Complex Quadratic Optimization and Semidefinite Programming
|
|
- Britton White
- 5 years ago
- Views:
Transcription
1 Coplex Quadratic Optiization and Seidefinite Prograing Shuzhong Zhang Yongwei Huang August 4 Abstract In this paper we study the approxiation algoriths for a class of discrete quadratic optiization probles in the Heritian coplex for. A special case of the proble that we study corresponds to the ax-3-cut odel used in a recent paper of Goeans and Williason. We first develop a closed-for forula to copute the probability of a coplex-valued norally distributed bivariate rando vector to be in a given angular region. This forula allows us to copute the expected value of a randoized with a specific rounding rule solution based on the optial solution of the coplex SDP relaxation proble. In particular, we study the liit of that odel, in which the proble reains NP-hard. We show that if the objective is to axiize a positive seidefinite Heritian for, then the randoization-rounding procedure guarantees a worst-case perforance ratio of π/4.7854, which is better than the ratio of /π.6366 for its counter-part in the real case due to Nesterov. Furtherore, if the objective atrix is real-valued positive seidefinite with non-positive off-diagonal eleents, then the perforance ratio iproves to Keywords: Heritian quadratic functions, approxiation ratio, randoized algoriths, coplex SDP relaxation. MSC subject classification: 9C, 9C. Departent of Systes Engineering and Engineering Manageent, The Chinese University of Hong Kong, Shatin, Hong Kong zhang@se.cuhk.edu.hk. Research supported by Hong Kong RGC Eararked Grants CUHK433/E and CUHK474/3E. Departent of Systes Engineering and Engineering Manageent, The Chinese University of Hong Kong, Shatin, Hong Kong ywhuang@se.cuhk.edu.hk.
2 Introduction The pioneering work of Goeans and Williason [5] has caused a great deal of exciteent in the field of optiization, as it used a new tool SDP in continuous optiization, through randoization and probabilistic analysis, to yield an excellent approxiation ratio for a classical cobinatorial optiization proble, known as the ax-cut proble. This ground-breaking work has been extended in various ways since its first appearance. Aong others, Frieze and Jerru [4] extended the ethod to solve the general ax-k-cut proble. Bertsias and Ye [3] introduced another randoization schee using noral distributions, to achieve the sae approxiation result as in Goeans and Williason s original paper [5]. The Bertsias-Ye analysis akes use of an iportant result in statistics, which states that the probability of a bivariate -diensional norally distributed rando vector to be in the first orthant can be expressed analytically using eleentary functions. This is ipossible however, for any diension higher than three; see []. Recently, Goeans and Williason [6] proposed another novel approach to solve the ax-3-cut proble, using the unit circle in the coplex plane as a key odelling ingredient. In this paper we show that it is possible to copute the probability of the bivariate coplex-valued norally distributed rando vector to be in a specific angular region in C. Using this result, we are able to copute the expected quality of a particular randoized solution for solving a general quadratic optiization odel, where the variables take values fro the roots of z = is an integer paraeter of the odel. The odel of Goeans and Williason for ax-3-cut = 3 is a special case of this general odel. It is interesting to study the liit of this odel; that is, the case where. It turns out that the proble reains NP-hard. However, the corresponding coplex SDP relaxation yields an approxiation ratio of π/4.7854, whereas for its counter-part in the real case, the ratio is /π.6366 as shown by Nesterov [8]. If the off-diagonal eleents of the objective atrix are real-valued and non-positive, then the approxiation ratio is actually This paper is organized as follows. In Section we discuss the coputation of the probability for the coplex-valued noral distributions. In Section 3 we apply the results developed in Section to solve coplex-valued quadratic optiization probles. In particular, Subsection 3. discusses the Heritian quadratic function iniization proble, where the coplex decision variables take discrete values. Subsection 3. considers the continuous version of the proble. Subsection 3.3 considers a special case where a sign restriction on the objective atrix is observed. Notation. Throughout, we denote by ā the conjugate of a coplex nuber a, by C n the space of n-diensional coplex vectors. For a given vector z C n, z H denotes the conjugate transpose of z. The space of n n real syetric and coplex Heritian atrices are denoted by S n and H n, respectively. For a atrix Z H n, we write Re Z and I Z for the real and iaginary part of Z, respectively. Matrix Z being Heritian iplies that Re Z is syetric and I Z is skew-syetric.
3 We denote by S+ n S++ n and H+ n H++ n the cones of real syetric positive seidefinite positive definite and coplex Heritian positive seidefinite positive definite atrices, respectively. The notation Z eans that Z is positive seidefinite positive definite. For two coplex atrices Y and Z, their inner product Y Z = Re tr Y H Z = tr [ Re Y T Re Z + I Y T I Z ], where tr denotes the trace of a atrix and T denotes the transpose of a atrix. Coplex Bivariate Noral Distribution It is well known that the density function of an n-diensional real-valued ultivariate noral distribution is given as follows fx = π n/ det Ω exp x µt Ω x µ, where µ R n is the ean and Ω S n ++ is the covariance atrix. Let us consider a coplex-valued norally distributed rando variable in C, with the ean value z C and variance σ R +. For ore inforation on the coplex-valued noral distributions, we refer the reader to []. Siilar as in the real-valued case, its density function can be written as fz = πσ exp z z /σ, z C. Denote by N c z, σ the coplex-valued noral distribution with ean z and variance σ. Using Euler s forula, i.e., letting z z = ρe iθ, we have fρ, θ = ρ πσ exp ρ σ, with ρ, θ [, + [, π, where the variable transforation is { Re z z = ρ cos θ I z z = ρ sin θ. As a atter of terinology, ρ is usually called the odulus of z z, also denoted as z z ; θ is called the arguent of z z, denoted as Arg z z. The density of the joint coplex-valued noral distribution z = z, z,..., z n, with z k C, k =,..., n, has the following for fz = π n det Ω exp z µh Ω z µ, where z, µ C n, and Ω H n ++; µ is the ean vector, and Ω is the covariance atrix. 3
4 Let us denote the above coplex-valued noral distribution as N c µ, Ω. The bivariate case is of particular interest to us. Consider a coplex-valued, bivariate noral rando vector. Suppose that it has zero-ean. Furtherore, suppose that its covariance atrix is [ ] λ Ω = λ where λ C denotes the conjugate of λ C. In particular, let λ = γe iα, and so λ = γe iα. Since Ω, it follows that γ >. Moreover, Ω = γ [ γe iα γe iα ]. Then, by letting z = ρ e iθ and z = ρ e iθ, we ay rewrite the density function as [ ] H [ ] [ fρ, ρ, θ, θ = 4π γ exp ρ e iθ γe iα γ ρ e iθ γe iα ρ ρ = 4π γ exp ρ + ρ ρ ρ γ cos α + θ θ γ, where the doain of the variables is given as ρ, ρ, θ, θ [, + [, π. ρ e iθ ρ e iθ ] Now let us further introduce a variable transforation { ρ = ρ cos ξ ρ = ρ sin ξ with the doain ρ, ξ [, + [, π/]. The density function can be further written as fρ, ξ, θ, θ = ρ3 cos ξ sin ξ 4π γ exp ρ γρ cos ξ sin ξ cos α + θ θ γ ρ 3 sin ξ = 8π γ exp ρ ρ γ sin ξ cos α + θ θ γ, and the doain is ρ, ξ, θ, θ [, + [, π/] [, π. Let us note the following two siple facts. Lea. Suppose that a > is a given real nuber. Then, it holds that ρ 3 exp aρ dρ = a. 4
5 Lea. Suppose that < b < is a given real nuber. Then, with respect to the variable θ, it holds that sin θ b sin θ dθ = cos θ b b sin θ + b tanθ/ b b arctan + C. 3/ b Consider θ b < θe π and θb < θe θ, θ [θ b, θe ] [θb, θe ]. π. Below we shall copute the probability that Let us denote P : = Prob {θ b θ θ; e θ b θ θ} e θ e θ e π/ [ ρ 3 sin ξ = θ b θ b 8π γ exp ρ ρ ] γ sin ξ cos α + θ θ γ dρ dξdθ dθ θ e [ θ e π/ γ = 6π γ sin ξ dξ] dθ dθ θ b θ b γ sin ξ cos α + θ θ = θ γ e [ θ e ] π/ sin ξ 4π γ cos α + θ θ sin ξ dξ dθ dθ = γ 4π θ b θ b θ e θ e θ b θ b [ γ cos α + θ θ + + γ cos α + θ θ arccos γ cos α + θ θ γ cos α + θ θ 3/ ] dθ dθ, where in the third equality we used Lea. and in the last equality we used Lea.. To further copute the above integration, we note the following two ore facts: Lea.3 With respect to the variable θ, it holds that [ γ cos θ + γ cos θ arccos γ cos θ γ cos θ 3/ ] dθ = γ θ + γ sin θ arccos γ cos θ + C. γ cos θ Lea.4 With respect to the variable θ, it holds that [ ] γ sinβ θ arccos γ cosθ β dθ = γ cos θ β arccos γ cosθ β + C. Using Lea.3 we obtain [ P = 4π θ e θθ b e θ b + θ e θ b θ e θ b γ sinθ b α θ arccos γ cosθ b α θ γ cos θ b α θ γ sinθ e α θ arccos γ cosθ e α θ γ cos θ e α θ dθ 5 dθ,
6 and further using Lea.4, we have P = θe θb θe θb 4π + 8π + arccos γ cosθ b θ b + α [ arccos γ cosθ e θ e + α arccos γ cosθ e θ b + α arccos γ cosθ b θ e + α ]. Suarizing, we have proven the following result by a liiting arguent. [ Theore.5 For the coplex-value bivariate noral rando vector [ ] [ µ = and Ω = it holds that ] γe iα γe iα H+, z z ] N c µ, Ω with Prob {θ b Arg z θ e ; θ b Arg z θ e } = θe θb θe θb 4π + 8π + arccos γ cosθ b θ b + α [ arccos γ cosθ e θ e + α arccos γ cosθ e θ b + α arccos γ cosθ b θ e + α ]. 3 Quadratic Progras and Coplex SDP Forulations 3. Discrete Coplex Quadratic Optiization Suppose that Q is a Heritian atrix. Consider the following quadratic prograing proble with discrete decision variables, P ax z H Qz s.t. z k {, ω,..., ω }, k =,..., n, where and ω = e i π = cos π + i sin π. Denote the optial value of P to be vp. Consider the following coplex-valued apping F F z := ω ω 8π j= ω j arccos Re ω j z. For a Heritian atrix Z with Z kl for all k, l, define the coponentwise atrix function F Z := F Z kl n n H n. It is easy to verify that F z = F z. Therefore, if Z is Heritian, then so is F Z. 6
7 Lea 3. We have = ω ω 8π j= ω j arccos Re ω j. Moreover, for any z {, ω,..., ω } it follows that F z = z. Proof. We observe that Moreover, we have j= ω ω 8π = ω ω 8π = ω ω 8 4 j= j= j= j ω j = ω ω ω and ω j arccos cos j π ω j π j j ω j 4 j= j= jω j = Substituting the above equations into yields the intended result. jω j. ω. Suppose z = ω j for soe j {,..., n}. Then, ω ω 8π = ω ω 8π = ω ω 8π j= j= j= = ω j ω ω 8π = ω j = z. ω j arccos Re ω j z ω j arccos cos j j π ω j arccos cos j j π j j= j ω j arccos cos j π This copletes the proof for Lea 3.. Hence we can rewrite P as ax Q F zz H s.t. z k {, ω,..., ω }, k =,..., n. 7
8 Consider the following nonlinear coplex seidefinite prograing proble SP ax Q F Z s.t. Z kk =, k =,..., n, Z. Let vsp denote the optial value of SP. Theore 3. It holds that vp = vsp. Proof. Let ẑ is optial to P, then, by Lea 3., Ẑ = ẑẑ H is a feasible solution to SP and F Ẑ = Ẑ. Therefore, vsp Q FẐ = Q Ẑ = vp. On the other hand, for every feasible solution Z of SP, we randoly generate a coplex vector ξ such that ξ N c, Z, and assign, if Arg ξ k [, π ω,. if Arg ξ k [ π, π y k = σξ k = ω j, if Arg ξ k [ j j+ π, π. ω, if Arg ξ k [ π, π and finally let z k = ȳ k, k =,..., n. Suppose that Z kl = γe iα. Then by Theore.5, we have Prob {y k = y l ω j, y l = ω r } = Prob {y k = ω j+r, y l = ω r } = Prob {Arg ξ k [ j + r π, j + r + π, Arg ξ l [ r π, r + π} = + 8π arccos γ cos j π + α arccos γ cos j π + α arccos γ cos j + π + α for any j, r {,,..., }. Therefore, for any given k and l we have = Prob {y k ȳ l = ω j } r= Prob {y k = y l ω j, y l = ω r } = + 8π arccos γ cos j π + α arccos γ cos j π + α arccos γ cos j + π + α. 3 8
9 It follows that = Consequently, E[y k ȳ l ] j= ω j Prob {y k ȳ l = ω j } = 8π ω j arccos γ cos j π + α arccos γ cos j π + α j= arccos γ cos j + π + α = 8π ω j ω j ω j+ arccos γ cos j π + α j= = ω ω 8π = ω ω 8π j= j= E[z k z l ] = E[y k ȳ l ] = ω ω 8π = ω ω 8π = ω ω 8π = ω ω 8π ω j arccos γ cos j π + α ω j arccos Re ω j Z kl. 4 j= j= j= j= By the linearity of atheatical expectation, we get ω j arccos γ cos j π + α ω j arccos γ cos j π + α ω j arccos γ cos j π + α ω j arccos Re ω j Z kl. E[z H Qz] = Q F Z. Since the solution z so generated is feasible to P, we have vp E[z H Qz] = Q Z, for every feasible solution Z of SP. This cobining with vsp vp yields the desired result. 9
10 In particular, if = then one can verify that proble P reduces to and proble SP reduces to ax x T Qx s.t. x k {±}, k =,..., n, ax π Q arcsinx s.t. X kk =, k =,..., n, X, where arcsinx := [arcsinx kl ] n n. In that case, Theore 3. specializes to Theore.9 in [5] or Theore in []. If = 3, then P is ax z H Qz s.t. z k {, ω, ω }, k =,..., n, with ω = e i π 3. In fact, Goeans and Williason [6] odel the ax-3-cut proble as M3C ax k<l n w klz k z l H z k z l s.t. z k = {, ω, ω }, k =,..., n, and they consider the following coplex SDP relaxation ax k<l n w kl Re Z kl s.t. Z kk =, k =,..., n Re Z kl /, Re ωz kl /, Re ω Z kl /, k < l n Z. Let the optial solution of the SDP relaxation be Z. Then, Theore 3. asserts that the expected value of the randoized solution based on Z is k<l n w kl Re F 3 Z kl where F 3 z = 9 8π [ arccos Re z + ωarccos Re ω z + ω arccos Re ωz ]. Since arccosx is a convex function, it follows that Re F 3 Zkl = 9 [ 8π arccos Re Zkl arccos Re ω Zkl + arccos Re ωzkl ] [ 9 8π arccos Re Zkl arccos Re ωzkl + ω Zkl ] = 9 8π [arccos Re Z kl arccos Re Z kl ].
11 Further noticing that [ 9 arccos ] x + 4π in arccos x = x< x the approxiation ratio of Goeans and Williason [6] thus follows fro the fact that k<l n w kl Re F 3 Z kl { w kl 9 [arccos Re 8π Z kl arccos ]} Re Z kl k<l n.836 w kl Re Zkl k<l n.836 v M3C. 3. Continuous Coplex Quadratic Optiization By taking the liit, i.e., the quadratic optiization odel P becoes CP ax z H Qz s.t. z k =, k =,..., n, where Q H n +. In that case, the proble is equivalent to with SCP ax Q F Z F z := li F z = 4π where γ = z and α = Arg z. π s.t. Z kk =, k =,..., n, Z e iθ arccos γ cosθ α dθ The applications of Heritian quadratic optiization odels such as CP can be found, e.g. in [7], although in [7] the iniization version of the proble was considered. Proposition 3.3 Proble CP is strongly NP-hard in general.
12 Proof. The optiization proble in the for of ax s.t. z T Az z k C, z k, k =,..., n is called coplex prograing, and was shown in [9] to be NP-hard in general. We thank André Tits for drawing our attention to coplex prograing. Proble CP is related to coplex prograing, but they are not the sae: the objective in CP takes the Heritian for, and is assued to be positive seidefinite. The proof for Proposition 3.3 to be presented below is due to To Luo of Minnesota University, who sketched this proof to us in a private counication. As a first step we shall prove that the following proble in z H Qz s.t. z k =, k =,..., n, is NP-hard in general, where Q H+. n To this end, we consider a reduction fro the following strongly NP-coplete atrix partition proble; i.e., given a atrix G = [G,..., G N ] R M N, decide whether or not a subset of {,..., N} exists, say I, such that G k = N G k. k I k= Let the decision vector be z = z, z,, z N, z N+,, z N T C N+. Let n = N +, and e N I N I N A := Ge N G T R M+N n, N where e N R N is the vector of all ones. Let Q := A T A. Next we show that a atrix partition exists is equivalent to the fact that there is z C n with z k = for all k, such that z H Qz =. Clearly, z H Qz = is equivalent to Az = ; that is, = z + z k + z N+k, k =,..., N 5 = N N G k z + G k z k. 6 k= k= Let z k /z = e iθ k for k =,..., N. Using 5 we have cos θ k + cos θ N+k = 7 sin θ k + sin θ N+k = 8
13 where k =,..., N. Equations 7 and 8 iply that θ k { π/3, π/3}. This in particular eans that cos θ k = cos θ N+k = / for k =,..., N. Since Re N N G k + G k z k /z = N N G k + G k cos θ k = k= k= is always satisfied, 6 is true if and only if I N G k + k= k= k= k= N N G k z k /z = G k sin θ k =, which aounts to the existence of a atrix partition. Let λ ax be the axiu eigenvalue of Q. By observing that in z H Qz s.t. z k =, k =,..., n, k= is equivalent to ax z H λ ax I Qz s.t. z k =, k =,..., n, where λ ax I Q H+, n the desired result follows. For a given z C with z = γe iα and z = γ, we have F z = π e iθ arccos γ cosθ α dθ 4π π = 4π eiα e iθ arccos γ cos θ dθ = [ π 4π eiα e iθ arccos γ cos θ dθ = eiα π e iθ π arccosγ cos θ dθ π ] e iθ arccosγ cos θ dθ = π eiα e iθ arcsinγ cos θdθ = π eiα e γ iθ k! cos θ + 4 k k! γ cos θk+ dθ k + k= = π 4 γeiα + π k! 4k+ k! 4 k + γk+ e iα k= = π 4 z + π k! 4k+ k! 4 k + z k z, 9 k= 3
14 where the second last step follows fro the fact that π sin θcos θ k+ dθ = and π cos θ k+ dθ = k + k π, k =,,... k + k Clearly, if Z H n + then Z T H n +. Furtherore, observe that the Hadaard product of any two positive seidefinite Heritian atrices reains Heritian positive seidefinite. Denote A B to k {}}{ be the Hadaard product of A and B, and denote A k to be A A A. It thus follows fro 9 that F Z = π 4 Z + π k= k! 4k+ k! 4 k + ZT Z k Z π 4 Z. Therefore, if Q, then we have Q F Z π 4 Q Z. Consider the following coplex SDP relaxation for CP CSDP ax Q Z s.t. Z kk =, k =,..., n, Z. Let the optial value of CP be v CP, and the optial value of CSDP be v CSDP. Let the expected value of the randoized solutions based on the optial solution of CSDP be vhc. Then vhc π 4 v CSDP π 4 v CP.7854 v CP. It is interesting to copare this ratio with that of its real counterpart: RP ax x T Qx s.t. x k =, k =,..., n. Nesterov [8] showed that the randoization solution based on the SDP relaxation RSDP ax Q X s.t. X kk =, k =,..., n, X, has the following approxiation ratio vhr π v RSDP π v RP.6366 v RP. 4
15 Therefore, the coplex SDP relaxation for the coplex quadratic optiization proble is ore effective than the real SDP relaxation for its real counter-part, in the sense that the forer has a slightly better approxiation ratio. Reark that siilar as the analysis in Nesterov [8], Ye [], and Zhang [] for the real case, we can extend all the approxiation results to the following ore general setting ax z H Qz s.t. z, z,, z n T F, where F is a closed convex set in R n. The corresponding coplex and convex SDP relaxation is ax s.t. Q Z diag Z F Z. It is also interesting to reark that if we regard CP as an equivalent real quadratic proble ax u T, v T Re Q I Q u I Q Re Q v s.t. u k + v k =, k =,..., n, then the approxiation ratio obtained that way would be /π, instead of π/4. This shows that the coplex SDP relaxation does have an advantage in this particular case. 3.3 Structured Continuous Coplex Quadratic Optiization In this subsection, we study a special case of CP with a sign structure on the object atrix, which is parallel to the original real ax-cut odel studied in [5]: CPS ax z H Qz s.t. z k =, k =,..., n, where we assue that Q = [q jl ] n n S n + and q jl for all j < l n. Using 9 we know that the expected value of the randoized solution based on the coplex SDP relaxation is vhc = j<l = j<l q jl Re F Z jl + n q jl π 4 + π k= j= q jj k! n 4k+ k! 4 k + Z jl k Re Zjl + q jj j= 5
16 where Z is the optial solution of the coplex SDP relaxation. Define the following real function gy := π 4 + π k= k! 4k+ k! 4 k + yk on y [, ]. We have gy for all y [, ]. Suppose that x is real, and x y. Then, gyx in = in gy + gy = + gyy. x y x x y x + y One coputes that Therefore, + gyy in y + y.9349 =: β. gyx β βx, for all y [, ] and x y, or equivalently, k= gyx β + βx for all y [, ] and x y. Using, we have π 4 + π k! 4k+ k! 4 k + Z jl k Re Zjl β + βre Z jl. Now we apply in a coponentwise fashion to, and obtain, thanks to the sign restriction, the following inequalities vhc = j<l j<l q jl π 4 + π k= q jl β + βre Z jl + n = βe T Qe + βq Z βv CSDP k! n 4k+ k! 4 k + Z jl k Re Zjl + j= q jj βv CP S. 3 j= q jj This yields an approxiation ratio of.9349 for CPS. Acknowledgeent: We would like to thank To Luo, Anthony So, Yinyu Ye, and Jiawei Zhang for stiulating discussions on the subject. 6
17 References [] I.G. Abrahason. Orthant probability for the quadrivariate noral distribution. The Annals of Matheatical Statistics 35: , 964. [] H.H. Andersen. Linear and graphical odels for the ultivariate coplex noral distribution. Springer-Verlag, 995. [3] D. Bertsias and Y. Ye. Seidefinite relaxations, ultivarite noral distribution, and order statistics. In D.Z. Du and P.M. Pardalos, editors, Handbook of Cobinatorial Optiization, volue 3, pages 9. Kluwer Acadeic Publishers, 998. [4] A. Frieze and M. Jerru. Iproved approxiation algoriths for MAX-k-CUT and Max BI- SECTION. Algorithica 8: 67 8, 997. [5] M.X. Goeans and D.P. Williason. Iproved approxiation algoriths for axiu cut and satisfiability probles using seidefinite prograing. Journal of the ACM 4: 5 45, 995. [6] M.X. Goeans and D.P. Williason. Approxiation algoriths for MAX-3-CUT and other probles via coplex seidefinite prograing. Journal of Coputer and Syste Sciences 68: 44 47, 4. [7] Z.Q. Luo, X.D. Luo, M. Kisialiou. An efficient quasi-axiu likelihood decoder for PSK signals. Proceedings IEEE International Conference on Acoustics, Speech, and Signal Processing ICASSP 3, Pages: VI vol. 6, 3. [8] Yu. Nesterov. Seidefinite relaxation and nonconvex quadratic optiization. Optiization Methods and Softwares 9: 4 6, 998. [9] O. Toker and H. Özbay. On the coplexity of purely coplex µ coputation and related probles in ultidiensional systes. IEEE Transactions on Autoatic Control 43: 49 44, 998. [] Y. Ye. Approxiating quadratic prograing with bound and quadratic constraints. Matheatical Prograing 84: 9 6, 999. [] S. Zhang. Quadratic axiization and seidefinite relaxation. Matheatical Prograing 87: ,. 7
Complex Quadratic Optimization and Semidefinite Programming
Complex Quadratic Optimization and Semidefinite Programming Shuzhong Zhang Yongwei Huang August 4; revised April 5 Abstract In this paper we study the approximation algorithms for a class of discrete quadratic
More informationConvex Programming for Scheduling Unrelated Parallel Machines
Convex Prograing for Scheduling Unrelated Parallel Machines Yossi Azar Air Epstein Abstract We consider the classical proble of scheduling parallel unrelated achines. Each job is to be processed by exactly
More informationChapter 6 1-D Continuous Groups
Chapter 6 1-D Continuous Groups Continuous groups consist of group eleents labelled by one or ore continuous variables, say a 1, a 2,, a r, where each variable has a well- defined range. This chapter explores:
More informationBipartite subgraphs and the smallest eigenvalue
Bipartite subgraphs and the sallest eigenvalue Noga Alon Benny Sudaov Abstract Two results dealing with the relation between the sallest eigenvalue of a graph and its bipartite subgraphs are obtained.
More informationRandomized Recovery for Boolean Compressed Sensing
Randoized Recovery for Boolean Copressed Sensing Mitra Fatei and Martin Vetterli Laboratory of Audiovisual Counication École Polytechnique Fédéral de Lausanne (EPFL) Eail: {itra.fatei, artin.vetterli}@epfl.ch
More informationFeature Extraction Techniques
Feature Extraction Techniques Unsupervised Learning II Feature Extraction Unsupervised ethods can also be used to find features which can be useful for categorization. There are unsupervised ethods that
More informationNon-Parametric Non-Line-of-Sight Identification 1
Non-Paraetric Non-Line-of-Sight Identification Sinan Gezici, Hisashi Kobayashi and H. Vincent Poor Departent of Electrical Engineering School of Engineering and Applied Science Princeton University, Princeton,
More informationShannon Sampling II. Connections to Learning Theory
Shannon Sapling II Connections to Learning heory Steve Sale oyota echnological Institute at Chicago 147 East 60th Street, Chicago, IL 60637, USA E-ail: sale@athberkeleyedu Ding-Xuan Zhou Departent of Matheatics,
More informationLecture 9 November 23, 2015
CSC244: Discrepancy Theory in Coputer Science Fall 25 Aleksandar Nikolov Lecture 9 Noveber 23, 25 Scribe: Nick Spooner Properties of γ 2 Recall that γ 2 (A) is defined for A R n as follows: γ 2 (A) = in{r(u)
More informationGary J. Balas Aerospace Engineering and Mechanics, University of Minnesota, Minneapolis, MN USA
μ-synthesis Gary J. Balas Aerospace Engineering and Mechanics, University of Minnesota, Minneapolis, MN 55455 USA Keywords: Robust control, ultivariable control, linear fractional transforation (LFT),
More informationKernel Methods and Support Vector Machines
Intelligent Systes: Reasoning and Recognition Jaes L. Crowley ENSIAG 2 / osig 1 Second Seester 2012/2013 Lesson 20 2 ay 2013 Kernel ethods and Support Vector achines Contents Kernel Functions...2 Quadratic
More informationUniform Approximation and Bernstein Polynomials with Coefficients in the Unit Interval
Unifor Approxiation and Bernstein Polynoials with Coefficients in the Unit Interval Weiang Qian and Marc D. Riedel Electrical and Coputer Engineering, University of Minnesota 200 Union St. S.E. Minneapolis,
More informationSupport Vector Machine Classification of Uncertain and Imbalanced data using Robust Optimization
Recent Researches in Coputer Science Support Vector Machine Classification of Uncertain and Ibalanced data using Robust Optiization RAGHAV PAT, THEODORE B. TRAFALIS, KASH BARKER School of Industrial Engineering
More informationThe Simplex Method is Strongly Polynomial for the Markov Decision Problem with a Fixed Discount Rate
The Siplex Method is Strongly Polynoial for the Markov Decision Proble with a Fixed Discount Rate Yinyu Ye April 20, 2010 Abstract In this note we prove that the classic siplex ethod with the ost-negativereduced-cost
More informationarxiv: v1 [cs.ds] 3 Feb 2014
arxiv:40.043v [cs.ds] 3 Feb 04 A Bound on the Expected Optiality of Rando Feasible Solutions to Cobinatorial Optiization Probles Evan A. Sultani The Johns Hopins University APL evan@sultani.co http://www.sultani.co/
More informationDetection and Estimation Theory
ESE 54 Detection and Estiation Theory Joseph A. O Sullivan Sauel C. Sachs Professor Electronic Systes and Signals Research Laboratory Electrical and Systes Engineering Washington University 11 Urbauer
More informationRecovering Data from Underdetermined Quadratic Measurements (CS 229a Project: Final Writeup)
Recovering Data fro Underdeterined Quadratic Measureents (CS 229a Project: Final Writeup) Mahdi Soltanolkotabi Deceber 16, 2011 1 Introduction Data that arises fro engineering applications often contains
More informationCS Lecture 13. More Maximum Likelihood
CS 6347 Lecture 13 More Maxiu Likelihood Recap Last tie: Introduction to axiu likelihood estiation MLE for Bayesian networks Optial CPTs correspond to epirical counts Today: MLE for CRFs 2 Maxiu Likelihood
More informationBlock designs and statistics
Bloc designs and statistics Notes for Math 447 May 3, 2011 The ain paraeters of a bloc design are nuber of varieties v, bloc size, nuber of blocs b. A design is built on a set of v eleents. Each eleent
More information1 Bounding the Margin
COS 511: Theoretical Machine Learning Lecturer: Rob Schapire Lecture #12 Scribe: Jian Min Si March 14, 2013 1 Bounding the Margin We are continuing the proof of a bound on the generalization error of AdaBoost
More informationA note on the multiplication of sparse matrices
Cent. Eur. J. Cop. Sci. 41) 2014 1-11 DOI: 10.2478/s13537-014-0201-x Central European Journal of Coputer Science A note on the ultiplication of sparse atrices Research Article Keivan Borna 12, Sohrab Aboozarkhani
More informationA Low-Complexity Congestion Control and Scheduling Algorithm for Multihop Wireless Networks with Order-Optimal Per-Flow Delay
A Low-Coplexity Congestion Control and Scheduling Algorith for Multihop Wireless Networks with Order-Optial Per-Flow Delay Po-Kai Huang, Xiaojun Lin, and Chih-Chun Wang School of Electrical and Coputer
More information13.2 Fully Polynomial Randomized Approximation Scheme for Permanent of Random 0-1 Matrices
CS71 Randoness & Coputation Spring 018 Instructor: Alistair Sinclair Lecture 13: February 7 Disclaier: These notes have not been subjected to the usual scrutiny accorded to foral publications. They ay
More informatione-companion ONLY AVAILABLE IN ELECTRONIC FORM
OPERATIONS RESEARCH doi 10.1287/opre.1070.0427ec pp. ec1 ec5 e-copanion ONLY AVAILABLE IN ELECTRONIC FORM infors 07 INFORMS Electronic Copanion A Learning Approach for Interactive Marketing to a Custoer
More informationNew Classes of Positive Semi-Definite Hankel Tensors
Miniax Theory and its Applications Volue 017, No., 1 xxx New Classes of Positive Sei-Definite Hankel Tensors Qun Wang Dept. of Applied Matheatics, The Hong Kong Polytechnic University, Hung Ho, Kowloon,
More informationIntelligent Systems: Reasoning and Recognition. Perceptrons and Support Vector Machines
Intelligent Systes: Reasoning and Recognition Jaes L. Crowley osig 1 Winter Seester 2018 Lesson 6 27 February 2018 Outline Perceptrons and Support Vector achines Notation...2 Linear odels...3 Lines, Planes
More informationUsing EM To Estimate A Probablity Density With A Mixture Of Gaussians
Using EM To Estiate A Probablity Density With A Mixture Of Gaussians Aaron A. D Souza adsouza@usc.edu Introduction The proble we are trying to address in this note is siple. Given a set of data points
More informationLecture 21. Interior Point Methods Setup and Algorithm
Lecture 21 Interior Point Methods In 1984, Kararkar introduced a new weakly polynoial tie algorith for solving LPs [Kar84a], [Kar84b]. His algorith was theoretically faster than the ellipsoid ethod and
More informationHybrid System Identification: An SDP Approach
49th IEEE Conference on Decision and Control Deceber 15-17, 2010 Hilton Atlanta Hotel, Atlanta, GA, USA Hybrid Syste Identification: An SDP Approach C Feng, C M Lagoa, N Ozay and M Sznaier Abstract The
More informationarxiv: v1 [math.na] 10 Oct 2016
GREEDY GAUSS-NEWTON ALGORITHM FOR FINDING SPARSE SOLUTIONS TO NONLINEAR UNDERDETERMINED SYSTEMS OF EQUATIONS MÅRTEN GULLIKSSON AND ANNA OLEYNIK arxiv:6.395v [ath.na] Oct 26 Abstract. We consider the proble
More informationLower Bounds for Quantized Matrix Completion
Lower Bounds for Quantized Matrix Copletion Mary Wootters and Yaniv Plan Departent of Matheatics University of Michigan Ann Arbor, MI Eail: wootters, yplan}@uich.edu Mark A. Davenport School of Elec. &
More informationA note on the realignment criterion
A note on the realignent criterion Chi-Kwong Li 1, Yiu-Tung Poon and Nung-Sing Sze 3 1 Departent of Matheatics, College of Willia & Mary, Williasburg, VA 3185, USA Departent of Matheatics, Iowa State University,
More informationThis model assumes that the probability of a gap has size i is proportional to 1/i. i.e., i log m e. j=1. E[gap size] = i P r(i) = N f t.
CS 493: Algoriths for Massive Data Sets Feb 2, 2002 Local Models, Bloo Filter Scribe: Qin Lv Local Models In global odels, every inverted file entry is copressed with the sae odel. This work wells when
More informationOn the Communication Complexity of Lipschitzian Optimization for the Coordinated Model of Computation
journal of coplexity 6, 459473 (2000) doi:0.006jco.2000.0544, available online at http:www.idealibrary.co on On the Counication Coplexity of Lipschitzian Optiization for the Coordinated Model of Coputation
More informationEstimating Parameters for a Gaussian pdf
Pattern Recognition and achine Learning Jaes L. Crowley ENSIAG 3 IS First Seester 00/0 Lesson 5 7 Noveber 00 Contents Estiating Paraeters for a Gaussian pdf Notation... The Pattern Recognition Proble...3
More informationCourse Notes for EE227C (Spring 2018): Convex Optimization and Approximation
Course Notes for EE7C (Spring 018: Convex Optiization and Approxiation Instructor: Moritz Hardt Eail: hardt+ee7c@berkeley.edu Graduate Instructor: Max Sichowitz Eail: sichow+ee7c@berkeley.edu October 15,
More informationDERIVING PROPER UNIFORM PRIORS FOR REGRESSION COEFFICIENTS
DERIVING PROPER UNIFORM PRIORS FOR REGRESSION COEFFICIENTS N. van Erp and P. van Gelder Structural Hydraulic and Probabilistic Design, TU Delft Delft, The Netherlands Abstract. In probles of odel coparison
More informationModel Fitting. CURM Background Material, Fall 2014 Dr. Doreen De Leon
Model Fitting CURM Background Material, Fall 014 Dr. Doreen De Leon 1 Introduction Given a set of data points, we often want to fit a selected odel or type to the data (e.g., we suspect an exponential
More informationarxiv: v1 [cs.ds] 29 Jan 2012
A parallel approxiation algorith for ixed packing covering seidefinite progras arxiv:1201.6090v1 [cs.ds] 29 Jan 2012 Rahul Jain National U. Singapore January 28, 2012 Abstract Penghui Yao National U. Singapore
More informationDepartment of Electronic and Optical Engineering, Ordnance Engineering College, Shijiazhuang, , China
6th International Conference on Machinery, Materials, Environent, Biotechnology and Coputer (MMEBC 06) Solving Multi-Sensor Multi-Target Assignent Proble Based on Copositive Cobat Efficiency and QPSO Algorith
More informationAPPROXIMATION BOUNDS FOR SPARSE PRINCIPAL COMPONENT ANALYSIS
APPROXIMATION BOUNDS FOR SPARSE PRINCIPAL COMPONENT ANALYSIS ALEXANDRE D ASPREMONT, FRANCIS BACH, AND LAURENT EL GHAOUI ABSTRACT. We produce approxiation bounds on a seidefinite prograing relaxation for
More informationAlgorithms for parallel processor scheduling with distinct due windows and unit-time jobs
BULLETIN OF THE POLISH ACADEMY OF SCIENCES TECHNICAL SCIENCES Vol. 57, No. 3, 2009 Algoriths for parallel processor scheduling with distinct due windows and unit-tie obs A. JANIAK 1, W.A. JANIAK 2, and
More informationApproximation in Stochastic Scheduling: The Power of LP-Based Priority Policies
Approxiation in Stochastic Scheduling: The Power of -Based Priority Policies Rolf Möhring, Andreas Schulz, Marc Uetz Setting (A P p stoch, r E( w and (B P p stoch E( w We will assue that the processing
More informationTEST OF HOMOGENEITY OF PARALLEL SAMPLES FROM LOGNORMAL POPULATIONS WITH UNEQUAL VARIANCES
TEST OF HOMOGENEITY OF PARALLEL SAMPLES FROM LOGNORMAL POPULATIONS WITH UNEQUAL VARIANCES S. E. Ahed, R. J. Tokins and A. I. Volodin Departent of Matheatics and Statistics University of Regina Regina,
More informationADVANCES ON THE BESSIS- MOUSSA-VILLANI TRACE CONJECTURE
ADVANCES ON THE BESSIS- MOUSSA-VILLANI TRACE CONJECTURE CHRISTOPHER J. HILLAR Abstract. A long-standing conjecture asserts that the polynoial p(t = Tr(A + tb ] has nonnegative coefficients whenever is
More informationOptimal Jamming Over Additive Noise: Vector Source-Channel Case
Fifty-first Annual Allerton Conference Allerton House, UIUC, Illinois, USA October 2-3, 2013 Optial Jaing Over Additive Noise: Vector Source-Channel Case Erah Akyol and Kenneth Rose Abstract This paper
More informationThe Methods of Solution for Constrained Nonlinear Programming
Research Inventy: International Journal Of Engineering And Science Vol.4, Issue 3(March 2014), PP 01-06 Issn (e): 2278-4721, Issn (p):2319-6483, www.researchinventy.co The Methods of Solution for Constrained
More informationExperimental Design For Model Discrimination And Precise Parameter Estimation In WDS Analysis
City University of New York (CUNY) CUNY Acadeic Works International Conference on Hydroinforatics 8-1-2014 Experiental Design For Model Discriination And Precise Paraeter Estiation In WDS Analysis Giovanna
More informationTHE AVERAGE NORM OF POLYNOMIALS OF FIXED HEIGHT
THE AVERAGE NORM OF POLYNOMIALS OF FIXED HEIGHT PETER BORWEIN AND KWOK-KWONG STEPHEN CHOI Abstract. Let n be any integer and ( n ) X F n : a i z i : a i, ± i be the set of all polynoials of height and
More informationE0 370 Statistical Learning Theory Lecture 6 (Aug 30, 2011) Margin Analysis
E0 370 tatistical Learning Theory Lecture 6 (Aug 30, 20) Margin Analysis Lecturer: hivani Agarwal cribe: Narasihan R Introduction In the last few lectures we have seen how to obtain high confidence bounds
More informationWeighted- 1 minimization with multiple weighting sets
Weighted- 1 iniization with ultiple weighting sets Hassan Mansour a,b and Özgür Yılaza a Matheatics Departent, University of British Colubia, Vancouver - BC, Canada; b Coputer Science Departent, University
More informationAsynchronous Gossip Algorithms for Stochastic Optimization
Asynchronous Gossip Algoriths for Stochastic Optiization S. Sundhar Ra ECE Dept. University of Illinois Urbana, IL 680 ssrini@illinois.edu A. Nedić IESE Dept. University of Illinois Urbana, IL 680 angelia@illinois.edu
More informationCurious Bounds for Floor Function Sums
1 47 6 11 Journal of Integer Sequences, Vol. 1 (018), Article 18.1.8 Curious Bounds for Floor Function Sus Thotsaporn Thanatipanonda and Elaine Wong 1 Science Division Mahidol University International
More informationPhysics 215 Winter The Density Matrix
Physics 215 Winter 2018 The Density Matrix The quantu space of states is a Hilbert space H. Any state vector ψ H is a pure state. Since any linear cobination of eleents of H are also an eleent of H, it
More informationList Scheduling and LPT Oliver Braun (09/05/2017)
List Scheduling and LPT Oliver Braun (09/05/207) We investigate the classical scheduling proble P ax where a set of n independent jobs has to be processed on 2 parallel and identical processors (achines)
More informationMinimum Rates Scheduling for MIMO OFDM Broadcast Channels
IEEE Ninth International Syposiu on Spread Spectru Techniques and Applications Miniu Rates Scheduling for MIMO OFDM Broadcast Channels Gerhard Wunder and Thoas Michel Fraunhofer Geran-Sino Mobile Counications
More informationIntroduction to Discrete Optimization
Prof. Friedrich Eisenbrand Martin Nieeier Due Date: March 9 9 Discussions: March 9 Introduction to Discrete Optiization Spring 9 s Exercise Consider a school district with I neighborhoods J schools and
More informationSoft Computing Techniques Help Assign Weights to Different Factors in Vulnerability Analysis
Soft Coputing Techniques Help Assign Weights to Different Factors in Vulnerability Analysis Beverly Rivera 1,2, Irbis Gallegos 1, and Vladik Kreinovich 2 1 Regional Cyber and Energy Security Center RCES
More informationOn Constant Power Water-filling
On Constant Power Water-filling Wei Yu and John M. Cioffi Electrical Engineering Departent Stanford University, Stanford, CA94305, U.S.A. eails: {weiyu,cioffi}@stanford.edu Abstract This paper derives
More informationOn Rough Interval Three Level Large Scale Quadratic Integer Programming Problem
J. Stat. Appl. Pro. 6, No. 2, 305-318 2017) 305 Journal of Statistics Applications & Probability An International Journal http://dx.doi.org/10.18576/jsap/060206 On Rough Interval Three evel arge Scale
More informationAn Improved Particle Filter with Applications in Ballistic Target Tracking
Sensors & ransducers Vol. 72 Issue 6 June 204 pp. 96-20 Sensors & ransducers 204 by IFSA Publishing S. L. http://www.sensorsportal.co An Iproved Particle Filter with Applications in Ballistic arget racing
More informationInspection; structural health monitoring; reliability; Bayesian analysis; updating; decision analysis; value of information
Cite as: Straub D. (2014). Value of inforation analysis with structural reliability ethods. Structural Safety, 49: 75-86. Value of Inforation Analysis with Structural Reliability Methods Daniel Straub
More informationResearch Article Robust ε-support Vector Regression
Matheatical Probles in Engineering, Article ID 373571, 5 pages http://dx.doi.org/10.1155/2014/373571 Research Article Robust ε-support Vector Regression Yuan Lv and Zhong Gan School of Mechanical Engineering,
More informationRESTARTED FULL ORTHOGONALIZATION METHOD FOR SHIFTED LINEAR SYSTEMS
BIT Nuerical Matheatics 43: 459 466, 2003. 2003 Kluwer Acadeic Publishers. Printed in The Netherlands 459 RESTARTED FULL ORTHOGONALIZATION METHOD FOR SHIFTED LINEAR SYSTEMS V. SIMONCINI Dipartiento di
More informationA new type of lower bound for the largest eigenvalue of a symmetric matrix
Linear Algebra and its Applications 47 7 9 9 www.elsevier.co/locate/laa A new type of lower bound for the largest eigenvalue of a syetric atrix Piet Van Mieghe Delft University of Technology, P.O. Box
More informationInteractive Markov Models of Evolutionary Algorithms
Cleveland State University EngagedScholarship@CSU Electrical Engineering & Coputer Science Faculty Publications Electrical Engineering & Coputer Science Departent 2015 Interactive Markov Models of Evolutionary
More informationDistributed Subgradient Methods for Multi-agent Optimization
1 Distributed Subgradient Methods for Multi-agent Optiization Angelia Nedić and Asuan Ozdaglar October 29, 2007 Abstract We study a distributed coputation odel for optiizing a su of convex objective functions
More informationSharp Time Data Tradeoffs for Linear Inverse Problems
Sharp Tie Data Tradeoffs for Linear Inverse Probles Saet Oyak Benjain Recht Mahdi Soltanolkotabi January 016 Abstract In this paper we characterize sharp tie-data tradeoffs for optiization probles used
More informationExtension of CSRSM for the Parametric Study of the Face Stability of Pressurized Tunnels
Extension of CSRSM for the Paraetric Study of the Face Stability of Pressurized Tunnels Guilhe Mollon 1, Daniel Dias 2, and Abdul-Haid Soubra 3, M.ASCE 1 LGCIE, INSA Lyon, Université de Lyon, Doaine scientifique
More informationThe Distribution of the Covariance Matrix for a Subset of Elliptical Distributions with Extension to Two Kurtosis Parameters
journal of ultivariate analysis 58, 96106 (1996) article no. 0041 The Distribution of the Covariance Matrix for a Subset of Elliptical Distributions with Extension to Two Kurtosis Paraeters H. S. Steyn
More informationŞtefan ŞTEFĂNESCU * is the minimum global value for the function h (x)
7Applying Nelder Mead s Optiization Algorith APPLYING NELDER MEAD S OPTIMIZATION ALGORITHM FOR MULTIPLE GLOBAL MINIMA Abstract Ştefan ŞTEFĂNESCU * The iterative deterinistic optiization ethod could not
More informationSparse beamforming in peer-to-peer relay networks Yunshan Hou a, Zhijuan Qi b, Jianhua Chenc
3rd International Conference on Machinery, Materials and Inforation echnology Applications (ICMMIA 015) Sparse beaforing in peer-to-peer relay networs Yunshan ou a, Zhijuan Qi b, Jianhua Chenc College
More informationHomework 3 Solutions CSE 101 Summer 2017
Hoework 3 Solutions CSE 0 Suer 207. Scheduling algoriths The following n = 2 jobs with given processing ties have to be scheduled on = 3 parallel and identical processors with the objective of iniizing
More informationSupport Vector Machines. Goals for the lecture
Support Vector Machines Mark Craven and David Page Coputer Sciences 760 Spring 2018 www.biostat.wisc.edu/~craven/cs760/ Soe of the slides in these lectures have been adapted/borrowed fro aterials developed
More informationPrediction by random-walk perturbation
Prediction by rando-walk perturbation Luc Devroye School of Coputer Science McGill University Gábor Lugosi ICREA and Departent of Econoics Universitat Popeu Fabra lucdevroye@gail.co gabor.lugosi@gail.co
More informationRANDOM GRADIENT EXTRAPOLATION FOR DISTRIBUTED AND STOCHASTIC OPTIMIZATION
RANDOM GRADIENT EXTRAPOLATION FOR DISTRIBUTED AND STOCHASTIC OPTIMIZATION GUANGHUI LAN AND YI ZHOU Abstract. In this paper, we consider a class of finite-su convex optiization probles defined over a distributed
More informationSupport recovery in compressed sensing: An estimation theoretic approach
Support recovery in copressed sensing: An estiation theoretic approach Ain Karbasi, Ali Horati, Soheil Mohajer, Martin Vetterli School of Coputer and Counication Sciences École Polytechnique Fédérale de
More informationAn improved self-adaptive harmony search algorithm for joint replenishment problems
An iproved self-adaptive harony search algorith for joint replenishent probles Lin Wang School of Manageent, Huazhong University of Science & Technology zhoulearner@gail.co Xiaojian Zhou School of Manageent,
More informationThe Simplex and Policy-Iteration Methods are Strongly Polynomial for the Markov Decision Problem with a Fixed Discount Rate
The Siplex and Policy-Iteration Methods are Strongly Polynoial for the Markov Decision Proble with a Fixed Discount Rate Yinyu Ye April 20, 2010; revised Noveber 30, 2010 Abstract We prove that the classic
More informationThe Hilbert Schmidt version of the commutator theorem for zero trace matrices
The Hilbert Schidt version of the coutator theore for zero trace atrices Oer Angel Gideon Schechtan March 205 Abstract Let A be a coplex atrix with zero trace. Then there are atrices B and C such that
More informationA Simplified Analytical Approach for Efficiency Evaluation of the Weaving Machines with Automatic Filling Repair
Proceedings of the 6th SEAS International Conference on Siulation, Modelling and Optiization, Lisbon, Portugal, Septeber -4, 006 0 A Siplified Analytical Approach for Efficiency Evaluation of the eaving
More informationON THE TWO-LEVEL PRECONDITIONING IN LEAST SQUARES METHOD
PROCEEDINGS OF THE YEREVAN STATE UNIVERSITY Physical and Matheatical Sciences 04,, p. 7 5 ON THE TWO-LEVEL PRECONDITIONING IN LEAST SQUARES METHOD M a t h e a t i c s Yu. A. HAKOPIAN, R. Z. HOVHANNISYAN
More informationOn Poset Merging. 1 Introduction. Peter Chen Guoli Ding Steve Seiden. Keywords: Merging, Partial Order, Lower Bounds. AMS Classification: 68W40
On Poset Merging Peter Chen Guoli Ding Steve Seiden Abstract We consider the follow poset erging proble: Let X and Y be two subsets of a partially ordered set S. Given coplete inforation about the ordering
More information. The univariate situation. It is well-known for a long tie that denoinators of Pade approxiants can be considered as orthogonal polynoials with respe
PROPERTIES OF MULTIVARIATE HOMOGENEOUS ORTHOGONAL POLYNOMIALS Brahi Benouahane y Annie Cuyt? Keywords Abstract It is well-known that the denoinators of Pade approxiants can be considered as orthogonal
More informationMulti-Dimensional Hegselmann-Krause Dynamics
Multi-Diensional Hegselann-Krause Dynaics A. Nedić Industrial and Enterprise Systes Engineering Dept. University of Illinois Urbana, IL 680 angelia@illinois.edu B. Touri Coordinated Science Laboratory
More informationOn Conditions for Linearity of Optimal Estimation
On Conditions for Linearity of Optial Estiation Erah Akyol, Kuar Viswanatha and Kenneth Rose {eakyol, kuar, rose}@ece.ucsb.edu Departent of Electrical and Coputer Engineering University of California at
More informationFairness via priority scheduling
Fairness via priority scheduling Veeraruna Kavitha, N Heachandra and Debayan Das IEOR, IIT Bobay, Mubai, 400076, India vavitha,nh,debayan}@iitbacin Abstract In the context of ulti-agent resource allocation
More informationOptimal quantum detectors for unambiguous detection of mixed states
PHYSICAL REVIEW A 69, 06318 (004) Optial quantu detectors for unabiguous detection of ixed states Yonina C. Eldar* Departent of Electrical Engineering, Technion Israel Institute of Technology, Haifa 3000,
More informationSupplementary Material for Fast and Provable Algorithms for Spectrally Sparse Signal Reconstruction via Low-Rank Hankel Matrix Completion
Suppleentary Material for Fast and Provable Algoriths for Spectrally Sparse Signal Reconstruction via Low-Ran Hanel Matrix Copletion Jian-Feng Cai Tianing Wang Ke Wei March 1, 017 Abstract We establish
More informationA remark on a success rate model for DPA and CPA
A reark on a success rate odel for DPA and CPA A. Wieers, BSI Version 0.5 andreas.wieers@bsi.bund.de Septeber 5, 2018 Abstract The success rate is the ost coon evaluation etric for easuring the perforance
More informationNonlinear Analysis. On the co-derivative of normal cone mappings to inequality systems
Nonlinear Analysis 71 2009) 1213 1226 Contents lists available at ScienceDirect Nonlinear Analysis journal hoepage: www.elsevier.co/locate/na On the co-derivative of noral cone appings to inequality systes
More informationMixed Robust/Average Submodular Partitioning
Mixed Robust/Average Subodular Partitioning Kai Wei 1 Rishabh Iyer 1 Shengjie Wang 2 Wenruo Bai 1 Jeff Biles 1 1 Departent of Electrical Engineering, University of Washington 2 Departent of Coputer Science,
More informationFast Montgomery-like Square Root Computation over GF(2 m ) for All Trinomials
Fast Montgoery-like Square Root Coputation over GF( ) for All Trinoials Yin Li a, Yu Zhang a, a Departent of Coputer Science and Technology, Xinyang Noral University, Henan, P.R.China Abstract This letter
More informationarxiv: v1 [cs.ds] 17 Mar 2016
Tight Bounds for Single-Pass Streaing Coplexity of the Set Cover Proble Sepehr Assadi Sanjeev Khanna Yang Li Abstract arxiv:1603.05715v1 [cs.ds] 17 Mar 2016 We resolve the space coplexity of single-pass
More informationDetermining OWA Operator Weights by Mean Absolute Deviation Minimization
Deterining OWA Operator Weights by Mean Absolute Deviation Miniization Micha l Majdan 1,2 and W lodziierz Ogryczak 1 1 Institute of Control and Coputation Engineering, Warsaw University of Technology,
More informationA Simple Regression Problem
A Siple Regression Proble R. M. Castro March 23, 2 In this brief note a siple regression proble will be introduced, illustrating clearly the bias-variance tradeoff. Let Y i f(x i ) + W i, i,..., n, where
More informationMathematical Model and Algorithm for the Task Allocation Problem of Robots in the Smart Warehouse
Aerican Journal of Operations Research, 205, 5, 493-502 Published Online Noveber 205 in SciRes. http://www.scirp.org/journal/ajor http://dx.doi.org/0.4236/ajor.205.56038 Matheatical Model and Algorith
More information3.3 Variational Characterization of Singular Values
3.3. Variational Characterization of Singular Values 61 3.3 Variational Characterization of Singular Values Since the singular values are square roots of the eigenvalues of the Heritian atrices A A and
More informationTight Information-Theoretic Lower Bounds for Welfare Maximization in Combinatorial Auctions
Tight Inforation-Theoretic Lower Bounds for Welfare Maxiization in Cobinatorial Auctions Vahab Mirrokni Jan Vondrák Theory Group, Microsoft Dept of Matheatics Research Princeton University Redond, WA 9805
More informationBayes Decision Rule and Naïve Bayes Classifier
Bayes Decision Rule and Naïve Bayes Classifier Le Song Machine Learning I CSE 6740, Fall 2013 Gaussian Mixture odel A density odel p(x) ay be ulti-odal: odel it as a ixture of uni-odal distributions (e.g.
More information