Efficient algorithms for the smallest enclosing ball problem

Size: px
Start display at page:

Download "Efficient algorithms for the smallest enclosing ball problem"

Transcription

1 Efficient algorithms for the smallest enclosing ball roblem Guanglu Zhou, Kim-Chuan Toh, Jie Sun November 27, 2002; Revised August 4, 2003 Abstract. Consider the roblem of comuting the smallest enclosing ball of a set of m balls in R n. Existing algorithms are known to be inefficient when n>30. In this aer we develo two algorithms that are articularly suitable for roblems where n is large. The first algorithm is based on log-exonential aggregation of the maximum function and reduces the roblem into an unconstrained convex rogram. The second algorithm is based on a second-order cone rogramming formulation, with secial structures taken into consideration. Our comutational exeriments show that both methods are efficient for large roblems, with the roduct mn in the order of Using the first algorithm, we are able to solve roblems with n = 100 and m = in about 1 hour. Keywords: Comutational geometry, smoothing aroximation, second order cone rogramming. AMS subject classification: 90C30. 1 Introduction A ball B i in R n with center c i and radius r i 0 is the closed set B i = {x R n : x c i r i }. Given a set of balls B = {B 1,B 2,,B m } in R n, the smallest enclosing ball mb(b) of the set B is the ball with the smallest radius that encloses all the balls in B. Given a set of oints P = { 1, 2,, m } in R n, the smallest enclosing ball m(p) of the oints is the ball with the smallest radius that encloses all the oints in P. Clearly, the roblem of smallest enclosing ball of oints is a secial case of the roblem Singaore-MIT Alliance, National University of Singaore, 4 Engineering Drive 3, Singaore smazgl@nus.edu.sg. Deartment of Mathematics, National University of Singaore, 2 Science Drive 2, Singaore mattohkc@nus.edu.sg. Research suorted in art by the Singaore-MIT Alliance. Deartment of Decision Sciences, National University of Singaore, Singaore. jsun@nus.edu.sg. 1

2 of smallest enclosing ball of balls. In what follows, unless otherwise stated, we will term the roblem of smallest enclosing ball of balls to be the smallest enclosing ball roblem. This roblem can be easily formulated as the following convex otimization roblem: min x R n max { x c i + r i }. (1) 1 i m Let f i (x) = x c i + r i, i =1, 2,,m, (2) and f(x) = max {f i(x)}. (3) 1 i m Then roblem (1) can be rewritten as min f(x). (4) x Rn It is readily seen that the solution to (1) exists since f(x) is coercive. Moreover, the solution is unique, for otherwise there would exist two different balls, C 1 and C 2,of the same radius, with m j=1 B j C i,i =1, 2. Then one can construct a smaller ball containing C 1 C 2 and thus B as well. Problem (1) arises in alications such as location analysis and military oerations and it is itself of interest as a roblem in comutational geometry; see [2, 3, 6, 9, 12, 13, 18, 20] for details. Many algorithms have been develoed for (1). In articular, for the roblem of the smallest enclosing ball of oints, Megiddo [9] resented a deterministic O(m) algorithm for the case where n 3. Welzl [18] develoed a simle randomized algorithm with exected linear time in m for the case where n is small, and Gärtner described a C++ imlementation thereof in [4]. For the roblem of the smallest enclosing ball of balls, we are aware of only a C++ rogram develoed by David White in [19], which is based on Welzl s algorithm [18] and Gärtner s imlementation [4]. The existing software ackages [4, 19], however, are not efficient for solving roblems with n>30. The goal of this aer is to resent efficient algorithms that can be used to solve roblems where n and m are large (say, in the order of thousands). Problem (1) is a non-differentiable (nonsmooth) convex otimization roblem. Due to the non-differentiability of the objective function, gradient-based algorithms cannot be used to solve this roblem. To overcome this difficulty, in Section 2 we give a smooth aroximation for the maximum function f. This aroximation is based on the socalled log-exonential aggregation function studied, for instance, in [8]. Based on this smooth aroximate function, in Section 3 we roose a limited-memory BFGS algorithm for solving the roblem. In Section 4, we reformulate (1) as a second order cone rogramming (SOCP) roblem. Thus, we can also aly interior-oint methods to solve (1). In Section 5, we reort some numerical results, which show the algorithms roosed here are able to solve large roblems. In this aer, unless otherwise stated, all vectors are column vectors. We let I d denote the d d identity matrix. To reresent a large matrix with several small matrices, we use semicolons ; for column concatenation and commas, for row concatenation. These notations also aly to vectors. We denote the convex hull of a set S R n by co(s). For a convex function f : R n R,welet f(x) denote the subdifferential of f at x. Note that f is minimized at x over R n if and only if 0 f(x). For calculus rules on subdifferentials of convex functions, lease refer to Chater VI in [7]. 2

3 2 Smooth aroximation For any > 0, define the smoothing log-exonential aggregation function of f in (3) as ( m ) f(x; ) = ln ex (g i (x; )/) where g i (x; ) =r i + x c i (5) Lemma 1 The function f(x; ) has the following roerties: (i) For any x R n, and 1, 2 satisfying 0 < 1 < 2, we have f(x; 1 ) <f(x; 2 ); (ii) For any x R n and >0, f(x) f(x; ) f(x)+(1 + ln m); (iii) For any > 0, f(x; ) is continuously differentiable and strictly convex. Proof. (i) For any x R n and 1, 2 satisfying 0 < 1 < 2, by Jensen s inequality [5], [ m 2 [ m ] (ex(g i (x; 2 ))) 2] 1/ 1 > (ex(g i (x; 2 ))) 1/ 1 [ m 1 > (ex(g i (x; 1 ))) 1] 1/. Hence, f(x; 1 ) <f(x; 2 ). (ii) Fix = 2 and let 1 0 in (i), we have f(x) <f(x; ). Let g (x; ) = max 1 i m {g i(x; )}. (6) It is readily roven that f(x) g (x; ) f(x)+. Thus, from (5), we have f(x; ) =g (x; )+ln ex [(g i (x; ) g (x; )) /] g (x; )+ln m. Hence f(x) f(x; ) f(x)+(1 + ln m). (iii) For any >0, clearly, f(x; ) is continuously differentiable. Now we rove that f(x; ) is strictly convex. From (5), where h i (x; ) = f(x; ) = x c i 2 + 2, τ(x; ) = λ i (x; ) h i (x; ) (x c i), (7) ex(g i (x; )/), (8) λ i (x; ) = ex(g i(x; )/). (9) τ(x; ) 3

4 Let Q ij = (x c i)(x c j ) T h i (x; )h j (x; ), i,j =1, 2,,m. From (7), we get 2 f(x; ) = λ i(x; ) h i (x; ) (I n Q ii )+ λ i(x; ) Q ii j=1 λ i (x; )λ j (x; ) Q ij. For any z R n with z 0, by the Cauchy-Schwartz inequality, Thus, z 2 z T Q ii z z 2 z 2 (x c i )/h i (x; ) 2 > 0, i =1,...,m. z T 2 f(x; )z = > λ i (x; ) ( z 2 z T Q ii z ) h i (x; ) + λ i(x; ) z T Q ii z λ i(x; ) z T Q ii z j=1 j=1 λ i (x; )λ j (x; ) λ i (x; )λ j (x; ) z T Q ij z z T Q ij z = 1 λ i (x; )a 2 i 1 ( m ) 2 λ i (x; )a i 0, (10) where a i = z T (x c i )/h i (x; ). This shows that 2 f(x; ) is ositive definite. Therefore, f(x; ) is strictly convex. Note that the inequality (10) follows from the fact that λ i (x; )a i m λ i (x; ) m λ i (x; )a 2 i, and m λ i (x; ) =1,λ i (x; ) 0 for i =1,...,m. 3 Log-exonential aggregation algorithm We now give an algorithm for roblem (1) based on Lemma 1, followed by a global convergence result. Algorithm 1 Let σ (0, 1), x 0 R n and 0 > 0 be given, and set k := 0. 4

5 For k =0, 1, 2,, do End 1. Use an unconstrained minimization method to solve min x R n f(x; k ). (11) Let the minimizer be x k. 2. Set k+1 = σ k, increment k by 1, and return to ste 1. Lemma 2 Let {x k } k 1 be the sequence of oints roduced by Algorithm 1. Then any limit oint of {x k } k 1 is an otimal solution of (1). Proof. Let x be a limit oint of {x k } k 1. Without loss of generality, we suose that x k x as k tends to +. From the fact that x k is a solution of (11), we have f(x k ; k ) = λ k i h i (x k ; k ) (xk c i ) = 0, where λ k i = λ i(x k ; k ), and h i (x; k ) is defined as in (8). Let By noting that I(x ) = {i : f i (x )=f(x ),, 2,,m}. λ k i = ex[(g i (x k ; k ) g (x k ; k ))/ k ] m ex[(g i (x k ; k ) g (x k ; k ))/ k ], we have lim k + λk i =0, i I(x ). (12) By (12) and the fact that m λ k i = 1 and λ k i > 0, i = 1, 2,,m, the sequence {λ k i,i =1, 2,,m} k 1 has a convergent subsequence. Without loss of generality, we suose that lim k + λk i = λ i, i I(x ). Then i I(x ) λ i = 1 and λ i 0, i I(x ). For i I(x ), t i = lim k + x k c i h i (x k ; k ) f i(x ). Thus lim k + f(xk ; k )= i I(x ) λ i t i =0. This shows that 0 co( f i (x ),i I(x )) = f(x ). Therefore, x is an otimal solution of (1). 5

6 Theorem 3 Let {x k } k 1 be the sequence of oints roduced by Algorithm 1 and x be the unique otimal solution of (1). Then lim k + xk = x. Proof. For any k 1, by Lemma 1, Since f(x) is coercive, the level set f(x 1 ; 1 ) >f(x 1 ; k ) f(x k ; k ) f(x k ). (13) L = {x R n : f(x) f(x 1 ; 1 )} is bounded. From (13), we have {x k } k 1 L. Hence {x k } k 1 is bounded. As x is the unique solution of (1), it follows from Lemma 2 that lim k + xk = x. In ractice, we would sto Algorithm 1 once some stoing criteria are met. Moreover, we will use a first-order, or gradient based, unconstrained minimization algorithm for solving (11) in ste 1. In the following algorithm, we will choose a limited-memory BFGS algorithm [1] to solve (11) as this algorithm can solve very large unconstrained otimization roblems efficiently. The following is a more ractical version of Algorithm 1. Algorithm 2 Let σ (0, 1), ɛ 1,ɛ 2 > 0, x 0 R n and 0 > 0 be given, and set k := 0. For k =0, 1, 2,, until k ɛ 1, do End 1. Use a version of the limited-memory BFGS algorithm to solve (11) aroximately, and obtain an x k such that f(x k ; k ) ɛ Set k+1 = σ k, increment k by 1, and return to ste 1. The actual arameter values used for 0,σ,ɛ 1,ɛ 2 are given in Section 5. 4 SOCP reformulation Problem (1) is equivalent to the following roblem. min r s.t. y c i + r i r, i =1,,m, y R n. (14) 6

7 It in turns can be reformulated as follows: max r s.t. t i + r = r i, s i + y = c i, t i s i, i =1,...,m. (15) Let N = m(n +1), K := {(t; s) R n+1 : t s }, y =(r; y) R n+1, c = (r 1 ; c 1 ; r 2 ; c 2 ; ; r m ; c m ) R N, A = (I n+1,,i n+1 ) R (n+1) N, b = (1; 0; ;0) R n+1, s =(t 1 ; s 1 ; t 2 ; s 2 ; ; t m ; s m ) R N, where I n+1 is an identity matrix. For later urose, we use the notation γ(α; β) = α 2 β 2 for (α; β) K. The roblem (15) can be written as a standard dual second order cone rogram as follows: (D) max b T y, s.t. s = c A T y, s K, where K = K K K = K m. Let The corresonding rimal roblem is x =(τ 1 ; x 1 ; τ 2 ; x 2 ; ; τ m ; x m ) R N. (P ) min c T x, s.t. Ax = b, x K. This air of roblems (P ) and (D) are studied in Nesterov and Nemirovskii [10], Nesterov and Todd [11], and Xue and Ye [21]. We can solve these roblems by using a rimal-dual interior-oint method (IPM). There are IPM based general urose softwares for solving SOCPs, examles are SeDuMi [14], SDPT3[16], and LOQO [17]. But the roblems (P ) and (D) have a very secial constraint matrix A, and it is beneficial to analyze how such a structure can be exloited when comuting the search direction in each IPM iteration. Suose that at a articular IPM iteration, the current iterate is (x, y, s) and the target barrier arameter is µ. To obtain the next iterate, the most exensive comutation lies in assembling and solving a linear system of n + 1 equations for the search direction y. By adating the results from Section 2.3 of [16] for the so-called Nesterov-Todd search direction, such a linear system takes the following form for our resent SOCPs: ( 1 m ) σ J + 1 g 2 wi 2 i g T i y = h, (16) }{{} M 7

8 where [ ] 1 J =, wi 2 I = γ(t i; s i ) n γ(τ i ; x i ), 1 σ = m 1, 2 wi 2 2 g i = γ(ρ i ; ξ i ) ( ρ i; ξ i ), (ρ i ; ξ i ) = 1 (t i ; s i ) w i ( τ i ; x i ), w i and h = b + 1 wi 2 ( J + gi g T i ) ((ri ; c i )+(t i ; s i ) y) µ γ(t i ; s i ) 2 ( t i; s i ). Note that the matrix M is symmetric ositive definite. In the above, we give a full descrition of the linear system (16) for the sake of comleteness. But our main aim is to study the matrix M, and exlore the ossibility of solving (16) via a Krylov subsace method such as the reconditioned conjugate gradient (PCG) method when n is large. Let I = {i {1,...,m} : σ g i w i > 1}, J = {1,...,m}\I. We can write the matrix M in (16) as follows: M = 1 ( I + σ 2 ) g σ 2 i I wi 2 i g T i + 1 σ 2 g σ 2 j J w 2 j g T j j }{{}}{{} M 0 M 1 2 σ 2 e 1e T 1, where M 0 is ositive definite, M 1 is ositive semidefinite, and e 1 = (1; 0; ;0) R n+1. Numerical exeriments have shown that the cardinality of I is usually much smaller than n when the barrier arameter µ is small. Thus M 0 is a diagonal matrix lus a lowrank matrix, and its inverse can be found readily via the Sherman-Morrison-Woodbury formula: 0 = σ [ 2 I G(I + G T G) 1 G ] T, G = }{{} B M 1 ( ) σ g i i I. wi Suose M 0 is used as a reconditioner in (16), then the reconditioned matrix 0 M has the form: M 1 M 1 0 M = I + 1 σ 2 M 1 0 M 1 2 σ 2 M 1 0 e 1e T 1 = I + BM 1 2Be 1 e T 1. (17) To analyze the sectrum of M 1 0 M 1, we make use of the following lemma. Lemma 4 Let = I. If<n+1, then the matrix B is ositive semidefinite and B 1 8

9 Proof. Let the singular value decomosition of G R (n+1) be G = UΣV T, where U R (n+1) has orthonormal columns, Σ R is diagonal, and V R is orthogonal. It is easily shown that Thus G(I + G T G)G T = UΣ(I +Σ 2 ) 1 ΣU T. B = I UΣ(I +Σ 2 ) 1 ΣU T (18) = (I UU T )+U(I +Σ 2 ) 1 U T, which is the sum of two ositive semidefinite matrices, and hence B is also ositive semidefinite. From (18), it is clear that x T Bx x T x for any x R n+1. Hence, B 1. Notice that BM 1 is similar to a ositive semidefinite matrix and thus all its eigenvalues lie in the interval [0, BM 1 ] [0, M 1 ]. For j J, σ g j /w j 1, thus M 1 J, and we would exect M 1 to be a moderate number. Hence, we can exect the reconditioned matrix M0 1 M to have a favorable sectral distribution (with all its eigenvalues clustered around 1 excet ossibly for an outlier due to the resence of a rank-1 erturbation in (17)) for the PCG to attain fast convergence. So far, we have discussed the reconditioning strategy to solve (16) when the barrier arameter µ is small. When µ is not small, the cardinality of I may be comarable to n, and it would be comutationally exensive to use M 0 as a reconditioner. Fortunately when µ is not small, the matrix M itself is quite well-conditioned and reconditioning by the diagonal of M would be good enough for PCG to obtain fast convergence. In ractice, we switch from the diagonal reconditioner to M 0 when 5 Comutational results max( g i /w i ) min( g i /w i ) (19) We imlemented Algorithm 2 described in Section 3 and the numerical exeriments were done by using a Pentium IV 2.2GHz ersonal comuter with 4GB of memory. To solve (11), we choose a limited-memory BFGS algorithm with strong Wolfe-Powell line-search and 7 limited-memory vector udates [1]. We imlemented our code in Fortran 77 and comared it with SeDuMi (version 1.05) [14] and SDPT3 (version 3.02) [16] (two high quality software ackages with Matlab interface for solving semidefinite-quadraticlinear rograms). We also imlemented an adatation of SDPT3 to exloit the secial structure resent in our SOCPs in (P) and (D). The main change lies in relacing the direct solution method in comuting the direction y in SDPT3 by the PCG iterative solution method discussed in the last section. For convenience, we will call the modified algorithm SDPT3-PCG. 9

10 Throughout the comutational exeriments, we use the following arameters in Algorithm 2: 0 =0.01, σ =0.1, ɛ 1 =10 6, and ɛ 2 =10 5. The starting oint is x 0 = 0. For SDPT3, we sto the interior-oint algorithm when φ is smaller than Here ( b Ax φ = max 1+ b, c s AT y x T ) s, 1+ c 1+0.5( b T. y + c T x ) We use the same stoing criterion for SDPT3-PCG. When solving the system (16), we sto the PCG iteration when the relative residual norm, h M y h, is less than For SeDuMi, we use all the default values excet that the arameter es is set to be The test roblems are generated randomly. We use the following seudo-random sequence: ψ 0 =7, ψ i+1 = (445ψ i + 1) mod 4096, i =1, 2,, ψ i = ψ i 40.96, i =1, 2,. (20) The elements of c i,, 2,,m, are successively set to ψ 1, ψ 2,, in the order: r 1,c 1 (1), c 1 (2),,c 1 (n), r 2, c 2 (1),,c 2 (n),,r m,c m (1),,c m (n). The numerical results we obtained are summarized in Tables 1 4. In these tables, n and m denote the dimension of the Euclidean sace and the number of balls, resectively, Obj Value denotes the value of the objective function at the final iteration, Iter denotes the number of iterations, Time denotes the CPU time in second for solving each roblem, and RAM denotes the random access memory usage in MB. The results reorted in Tables 1 and 2 show that algorithms resented in this aer erform very well. We can see from Table 1 that all algorithms are able to obtain good accuracy for moderately large roblems. Moreover, it aears from Table 2 that Algorithm 2 consistently takes more iterations than the SOCP algorithm and that for all tested roblems, Algorithm 2 and SDPT3-PCG consistently use less CPU time than SeDuMi and SDPT3. It is interesting to note that for the last 2 roblems, SeDuMi, SDPT3, and Alg. 2 are about 30, 8.5, and 2.8 times more exensive (in CPU time) than SDPT3-PCG, resectively. From the erformance data in Table 2, it is clear that for the smallest enclosing ball roblem, Alg.2 and SDPT3-PCG are much more efficient than the general urose algorithms in SeDuMi or SDPT3. Thus we shall concentrate only on Alg. 2 and SDPT3-PCG for the rest of this section. Next, we look at the erformance of Alg. 2 and SDPT3-PCG for roblems with large n but m is moderate. For each air of (n, m), we run the algorithms on 5 random roblems generated from (20) with the integer 445 relaced by 437, 441, 445, 449, and 453. The results (each number is the average over 5 roblems) reorted in Table 3 show that Algorithm 2 can solve roblems with very large n in a reasonable amount of CPU time that grows linearly with n. Its memory demand is just slightly more than the 10

11 amount required to store the data oints {c i R n : i =1,...,m}. However, SDPT3- PCG fails for the last 2 larger roblems in Table 3 due to excessive memory requirement. In fact, SDPT3-PCG requires about 18 times more memory sace than Alg. 2. It is interesting to note however, that when there is enough comuter memory, SDPT3-PCG aears to be more efficient (in terms of CPU time) than Alg. 2. In Table 4, we comare the the erformance of Alg. 2 and SDPT3-PCG for roblems with large m but n is moderate. Again, we test the algorithms on 5 randomly generated roblems for each air of (n, m). Both Alg. 2 and SDPT3-PCG can solve these roblems very efficiently. The required CPU time and memory for Alg. 2 grow almost linearly with m. For SDPT3-PCG, the required CPU time grows like m 1.13 with m. Again, its memory demand is about 18 times more than that required by Alg. 2. From the results in Tables 2 to 4, we conclude that Algorithm 2 is better than SeDuMi, SDPT3 and SDPT3-PCG for very large roblems, since it consumes less memory. For moderately large roblems, SDPT3-PCG is more efficient (in CPU time) than Algorithm 2. Problem Obj Value n m SeDuMi SDPT3 Alg2 SDPT3-PCG e e e e e e e e e e e e e e e e e e e e3 Table 1: Objective function value at the final iteration. Problem SeDuMi SDPT3 Alg2 SDPT3-PCG n m Time Iter Time Iter Time Iter Time Iter Table 2: Performance comarison of SeDuMi, SDPT3, Algorithm 2 and SDPT3-PCG. The numbers in the last two columns corresond to average value of I and the average number of PCG iterations in each interior-oint iteration, resectively. References [1] R.H. Byrd, P. Lu, J. Nocedal and C. Zhu, A limited memory algorithm for bound constrained otimization, SIAM J. Sci. Comut., 16 (1995)

12 Problem Alg2 SDPT3-PCG n m Time Iter RAM Time Iter RAM out of memory out of memory Table 3: Performance of Algorithm 2 and SDPT3-PCG on roblems with large n. Problem Alg2 SDPT3-PCG n m Time Iter RAM Time Iter RAM out of memory out of memory Table 4: Performance of Algorithm 2 and SDPT3-PCG on roblems with large m. [2] J. Elzinga and D. Hearn, The minimum covering shere roblem, Management Sci., 19 (1972) [3] B. Gärter, Fast and robust smallest enclosing balls, J. Nestril, editor, Algorithms- ESA 99: 7th Annual Euroean Symosium Proceedings, Lecture Notes in Comuter Science 1643, , Sringer-Verlag. [4] B. Gärter, Smallest enclosing ball - fast and robust in C++, htt:// [5] G.H. Hardy, J.E. Littlewood and G. Polya, Inequalities, Cambridge University Press, [6] D.W. Hearn and J. Vijan, Efficient algorithms for the minimum circle roblem, Oer. Res., 30 (1982) [7] J.B. Hiriart-Urruty and C. Lemarechal, Convex Analysis and Minimization Algorithm, Sringer-Verlag Berlin Heidelberg, [8] X.S. Li, An aggregate function method for nonlinear rogramming, Sci. China Ser. A, 34 (1991) [9] N. Megiddo, Linear-time algorithms for linear rogramming in R 3 and related roblems, SIAM J. Comut., 12 (1983)

13 [10] Yu. E. Nesterov and A. Nemirovskii, Interior Polynomial Algorithms in Convex Programming, SIAM, Philadelhia, [11] Yu. E. Nesterov and M. J. Todd, Self-scaled barriers and interior-oint methods for convex rogramming, Math. Oer. Res., 22 (1997) [12] F.P. Prearata and M.I. Shamos, Comutational geometry: an introduction, Texts and Monograhs in Comuter Science, Sringer-Verlag, New York, [13] M.I. Shamos and D. Hoey, Closest-oint roblems, in 16th Annual Symosium on Foundations of Comuter Science (Berkeley, Calif., 1975), IEEE Comuter Society, Long Beach, Calif., [14] J.F. Sturm, Using SeDuMi 1.02, a MATLAB toolbox for otimization over symmetric cones, Otim. Methods Softw., 11&12 (1999) [15] P. Sun, and R.M. Freund, Comutation of minimum volume covering ellisoids, MIT Oerations Research Center Working Paer OR , July, [16] R.H Tütüncü, K.C. Toh, and M.J. Todd, Solving semidefinite-quadratic-linear rograms using SDPT3, Math. Programming, to aear. [17] R.J. Vanderbei, LOQO user s manual version 3.10, Otim. Methods Softw., 11&12 (1999) [18] E. Welzl, Smallest enclosing disks (balls and ellisoids), H. Maurer, editor, New Results and New Trends in Comuter Science, , Sringer-Verlag, [19] D. White, Enclosing ball software, htt://vision.ucsd.edu/~dwhite/ball.html. [20] S. Xu, R. Freund and J. Sun, Solution methodologies for the smallest enclosing circle roblem, Comut. Otim. Al., to aear. [21] G.L. Xue and Y.Y. Ye, An efficient algorithm for minimizing a sum of Euclidean norms with alications, SIAM J. Otim., 7 (1997)

Notes on duality in second order and -order cone otimization E. D. Andersen Λ, C. Roos y, and T. Terlaky z Aril 6, 000 Abstract Recently, the so-calle

Notes on duality in second order and -order cone otimization E. D. Andersen Λ, C. Roos y, and T. Terlaky z Aril 6, 000 Abstract Recently, the so-calle McMaster University Advanced Otimization Laboratory Title: Notes on duality in second order and -order cone otimization Author: Erling D. Andersen, Cornelis Roos and Tamás Terlaky AdvOl-Reort No. 000/8

More information

Figure : An 8 bridge design grid. (a) Run this model using LOQO. What is the otimal comliance? What is the running time?

Figure : An 8 bridge design grid. (a) Run this model using LOQO. What is the otimal comliance? What is the running time? 5.094/SMA53 Systems Otimization: Models and Comutation Assignment 5 (00 o i n ts) Due Aril 7, 004 Some Convex Analysis (0 o i n ts) (a) Given ositive scalars L and E, consider the following set in three-dimensional

More information

Research Article An iterative Algorithm for Hemicontractive Mappings in Banach Spaces

Research Article An iterative Algorithm for Hemicontractive Mappings in Banach Spaces Abstract and Alied Analysis Volume 2012, Article ID 264103, 11 ages doi:10.1155/2012/264103 Research Article An iterative Algorithm for Hemicontractive Maings in Banach Saces Youli Yu, 1 Zhitao Wu, 2 and

More information

216 S. Chandrasearan and I.C.F. Isen Our results dier from those of Sun [14] in two asects: we assume that comuted eigenvalues or singular values are

216 S. Chandrasearan and I.C.F. Isen Our results dier from those of Sun [14] in two asects: we assume that comuted eigenvalues or singular values are Numer. Math. 68: 215{223 (1994) Numerische Mathemati c Sringer-Verlag 1994 Electronic Edition Bacward errors for eigenvalue and singular value decomositions? S. Chandrasearan??, I.C.F. Isen??? Deartment

More information

Approximating min-max k-clustering

Approximating min-max k-clustering Aroximating min-max k-clustering Asaf Levin July 24, 2007 Abstract We consider the roblems of set artitioning into k clusters with minimum total cost and minimum of the maximum cost of a cluster. The cost

More information

Convex Optimization methods for Computing Channel Capacity

Convex Optimization methods for Computing Channel Capacity Convex Otimization methods for Comuting Channel Caacity Abhishek Sinha Laboratory for Information and Decision Systems (LIDS), MIT sinhaa@mit.edu May 15, 2014 We consider a classical comutational roblem

More information

Recursive Estimation of the Preisach Density function for a Smart Actuator

Recursive Estimation of the Preisach Density function for a Smart Actuator Recursive Estimation of the Preisach Density function for a Smart Actuator Ram V. Iyer Deartment of Mathematics and Statistics, Texas Tech University, Lubbock, TX 7949-142. ABSTRACT The Preisach oerator

More information

Various Proofs for the Decrease Monotonicity of the Schatten s Power Norm, Various Families of R n Norms and Some Open Problems

Various Proofs for the Decrease Monotonicity of the Schatten s Power Norm, Various Families of R n Norms and Some Open Problems Int. J. Oen Problems Comt. Math., Vol. 3, No. 2, June 2010 ISSN 1998-6262; Coyright c ICSRS Publication, 2010 www.i-csrs.org Various Proofs for the Decrease Monotonicity of the Schatten s Power Norm, Various

More information

Chater Matrix Norms and Singular Value Decomosition Introduction In this lecture, we introduce the notion of a norm for matrices The singular value de

Chater Matrix Norms and Singular Value Decomosition Introduction In this lecture, we introduce the notion of a norm for matrices The singular value de Lectures on Dynamic Systems and Control Mohammed Dahleh Munther A Dahleh George Verghese Deartment of Electrical Engineering and Comuter Science Massachuasetts Institute of Technology c Chater Matrix Norms

More information

Some results of convex programming complexity

Some results of convex programming complexity 2012c12 $ Ê Æ Æ 116ò 14Ï Dec., 2012 Oerations Research Transactions Vol.16 No.4 Some results of convex rogramming comlexity LOU Ye 1,2 GAO Yuetian 1 Abstract Recently a number of aers were written that

More information

Sharp gradient estimate and spectral rigidity for p-laplacian

Sharp gradient estimate and spectral rigidity for p-laplacian Shar gradient estimate and sectral rigidity for -Lalacian Chiung-Jue Anna Sung and Jiaing Wang To aear in ath. Research Letters. Abstract We derive a shar gradient estimate for ositive eigenfunctions of

More information

Preconditioning techniques for Newton s method for the incompressible Navier Stokes equations

Preconditioning techniques for Newton s method for the incompressible Navier Stokes equations Preconditioning techniques for Newton s method for the incomressible Navier Stokes equations H. C. ELMAN 1, D. LOGHIN 2 and A. J. WATHEN 3 1 Deartment of Comuter Science, University of Maryland, College

More information

Linear diophantine equations for discrete tomography

Linear diophantine equations for discrete tomography Journal of X-Ray Science and Technology 10 001 59 66 59 IOS Press Linear diohantine euations for discrete tomograhy Yangbo Ye a,gewang b and Jiehua Zhu a a Deartment of Mathematics, The University of Iowa,

More information

Improving AOR Method for a Class of Two-by-Two Linear Systems

Improving AOR Method for a Class of Two-by-Two Linear Systems Alied Mathematics 2 2 236-24 doi:4236/am22226 Published Online February 2 (htt://scirporg/journal/am) Imroving AOR Method for a Class of To-by-To Linear Systems Abstract Cuixia Li Shiliang Wu 2 College

More information

ON THE DEVELOPMENT OF PARAMETER-ROBUST PRECONDITIONERS AND COMMUTATOR ARGUMENTS FOR SOLVING STOKES CONTROL PROBLEMS

ON THE DEVELOPMENT OF PARAMETER-ROBUST PRECONDITIONERS AND COMMUTATOR ARGUMENTS FOR SOLVING STOKES CONTROL PROBLEMS Electronic Transactions on Numerical Analysis. Volume 44,. 53 72, 25. Coyright c 25,. ISSN 68 963. ETNA ON THE DEVELOPMENT OF PARAMETER-ROBUST PRECONDITIONERS AND COMMUTATOR ARGUMENTS FOR SOLVING STOKES

More information

Radial Basis Function Networks: Algorithms

Radial Basis Function Networks: Algorithms Radial Basis Function Networks: Algorithms Introduction to Neural Networks : Lecture 13 John A. Bullinaria, 2004 1. The RBF Maing 2. The RBF Network Architecture 3. Comutational Power of RBF Networks 4.

More information

where x i is the ith coordinate of x R N. 1. Show that the following upper bound holds for the growth function of H:

where x i is the ith coordinate of x R N. 1. Show that the following upper bound holds for the growth function of H: Mehryar Mohri Foundations of Machine Learning Courant Institute of Mathematical Sciences Homework assignment 2 October 25, 2017 Due: November 08, 2017 A. Growth function Growth function of stum functions.

More information

Iteration with Stepsize Parameter and Condition Numbers for a Nonlinear Matrix Equation

Iteration with Stepsize Parameter and Condition Numbers for a Nonlinear Matrix Equation Electronic Journal of Linear Algebra Volume 34 Volume 34 2018) Article 16 2018 Iteration with Stesize Parameter and Condition Numbers for a Nonlinear Matrix Equation Syed M Raza Shah Naqvi Pusan National

More information

DEPARTMENT OF ECONOMICS ISSN DISCUSSION PAPER 20/07 TWO NEW EXPONENTIAL FAMILIES OF LORENZ CURVES

DEPARTMENT OF ECONOMICS ISSN DISCUSSION PAPER 20/07 TWO NEW EXPONENTIAL FAMILIES OF LORENZ CURVES DEPARTMENT OF ECONOMICS ISSN 1441-549 DISCUSSION PAPER /7 TWO NEW EXPONENTIAL FAMILIES OF LORENZ CURVES ZuXiang Wang * & Russell Smyth ABSTRACT We resent two new Lorenz curve families by using the basic

More information

MODELING THE RELIABILITY OF C4ISR SYSTEMS HARDWARE/SOFTWARE COMPONENTS USING AN IMPROVED MARKOV MODEL

MODELING THE RELIABILITY OF C4ISR SYSTEMS HARDWARE/SOFTWARE COMPONENTS USING AN IMPROVED MARKOV MODEL Technical Sciences and Alied Mathematics MODELING THE RELIABILITY OF CISR SYSTEMS HARDWARE/SOFTWARE COMPONENTS USING AN IMPROVED MARKOV MODEL Cezar VASILESCU Regional Deartment of Defense Resources Management

More information

The Algebraic Structure of the p-order Cone

The Algebraic Structure of the p-order Cone The Algebraic Structure of the -Order Cone Baha Alzalg Abstract We study and analyze the algebraic structure of the -order cones We show that, unlike oularly erceived, the -order cone is self-dual for

More information

Elementary Analysis in Q p

Elementary Analysis in Q p Elementary Analysis in Q Hannah Hutter, May Szedlák, Phili Wirth November 17, 2011 This reort follows very closely the book of Svetlana Katok 1. 1 Sequences and Series In this section we will see some

More information

For q 0; 1; : : : ; `? 1, we have m 0; 1; : : : ; q? 1. The set fh j(x) : j 0; 1; ; : : : ; `? 1g forms a basis for the tness functions dened on the i

For q 0; 1; : : : ; `? 1, we have m 0; 1; : : : ; q? 1. The set fh j(x) : j 0; 1; ; : : : ; `? 1g forms a basis for the tness functions dened on the i Comuting with Haar Functions Sami Khuri Deartment of Mathematics and Comuter Science San Jose State University One Washington Square San Jose, CA 9519-0103, USA khuri@juiter.sjsu.edu Fax: (40)94-500 Keywords:

More information

An Estimate For Heilbronn s Exponential Sum

An Estimate For Heilbronn s Exponential Sum An Estimate For Heilbronn s Exonential Sum D.R. Heath-Brown Magdalen College, Oxford For Heini Halberstam, on his retirement Let be a rime, and set e(x) = ex(2πix). Heilbronn s exonential sum is defined

More information

Approximation of the Euclidean Distance by Chamfer Distances

Approximation of the Euclidean Distance by Chamfer Distances Acta Cybernetica 0 (0 399 47. Aroximation of the Euclidean Distance by Chamfer Distances András Hajdu, Lajos Hajdu, and Robert Tijdeman Abstract Chamfer distances lay an imortant role in the theory of

More information

A numerical implementation of a predictor-corrector algorithm for sufcient linear complementarity problem

A numerical implementation of a predictor-corrector algorithm for sufcient linear complementarity problem A numerical imlementation of a redictor-corrector algorithm for sufcient linear comlementarity roblem BENTERKI DJAMEL University Ferhat Abbas of Setif-1 Faculty of science Laboratory of fundamental and

More information

Positive Definite Uncertain Homogeneous Matrix Polynomials: Analysis and Application

Positive Definite Uncertain Homogeneous Matrix Polynomials: Analysis and Application BULGARIA ACADEMY OF SCIECES CYBEREICS AD IFORMAIO ECHOLOGIES Volume 9 o 3 Sofia 009 Positive Definite Uncertain Homogeneous Matrix Polynomials: Analysis and Alication Svetoslav Savov Institute of Information

More information

MATH 6210: SOLUTIONS TO PROBLEM SET #3

MATH 6210: SOLUTIONS TO PROBLEM SET #3 MATH 6210: SOLUTIONS TO PROBLEM SET #3 Rudin, Chater 4, Problem #3. The sace L (T) is searable since the trigonometric olynomials with comlex coefficients whose real and imaginary arts are rational form

More information

Sums of independent random variables

Sums of independent random variables 3 Sums of indeendent random variables This lecture collects a number of estimates for sums of indeendent random variables with values in a Banach sace E. We concentrate on sums of the form N γ nx n, where

More information

An Ant Colony Optimization Approach to the Probabilistic Traveling Salesman Problem

An Ant Colony Optimization Approach to the Probabilistic Traveling Salesman Problem An Ant Colony Otimization Aroach to the Probabilistic Traveling Salesman Problem Leonora Bianchi 1, Luca Maria Gambardella 1, and Marco Dorigo 2 1 IDSIA, Strada Cantonale Galleria 2, CH-6928 Manno, Switzerland

More information

RIEMANN-STIELTJES OPERATORS BETWEEN WEIGHTED BERGMAN SPACES

RIEMANN-STIELTJES OPERATORS BETWEEN WEIGHTED BERGMAN SPACES RIEMANN-STIELTJES OPERATORS BETWEEN WEIGHTED BERGMAN SPACES JIE XIAO This aer is dedicated to the memory of Nikolaos Danikas 1947-2004) Abstract. This note comletely describes the bounded or comact Riemann-

More information

Verifying Two Conjectures on Generalized Elite Primes

Verifying Two Conjectures on Generalized Elite Primes 1 2 3 47 6 23 11 Journal of Integer Sequences, Vol. 12 (2009), Article 09.4.7 Verifying Two Conjectures on Generalized Elite Primes Xiaoqin Li 1 Mathematics Deartment Anhui Normal University Wuhu 241000,

More information

Uniformly best wavenumber approximations by spatial central difference operators: An initial investigation

Uniformly best wavenumber approximations by spatial central difference operators: An initial investigation Uniformly best wavenumber aroximations by satial central difference oerators: An initial investigation Vitor Linders and Jan Nordström Abstract A characterisation theorem for best uniform wavenumber aroximations

More information

Numerical Linear Algebra

Numerical Linear Algebra Numerical Linear Algebra Numerous alications in statistics, articularly in the fitting of linear models. Notation and conventions: Elements of a matrix A are denoted by a ij, where i indexes the rows and

More information

State Estimation with ARMarkov Models

State Estimation with ARMarkov Models Deartment of Mechanical and Aerosace Engineering Technical Reort No. 3046, October 1998. Princeton University, Princeton, NJ. State Estimation with ARMarkov Models Ryoung K. Lim 1 Columbia University,

More information

Extremal Polynomials with Varying Measures

Extremal Polynomials with Varying Measures International Mathematical Forum, 2, 2007, no. 39, 1927-1934 Extremal Polynomials with Varying Measures Rabah Khaldi Deartment of Mathematics, Annaba University B.P. 12, 23000 Annaba, Algeria rkhadi@yahoo.fr

More information

OPTIMAL AFFINE INVARIANT SMOOTH MINIMIZATION ALGORITHMS

OPTIMAL AFFINE INVARIANT SMOOTH MINIMIZATION ALGORITHMS 1 OPTIMAL AFFINE INVARIANT SMOOTH MINIMIZATION ALGORITHMS ALEXANDRE D ASPREMONT, CRISTÓBAL GUZMÁN, AND MARTIN JAGGI ABSTRACT. We formulate an affine invariant imlementation of the accelerated first-order

More information

arxiv: v2 [quant-ph] 2 Aug 2012

arxiv: v2 [quant-ph] 2 Aug 2012 Qcomiler: quantum comilation with CSD method Y. G. Chen a, J. B. Wang a, a School of Physics, The University of Western Australia, Crawley WA 6009 arxiv:208.094v2 [quant-h] 2 Aug 202 Abstract In this aer,

More information

Estimation of the large covariance matrix with two-step monotone missing data

Estimation of the large covariance matrix with two-step monotone missing data Estimation of the large covariance matrix with two-ste monotone missing data Masashi Hyodo, Nobumichi Shutoh 2, Takashi Seo, and Tatjana Pavlenko 3 Deartment of Mathematical Information Science, Tokyo

More information

Spectral Clustering based on the graph p-laplacian

Spectral Clustering based on the graph p-laplacian Sectral Clustering based on the grah -Lalacian Thomas Bühler tb@cs.uni-sb.de Matthias Hein hein@cs.uni-sb.de Saarland University Comuter Science Deartment Camus E 663 Saarbrücken Germany Abstract We resent

More information

A Recursive Block Incomplete Factorization. Preconditioner for Adaptive Filtering Problem

A Recursive Block Incomplete Factorization. Preconditioner for Adaptive Filtering Problem Alied Mathematical Sciences, Vol. 7, 03, no. 63, 3-3 HIKARI Ltd, www.m-hiari.com A Recursive Bloc Incomlete Factorization Preconditioner for Adative Filtering Problem Shazia Javed School of Mathematical

More information

GENERALIZED NORMS INEQUALITIES FOR ABSOLUTE VALUE OPERATORS

GENERALIZED NORMS INEQUALITIES FOR ABSOLUTE VALUE OPERATORS International Journal of Analysis Alications ISSN 9-8639 Volume 5, Number (04), -9 htt://www.etamaths.com GENERALIZED NORMS INEQUALITIES FOR ABSOLUTE VALUE OPERATORS ILYAS ALI, HU YANG, ABDUL SHAKOOR Abstract.

More information

On Doob s Maximal Inequality for Brownian Motion

On Doob s Maximal Inequality for Brownian Motion Stochastic Process. Al. Vol. 69, No., 997, (-5) Research Reort No. 337, 995, Det. Theoret. Statist. Aarhus On Doob s Maximal Inequality for Brownian Motion S. E. GRAVERSEN and G. PESKIR If B = (B t ) t

More information

#A64 INTEGERS 18 (2018) APPLYING MODULAR ARITHMETIC TO DIOPHANTINE EQUATIONS

#A64 INTEGERS 18 (2018) APPLYING MODULAR ARITHMETIC TO DIOPHANTINE EQUATIONS #A64 INTEGERS 18 (2018) APPLYING MODULAR ARITHMETIC TO DIOPHANTINE EQUATIONS Ramy F. Taki ElDin Physics and Engineering Mathematics Deartment, Faculty of Engineering, Ain Shams University, Cairo, Egyt

More information

Intrinsic Approximation on Cantor-like Sets, a Problem of Mahler

Intrinsic Approximation on Cantor-like Sets, a Problem of Mahler Intrinsic Aroximation on Cantor-like Sets, a Problem of Mahler Ryan Broderick, Lior Fishman, Asaf Reich and Barak Weiss July 200 Abstract In 984, Kurt Mahler osed the following fundamental question: How

More information

ETNA Kent State University

ETNA Kent State University Electronic Transactions on Numerical Analysis. Volume 9,. 29-36, 25. Coyright 25,. ISSN 68-963. ETNA ASYMPTOTICS FOR EXTREMAL POLYNOMIALS WITH VARYING MEASURES M. BELLO HERNÁNDEZ AND J. MíNGUEZ CENICEROS

More information

Generalized Coiflets: A New Family of Orthonormal Wavelets

Generalized Coiflets: A New Family of Orthonormal Wavelets Generalized Coiflets A New Family of Orthonormal Wavelets Dong Wei, Alan C Bovik, and Brian L Evans Laboratory for Image and Video Engineering Deartment of Electrical and Comuter Engineering The University

More information

Towards understanding the Lorenz curve using the Uniform distribution. Chris J. Stephens. Newcastle City Council, Newcastle upon Tyne, UK

Towards understanding the Lorenz curve using the Uniform distribution. Chris J. Stephens. Newcastle City Council, Newcastle upon Tyne, UK Towards understanding the Lorenz curve using the Uniform distribution Chris J. Stehens Newcastle City Council, Newcastle uon Tyne, UK (For the Gini-Lorenz Conference, University of Siena, Italy, May 2005)

More information

ON FREIMAN S 2.4-THEOREM

ON FREIMAN S 2.4-THEOREM ON FREIMAN S 2.4-THEOREM ØYSTEIN J. RØDSETH Abstract. Gregory Freiman s celebrated 2.4-Theorem says that if A is a set of residue classes modulo a rime satisfying 2A 2.4 A 3 and A < /35, then A is contained

More information

Hidden Predictors: A Factor Analysis Primer

Hidden Predictors: A Factor Analysis Primer Hidden Predictors: A Factor Analysis Primer Ryan C Sanchez Western Washington University Factor Analysis is a owerful statistical method in the modern research sychologist s toolbag When used roerly, factor

More information

Distribution of Matrices with Restricted Entries over Finite Fields

Distribution of Matrices with Restricted Entries over Finite Fields Distribution of Matrices with Restricted Entries over Finite Fields Omran Ahmadi Deartment of Electrical and Comuter Engineering University of Toronto, Toronto, ON M5S 3G4, Canada oahmadid@comm.utoronto.ca

More information

Multiplicity of weak solutions for a class of nonuniformly elliptic equations of p-laplacian type

Multiplicity of weak solutions for a class of nonuniformly elliptic equations of p-laplacian type Nonlinear Analysis 7 29 536 546 www.elsevier.com/locate/na Multilicity of weak solutions for a class of nonuniformly ellitic equations of -Lalacian tye Hoang Quoc Toan, Quô c-anh Ngô Deartment of Mathematics,

More information

Two unconstrained optimization approaches for the Euclidean κ-centrum location problem

Two unconstrained optimization approaches for the Euclidean κ-centrum location problem Noname manuscript No (will be inserted by the editor) Shaohua Pan and Jein-Shan Chen Two unconstrained optimization approaches for the Euclidean κ-centrum location problem Abstract Consider the single-facility

More information

Stochastic integration II: the Itô integral

Stochastic integration II: the Itô integral 13 Stochastic integration II: the Itô integral We have seen in Lecture 6 how to integrate functions Φ : (, ) L (H, E) with resect to an H-cylindrical Brownian motion W H. In this lecture we address the

More information

NONLINEAR OPTIMIZATION WITH CONVEX CONSTRAINTS. The Goldstein-Levitin-Polyak algorithm

NONLINEAR OPTIMIZATION WITH CONVEX CONSTRAINTS. The Goldstein-Levitin-Polyak algorithm - (23) NLP - NONLINEAR OPTIMIZATION WITH CONVEX CONSTRAINTS The Goldstein-Levitin-Polya algorithm We consider an algorithm for solving the otimization roblem under convex constraints. Although the convexity

More information

An Algorithm for Total Variation Minimization and Applications

An Algorithm for Total Variation Minimization and Applications Journal of Mathematical Imaging and Vision 0: 89 97, 004 c 004 Kluwer Academic Publishers. Manufactured in The Netherlands. An Algorithm for Total Variation Minimization and Alications ANTONIN CHAMBOLLE

More information

ON UNIFORM BOUNDEDNESS OF DYADIC AVERAGING OPERATORS IN SPACES OF HARDY-SOBOLEV TYPE. 1. Introduction

ON UNIFORM BOUNDEDNESS OF DYADIC AVERAGING OPERATORS IN SPACES OF HARDY-SOBOLEV TYPE. 1. Introduction ON UNIFORM BOUNDEDNESS OF DYADIC AVERAGING OPERATORS IN SPACES OF HARDY-SOBOLEV TYPE GUSTAVO GARRIGÓS ANDREAS SEEGER TINO ULLRICH Abstract We give an alternative roof and a wavelet analog of recent results

More information

A direct formulation for sparse PCA using semidefinite programming

A direct formulation for sparse PCA using semidefinite programming A direct formulation for sparse PCA using semidefinite programming A. d Aspremont, L. El Ghaoui, M. Jordan, G. Lanckriet ORFE, Princeton University & EECS, U.C. Berkeley Available online at www.princeton.edu/~aspremon

More information

DETC2003/DAC AN EFFICIENT ALGORITHM FOR CONSTRUCTING OPTIMAL DESIGN OF COMPUTER EXPERIMENTS

DETC2003/DAC AN EFFICIENT ALGORITHM FOR CONSTRUCTING OPTIMAL DESIGN OF COMPUTER EXPERIMENTS Proceedings of DETC 03 ASME 003 Design Engineering Technical Conferences and Comuters and Information in Engineering Conference Chicago, Illinois USA, Setember -6, 003 DETC003/DAC-48760 AN EFFICIENT ALGORITHM

More information

Location of solutions for quasi-linear elliptic equations with general gradient dependence

Location of solutions for quasi-linear elliptic equations with general gradient dependence Electronic Journal of Qualitative Theory of Differential Equations 217, No. 87, 1 1; htts://doi.org/1.14232/ejqtde.217.1.87 www.math.u-szeged.hu/ejqtde/ Location of solutions for quasi-linear ellitic equations

More information

Monopolist s mark-up and the elasticity of substitution

Monopolist s mark-up and the elasticity of substitution Croatian Oerational Research Review 377 CRORR 8(7), 377 39 Monoolist s mark-u and the elasticity of substitution Ilko Vrankić, Mira Kran, and Tomislav Herceg Deartment of Economic Theory, Faculty of Economics

More information

Generalized Convex Functions

Generalized Convex Functions Chater Generalized Convex Functions Convexity is one of the most frequently used hyotheses in otimization theory. It is usually introduced to give global validity to roositions otherwise only locally true,

More information

Applicable Analysis and Discrete Mathematics available online at HENSEL CODES OF SQUARE ROOTS OF P-ADIC NUMBERS

Applicable Analysis and Discrete Mathematics available online at   HENSEL CODES OF SQUARE ROOTS OF P-ADIC NUMBERS Alicable Analysis and Discrete Mathematics available online at htt://efmath.etf.rs Al. Anal. Discrete Math. 4 (010), 3 44. doi:10.98/aadm1000009m HENSEL CODES OF SQUARE ROOTS OF P-ADIC NUMBERS Zerzaihi

More information

Generation of Linear Models using Simulation Results

Generation of Linear Models using Simulation Results 4. IMACS-Symosium MATHMOD, Wien, 5..003,. 436-443 Generation of Linear Models using Simulation Results Georg Otte, Sven Reitz, Joachim Haase Fraunhofer Institute for Integrated Circuits, Branch Lab Design

More information

Research Article Controllability of Linear Discrete-Time Systems with Both Delayed States and Delayed Inputs

Research Article Controllability of Linear Discrete-Time Systems with Both Delayed States and Delayed Inputs Abstract and Alied Analysis Volume 203 Article ID 97546 5 ages htt://dxdoiorg/055/203/97546 Research Article Controllability of Linear Discrete-Time Systems with Both Delayed States and Delayed Inuts Hong

More information

On Wald-Type Optimal Stopping for Brownian Motion

On Wald-Type Optimal Stopping for Brownian Motion J Al Probab Vol 34, No 1, 1997, (66-73) Prerint Ser No 1, 1994, Math Inst Aarhus On Wald-Tye Otimal Stoing for Brownian Motion S RAVRSN and PSKIR The solution is resented to all otimal stoing roblems of

More information

Discrete Calderón s Identity, Atomic Decomposition and Boundedness Criterion of Operators on Multiparameter Hardy Spaces

Discrete Calderón s Identity, Atomic Decomposition and Boundedness Criterion of Operators on Multiparameter Hardy Spaces J Geom Anal (010) 0: 670 689 DOI 10.1007/s10-010-913-6 Discrete Calderón s Identity, Atomic Decomosition and Boundedness Criterion of Oerators on Multiarameter Hardy Saces Y. Han G. Lu K. Zhao Received:

More information

Deriving Indicator Direct and Cross Variograms from a Normal Scores Variogram Model (bigaus-full) David F. Machuca Mory and Clayton V.

Deriving Indicator Direct and Cross Variograms from a Normal Scores Variogram Model (bigaus-full) David F. Machuca Mory and Clayton V. Deriving ndicator Direct and Cross Variograms from a Normal Scores Variogram Model (bigaus-full) David F. Machuca Mory and Clayton V. Deutsch Centre for Comutational Geostatistics Deartment of Civil &

More information

IMPROVED BOUNDS IN THE SCALED ENFLO TYPE INEQUALITY FOR BANACH SPACES

IMPROVED BOUNDS IN THE SCALED ENFLO TYPE INEQUALITY FOR BANACH SPACES IMPROVED BOUNDS IN THE SCALED ENFLO TYPE INEQUALITY FOR BANACH SPACES OHAD GILADI AND ASSAF NAOR Abstract. It is shown that if (, ) is a Banach sace with Rademacher tye 1 then for every n N there exists

More information

Solved Problems. (a) (b) (c) Figure P4.1 Simple Classification Problems First we draw a line between each set of dark and light data points.

Solved Problems. (a) (b) (c) Figure P4.1 Simple Classification Problems First we draw a line between each set of dark and light data points. Solved Problems Solved Problems P Solve the three simle classification roblems shown in Figure P by drawing a decision boundary Find weight and bias values that result in single-neuron ercetrons with the

More information

Minimax Design of Nonnegative Finite Impulse Response Filters

Minimax Design of Nonnegative Finite Impulse Response Filters Minimax Design of Nonnegative Finite Imulse Resonse Filters Xiaoing Lai, Anke Xue Institute of Information and Control Hangzhou Dianzi University Hangzhou, 3118 China e-mail: laix@hdu.edu.cn; akxue@hdu.edu.cn

More information

The inverse Goldbach problem

The inverse Goldbach problem 1 The inverse Goldbach roblem by Christian Elsholtz Submission Setember 7, 2000 (this version includes galley corrections). Aeared in Mathematika 2001. Abstract We imrove the uer and lower bounds of the

More information

Research Note. A New Infeasible Interior-Point Algorithm with Full Nesterov-Todd Step for Semi-Definite Optimization

Research Note. A New Infeasible Interior-Point Algorithm with Full Nesterov-Todd Step for Semi-Definite Optimization Iranian Journal of Operations Research Vol. 4, No. 1, 2013, pp. 88-107 Research Note A New Infeasible Interior-Point Algorithm with Full Nesterov-Todd Step for Semi-Definite Optimization B. Kheirfam We

More information

Robustness of classifiers to uniform l p and Gaussian noise Supplementary material

Robustness of classifiers to uniform l p and Gaussian noise Supplementary material Robustness of classifiers to uniform l and Gaussian noise Sulementary material Jean-Yves Franceschi Ecole Normale Suérieure de Lyon LIP UMR 5668 Omar Fawzi Ecole Normale Suérieure de Lyon LIP UMR 5668

More information

Positive decomposition of transfer functions with multiple poles

Positive decomposition of transfer functions with multiple poles Positive decomosition of transfer functions with multile oles Béla Nagy 1, Máté Matolcsi 2, and Márta Szilvási 1 Deartment of Analysis, Technical University of Budaest (BME), H-1111, Budaest, Egry J. u.

More information

E-companion to A risk- and ambiguity-averse extension of the max-min newsvendor order formula

E-companion to A risk- and ambiguity-averse extension of the max-min newsvendor order formula e-comanion to Han Du and Zuluaga: Etension of Scarf s ma-min order formula ec E-comanion to A risk- and ambiguity-averse etension of the ma-min newsvendor order formula Qiaoming Han School of Mathematics

More information

ALTERNATIVE SOLUTION TO THE QUARTIC EQUATION by Farid A. Chouery 1, P.E. 2006, All rights reserved

ALTERNATIVE SOLUTION TO THE QUARTIC EQUATION by Farid A. Chouery 1, P.E. 2006, All rights reserved ALTERNATIVE SOLUTION TO THE QUARTIC EQUATION b Farid A. Chouer, P.E. 006, All rights reserved Abstract A new method to obtain a closed form solution of the fourth order olnomial equation is roosed in this

More information

Uncorrelated Multilinear Principal Component Analysis for Unsupervised Multilinear Subspace Learning

Uncorrelated Multilinear Principal Component Analysis for Unsupervised Multilinear Subspace Learning TNN-2009-P-1186.R2 1 Uncorrelated Multilinear Princial Comonent Analysis for Unsuervised Multilinear Subsace Learning Haiing Lu, K. N. Plataniotis and A. N. Venetsanooulos The Edward S. Rogers Sr. Deartment

More information

Specialized and hybrid Newton schemes for the matrix pth root

Specialized and hybrid Newton schemes for the matrix pth root Secialized and hybrid Newton schemes for the matrix th root Braulio De Abreu Marlliny Monsalve Marcos Raydan June 15, 2006 Abstract We discuss different variants of Newton s method for comuting the th

More information

AN ALGORITHM FOR MATRIX EXTENSION AND WAVELET CONSTRUCTION W. LAWTON, S. L. LEE AND ZUOWEI SHEN Abstract. This paper gives a practical method of exten

AN ALGORITHM FOR MATRIX EXTENSION AND WAVELET CONSTRUCTION W. LAWTON, S. L. LEE AND ZUOWEI SHEN Abstract. This paper gives a practical method of exten AN ALGORITHM FOR MATRIX EXTENSION AND WAVELET ONSTRUTION W. LAWTON, S. L. LEE AND ZUOWEI SHEN Abstract. This aer gives a ractical method of extending an nr matrix P (z), r n, with Laurent olynomial entries

More information

Spectral Analysis by Stationary Time Series Modeling

Spectral Analysis by Stationary Time Series Modeling Chater 6 Sectral Analysis by Stationary Time Series Modeling Choosing a arametric model among all the existing models is by itself a difficult roblem. Generally, this is a riori information about the signal

More information

SUPER-GEOMETRIC CONVERGENCE OF A SPECTRAL ELEMENT METHOD FOR EIGENVALUE PROBLEMS WITH JUMP COEFFICIENTS *

SUPER-GEOMETRIC CONVERGENCE OF A SPECTRAL ELEMENT METHOD FOR EIGENVALUE PROBLEMS WITH JUMP COEFFICIENTS * Journal of Comutational Mathematics Vol.8, No.,, 48 48. htt://www.global-sci.org/jcm doi:.48/jcm.9.-m6 SUPER-GEOMETRIC CONVERGENCE OF A SPECTRAL ELEMENT METHOD FOR EIGENVALUE PROBLEMS WITH JUMP COEFFICIENTS

More information

Math 4400/6400 Homework #8 solutions. 1. Let P be an odd integer (not necessarily prime). Show that modulo 2,

Math 4400/6400 Homework #8 solutions. 1. Let P be an odd integer (not necessarily prime). Show that modulo 2, MATH 4400 roblems. Math 4400/6400 Homework # solutions 1. Let P be an odd integer not necessarily rime. Show that modulo, { P 1 0 if P 1, 7 mod, 1 if P 3, mod. Proof. Suose that P 1 mod. Then we can write

More information

4. Score normalization technical details We now discuss the technical details of the score normalization method.

4. Score normalization technical details We now discuss the technical details of the score normalization method. SMT SCORING SYSTEM This document describes the scoring system for the Stanford Math Tournament We begin by giving an overview of the changes to scoring and a non-technical descrition of the scoring rules

More information

Elementary theory of L p spaces

Elementary theory of L p spaces CHAPTER 3 Elementary theory of L saces 3.1 Convexity. Jensen, Hölder, Minkowski inequality. We begin with two definitions. A set A R d is said to be convex if, for any x 0, x 1 2 A x = x 0 + (x 1 x 0 )

More information

On the Chvatál-Complexity of Knapsack Problems

On the Chvatál-Complexity of Knapsack Problems R u t c o r Research R e o r t On the Chvatál-Comlexity of Knasack Problems Gergely Kovács a Béla Vizvári b RRR 5-08, October 008 RUTCOR Rutgers Center for Oerations Research Rutgers University 640 Bartholomew

More information

ON JOINT CONVEXITY AND CONCAVITY OF SOME KNOWN TRACE FUNCTIONS

ON JOINT CONVEXITY AND CONCAVITY OF SOME KNOWN TRACE FUNCTIONS ON JOINT CONVEXITY ND CONCVITY OF SOME KNOWN TRCE FUNCTIONS MOHMMD GHER GHEMI, NHID GHRKHNLU and YOEL JE CHO Communicated by Dan Timotin In this aer, we rovide a new and simle roof for joint convexity

More information

On a class of Rellich inequalities

On a class of Rellich inequalities On a class of Rellich inequalities G. Barbatis A. Tertikas Dedicated to Professor E.B. Davies on the occasion of his 60th birthday Abstract We rove Rellich and imroved Rellich inequalities that involve

More information

On the capacity of the general trapdoor channel with feedback

On the capacity of the general trapdoor channel with feedback On the caacity of the general tradoor channel with feedback Jui Wu and Achilleas Anastasooulos Electrical Engineering and Comuter Science Deartment University of Michigan Ann Arbor, MI, 48109-1 email:

More information

Paper C Exact Volume Balance Versus Exact Mass Balance in Compositional Reservoir Simulation

Paper C Exact Volume Balance Versus Exact Mass Balance in Compositional Reservoir Simulation Paer C Exact Volume Balance Versus Exact Mass Balance in Comositional Reservoir Simulation Submitted to Comutational Geosciences, December 2005. Exact Volume Balance Versus Exact Mass Balance in Comositional

More information

Best approximation by linear combinations of characteristic functions of half-spaces

Best approximation by linear combinations of characteristic functions of half-spaces Best aroximation by linear combinations of characteristic functions of half-saces Paul C. Kainen Deartment of Mathematics Georgetown University Washington, D.C. 20057-1233, USA Věra Kůrková Institute of

More information

A direct formulation for sparse PCA using semidefinite programming

A direct formulation for sparse PCA using semidefinite programming A direct formulation for sparse PCA using semidefinite programming A. d Aspremont, L. El Ghaoui, M. Jordan, G. Lanckriet ORFE, Princeton University & EECS, U.C. Berkeley A. d Aspremont, INFORMS, Denver,

More information

ANALYTIC APPROXIMATE SOLUTIONS FOR FLUID-FLOW IN THE PRESENCE OF HEAT AND MASS TRANSFER

ANALYTIC APPROXIMATE SOLUTIONS FOR FLUID-FLOW IN THE PRESENCE OF HEAT AND MASS TRANSFER Kilicman, A., et al.: Analytic Aroximate Solutions for Fluid-Flow in the Presence... THERMAL SCIENCE: Year 8, Vol., Sul.,. S59-S64 S59 ANALYTIC APPROXIMATE SOLUTIONS FOR FLUID-FLOW IN THE PRESENCE OF HEAT

More information

Introduction to MVC. least common denominator of all non-identical-zero minors of all order of G(s). Example: The minor of order 2: 1 2 ( s 1)

Introduction to MVC. least common denominator of all non-identical-zero minors of all order of G(s). Example: The minor of order 2: 1 2 ( s 1) Introduction to MVC Definition---Proerness and strictly roerness A system G(s) is roer if all its elements { gij ( s)} are roer, and strictly roer if all its elements are strictly roer. Definition---Causal

More information

Information collection on a graph

Information collection on a graph Information collection on a grah Ilya O. Ryzhov Warren Powell February 10, 2010 Abstract We derive a knowledge gradient olicy for an otimal learning roblem on a grah, in which we use sequential measurements

More information

Multi-instance Support Vector Machine Based on Convex Combination

Multi-instance Support Vector Machine Based on Convex Combination The Eighth International Symosium on Oerations Research and Its Alications (ISORA 09) Zhangjiajie, China, Setember 20 22, 2009 Coyright 2009 ORSC & APORC,. 48 487 Multi-instance Suort Vector Machine Based

More information

AR PROCESSES AND SOURCES CAN BE RECONSTRUCTED FROM. Radu Balan, Alexander Jourjine, Justinian Rosca. Siemens Corporation Research

AR PROCESSES AND SOURCES CAN BE RECONSTRUCTED FROM. Radu Balan, Alexander Jourjine, Justinian Rosca. Siemens Corporation Research AR PROCESSES AND SOURCES CAN BE RECONSTRUCTED FROM DEGENERATE MIXTURES Radu Balan, Alexander Jourjine, Justinian Rosca Siemens Cororation Research 7 College Road East Princeton, NJ 8 fradu,jourjine,roscag@scr.siemens.com

More information

SPARSE SECOND ORDER CONE PROGRAMMING FORMULATIONS FOR CONVEX OPTIMIZATION PROBLEMS

SPARSE SECOND ORDER CONE PROGRAMMING FORMULATIONS FOR CONVEX OPTIMIZATION PROBLEMS Journal of the Operations Research Society of Japan 2008, Vol. 51, No. 3, 241-264 SPARSE SECOND ORDER CONE PROGRAMMING FORMULATIONS FOR CONVEX OPTIMIZATION PROBLEMS Kazuhiro Kobayashi Sunyoung Kim Masakazu

More information

AN IMPROVED BABY-STEP-GIANT-STEP METHOD FOR CERTAIN ELLIPTIC CURVES. 1. Introduction

AN IMPROVED BABY-STEP-GIANT-STEP METHOD FOR CERTAIN ELLIPTIC CURVES. 1. Introduction J. Al. Math. & Comuting Vol. 20(2006), No. 1-2,. 485-489 AN IMPROVED BABY-STEP-GIANT-STEP METHOD FOR CERTAIN ELLIPTIC CURVES BYEONG-KWEON OH, KIL-CHAN HA AND JANGHEON OH Abstract. In this aer, we slightly

More information

A Simple Weight Decay Can Improve. Abstract. It has been observed in numerical simulations that a weight decay can improve

A Simple Weight Decay Can Improve. Abstract. It has been observed in numerical simulations that a weight decay can improve In Advances in Neural Information Processing Systems 4, J.E. Moody, S.J. Hanson and R.P. Limann, eds. Morgan Kaumann Publishers, San Mateo CA, 1995,. 950{957. A Simle Weight Decay Can Imrove Generalization

More information