Roth, Member, IEEE Abstract The crisscross error model in data arrays is considered, where the corrupted symbols are confined to a prescribed number of rows or columns (or both). Under the additional assumption that the corrupted entries are uniformly distributed over the channel alphabet, and by allowing a small decoding error probability, a coding scheme is presented where the redundancy can get close to one half the redundancy required in minimumdistance decoding of crisscross errors. Index Terms Crisscross errors, probabilistic coding, rank metric. I. INTRODUCTION ONSIDER an application where information symbols C(such as bits or bytes) are stored in arrays, with the possibility of some of the symbols recorded erroneously. The error patterns are such that all corrupted symbols are confined to a prescribed number of rows or columns (or both). We refer to such an error model as crisscross errors. A crisscross error pattern that is confined to two rows and three columns Fig. 1. Typical crisscross error pattern. is shown in Fig. 1. Crisscross Dataerrorstorage can be found applications: in various data storage Data stored in arrays applications; see, for instance, [3], [5], [6], [11], [16] [19]. arrays over is the cover weight of their difference. An Such errors may occur in memory chip arrays, where row array code over is a -dimensional linear or column failures occur due to the malfunctioning of row subspace of the vector space of all matrices over Error patterns confined to two rows and three columns above drivers or column amplifiers. Crisscross errors can also be such that is the smallest cover distance between any found in helical tapes, where the tracks are recorded in a two distinct elements of or, equivalently, the smallest cover direction which is (conceptually) perpendicular to the direction weight of any nonzero element of. The parameter is of the movement Low-rank of the tape; errors misalignment of the reading head referred to as the minimum cover distance of and the term causes burst errors to occur along the track (and across the stands for the redundancy of. tape), whereas scratches on the tape usually occur along the The Singleton bound on the minimum cover distance states tape Vincent (and across Tan the (UW-Madison) tracks). Crisscross error-correcting Rank Minimization codes thatover the minimum Finite Fields cover distance and the redundancy ISITof 2011 any 4 / 20

18 Problem Setup: Rank minimization over finite fields Square matrix X F n n q has rank r n Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

19 Problem Setup: Rank minimization over finite fields Square matrix X F n n q has rank r n!!!" Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

20 Problem Setup: Rank minimization over finite fields Square matrix X F n n q has rank r n!!!" Assume that the sensing matrices H 1,..., H k F n n q are random. Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

21 Problem Setup: Rank minimization over finite fields Square matrix X F n n q has rank r n!!!" Assume that the sensing matrices H 1,..., H k F n n q are random. Arithmetic is performed in the field F q y a F q Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

22 Problem Setup: Rank minimization over finite fields Square matrix X F n n q has rank r n!!!" Assume that the sensing matrices H 1,..., H k F n n q are random. Arithmetic is performed in the field F q y a F q Given (y k, H k ), find necessary and sufficient conditions on k and sensing model such that recovery is reliable, i.e., P(E n ) 0 Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

23 Main Results k: Num. of linear measurements n: Dim. of matrix X r: Max. rank of matrix X Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

24 Main Results k: Num. of linear measurements n: Dim. of matrix X r: Max. rank of matrix X γ = r n : Rank-dimension ratio Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

25 Main Results k: Num. of linear measurements n: Dim. of matrix X r: Max. rank of matrix X γ = r n : Rank-dimension ratio Critical Quantity 2γ (1 γ/2) n 2 = 2rn r 2 Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

26 Main Results k: Num. of linear measurements n: Dim. of matrix X r: Max. rank of matrix X γ = r n : Rank-dimension ratio Critical Quantity 2γ (1 γ/2) n 2 = 2rn r 2 Result Statement Consequence Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

27 Main Results k: Num. of linear measurements n: Dim. of matrix X r: Max. rank of matrix X γ = r n : Rank-dimension ratio Critical Quantity 2γ (1 γ/2) n 2 = 2rn r 2 Result Statement Consequence Converse k < (2 ε)γ(1 γ/2)n 2 P(E n ) 0 Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

28 Main Results k: Num. of linear measurements n: Dim. of matrix X r: Max. rank of matrix X γ = r n : Rank-dimension ratio Critical Quantity 2γ (1 γ/2) n 2 = 2rn r 2 Result Statement Consequence Converse k < (2 ε)γ(1 γ/2)n 2 P(E n ) 0 P(E n ) 0 Achievability (Uniform) k > (2 + ε)γ(1 γ/2)n 2 P(E n ) q n2 E(R) Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

29 Main Results k: Num. of linear measurements n: Dim. of matrix X r: Max. rank of matrix X γ = r n : Rank-dimension ratio Critical Quantity 2γ (1 γ/2) n 2 = 2rn r 2 Result Statement Consequence Converse k < (2 ε)γ(1 γ/2)n 2 P(E n ) 0 P(E n ) 0 Achievability (Uniform) k > (2 + ε)γ(1 γ/2)n 2 P(E n ) q n2 E(R) Achievability (Sparse) k > (2 + ε)γ(1 γ/2)n 2 P(E n ) 0 Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

30 Main Results k: Num. of linear measurements n: Dim. of matrix X r: Max. rank of matrix X γ = r n : Rank-dimension ratio Critical Quantity 2γ (1 γ/2) n 2 = 2rn r 2 Result Statement Consequence Converse k < (2 ε)γ(1 γ/2)n 2 P(E n ) 0 P(E n ) 0 Achievability (Uniform) k > (2 + ε)γ(1 γ/2)n 2 P(E n ) q n2 E(R) Achievability (Sparse) k > (2 + ε)γ(1 γ/2)n 2 P(E n ) 0 Achievability (Noisy) k (3 + ε)(γ + σ)n 2 P(E n ) 0 (q assumed large) Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

31 A necessary condition on number of measurements Given k measurements y a F q and sensing matrices H a F n n q, we want a necessary condition for reliable recovery of X. Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

32 A necessary condition on number of measurements Given k measurements y a F q and sensing matrices H a F n n q, we want a necessary condition for reliable recovery of X. Proposition (Converse) Assume X drawn uniformly at random from all matrices in F n n q Sensing matrices H a, a = 1,..., k jointly independent of X r/n γ (constant) If the number of measurements satisfies ( k < (2 ε)γ 1 γ ) n 2 2 then P(ˆX X) ε/2 for all n sufficiently large. of rank r Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

33 A sensing model: Uniform model Now we assume that X is non-random Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

34 A sensing model: Uniform model Now we assume that X is non-random rank(x) r = γn Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

35 A sensing model: Uniform model Now we assume that X is non-random rank(x) r = γn!!!" Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

36 A sensing model: Uniform model Now we assume that X is non-random rank(x) r = γn!!!" Each entry of each sensing matrix H a is i.i.d. and has a uniform distribution in F q : P([H a ] i,j = h) = 1 q, h F q Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

37 The min-rank decoder We employ the min-rank decoder minimize rank( X) subject to H a, X = y a, a = 1,..., k Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

38 The min-rank decoder We employ the min-rank decoder minimize rank( X) subject to H a, X = y a, a = 1,..., k NP-hard, combinatorial Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

39 The min-rank decoder We employ the min-rank decoder minimize rank( X) subject to H a, X = y a, a = 1,..., k NP-hard, combinatorial Denote the set of optimizers as S Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

40 The min-rank decoder We employ the min-rank decoder minimize rank( X) subject to H a, X = y a, a = 1,..., k NP-hard, combinatorial Denote the set of optimizers as S Define the error event: E n := { S > 1} ({ S = 1} {X X}) Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

41 The min-rank decoder We employ the min-rank decoder minimize rank( X) subject to H a, X = y a, a = 1,..., k NP-hard, combinatorial Denote the set of optimizers as S Define the error event: E n := { S > 1} ({ S = 1} {X X}) We want the solution to be unique and correct Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

42 Achievability under uniform model Proposition (Achievability under uniform model) Assume Sensing matrices H a drawn uniformly Min-rank decoder is used r/n γ (constant) If the number of measurements satisfies ( k > (2 + ε)γ 1 γ ) n 2 2 then P(E n ) 0. Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

43 Proof Sketch E n = { Z, H a = X, H a, a = 1,..., k} Z X:rank(Z) rank(x) Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

44 Proof Sketch E n = { Z, H a = X, H a, a = 1,..., k} Z X:rank(Z) rank(x) By the union bound, the error probability can be bounded as P(E n ) P( Z, H a = X, H a, a = 1,..., k) Z X:rank(Z) rank(x) Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

45 Proof Sketch E n = { Z, H a = X, H a, a = 1,..., k} Z X:rank(Z) rank(x) By the union bound, the error probability can be bounded as P(E n ) P( Z, H a = X, H a, a = 1,..., k) Z X:rank(Z) rank(x) At the same time, by uniformity and independence, P( Z, H a = X, H a, a = 1,..., k) = q k Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

46 Proof Sketch E n = Z X:rank(Z) rank(x) { Z, H a = X, H a, a = 1,..., k} By the union bound, the error probability can be bounded as P(E n ) P( Z, H a = X, H a, a = 1,..., k) Z X:rank(Z) rank(x) At the same time, by uniformity and independence, P( Z, H a = X, H a, a = 1,..., k) = q k That the number of n n matrices over F q of rank r is bounded above by 4q 2γ(1 γ/2)n2 Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

47 Proof Sketch E n = Z X:rank(Z) rank(x) { Z, H a = X, H a, a = 1,..., k} By the union bound, the error probability can be bounded as P(E n ) P( Z, H a = X, H a, a = 1,..., k) Z X:rank(Z) rank(x) At the same time, by uniformity and independence, P( Z, H a = X, H a, a = 1,..., k) = q k That the number of n n matrices over F q of rank r is bounded above by 4q 2γ(1 γ/2)n2 Remark: Can be extended to noisy case Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

48 The Reliability Function Definition The rate of a sequence of linear measurement models is defined as R := lim n 1 k n 2, k = # measurements Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

49 The Reliability Function Definition The rate of a sequence of linear measurement models is defined as R := lim n 1 k n 2, k = # measurements Analogy to coding: The rate of the code is lower bounded by 1 k/n 2. C = {C : C, H a = 0, a = 1,..., k} Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

50 The Reliability Function Definition The rate of a sequence of linear measurement models is defined as R := lim n 1 k n 2, k = # measurements Analogy to coding: The rate of the code C = {C : C, H a = 0, a = 1,..., k} is lower bounded by 1 k/n 2. Definition The reliability function of the min-rank decoder is defined as E(R) := lim n 1 n 2 log q P(E n) Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

51 The Reliability Function Our achievability proof gives a lower bound on E(R) Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

52 The Reliability Function Our achievability proof gives a lower bound on E(R) Upper bound on E(R) is more tricky. But... Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

53 The Reliability Function Our achievability proof gives a lower bound on E(R) Upper bound on E(R) is more tricky. But... Proposition (Reliability Function of Min-Rank Decoder) Assume Sensing matrices H a drawn uniformly Min-rank decoder is used r/n γ (constant) Then, E(R) = ( (1 R) 2γ 1 γ ) + 2 Note x + := max{x, 0}. Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

54 Interpretation E(R) k (1 n 2 2γ γ ) + 2 Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

55 Interpretation E(R) k (1 n 2 2γ γ ) + 2 The more the ratio k exceeds 2γ ( 1 γ ) n 2 2 Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

56 Interpretation E(R) k (1 n 2 2γ γ ) + 2 The more the ratio k exceeds 2γ ( 1 γ ) n 2 2 The larger E(R) Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

57 Interpretation E(R) k (1 n 2 2γ γ ) + 2 The more the ratio k exceeds 2γ ( 1 γ ) n 2 2 The larger E(R) The faster P(E n ) decays Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

58 Interpretation E(R) k (1 n 2 2γ γ ) + 2 The more the ratio k exceeds 2γ ( 1 γ ) n 2 2 The larger E(R) The faster P(E n ) decays Linear relationship for the min-rank decoder Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

59 Interpretation E(R) k (1 n 2 2γ γ ) + 2 The more the ratio k exceeds 2γ ( 1 γ ) n 2 2 The larger E(R) The faster P(E n ) decays Linear relationship for the min-rank decoder de Caen s lower bound: Let B 1,..., B M be events: ( M ) M P(B m ) 2 P B m M m =1 P(B m B m ). m=1 m=1 Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

60 Interpretation E(R) k (1 n 2 2γ γ ) + 2 The more the ratio k exceeds 2γ ( 1 γ ) n 2 2 The larger E(R) The faster P(E n ) decays Linear relationship for the min-rank decoder de Caen s lower bound: Let B 1,..., B M be events: ( M ) M P(B m ) 2 P B m M m =1 P(B m B m ). m=1 m=1 Exploit pairwise independence to make statements about error exponents (linear codes achieve capacity in symmetric DMCs) Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

61 Sparse sensing model Assume as usual that X is non-random Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

62 Sparse sensing model Assume as usual that X is non-random rank(x) r = γn Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

63 Sparse sensing model Assume as usual that X is non-random rank(x) r = γn Sensing matrices are sparse!!!" Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

64 Sparse sensing model Assume as usual that X is non-random rank(x) r = γn Sensing matrices are sparse!!!" Arithmetic still performed in F q Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

65 Sparse sensing model Each entry of each sensing matrix H a is i.i.d. and has a δ-sparse distribution in F q : 1 δ 0 δ/(q 1) δ/(q 1) δ/(q 1) δ/(q 1) { 1 δ h = 0 P([H a ] ij = h) = δ q 1 h 0 Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

66 Sparse sensing model Each entry of each sensing matrix H a is i.i.d. and has a δ-sparse distribution in F q : 1 δ 0 δ/(q 1) δ/(q 1) δ/(q 1) δ/(q 1) { 1 δ h = 0 P([H a ] ij = h) = δ q 1 h 0 Fewer adds and multiplies since H a sparse Encoding cheaper Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

67 Sparse sensing model Each entry of each sensing matrix H a is i.i.d. and has a δ-sparse distribution in F q : 1 δ 0 δ/(q 1) δ/(q 1) δ/(q 1) δ/(q 1) { 1 δ h = 0 P([H a ] ij = h) = δ q 1 h 0 Fewer adds and multiplies since H a sparse Encoding cheaper May help in decoding via message-passing algorithms Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

68 Sparse sensing model Each entry of each sensing matrix H a is i.i.d. and has a δ-sparse distribution in F q : 1 δ 0 δ/(q 1) δ/(q 1) δ/(q 1) δ/(q 1) { 1 δ h = 0 P([H a ] ij = h) = δ q 1 h 0 Fewer adds and multiplies since H a sparse Encoding cheaper May help in decoding via message-passing algorithms How fast can δ, the sparsity factor, decay with n for reliable recovery? Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

69 Sparse sensing model Problem becomes more challenging Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

70 Sparse sensing model Problem becomes more challenging X is not sensed as much Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

71 Sparse sensing model Problem becomes more challenging X is not sensed as much Measurements y k do not contain as much information about X Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

72 Sparse sensing model Problem becomes more challenging X is not sensed as much Measurements y k do not contain as much information about X The equality P( Z, H a = X, H a, a = 1,..., k) = q k no longer holds Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

73 Sparse sensing model Problem becomes more challenging X is not sensed as much Measurements y k do not contain as much information about X The equality P( Z, H a = X, H a, a = 1,..., k) = q k no longer holds Nevertheless... Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

74 Achievability under sparse model Theorem (Achievability under sparse model) Assume Sensing matrices H a drawn according to δ-sparse distribution Min-rank decoder is used r/n γ (constant) Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

75 Achievability under sparse model Theorem (Achievability under sparse model) Assume Sensing matrices H a drawn according to δ-sparse distribution Min-rank decoder is used r/n γ (constant) If the sparsity factor δ = δ n satisfies ( ) log n δ Ω o(1) n Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

76 Achievability under sparse model Theorem (Achievability under sparse model) Assume Sensing matrices H a drawn according to δ-sparse distribution Min-rank decoder is used r/n γ (constant) If the sparsity factor δ = δ n satisfies ( ) log n δ Ω o(1) n and the number of measurements satisfies ( k > (2 + ε)γ 1 γ ) n 2 2 then P(E n ) 0. Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

77 Proof Strategy Approximate with uniform model Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

78 Proof Strategy Approximate with uniform model P(E n ) P( Z, H a = X, H a, a = 1,..., k) Z X:rank(Z) rank(x) Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

79 Proof Strategy Approximate with uniform model P(E n ) P( Z, H a = X, H a, a = 1,..., k) Z X:rank(Z) rank(x) Z X 0 βn 2 Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

80 Proof Strategy Approximate with uniform model P(E n ) P( Z, H a = X, H a, a = 1,..., k) Z X:rank(Z) rank(x) Z X 0 βn 2 + Z X:rank(Z) rank(x) Z X 0 >βn 2 P( Z, H a = X, H a, a = 1,..., k) Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

81 Proof Strategy Approximate with uniform model P(E n ) P( Z, H a = X, H a, a = 1,..., k) Z X:rank(Z) rank(x) Z X 0 βn 2 + Z X:rank(Z) rank(x) Z X 0 >βn 2 P( Z, H a = X, H a, a = 1,..., k) First term Few terms Low Hamming weight can afford loose bound on probability Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

82 Proof Strategy Approximate with uniform model P(E n ) P( Z, H a = X, H a, a = 1,..., k) Z X:rank(Z) rank(x) Z X 0 βn 2 + Z X:rank(Z) rank(x) Z X 0 >βn 2 P( Z, H a = X, H a, a = 1,..., k) First term Few terms Low Hamming weight can afford loose bound on probability Second term Many terms Tight bound on probability (circular convolution of δ-sparse pmf) Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

83 Proof Strategy Approximate with uniform model P(E n ) P( Z, H a = X, H a, a = 1,..., k) Z X:rank(Z) rank(x) Z X 0 βn 2 + Z X:rank(Z) rank(x) Z X 0 >βn 2 P( Z, H a = X, H a, a = 1,..., k) First term Few terms Low Hamming weight can afford loose bound on probability Second term Many terms Tight bound on probability (circular convolution of δ-sparse pmf) Set β = Θ( δ log n ) to be the optimal split Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

84 Concluding remarks Derived necessary and sufficient conditions for rank minimization over finite fields Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

85 Concluding remarks Derived necessary and sufficient conditions for rank minimization over finite fields Analyzed the sparse sensing model Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

86 Concluding remarks Derived necessary and sufficient conditions for rank minimization over finite fields Analyzed the sparse sensing model Minimum rank distance properties of rank-metric codes [analogous to Barg and Forney (2002)] were derived Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

87 Concluding remarks Derived necessary and sufficient conditions for rank minimization over finite fields Analyzed the sparse sensing model Minimum rank distance properties of rank-metric codes [analogous to Barg and Forney (2002)] were derived Drew analogies between number of measurements and minimum distance properties Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

88 Concluding remarks Derived necessary and sufficient conditions for rank minimization over finite fields Analyzed the sparse sensing model Minimum rank distance properties of rank-metric codes [analogous to Barg and Forney (2002)] were derived Drew analogies between number of measurements and minimum distance properties Open Problem: Can the sparsity level of ( ) log n Ω n be improved (reduced) further? Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

89 Concluding remarks Derived necessary and sufficient conditions for rank minimization over finite fields Analyzed the sparse sensing model Minimum rank distance properties of rank-metric codes [analogous to Barg and Forney (2002)] were derived Drew analogies between number of measurements and minimum distance properties Open Problem: Can the sparsity level of ( ) log n Ω n be improved (reduced) further? Vincent Tan (UW-Madison) Rank Minimization over Finite Fields ISIT / 20

A Connection between Network Coding and. Convolutional Codes

A Connection between Network Coding and. Convolutional Codes A Connection between Network Coding and 1 Convolutional Codes Christina Fragouli, Emina Soljanin christina.fragouli@epfl.ch, emina@lucent.com Abstract The min-cut, max-flow theorem states that a source

More information

Outline Introduction Problem Formulation Proposed Solution Applications Conclusion. Compressed Sensing. David L Donoho Presented by: Nitesh Shroff

Outline Introduction Problem Formulation Proposed Solution Applications Conclusion. Compressed Sensing. David L Donoho Presented by: Nitesh Shroff Compressed Sensing David L Donoho Presented by: Nitesh Shroff University of Maryland Outline 1 Introduction Compressed Sensing 2 Problem Formulation Sparse Signal Problem Statement 3 Proposed Solution

More information

Summary of Raptor Codes

Summary of Raptor Codes Summary of Raptor Codes Tracey Ho October 29, 2003 1 Introduction This summary gives an overview of Raptor Codes, the latest class of codes proposed for reliable multicast in the Digital Fountain model.

More information

Numerical Optimization

Numerical Optimization Convex Sets Computer Science and Automation Indian Institute of Science Bangalore 560 012, India. NPTEL Course on Let x 1, x 2 R n, x 1 x 2. Line and line segment Line passing through x 1 and x 2 : {y

More information

Polar Codes for Noncoherent MIMO Signalling

Polar Codes for Noncoherent MIMO Signalling ICC Polar Codes for Noncoherent MIMO Signalling Philip R. Balogun, Ian Marsland, Ramy Gohary, and Halim Yanikomeroglu Department of Systems and Computer Engineering, Carleton University, Canada WCS IS6

More information

Linear Block Codes. Allen B. MacKenzie Notes for February 4, 9, & 11, Some Definitions

Linear Block Codes. Allen B. MacKenzie Notes for February 4, 9, & 11, Some Definitions Linear Block Codes Allen B. MacKenzie Notes for February 4, 9, & 11, 2015 This handout covers our in-class study of Chapter 3 of your textbook. We ll introduce some notation and then discuss the generator

More information

THREE LECTURES ON BASIC TOPOLOGY. 1. Basic notions.

THREE LECTURES ON BASIC TOPOLOGY. 1. Basic notions. THREE LECTURES ON BASIC TOPOLOGY PHILIP FOTH 1. Basic notions. Let X be a set. To make a topological space out of X, one must specify a collection T of subsets of X, which are said to be open subsets of

More information

The AWGN Red Alert Problem

The AWGN Red Alert Problem The AWGN Red Alert Problem Bobak Nazer, Yanina Shkel, and Stark C. Draper ECE Department Boston University ECE Department University of Wisconsin, Madison ITA 2011 February 10, 2011 Nazer, Shkel, Draper

More information

Math 5593 Linear Programming Lecture Notes

Math 5593 Linear Programming Lecture Notes Math 5593 Linear Programming Lecture Notes Unit II: Theory & Foundations (Convex Analysis) University of Colorado Denver, Fall 2013 Topics 1 Convex Sets 1 1.1 Basic Properties (Luenberger-Ye Appendix B.1).........................

More information

However, this is not always true! For example, this fails if both A and B are closed and unbounded (find an example).

However, this is not always true! For example, this fails if both A and B are closed and unbounded (find an example). 98 CHAPTER 3. PROPERTIES OF CONVEX SETS: A GLIMPSE 3.2 Separation Theorems It seems intuitively rather obvious that if A and B are two nonempty disjoint convex sets in A 2, then there is a line, H, separating

More information

An incomplete survey of matrix completion. Benjamin Recht Department of Computer Sciences University of Wisconsin-Madison

An incomplete survey of matrix completion. Benjamin Recht Department of Computer Sciences University of Wisconsin-Madison An incomplete survey of matrix completion Benjamin Recht Department of Computer Sciences University of Wisconsin-Madison Abstract Setup: Matrix Completion M = M ij known for black cells M ij unknown for

More information

Maths for Signals and Systems Linear Algebra in Engineering. Some problems by Gilbert Strang

Maths for Signals and Systems Linear Algebra in Engineering. Some problems by Gilbert Strang Maths for Signals and Systems Linear Algebra in Engineering Some problems by Gilbert Strang Problems. Consider u, v, w to be non-zero vectors in R 7. These vectors span a vector space. What are the possible

More information

7. The Gauss-Bonnet theorem

7. The Gauss-Bonnet theorem 7. The Gauss-Bonnet theorem 7.1 Hyperbolic polygons In Euclidean geometry, an n-sided polygon is a subset of the Euclidean plane bounded by n straight lines. Thus the edges of a Euclidean polygon are formed

More information

MOST attention in the literature of network codes has

MOST attention in the literature of network codes has 3862 IEEE TRANSACTIONS ON INFORMATION THEORY, VOL. 56, NO. 8, AUGUST 2010 Efficient Network Code Design for Cyclic Networks Elona Erez, Member, IEEE, and Meir Feder, Fellow, IEEE Abstract This paper introduces

More information

Measurements and Bits: Compressed Sensing meets Information Theory. Dror Baron ECE Department Rice University dsp.rice.edu/cs

Measurements and Bits: Compressed Sensing meets Information Theory. Dror Baron ECE Department Rice University dsp.rice.edu/cs Measurements and Bits: Compressed Sensing meets Information Theory Dror Baron ECE Department Rice University dsp.rice.edu/cs Sensing by Sampling Sample data at Nyquist rate Compress data using model (e.g.,

More information

Detecting Burnscar from Hyperspectral Imagery via Sparse Representation with Low-Rank Interference

Detecting Burnscar from Hyperspectral Imagery via Sparse Representation with Low-Rank Interference Detecting Burnscar from Hyperspectral Imagery via Sparse Representation with Low-Rank Interference Minh Dao 1, Xiang Xiang 1, Bulent Ayhan 2, Chiman Kwan 2, Trac D. Tran 1 Johns Hopkins Univeristy, 3400

More information

Analyzing the Peeling Decoder

Analyzing the Peeling Decoder Analyzing the Peeling Decoder Supplemental Material for Advanced Channel Coding Henry D. Pfister January 5th, 01 1 Introduction The simplest example of iterative decoding is the peeling decoder introduced

More information

4. Simplicial Complexes and Simplicial Homology

4. Simplicial Complexes and Simplicial Homology MATH41071/MATH61071 Algebraic topology Autumn Semester 2017 2018 4. Simplicial Complexes and Simplicial Homology Geometric simplicial complexes 4.1 Definition. A finite subset { v 0, v 1,..., v r } R n

More information

1. Introduction. performance of numerical methods. complexity bounds. structural convex optimization. course goals and topics

1. Introduction. performance of numerical methods. complexity bounds. structural convex optimization. course goals and topics 1. Introduction EE 546, Univ of Washington, Spring 2016 performance of numerical methods complexity bounds structural convex optimization course goals and topics 1 1 Some course info Welcome to EE 546!

More information

Combinatorial Selection and Least Absolute Shrinkage via The CLASH Operator

Combinatorial Selection and Least Absolute Shrinkage via The CLASH Operator Combinatorial Selection and Least Absolute Shrinkage via The CLASH Operator Volkan Cevher Laboratory for Information and Inference Systems LIONS / EPFL http://lions.epfl.ch & Idiap Research Institute joint

More information

Lecture 15: The subspace topology, Closed sets

Lecture 15: The subspace topology, Closed sets Lecture 15: The subspace topology, Closed sets 1 The Subspace Topology Definition 1.1. Let (X, T) be a topological space with topology T. subset of X, the collection If Y is a T Y = {Y U U T} is a topology

More information

The convex geometry of inverse problems

The convex geometry of inverse problems The convex geometry of inverse problems Benjamin Recht Department of Computer Sciences University of Wisconsin-Madison Joint work with Venkat Chandrasekaran Pablo Parrilo Alan Willsky Linear Inverse Problems

More information

Lecture 19. Lecturer: Aleksander Mądry Scribes: Chidambaram Annamalai and Carsten Moldenhauer

Lecture 19. Lecturer: Aleksander Mądry Scribes: Chidambaram Annamalai and Carsten Moldenhauer CS-621 Theory Gems November 21, 2012 Lecture 19 Lecturer: Aleksander Mądry Scribes: Chidambaram Annamalai and Carsten Moldenhauer 1 Introduction We continue our exploration of streaming algorithms. First,

More information

In what follows, we will focus on Voronoi diagrams in Euclidean space. Later, we will generalize to other distance spaces.

In what follows, we will focus on Voronoi diagrams in Euclidean space. Later, we will generalize to other distance spaces. Voronoi Diagrams 4 A city builds a set of post offices, and now needs to determine which houses will be served by which office. It would be wasteful for a postman to go out of their way to make a delivery

More information

Unlabeled equivalence for matroids representable over finite fields

Unlabeled equivalence for matroids representable over finite fields Unlabeled equivalence for matroids representable over finite fields November 16, 2012 S. R. Kingan Department of Mathematics Brooklyn College, City University of New York 2900 Bedford Avenue Brooklyn,

More information

Lecture 2 September 3

Lecture 2 September 3 EE 381V: Large Scale Optimization Fall 2012 Lecture 2 September 3 Lecturer: Caramanis & Sanghavi Scribe: Hongbo Si, Qiaoyang Ye 2.1 Overview of the last Lecture The focus of the last lecture was to give

More information

Lecture 19 Subgradient Methods. November 5, 2008

Lecture 19 Subgradient Methods. November 5, 2008 Subgradient Methods November 5, 2008 Outline Lecture 19 Subgradients and Level Sets Subgradient Method Convergence and Convergence Rate Convex Optimization 1 Subgradients and Level Sets A vector s is a

More information

MDS (maximum-distance separable) codes over large

MDS (maximum-distance separable) codes over large IEEE TRANSACTIONS ON INFORMATION THEORY, VOL 55, NO 4, APRIL 2009 1721 Cyclic Lowest Density MDS Array Codes Yuval Cassuto, Member, IEEE, and Jehoshua Bruck, Fellow, IEEE Abstract Three new families of

More information

Lecture and notes by: Nate Chenette, Brent Myers, Hari Prasad November 8, Property Testing

Lecture and notes by: Nate Chenette, Brent Myers, Hari Prasad November 8, Property Testing Property Testing 1 Introduction Broadly, property testing is the study of the following class of problems: Given the ability to perform (local) queries concerning a particular object (e.g., a function,

More information

1. Represent each of these relations on {1, 2, 3} with a matrix (with the elements of this set listed in increasing order).

1. Represent each of these relations on {1, 2, 3} with a matrix (with the elements of this set listed in increasing order). Exercises Exercises 1. Represent each of these relations on {1, 2, 3} with a matrix (with the elements of this set listed in increasing order). a) {(1, 1), (1, 2), (1, 3)} b) {(1, 2), (2, 1), (2, 2), (3,

More information

Chapter S:V. V. Formal Properties of A*

Chapter S:V. V. Formal Properties of A* Chapter S:V V. Formal Properties of A* Properties of Search Space Graphs Auxiliary Concepts Roadmap Completeness of A* Admissibility of A* Efficiency of A* Monotone Heuristic Functions S:V-1 Formal Properties

More information

Chapter 4. Matrix and Vector Operations

Chapter 4. Matrix and Vector Operations 1 Scope of the Chapter Chapter 4 This chapter provides procedures for matrix and vector operations. This chapter (and Chapters 5 and 6) can handle general matrices, matrices with special structure and

More information

Error-Correcting Codes

Error-Correcting Codes Error-Correcting Codes Michael Mo 10770518 6 February 2016 Abstract An introduction to error-correcting codes will be given by discussing a class of error-correcting codes, called linear block codes. The

More information

Homework #5 Solutions Due: July 17, 2012 G = G = Find a standard form generator matrix for a code equivalent to C.

Homework #5 Solutions Due: July 17, 2012 G = G = Find a standard form generator matrix for a code equivalent to C. Homework #5 Solutions Due: July 7, Do the following exercises from Lax: Page 4: 4 Page 34: 35, 36 Page 43: 44, 45, 46 4 Let C be the (5, 3) binary code with generator matrix G = Find a standard form generator

More information

Polytopes Course Notes

Polytopes Course Notes Polytopes Course Notes Carl W. Lee Department of Mathematics University of Kentucky Lexington, KY 40506 lee@ms.uky.edu Fall 2013 i Contents 1 Polytopes 1 1.1 Convex Combinations and V-Polytopes.....................

More information

Efficiently decodable insertion/deletion codes for high-noise and high-rate regimes

Efficiently decodable insertion/deletion codes for high-noise and high-rate regimes Efficiently decodable insertion/deletion codes for high-noise and high-rate regimes Venkatesan Guruswami Carnegie Mellon University Pittsburgh, PA 53 Email: guruswami@cmu.edu Ray Li Carnegie Mellon University

More information

60 2 Convex sets. {x a T x b} {x ã T x b}

60 2 Convex sets. {x a T x b} {x ã T x b} 60 2 Convex sets Exercises Definition of convexity 21 Let C R n be a convex set, with x 1,, x k C, and let θ 1,, θ k R satisfy θ i 0, θ 1 + + θ k = 1 Show that θ 1x 1 + + θ k x k C (The definition of convexity

More information

Giovanni De Micheli. Integrated Systems Centre EPF Lausanne

Giovanni De Micheli. Integrated Systems Centre EPF Lausanne Two-level Logic Synthesis and Optimization Giovanni De Micheli Integrated Systems Centre EPF Lausanne This presentation can be used for non-commercial purposes as long as this note and the copyright footers

More information

Lecture 17. Lower bound for variable-length source codes with error. Coding a sequence of symbols: Rates and scheme (Arithmetic code)

Lecture 17. Lower bound for variable-length source codes with error. Coding a sequence of symbols: Rates and scheme (Arithmetic code) Lecture 17 Agenda for the lecture Lower bound for variable-length source codes with error Coding a sequence of symbols: Rates and scheme (Arithmetic code) Introduction to universal codes 17.1 variable-length

More information

Mathematics and Computer Science

Mathematics and Computer Science Technical Report TR-2006-010 Revisiting hypergraph models for sparse matrix decomposition by Cevdet Aykanat, Bora Ucar Mathematics and Computer Science EMORY UNIVERSITY REVISITING HYPERGRAPH MODELS FOR

More information

Non-exhaustive, Overlapping k-means

Non-exhaustive, Overlapping k-means Non-exhaustive, Overlapping k-means J. J. Whang, I. S. Dhilon, and D. F. Gleich Teresa Lebair University of Maryland, Baltimore County October 29th, 2015 Teresa Lebair UMBC 1/38 Outline Introduction NEO-K-Means

More information

DM545 Linear and Integer Programming. Lecture 2. The Simplex Method. Marco Chiarandini

DM545 Linear and Integer Programming. Lecture 2. The Simplex Method. Marco Chiarandini DM545 Linear and Integer Programming Lecture 2 The Marco Chiarandini Department of Mathematics & Computer Science University of Southern Denmark Outline 1. 2. 3. 4. Standard Form Basic Feasible Solutions

More information

A Toolkit for List Recovery. Nick Wasylyshyn A THESIS. Mathematics

A Toolkit for List Recovery. Nick Wasylyshyn A THESIS. Mathematics A Toolkit for List Recovery Nick Wasylyshyn A THESIS in Mathematics Presented to the Faculties of the University of Pennsylvania in Partial Fulfillment of the Requirements for the Degree of Master of Arts

More information

EC500. Design of Secure and Reliable Hardware. Lecture 9. Mark Karpovsky

EC500. Design of Secure and Reliable Hardware. Lecture 9. Mark Karpovsky EC500 Design of Secure and Reliable Hardware Lecture 9 Mark Karpovsky 1 1 Arithmetical Codes 1.1 Detection and Correction of errors in arithmetical channels (adders, multipliers, etc) Let = 0,1,,2 1 and

More information

Topological properties of convex sets

Topological properties of convex sets Division of the Humanities and Social Sciences Ec 181 KC Border Convex Analysis and Economic Theory Winter 2018 Topic 5: Topological properties of convex sets 5.1 Interior and closure of convex sets Let

More information

Fast Approximation of Matrix Coherence and Statistical Leverage

Fast Approximation of Matrix Coherence and Statistical Leverage Fast Approximation of Matrix Coherence and Statistical Leverage Michael W. Mahoney Stanford University ( For more info, see: http:// cs.stanford.edu/people/mmahoney/ or Google on Michael Mahoney ) Statistical

More information

Generell Topologi. Richard Williamson. May 6, 2013

Generell Topologi. Richard Williamson. May 6, 2013 Generell Topologi Richard Williamson May 6, 2013 1 8 Thursday 7th February 8.1 Using connectedness to distinguish between topological spaces I Proposition 8.1. Let (, O ) and (Y, O Y ) be topological spaces.

More information

LOW-density parity-check (LDPC) codes are widely

LOW-density parity-check (LDPC) codes are widely 1460 IEEE TRANSACTIONS ON INFORMATION THEORY, VOL 53, NO 4, APRIL 2007 Tree-Based Construction of LDPC Codes Having Good Pseudocodeword Weights Christine A Kelley, Member, IEEE, Deepak Sridhara, Member,

More information

arxiv: v2 [math.oc] 29 Jan 2015

arxiv: v2 [math.oc] 29 Jan 2015 Minimum Cost Input/Output Design for Large-Scale Linear Structural Systems Sérgio Pequito, Soummya Kar A. Pedro Aguiar, 1 arxiv:1407.3327v2 [math.oc] 29 Jan 2015 Abstract In this paper, we provide optimal

More information

3 No-Wait Job Shops with Variable Processing Times

3 No-Wait Job Shops with Variable Processing Times 3 No-Wait Job Shops with Variable Processing Times In this chapter we assume that, on top of the classical no-wait job shop setting, we are given a set of processing times for each operation. We may select

More information

Interleaving Schemes on Circulant Graphs with Two Offsets

Interleaving Schemes on Circulant Graphs with Two Offsets Interleaving Schemes on Circulant raphs with Two Offsets Aleksandrs Slivkins Department of Computer Science Cornell University Ithaca, NY 14853 slivkins@cs.cornell.edu Jehoshua Bruck Department of Electrical

More information

Section 6.3: Further Rules for Counting Sets

Section 6.3: Further Rules for Counting Sets Section 6.3: Further Rules for Counting Sets Often when we are considering the probability of an event, that event is itself a union of other events. For example, suppose there is a horse race with three

More information

Multiresponse Sparse Regression with Application to Multidimensional Scaling

Multiresponse Sparse Regression with Application to Multidimensional Scaling Multiresponse Sparse Regression with Application to Multidimensional Scaling Timo Similä and Jarkko Tikka Helsinki University of Technology, Laboratory of Computer and Information Science P.O. Box 54,

More information

Arithmetic in Quaternion Algebras

Arithmetic in Quaternion Algebras Arithmetic in Quaternion Algebras 31st Automorphic Forms Workshop Jordan Wiebe University of Oklahoma March 6, 2017 Jordan Wiebe (University of Oklahoma) Arithmetic in Quaternion Algebras March 6, 2017

More information

Computing over Multiple-Access Channels with Connections to Wireless Network Coding

Computing over Multiple-Access Channels with Connections to Wireless Network Coding ISIT 06: Computing over MACS 1 / 20 Computing over Multiple-Access Channels with Connections to Wireless Network Coding Bobak Nazer and Michael Gastpar Wireless Foundations Center Department of Electrical

More information

Advanced Operations Research Techniques IE316. Quiz 1 Review. Dr. Ted Ralphs

Advanced Operations Research Techniques IE316. Quiz 1 Review. Dr. Ted Ralphs Advanced Operations Research Techniques IE316 Quiz 1 Review Dr. Ted Ralphs IE316 Quiz 1 Review 1 Reading for The Quiz Material covered in detail in lecture. 1.1, 1.4, 2.1-2.6, 3.1-3.3, 3.5 Background material

More information

Two-graphs revisited. Peter J. Cameron University of St Andrews Modern Trends in Algebraic Graph Theory Villanova, June 2014

Two-graphs revisited. Peter J. Cameron University of St Andrews Modern Trends in Algebraic Graph Theory Villanova, June 2014 Two-graphs revisited Peter J. Cameron University of St Andrews Modern Trends in Algebraic Graph Theory Villanova, June 2014 History The icosahedron has six diagonals, any two making the same angle (arccos(1/

More information

The Probabilistic Method

The Probabilistic Method The Probabilistic Method Po-Shen Loh June 2010 1 Warm-up 1. (Russia 1996/4 In the Duma there are 1600 delegates, who have formed 16000 committees of 80 persons each. Prove that one can find two committees

More information

Polynomial Codes: an Optimal Design for High-Dimensional Coded Matrix Multiplication

Polynomial Codes: an Optimal Design for High-Dimensional Coded Matrix Multiplication Polynomial Codes: an Optimal Design for High-Dimensional Coded Matrix Multiplication Qian Yu, Mohammad Ali Maddah-Ali, and A. Salman Avestimehr Department of Electrical Engineering, University of Southern

More information

Embeddability of Arrangements of Pseudocircles into the Sphere

Embeddability of Arrangements of Pseudocircles into the Sphere Embeddability of Arrangements of Pseudocircles into the Sphere Ronald Ortner Department Mathematik und Informationstechnologie, Montanuniversität Leoben, Franz-Josef-Straße 18, 8700-Leoben, Austria Abstract

More information

Error correction guarantees

Error correction guarantees Error correction guarantees Drawback of asymptotic analyses Valid only as long as the incoming messages are independent. (independence assumption) The messages are independent for l iterations only if

More information

Describe the two most important ways in which subspaces of F D arise. (These ways were given as the motivation for looking at subspaces.

Describe the two most important ways in which subspaces of F D arise. (These ways were given as the motivation for looking at subspaces. Quiz Describe the two most important ways in which subspaces of F D arise. (These ways were given as the motivation for looking at subspaces.) What are the two subspaces associated with a matrix? Describe

More information

BELOW, we consider decoding algorithms for Reed Muller

BELOW, we consider decoding algorithms for Reed Muller 4880 IEEE TRANSACTIONS ON INFORMATION THEORY, VOL. 52, NO. 11, NOVEMBER 2006 Error Exponents for Recursive Decoding of Reed Muller Codes on a Binary-Symmetric Channel Marat Burnashev and Ilya Dumer, Senior

More information

arxiv: v1 [math.co] 22 Dec 2016

arxiv: v1 [math.co] 22 Dec 2016 THE LERAY DIMENSION OF A CONVEX CODE CARINA CURTO AND RAMÓN VERA arxiv:607797v [mathco] Dec 06 ABSTRACT Convex codes were recently introduced as models for neural codes in the brain Any convex code C has

More information

Graphs and Network Flows IE411. Lecture 21. Dr. Ted Ralphs

Graphs and Network Flows IE411. Lecture 21. Dr. Ted Ralphs Graphs and Network Flows IE411 Lecture 21 Dr. Ted Ralphs IE411 Lecture 21 1 Combinatorial Optimization and Network Flows In general, most combinatorial optimization and integer programming problems are

More information

The convex geometry of inverse problems

The convex geometry of inverse problems The convex geometry of inverse problems Benjamin Recht Department of Computer Sciences University of Wisconsin-Madison Joint work with Venkat Chandrasekaran Pablo Parrilo Alan Willsky Linear Inverse Problems

More information

Subspace Clustering by Block Diagonal Representation

Subspace Clustering by Block Diagonal Representation IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE Subspace Clustering by Block Diagonal Representation Canyi Lu, Student Member, IEEE, Jiashi Feng, Zhouchen Lin, Senior Member, IEEE, Tao Mei,

More information

TOPOLOGY, DR. BLOCK, FALL 2015, NOTES, PART 3.

TOPOLOGY, DR. BLOCK, FALL 2015, NOTES, PART 3. TOPOLOGY, DR. BLOCK, FALL 2015, NOTES, PART 3. 301. Definition. Let m be a positive integer, and let X be a set. An m-tuple of elements of X is a function x : {1,..., m} X. We sometimes use x i instead

More information

Treewidth and graph minors

Treewidth and graph minors Treewidth and graph minors Lectures 9 and 10, December 29, 2011, January 5, 2012 We shall touch upon the theory of Graph Minors by Robertson and Seymour. This theory gives a very general condition under

More information

Matrix algorithms: fast, stable, communication-optimizing random?!

Matrix algorithms: fast, stable, communication-optimizing random?! Matrix algorithms: fast, stable, communication-optimizing random?! Ioana Dumitriu Department of Mathematics University of Washington (Seattle) Joint work with Grey Ballard, James Demmel, Olga Holtz, Robert

More information

Subspace Clustering by Block Diagonal Representation

Subspace Clustering by Block Diagonal Representation IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE Subspace Clustering by Block Diagonal Representation Canyi Lu, Student Member, IEEE, Jiashi Feng, Zhouchen Lin, Senior Member, IEEE, Tao Mei,

More information

The External Network Problem

The External Network Problem The External Network Problem Jan van den Heuvel and Matthew Johnson CDAM Research Report LSE-CDAM-2004-15 December 2004 Abstract The connectivity of a communications network can often be enhanced if the

More information

TELCOM2125: Network Science and Analysis

TELCOM2125: Network Science and Analysis School of Information Sciences University of Pittsburgh TELCOM2125: Network Science and Analysis Konstantinos Pelechrinis Spring 2015 2 Part 4: Dividing Networks into Clusters The problem l Graph partitioning

More information

Section 5 Convex Optimisation 1. W. Dai (IC) EE4.66 Data Proc. Convex Optimisation page 5-1

Section 5 Convex Optimisation 1. W. Dai (IC) EE4.66 Data Proc. Convex Optimisation page 5-1 Section 5 Convex Optimisation 1 W. Dai (IC) EE4.66 Data Proc. Convex Optimisation 1 2018 page 5-1 Convex Combination Denition 5.1 A convex combination is a linear combination of points where all coecients

More information

CHAPTER 3 FUZZY RELATION and COMPOSITION

CHAPTER 3 FUZZY RELATION and COMPOSITION CHAPTER 3 FUZZY RELATION and COMPOSITION The concept of fuzzy set as a generalization of crisp set has been introduced in the previous chapter. Relations between elements of crisp sets can be extended

More information

Exact adaptive parallel algorithms for data depth problems. Vera Rosta Department of Mathematics and Statistics McGill University, Montreal

Exact adaptive parallel algorithms for data depth problems. Vera Rosta Department of Mathematics and Statistics McGill University, Montreal Exact adaptive parallel algorithms for data depth problems Vera Rosta Department of Mathematics and Statistics McGill University, Montreal joint work with Komei Fukuda School of Computer Science McGill

More information

arxiv: v1 [math.co] 27 Feb 2015

arxiv: v1 [math.co] 27 Feb 2015 Mode Poset Probability Polytopes Guido Montúfar 1 and Johannes Rauh 2 arxiv:1503.00572v1 [math.co] 27 Feb 2015 1 Max Planck Institute for Mathematics in the Sciences, Inselstraße 22, 04103 Leipzig, Germany,

More information

Convex Optimization MLSS 2015

Convex Optimization MLSS 2015 Convex Optimization MLSS 2015 Constantine Caramanis The University of Texas at Austin The Optimization Problem minimize : f (x) subject to : x X. The Optimization Problem minimize : f (x) subject to :

More information

Multidimensional Decoding Networks for Trapping Set Analysis

Multidimensional Decoding Networks for Trapping Set Analysis Multidimensional Decoding Networks for Trapping Set Analysis Allison Beemer* and Christine A. Kelley University of Nebraska-Lincoln, Lincoln, NE, U.S.A. allison.beemer@huskers.unl.edu Abstract. We present

More information

Graph drawing in spectral layout

Graph drawing in spectral layout Graph drawing in spectral layout Maureen Gallagher Colleen Tygh John Urschel Ludmil Zikatanov Beginning: July 8, 203; Today is: October 2, 203 Introduction Our research focuses on the use of spectral graph

More information

Pebble Sets in Convex Polygons

Pebble Sets in Convex Polygons 2 1 Pebble Sets in Convex Polygons Kevin Iga, Randall Maddox June 15, 2005 Abstract Lukács and András posed the problem of showing the existence of a set of n 2 points in the interior of a convex n-gon

More information

Computing the Minimum Hamming Distance for Z 2 Z 4 -Linear Codes

Computing the Minimum Hamming Distance for Z 2 Z 4 -Linear Codes Computing the Minimum Hamming Distance for Z 2 Z 4 -Linear Codes Marta Pujol and Mercè Villanueva Combinatorics, Coding and Security Group (CCSG) Universitat Autònoma de Barcelona (UAB) VIII JMDA, Almería

More information

Efficient Universal Recovery in Broadcast Networks

Efficient Universal Recovery in Broadcast Networks Efficient Universal Recovery in Broadcast Networks Thomas Courtade and Rick Wesel UCLA September 30, 2010 Courtade and Wesel (UCLA) Efficient Universal Recovery Allerton 2010 1 / 19 System Model and Problem

More information

An I/O-Complexity Lower Bound for All Recursive Matrix Multiplication Algorithms by Path-Routing. Jacob N. Scott

An I/O-Complexity Lower Bound for All Recursive Matrix Multiplication Algorithms by Path-Routing. Jacob N. Scott An I/O-Complexity Lower Bound for All Recursive Matrix Multiplication Algorithms by Path-Routing by Jacob N. Scott A dissertation submitted in partial satisfaction of the requirements for the degree of

More information

Robust Face Recognition via Sparse Representation Authors: John Wright, Allen Y. Yang, Arvind Ganesh, S. Shankar Sastry, and Yi Ma

Robust Face Recognition via Sparse Representation Authors: John Wright, Allen Y. Yang, Arvind Ganesh, S. Shankar Sastry, and Yi Ma Robust Face Recognition via Sparse Representation Authors: John Wright, Allen Y. Yang, Arvind Ganesh, S. Shankar Sastry, and Yi Ma Presented by Hu Han Jan. 30 2014 For CSE 902 by Prof. Anil K. Jain: Selected

More information

POLYHEDRAL GEOMETRY. Convex functions and sets. Mathematical Programming Niels Lauritzen Recall that a subset C R n is convex if

POLYHEDRAL GEOMETRY. Convex functions and sets. Mathematical Programming Niels Lauritzen Recall that a subset C R n is convex if POLYHEDRAL GEOMETRY Mathematical Programming Niels Lauritzen 7.9.2007 Convex functions and sets Recall that a subset C R n is convex if {λx + (1 λ)y 0 λ 1} C for every x, y C and 0 λ 1. A function f :

More information

Introduction to Topics in Machine Learning

Introduction to Topics in Machine Learning Introduction to Topics in Machine Learning Namrata Vaswani Department of Electrical and Computer Engineering Iowa State University Namrata Vaswani 1/ 27 Compressed Sensing / Sparse Recovery: Given y :=

More information

Flexible Coloring. Xiaozhou Li a, Atri Rudra b, Ram Swaminathan a. Abstract

Flexible Coloring. Xiaozhou Li a, Atri Rudra b, Ram Swaminathan a. Abstract Flexible Coloring Xiaozhou Li a, Atri Rudra b, Ram Swaminathan a a firstname.lastname@hp.com, HP Labs, 1501 Page Mill Road, Palo Alto, CA 94304 b atri@buffalo.edu, Computer Sc. & Engg. dept., SUNY Buffalo,

More information

T325 Summary T305 T325 B BLOCK 4 T325. Session 3. Dr. Saatchi, Seyed Mohsen. Prepared by:

T325 Summary T305 T325 B BLOCK 4 T325. Session 3. Dr. Saatchi, Seyed Mohsen. Prepared by: T305 T325 B BLOCK 4 T325 Summary Prepared by: Session 3 [Type Dr. Saatchi, your address] Seyed Mohsen [Type your phone number] [Type your e-mail address] Dr. Saatchi, Seyed Mohsen T325 Error Control Coding

More information

Lecture 5: Simplicial Complex

Lecture 5: Simplicial Complex Lecture 5: Simplicial Complex 2-Manifolds, Simplex and Simplicial Complex Scribed by: Lei Wang First part of this lecture finishes 2-Manifolds. Rest part of this lecture talks about simplicial complex.

More information

MATH 423 Linear Algebra II Lecture 17: Reduced row echelon form (continued). Determinant of a matrix.

MATH 423 Linear Algebra II Lecture 17: Reduced row echelon form (continued). Determinant of a matrix. MATH 423 Linear Algebra II Lecture 17: Reduced row echelon form (continued). Determinant of a matrix. Row echelon form A matrix is said to be in the row echelon form if the leading entries shift to the

More information

Section 17. Closed Sets and Limit Points

Section 17. Closed Sets and Limit Points 17. Closed Sets and Limit Points 1 Section 17. Closed Sets and Limit Points Note. In this section, we finally define a closed set. We also introduce several traditional topological concepts, such as limit

More information

Minimum Delay Packet-sizing for Linear Multi-hop Networks with Cooperative Transmissions

Minimum Delay Packet-sizing for Linear Multi-hop Networks with Cooperative Transmissions Minimum Delay acket-sizing for inear Multi-hop Networks with Cooperative Transmissions Ning Wen and Randall A. Berry Department of Electrical Engineering and Computer Science Northwestern University, Evanston,

More information

1 Undirected Vertex Geography UVG

1 Undirected Vertex Geography UVG Geography Start with a chip sitting on a vertex v of a graph or digraph G. A move consists of moving the chip to a neighbouring vertex. In edge geography, moving the chip from x to y deletes the edge (x,

More information

Integrated Math I. IM1.1.3 Understand and use the distributive, associative, and commutative properties.

Integrated Math I. IM1.1.3 Understand and use the distributive, associative, and commutative properties. Standard 1: Number Sense and Computation Students simplify and compare expressions. They use rational exponents and simplify square roots. IM1.1.1 Compare real number expressions. IM1.1.2 Simplify square

More information

Voronoi Region. K-means method for Signal Compression: Vector Quantization. Compression Formula 11/20/2013

Voronoi Region. K-means method for Signal Compression: Vector Quantization. Compression Formula 11/20/2013 Voronoi Region K-means method for Signal Compression: Vector Quantization Blocks of signals: A sequence of audio. A block of image pixels. Formally: vector example: (0.2, 0.3, 0.5, 0.1) A vector quantizer

More information

Lecture Notes 2: The Simplex Algorithm

Lecture Notes 2: The Simplex Algorithm Algorithmic Methods 25/10/2010 Lecture Notes 2: The Simplex Algorithm Professor: Yossi Azar Scribe:Kiril Solovey 1 Introduction In this lecture we will present the Simplex algorithm, finish some unresolved

More information

Phil 320 Chapter 1: Sets, Functions and Enumerability I. Sets Informally: a set is a collection of objects. The objects are called members or

Phil 320 Chapter 1: Sets, Functions and Enumerability I. Sets Informally: a set is a collection of objects. The objects are called members or Phil 320 Chapter 1: Sets, Functions and Enumerability I. Sets Informally: a set is a collection of objects. The objects are called members or elements of the set. a) Use capital letters to stand for sets

More information

CHAPTER 8. Copyright Cengage Learning. All rights reserved.

CHAPTER 8. Copyright Cengage Learning. All rights reserved. CHAPTER 8 RELATIONS Copyright Cengage Learning. All rights reserved. SECTION 8.3 Equivalence Relations Copyright Cengage Learning. All rights reserved. The Relation Induced by a Partition 3 The Relation

More information

Lecture 2. Topology of Sets in R n. August 27, 2008

Lecture 2. Topology of Sets in R n. August 27, 2008 Lecture 2 Topology of Sets in R n August 27, 2008 Outline Vectors, Matrices, Norms, Convergence Open and Closed Sets Special Sets: Subspace, Affine Set, Cone, Convex Set Special Convex Sets: Hyperplane,

More information