Distributed and Secure Computation of Convex Programs over a Network of Connected Processors

Size: px
Start display at page:

Download "Distributed and Secure Computation of Convex Programs over a Network of Connected Processors"

Transcription

1 DCDIS CONFERENCE GUELPH, ONTARIO, CANADA, JULY Disribued and Secure Compuaion of Convex Programs over a Newor of Conneced Processors Michael J. Neely Universiy of Souhern California hp://www-rcf.usc.edu/ mjneely Absrac We consider he fundamenal problem of opimizing a convex funcion subjec o a collecion of convex inequaliy consrains and se consrains. An ieraive algorihm is developed ha solves he problem o wihin any desired accuracy using a newor of disribued processors. The newor is assumed o form a conneced graph. Each processing node of he graph aes charge of a porion of he original consrains and solves a correspondingly less complex problem, passing ey values o neighboring nodes. The consrains can be assigned o nodes arbirarily, and individual nodes do no require nowledge of he newor opology or he consrains assigned o oher nodes. Furher, we assume ha each node has a se of privae opimizaion variables ha paricipae in he global opimizaion problem bu are unnown o oher nodes of he graph. This esablishes a general framewor for compuaional load sharing and secure opimizaion over a newor. Index Terms Disribued Compuing, Opimizaion, Privacy, Newor Securiy I. INTRODUCTION We consider he fundamenal problem of finding he minimum value of a muli-variable convex funcion subjec o a se of convex inequaliy consrains. Such convex programs have numerous applicaions, paricularly in he area of daa newors, and a variey of compuaional algorihms exis for solving hem [1][2][3][4][5]. In his paper, we develop a novel echnique for disribuing he compuaion over a conneced graph of newor processors. Each processing node aes charge of a subse of he original problem consrains, and ieraively solves a simplified problem involving only his subse. Problem parameers are updaed a every ieraion based on message passing beween neighboring nodes. Furher, we assume each processing node has a se of privae opimizaion variables ha paricipae in he global opimizaion problem bu mus be ep hidden from oher nodes of he graph. The resuling algorihm yields a soluion ha is arbirarily close o he global opimal soluion, where proximiy o opimaliy is conrolled by a parameer V ha affecs a radeoff in he required compuaion ime. These resuls conribue o he growing field of disribued compuing. This field has received a considerable amoun of aenion in recen years due o he inheren iner-neworing capabiliies of modern compuers and he numerous compuaionally inensive experimens performed by he scienific communiy. Recen heoreical resuls consider soring, ordering, and averaging over graphs [6] [7], and disribued compuaion of marix eigenvalues is considered in [8]. Relaed wor considers implemenaion of parallel algorihms over mesh newors [9] [10] [11] and disribued compuaion wih asynchronous updaes [12]. Algorihms for solving convex programs over muliple parallel processors have been developed previously in [5] [3] using a primal-dual mehodology, and disribued algorihms for solving linear programs on a newor have been recenly considered in [2] in he case when each linear inequaliy consrain involves only local variables. Disribued nonlinear opimizaion for sochasic newor flow problems is considered in [13] [14] [15]. In his paper, we develop disribued soluions o convex opimizaion problems wih any number of variables and wih general consrain ses. Our conribuions are hreefold: Firs, we develop a disribued algorihm ha operaes over any conneced graph of processors. Inequaliy consrains can be assigned o processors arbirarily o ensure an equiable sharing of sysem resources, and individual processors do no require nowledge of he consrains of oher processors. Second, we consider he issue of secure opimizaion, where a global opimum is aained wihou requiring individual processors o reveal heir privae opimizaion variables. Third, we presen our resuls in erms of Lyapunov drif heory, which deviaes significanly from he radiional primal-dual approach o convex opimizaion and simplifies he analysis. The ouline of his paper is as follows. In he nex secion we inroduce he opimizaion problem, and in Secion III we presen he disribued opimizaion algorihm. In Secion IV we inroduce Lyapunov drif heory and prove he performance bounds.

2 DCDIS CONFERENCE GUELPH, ONTARIO, CANADA, JULY II. PROBLEM FORMULATION Consider an undireced graph wih K nodes and L lins, where nodes represen processors and lins represen iner-processor communicaion channels. We assume he graph is conneced, so ha here is a pah from any node o any oher node. For each node {1,..., K}, le N represen he se of neighboring nodes, ha is, N consiss of all nodes j such ha here is a lin beween nodes and j. We desire o use his graph of processors o compue he soluion o he following convex opimizaion problem: Problem A: Minimize: K =1 g ( x, p ) Subjec o: f ( x, p ) b for {1,..., K} x Ω Ω 1... Ω K (1) p Θ for {1,..., K} (2) where x is a vecor of public variables in R M for some ineger M, p is a vecor of privae variables in R M for some ineger M for each, g ( ), f ( ) are convex funcions of heir muli-variable argumens (where we define a vecor valued funcion o be convex if each componen funcion is convex), Ω, Ω 1,..., Ω K are convex subses of R M, and Θ 1,..., Θ K are convex subses of R M1,..., R MK, respecively. To preven infinie soluions o he above opimizaion, we assume ha all ses are compac and all funcions are bounded. Furher, we mae he following non-negaiviy assumpions: We assume ha b > 0 and g ( ) 0, f ( ) 0 for all {1,..., K} (where he inequaliies are aen enrywise), and ha he se Ω resrics he public variables x o having non-negaive enries. I is no difficul o show ha general convex opimizaion problems wih bounded funcions over compac ses can be wrien as opimizaions ha conform o he nonnegaiviy assumpions. 1 Noe ha we have defined K ses of inequaliies so ha we can assign each se o a paricular processor. In he case when here are no privae opimizaion variables, he paricular assignmen of inequaliies o processors is arbirary. However, he privae opimizaion variables p and heir consrain ses Θ represen decision variables and consrains ha paricipae in he global opimizaion, bu which mus be ep hidden from all oher processors. Such a problem arises, for example, when he privae variables represen prices or consumpion levels ha a paricular individual does no wish o reveal. We 1 Indeed, all ha is required o modify he general problem o mee he non-negaiviy assumpions is o add a sufficienly large posiive consan o boh sides of every inequaliy and mae an appropriae change of variables. now ransform he opimizaion problem ino a form ha is more conducive o disribued implemenaion. We firs designae node 1 as he roo node of he graph, and form he shores pah ree from all nodes o his roo node 1. Specifically, each node i 2 is assigned a paren node P ar(i) from is se of neighbors, and he sequence of successive parens of a given node i erminaes a he roo node 1 and forms a shores hop pah from node i o node 1. Such rees always exis in conneced graphs, and simple disribued algorihms for consrucing hem are given in [16]. We le Child(i) represen he se of all children nodes of a given node i, ha is, Child(i) is he se of all nodes j such ha P ar(j) = i. Problem B: Minimize: K =1 g ( x, p ) Subjec o: f ( x, p ) b for {1,..., K} (3) x Ω Ω for {1,..., K} p Θ for {1,..., K} x x P ar() for {2,..., K} (4) x x P ar() for {2,..., K} (5) The consrains (4) and (5) imply ha children and parens have he same x values. Because he graph is conneced, his implies ha x 1 = x 2 =... = x K. We hus have he following simple lemma, he proof of which is sraighforward and omied for breviy. Lemma 1: The vecor ( x, p 1,..., p K ) is an opimal soluion o Problem A if and only if ( x,..., x, p 1,..., p K ) is an opimal soluion o Problem B. A. The Inerior Poin Assumpion To faciliae analysis, i is useful o assume ha here exiss a vecor ( x, p 1,..., p K ) saisfying he se consrains (1) and (2), and such ha f ( x, p ) < b for all. Tha is, here exiss a poin ha saisfies all inequaliies wih sric inequaliy (noe ha we do no require he opimal poin o have his propery). We define ɛ max as he maximum value of ɛ such ha here exiss a vecor ( x, p 1,..., p K ) saisfying (1) and (2) and addiionally saisfying f ( x, p ) b ɛ for all (where ɛ is a vecor wih all enries equal o ɛ). I is no difficul o show ha, given he exisence of a posiive value of ɛ max, here mus exis a sequence of vecors x (ɛ), p (ɛ) parameerized by posiive values ɛ ɛ max ha saisfy he se consrains (1) and (2) and such ha: f ( x (ɛ), p (ɛ) ) b ɛ for 0 < ɛ ɛ max (6) while x (ɛ) x, p (ɛ) p as ɛ 0, where x and p represen he opimal soluion vecors for Problem A.

3 DCDIS CONFERENCE GUELPH, ONTARIO, CANADA, JULY III. DISTRIBUTED AND SECURE OPTIMIZATION We now presen a disribued algorihm for compuing a soluion ha is arbirarily close o he opimal soluion of Problem B (and hence, Problem A). In paricular, each newor node aes charge of he inequaliy consrains f ( x, p ) b and he se consrains x Ω Ω and p Θ. A sequence of vecor values { x [0], x [1],..., x []} and { p [0],..., p []} is compued over ieraions, he average of which approaches he desired soluion. We noe ha he vecor x [] can be viewed as he esimae of he public variables a node a ime. To define he algorihm, le V > 0 be a conrol parameer ha affecs algorihm performance, and define he sequence δ[] =1/ 1 + for {0, 1,...}. Furhermore, we define violaion sequences U [], Y [], and Z [] as follows: Le U [0] = 0 for {1,..., K}, and le Y [0] = Z [0] = 0 for {2,..., K}. On every ieraion we updae he U [] sequences for each node {1,..., K} as follows: U [ + 1] = max[ U [] b, 0] + f ( x [], p []) (7) Liewise, for nodes {2,..., K} we have: Y [ + 1] = max[ Y [] x [] δ[], 0] + x P ar() [] (8) Z [ + 1] = max[ Z [] x P ar() [] δ[], 0] + x [] (9) where he values of p [] and x [] are compued as defined below, and where δ[] represens a vecor wih all enries equal o δ[]. The U [], Y [], and Z [] vecors are analogous o a sequence of slac variables in a dual soluion o he convex opimizaion problem of ineres, bu can inuiively be viewed as queue baclogs in a sloed queueing sysem wih arrivals and deparures deermined by he conrol decision variables x [], p []. Indeed, our algorihm below is inspired by he sable queue conrol policies of [17] [14] [18] [15]. Specifically, if he U [], Y [], and Z [] queue baclogs are ep bounded, hen i mus be he case ha he ime average inpu rae o each queue is less han or equal o he ime average service rae, so ha inequaliy consrains (3), (4), and (5) are saisfied. We noe ha he posiive δ[] sequence is defined o allow he server rae of he Y [] and Z [] queues of (8) and (9) o be slighly larger han he inpu rae o allow for sabiliy, alhough his margin decreases o zero wih increased ieraions. For he following ieraive algorihm, i is useful o define he vecor H [] for each node 2 as follows: H [] = Y [] Z [] (10) The Ieraive Algorihm: On ieraion, every node 2 ransmis is H [] vecor o is paren. Each node {1,..., K} hen compues x [] and p [] as soluions o he following opimizaion: Minimize: V g ( x, p ) + 2U [] f ( x, p ) ( 2 x H [] ) j Child() H j [] Subjec o: p Θ x Ω Ω where he vecor muliplicaion represens he sandard do produc (i.e., a sum of he producs of each enry of he wo vecors being muliplied). Each node hen ransmis is vecor x [] o all of is children (defined as he se Child()). Noe ha Child() is defined as he empy se if a node has no children. The violaion sequences U [], Y [], Z [] are hen updaed according o (7)-(9). Noe ha each node only requires nowledge of is own consrains and consrain ses, and ha he privae variables p [] are nown only o heir corresponding nodes. I is no difficul o show ha hese privae variables canno be inferred by oher nodes if hese nodes do no have exac nowledge of he individual f ( x, p ) funcions. Indeed, if a paricular node j replaces f j ( x j, p j ) and Θ j by a new funcion f j ( x j, pj 2 ) and a new consrain se 2Θ j, he resuling message passing beween neighbors will be exacly he same, alhough he magniudes of he privae variables p j [] will be doubled. I is ineresing o noe ha he above ieraive algorihm is similar o a classical subgradien search algorihm on he dual opimizaion of Problem B (see, for example, [1]), where he sepsize is normalized o 1 uni and he cos funcion is scaled by he conrol parameer V. However, he algorihm is inspired by minimizing he drif of a quadraic Lyapunov funcion of he violaion sequences U [], Y [], Z [], raher han by he classical primal-dual mehodology. One advanage of he Lyapunov approach is ha i yields a sequence of improving soluion esimaes obained by ime averages of he x [], p [] variables, and does no require a global evaluaion of he cos funcion. Specifically, we define empirical averages of he x [] and p [] sequences as follows: x av [] = 1 1 x [τ], p av [] = 1 1 p [τ] (11) Define ( x, p 1,..., p K ) as he opimal soluion vecor of Problem A, and define g as he corresponding opimal cos. Le G max represen he maximum value of K =1 g ( x, p ) over he feasible vecors saisfying he consrains of Problem A. Furher define F max as

4 DCDIS CONFERENCE GUELPH, ONTARIO, CANADA, JULY he maximum value of f ( x, p ) 2 over all {1,..., K} and over all vecors ( x, p ) saisfying x Ω Ω, p Θ. Define B max= max b 2. Theorem 1: (Algorihm Performance) If he opimizaion Problem A saisfies he inerior poin assumpion of Secion II-A, hen for all ieraions, he above ieraive algorihm yields vecors ( x av saisfies: K =1 g ( x av []) g + C V []) wih a cos ha (12) Furher, x av [] Ω Ω and p av [] Θ for all and all, and he following inequaliy consrains are saisfied: f ( x av []) C+V KGmax b + 2ɛ max + ɛmax (13) x av [] xav P ar() [] 2 + (C+V KGmax) +1 2 (14) where C is defined: ( ) C=K 2 max 1 + x max x Ω x Ω x 2 + B max + F max Hence, a value of V can be seleced so ha he vecors x av [] have a resuling cos ha is arbirarily close o he minimizing cos g. The algorihm can be run for a number of ieraions unil he consrains (13) and (14) are arbirarily close o he consrains (3)-(5). The proof of his heorem is provided in he nex secion. A. Discussion Noe ha he above algorihm is inherenly disribued, where each processor communicaes only wih is neighboring processor, as specified by he underlying graph srucure. Indeed, each node mainains is own esimae x [] of he public variables x, and he se consrain x Ω Ω 1... Ω K is enforced locally by resricing each esimae x [] o he local se consrain Ω Ω wihou requiring nowledge of he oher se consrains. The algorihm yields a sequence of soluions x av [] wih improved error bounds a each imesep. This is an imporan feaure for disribued applicaions, and a significan deparure from he primaldual opimizaion resuls of [1] which involve mainaining a running bes soluion for he global problem as compuaions proceed. Evaluaing he bes soluion compued so far is no always possible in disribued seings, as i involves complee nowledge of he global problem and usually requires all processors o share all of heir sae variables and consrain ses wih all oher processors. In our algorihm, each node passes vecors H [] only o neighboring nodes, and his informaion is sufficien o ensure ha local esimaes of he public variables ge progressively closer and closer o saisfying he global consrains. We noe ha he inerior poin assumpion leads o an inequaliy consrain (13) ha converges o (3) lie O(1/). In he case when no inerior poin exiss, he updae equaion (7) can be modified by he sequence δ[] as in (8) and (9). However, his would yield a convergence of O(1/ ). IV. PERFORMANCE ANALYSIS To prove he heorem, we firs presen a fundamenal resul concerning Lyapunov drif. Lyapunov drif heory has been useful in developing sable conrol policies for queueing sysems [17] [19] [20] [21] [14], and he heory has recenly been exended o allow for performance opimizaion of sochasic newors [15] [18]. Here, we consider a deerminisic varian of he Lyapunov resul in [15] applied o he violaion sequences of he previous secion. Specifically, le U[] represen a vecor sequence of non-negaive variables indexed by ime {0, 1, 2,...}. Define he Lyapunov funcion L( U[]) = U[] 2. Here we use he Euclidean norm, so ha U 2 represens he sum of squares of he individual enries of vecor U. Define he single-sep Lyapunov drif as follows: ( U[]) = L( U[ + 1]) L( U[]) A any ime, he U[] vecor can be viewed as he curren sae of a dynamic sysem. Le x[], p[] be vecor sequences represening conrol decision variables ha effec he evoluion of U[]. Le g( x, p) be a nonnegaive cos funcion, assumed o be convex in he composie vecor ( x, p). We assume he cos funcion is bounded and define G max as an upper bound on he maximum cos over all possible vecors x and p. Consider any paricular arge vecors x, p ha yield a desired cos. Theorem 2: (Opimizaion via Lyapunov Drif) If U[0] = 0 and if here are posiive consans V, ɛ, C such ha for all imeslos and all sequences U[] we have: ( U[]) C ɛ U[] + V g( x, p ) V g( x[], p[]) hen for all {1, 2,...} we have: (a) g( x av [], p av []) g( x, p ) + C/V (b) U[] C+V Gmax ɛ + ɛ 2 where x av [] and p av [] are empirical averages of he conrol variables, defined as in (11). Proof: To prove par (a), noe ha he drif condiion of he heorem ogeher wih non-negaiviy of he U[] values imply ha for all : ( U[]) + V g( x[], p[]) C + V g( x, p )

5 DCDIS CONFERENCE GUELPH, ONTARIO, CANADA, JULY Summing over all imes τ dividing by, we have: L( U[]) {0,..., 1} and + V 1 1 g( x[τ], p[τ]) C + V g( x, p ) (15) Because g( x, p) is convex in ( x, p), by Jensen s inequaliy we have 1 1 g( x[τ], p[τ]) g( xav [], p av []). Using his bound ogeher wih non-negaiviy of he Lyapunov funcion in he inequaliy (15) yields par (a) of he heorem. To prove (b), noe ha he drif condiion implies: ( U[]) C + V G max ɛ U[] Because he U[] vecors have non-negaive enries, i follows ha he maximum incremen in L( U[]) is C + V G max. Furher, L( U[]) canno increase if ɛ U [] C + V G max. Therefore, we have for all imes : L( U[]) L + C + V G max (16) where he value of L is he maximum value of U 2 subjec o ɛ U C+V G max. The maximum is achieved when all weigh is placed on a single enry of U, and hence L = (C + V G max ) 2 /ɛ 2. I follows from (16) (C+V G max) 2 ha U[] ɛ + C + V G 2 max. I is no A difficul o show ha 2 ɛ + A A 2 ɛ + ɛ 2 for all posiive values A, ɛ, and he resul of par (b) follows. A. Proof of Theorem 1 We now compue he Lyapunov drif associaed wih he ieraive algorihm of he previous secion. Firs, le U[], Y [], and Z[] represen composie vecors consising of concaenaed U [], Y [], and Z [] vecors for {1,..., K}. We define he Lyapunov funcion L( U, Y, Z) = U 2 + Y 2 + Z 2. The one-sep Lyapunov drif is hus: ( U[], Y [], Z[]) =L( U[ + 1], Y [ + 1], Z[ + 1]) L( U[], Y [], Z[])) (17) Consider he updae equaions (7)-(9) for he violaion sequences. Taing he squared norm of boh sides of (7) yields: U [ + 1] 2 U [] 2 + b 2 + F max 2U [] ( b f ( x [], p [])) Liewise, aing he squared norm of boh sides of (8) yields: Y [ + 1] 2 Y [] 2 + x [] + δ[] 2 + x P ar() [] 2 2 Y [] ( x [] + δ[] x P ar() []) Similarly, squaring (9) yields: Z [ + 1] 2 Z [] 2 + x P ar() [] + δ[] 2 + x [] 2 2 Z [] ( x P ar() [] + δ[] x []) Using he abbreviaed noaion o represen he Lyapunov drif defined in (17), i follows ha he drif saisfies: C 2 U [] b f ( x [], p [])) 2 ( Y [] x [] + ) δ[] x P ar() [] 2 ( Z [] x P ar() [] + ) δ[] x [] +V g ( x [], p []) V g ( x [], p []) (18) where C is defined in Theorem 1, and where we have added and subraced he opimizaion meric. By shifing he sums, using he definiion H [] = Y [] Z [], and recalling ha Child() is he se of all nodes j such ha P ar(j) =, we have: C 2 U [] ( b f ( x [], p [])) 2 x [] ( H [] j Child() H j []) 2 δ[] ( Y [] + Z []) +V g ( x [], p []) V g ( x [], p []) (19) The righ hand sides of (18) and (19) are idenical. However, from he laer expression i is clear ha he ieraive algorihm defined in he previous secion is precisely designed o minimize he sum of he second, hird, fourh, and fifh erms on he righ hand side of he above inequaliy (19) over all feasible conrol vecors x, p. Hence, he drif is less han or equal o he resuling righ hand side if a paricular se of feasible conrol vecors are plugged ino he second, hird, fourh, and fifh erms. Now recall from he inerior poin assumpion of Secion II-A ha feasible vecors x (ɛ), p (ɛ) exis and saisfy f ( x (ɛ), p (ɛ) ) b ɛ for all and for all ɛ such ha 0 < ɛ ɛ max. Hence, using he righ hand side as expressed in (18), we have: C 2 U [] ɛ 2 δ[] ( Y [] + Z []) +V g ( x (ɛ), p (ɛ) ) V g ( x [], p []) The above drif expression is in he form specified by he Lyapunov drif heorem (Theorem 2). Hence, we have he following for all imes : g ( x av []) g ( x (ɛ), p (ɛ) ) + C/V (20) U[] C+V KGmax 2ɛ + ɛ (21) Y [] + Z[] C+V KGmax 2δ[] + δ[] (22)

6 DCDIS CONFERENCE GUELPH, ONTARIO, CANADA, JULY where we have used he fac ha δ[] decreases wih, and where he norm of he composie vecors U[] and Y [] + Z[] is given by: U[] = K U [] 2 Y [] + Z[] = =1 K Y [] + Z [] 2 =2 The inequaliies (20)-(22) hold for any value ɛ such ha 0 < ɛ ɛ max. Hence, hey can be opimized separaely by choosing he bes ɛ value. Recall ha x (ɛ) and p (ɛ) converge o he opimal operaing poin as ɛ 0. Hence, aing a limi in (20) as ɛ 0 proves he performance bound (12). Conversely, he bound in (21) is minimized by seing ɛ = ɛ max, proving ha: U[] C + V KG max 2ɛ max + ɛ max (23) Recall ha he U [] values represen queue baclogs (see (7)). I follows ha he accumulaed arrivals o he queue during he firs slos are less han or equal o he maximum possible deparures plus he baclog a ime 1: 1 f ( x [τ], p [τ]) b + U [ 1] Dividing he above inequaliy by and using Jensen s inequaliy o push he ime average summaion inside he convex funcion f ( ) yields: f ( x av []) U b + [ 1] (24) Combining (24) and (23) proves he performance bound (13). Similarly, he performance bound (14) can be proven direcly from (22), using he fac ha 1 δ[τ] 2/ 1/ for 1, as well as he 1 fac ha he accumulaed arrivals o he Y [] queues over he firs slos are bounded by he oal service opporuniies over his inerval plus he baclog a ime 1. This proves Theorem 1. V. CONCLUSIONS We have developed a framewor for disribued and secure compuaion of convex programs using an arbirary conneced graph of processors. Each node of he graph paricipaes in he opimizaion by solving a simplified problem involving boh public and privae variables, and he resuling algorihm mainains privacy while achieving global opimaliy. Our analysis uses a Lyapunov drif echnique ha ransforms he opimizaion ino a corresponding queue conrol sraegy. The echnique is quie general and can be exended o rea sochasic versions of his problem. REFERENCES [1] D. P. Berseas, A. Nedic, and A. E. Ozdaglar. Convex Analysis and Opimizaion. Boson: Ahena Scienific, [2] Y. Baral, J. W. Byers, and D. Raz. Fas, disribued approximaion algorihms for posiive linear programming wih applicaions o flow conrol. Siam Journal of Compuing, vol. 33, no. 6, pp , [3] D. P. Berseas and P. Tseng. Parial proximal minimizaion algorihms for convex programming. Massachuses Insiue of Technology Technical Repor, [4] D. P. Berseas and J. N. Tsisilis. Parallel and Disribued Compuaion: Numerical Mehods. Prenice-Hall, Englewood Cliffs, NJ, [5] M. C. Ferris and O. L. Mangasarian. Parallel consrain disribuion. SIAM Journal on Opimizaion, vol. 1, pp , [6] D. Kempe, A. Dobra, and J. Gehre. Gossip-based compuaion of aggregae informaion. Proceedings of FOCS, [7] J. K. Bordim, K. Naano, and H. Shen. Soring on singlechannel wireless sensor newors. Inernaional Symposium on Parallel Archiecures, Algorihms, and Newors, May [8] D. Kempe and F. McSherry. A decenralized algorihm for specral analysis. Proc. of STOC, [9] M. Singh, V. K. Prasanna, and J. D. P. Rolim. Collaboraive and disribued compuaion in mesh-lie wireless sensor arrays. Personal Wireless Communicaions, Sep [10] R. Miller and Q. F. Sou. Parallel algorihms for regular archiecures: Meshes and pyramids. MIT Press, [11] D. Nassimi and S. Sahni. Bionic sor on mesh-conneced compuers. IEEE Trans. on Compuers, vol. c-27, Jan [12] J. N. Tsisilis, D. P. Berseas, and M. Ahens. Disribued asynchronous deerminisic and sochasic gradien opimizaion algorihms. IEEE Transacions on Auomaic Conrol, Vol. AC- 31, no. 9, Sepember [13] M. J. Neely. Dynamic Power Allocaion and Rouing for Saellie and Wireless Newors wih Time Varying Channels. PhD hesis, Massachuses Insiue of Technology, LIDS, [14] M. J. Neely, E. Modiano, and C. E Rohrs. Dynamic power allocaion and rouing for ime varying wireless newors. IEEE Journal on Seleced Areas in Communicaions, January [15] M. J. Neely, E. Modiano, and C. Li. Fairness and opimal sochasic conrol for heerogeneous newors. Proceedings of IEEE INFOCOM, March [16] D. P. Berseas and R. Gallager. Daa Newors. New Jersey: Prenice-Hall, Inc., [17] L. Tassiulas and A. Ephremides. Sabiliy properies of consrained queueing sysems and scheduling policies for maximum hroughpu in mulihop radio newors. IEEE Transacaions on Auomaic Conrol, Vol. 37, no. 12, Dec [18] M. J. Neely. Energy opimal conrol for ime varying wireless newors. Proceedings of IEEE INFOCOM, March [19] L. Tassiulas and A. Ephremides. Dynamic server allocaion o parallel queues wih randomly varying conneciviy. IEEE Trans. on Inform. Theory, vol. 39, pp , March [20] N. McKeown, V. Ananharam, and J. Walrand. Achieving 100% hroughpu in an inpu-queued swich. Proc. INFOCOM, [21] E. Leonardi, M. Melia, F. Neri, and M. Ajmone Marson. Bounds on average delays and queue size averages and variances in inpu-queued cell-based swiches. Proc. INFOCOM, 2001.

Multiprocessor Systems-on-Chips

Multiprocessor Systems-on-Chips Par of: Muliprocessor Sysems-on-Chips Edied by: Ahmed Amine Jerraya and Wayne Wolf Morgan Kaufmann Publishers, 2005 2 Modeling Shared Resources Conex swiching implies overhead. On a processing elemen,

More information

The Transport Equation

The Transport Equation The Transpor Equaion Consider a fluid, flowing wih velociy, V, in a hin sraigh ube whose cross secion will be denoed by A. Suppose he fluid conains a conaminan whose concenraion a posiion a ime will be

More information

Single-machine Scheduling with Periodic Maintenance and both Preemptive and. Non-preemptive jobs in Remanufacturing System 1

Single-machine Scheduling with Periodic Maintenance and both Preemptive and. Non-preemptive jobs in Remanufacturing System 1 Absrac number: 05-0407 Single-machine Scheduling wih Periodic Mainenance and boh Preempive and Non-preempive jobs in Remanufacuring Sysem Liu Biyu hen Weida (School of Economics and Managemen Souheas Universiy

More information

The Application of Multi Shifts and Break Windows in Employees Scheduling

The Application of Multi Shifts and Break Windows in Employees Scheduling The Applicaion of Muli Shifs and Brea Windows in Employees Scheduling Evy Herowai Indusrial Engineering Deparmen, Universiy of Surabaya, Indonesia Absrac. One mehod for increasing company s performance

More information

ANALYSIS AND COMPARISONS OF SOME SOLUTION CONCEPTS FOR STOCHASTIC PROGRAMMING PROBLEMS

ANALYSIS AND COMPARISONS OF SOME SOLUTION CONCEPTS FOR STOCHASTIC PROGRAMMING PROBLEMS ANALYSIS AND COMPARISONS OF SOME SOLUTION CONCEPTS FOR STOCHASTIC PROGRAMMING PROBLEMS R. Caballero, E. Cerdá, M. M. Muñoz and L. Rey () Deparmen of Applied Economics (Mahemaics), Universiy of Málaga,

More information

Map Task Scheduling in MapReduce with Data Locality: Throughput and Heavy-Traffic Optimality

Map Task Scheduling in MapReduce with Data Locality: Throughput and Heavy-Traffic Optimality Map Task Scheduling in MapReduce wih Daa Localiy: Throughpu and Heavy-Traffic Opimaliy Weina Wang, Kai Zhu and Lei Ying Elecrical, Compuer and Energy Engineering Arizona Sae Universiy Tempe, Arizona 85287

More information

Option Put-Call Parity Relations When the Underlying Security Pays Dividends

Option Put-Call Parity Relations When the Underlying Security Pays Dividends Inernaional Journal of Business and conomics, 26, Vol. 5, No. 3, 225-23 Opion Pu-all Pariy Relaions When he Underlying Securiy Pays Dividends Weiyu Guo Deparmen of Finance, Universiy of Nebraska Omaha,

More information

17 Laplace transform. Solving linear ODE with piecewise continuous right hand sides

17 Laplace transform. Solving linear ODE with piecewise continuous right hand sides 7 Laplace ransform. Solving linear ODE wih piecewise coninuous righ hand sides In his lecure I will show how o apply he Laplace ransform o he ODE Ly = f wih piecewise coninuous f. Definiion. A funcion

More information

Duration and Convexity ( ) 20 = Bond B has a maturity of 5 years and also has a required rate of return of 10%. Its price is $613.

Duration and Convexity ( ) 20 = Bond B has a maturity of 5 years and also has a required rate of return of 10%. Its price is $613. Graduae School of Business Adminisraion Universiy of Virginia UVA-F-38 Duraion and Convexiy he price of a bond is a funcion of he promised paymens and he marke required rae of reurn. Since he promised

More information

Efficient One-time Signature Schemes for Stream Authentication *

Efficient One-time Signature Schemes for Stream Authentication * JOURNAL OF INFORMATION SCIENCE AND ENGINEERING, 611-64 (006) Efficien One-ime Signaure Schemes for Sream Auhenicaion * YONGSU PARK AND YOOKUN CHO + College of Informaion and Communicaions Hanyang Universiy

More information

Network Discovery: An Estimation Based Approach

Network Discovery: An Estimation Based Approach Nework Discovery: An Esimaion Based Approach Girish Chowdhary, Magnus Egersed, and Eric N. Johnson Absrac We consider he unaddressed problem of nework discovery, in which, an agen aemps o formulae an esimae

More information

Stochastic Optimal Control Problem for Life Insurance

Stochastic Optimal Control Problem for Life Insurance Sochasic Opimal Conrol Problem for Life Insurance s. Basukh 1, D. Nyamsuren 2 1 Deparmen of Economics and Economerics, Insiue of Finance and Economics, Ulaanbaaar, Mongolia 2 School of Mahemaics, Mongolian

More information

Journal Of Business & Economics Research September 2005 Volume 3, Number 9

Journal Of Business & Economics Research September 2005 Volume 3, Number 9 Opion Pricing And Mone Carlo Simulaions George M. Jabbour, (Email: jabbour@gwu.edu), George Washingon Universiy Yi-Kang Liu, (yikang@gwu.edu), George Washingon Universiy ABSTRACT The advanage of Mone Carlo

More information

Real-time Particle Filters

Real-time Particle Filters Real-ime Paricle Filers Cody Kwok Dieer Fox Marina Meilă Dep. of Compuer Science & Engineering, Dep. of Saisics Universiy of Washingon Seale, WA 9895 ckwok,fox @cs.washingon.edu, mmp@sa.washingon.edu Absrac

More information

DETERMINISTIC INVENTORY MODEL FOR ITEMS WITH TIME VARYING DEMAND, WEIBULL DISTRIBUTION DETERIORATION AND SHORTAGES KUN-SHAN WU

DETERMINISTIC INVENTORY MODEL FOR ITEMS WITH TIME VARYING DEMAND, WEIBULL DISTRIBUTION DETERIORATION AND SHORTAGES KUN-SHAN WU Yugoslav Journal of Operaions Research 2 (22), Number, 6-7 DEERMINISIC INVENORY MODEL FOR IEMS WIH IME VARYING DEMAND, WEIBULL DISRIBUION DEERIORAION AND SHORAGES KUN-SHAN WU Deparmen of Bussines Adminisraion

More information

Inventory Planning with Forecast Updates: Approximate Solutions and Cost Error Bounds

Inventory Planning with Forecast Updates: Approximate Solutions and Cost Error Bounds OPERATIONS RESEARCH Vol. 54, No. 6, November December 2006, pp. 1079 1097 issn 0030-364X eissn 1526-5463 06 5406 1079 informs doi 10.1287/opre.1060.0338 2006 INFORMS Invenory Planning wih Forecas Updaes:

More information

Optimal Power Cost Management Using Stored Energy in Data Centers

Optimal Power Cost Management Using Stored Energy in Data Centers Opimal Power Cos Managemen Using Sored Energy in Daa Ceners Rahul Urgaonkar, Bhuvan Urgaonkar, Michael J. Neely, Anand Sivasubramanian Advanced Neworking Dep., Dep. of CSE, Dep. of EE Rayheon BBN Technologies,

More information

Task is a schedulable entity, i.e., a thread

Task is a schedulable entity, i.e., a thread Real-Time Scheduling Sysem Model Task is a schedulable eniy, i.e., a hread Time consrains of periodic ask T: - s: saring poin - e: processing ime of T - d: deadline of T - p: period of T Periodic ask T

More information

DYNAMIC MODELS FOR VALUATION OF WRONGFUL DEATH PAYMENTS

DYNAMIC MODELS FOR VALUATION OF WRONGFUL DEATH PAYMENTS DYNAMIC MODELS FOR VALUATION OF WRONGFUL DEATH PAYMENTS Hong Mao, Shanghai Second Polyechnic Universiy Krzyszof M. Osaszewski, Illinois Sae Universiy Youyu Zhang, Fudan Universiy ABSTRACT Liigaion, exper

More information

The option pricing framework

The option pricing framework Chaper 2 The opion pricing framework The opion markes based on swap raes or he LIBOR have become he larges fixed income markes, and caps (floors) and swapions are he mos imporan derivaives wihin hese markes.

More information

How To Predict A Person'S Behavior

How To Predict A Person'S Behavior Informaion Theoreic Approaches for Predicive Models: Resuls and Analysis Monica Dinculescu Supervised by Doina Precup Absrac Learning he inernal represenaion of parially observable environmens has proven

More information

Accelerated Gradient Methods for Stochastic Optimization and Online Learning

Accelerated Gradient Methods for Stochastic Optimization and Online Learning Acceleraed Gradien Mehods for Sochasic Opimizaion and Online Learning Chonghai Hu, James T. Kwok, Weike Pan Deparmen of Compuer Science and Engineering Hong Kong Universiy of Science and Technology Clear

More information

Optimal Investment and Consumption Decision of Family with Life Insurance

Optimal Investment and Consumption Decision of Family with Life Insurance Opimal Invesmen and Consumpion Decision of Family wih Life Insurance Minsuk Kwak 1 2 Yong Hyun Shin 3 U Jin Choi 4 6h World Congress of he Bachelier Finance Sociey Torono, Canada June 25, 2010 1 Speaker

More information

Online Learning with Sample Path Constraints

Online Learning with Sample Path Constraints Journal of Machine Learning Research 0 (2009) 569-590 Submied 7/08; Revised /09; Published 3/09 Online Learning wih Sample Pah Consrains Shie Mannor Deparmen of Elecrical and Compuer Engineering McGill

More information

Niche Market or Mass Market?

Niche Market or Mass Market? Niche Marke or Mass Marke? Maxim Ivanov y McMaser Universiy July 2009 Absrac The de niion of a niche or a mass marke is based on he ranking of wo variables: he monopoly price and he produc mean value.

More information

CPU Provisioning Algorithms for Service Differentiation in Cloud-based Environments

CPU Provisioning Algorithms for Service Differentiation in Cloud-based Environments CPU Provisioning Algorihms for Service Differeniaion in Cloud-based Environmens Kosas Kasalis, Georgios S. Paschos, Yannis Viniois, Leandros Tassiulas Absrac This work focuses on he design, analysis and

More information

Why Did the Demand for Cash Decrease Recently in Korea?

Why Did the Demand for Cash Decrease Recently in Korea? Why Did he Demand for Cash Decrease Recenly in Korea? Byoung Hark Yoo Bank of Korea 26. 5 Absrac We explores why cash demand have decreased recenly in Korea. The raio of cash o consumpion fell o 4.7% in

More information

PATHWISE PROPERTIES AND PERFORMANCE BOUNDS FOR A PERISHABLE INVENTORY SYSTEM

PATHWISE PROPERTIES AND PERFORMANCE BOUNDS FOR A PERISHABLE INVENTORY SYSTEM PATHWISE PROPERTIES AND PERFORMANCE BOUNDS FOR A PERISHABLE INVENTORY SYSTEM WILLIAM L. COOPER Deparmen of Mechanical Engineering, Universiy of Minnesoa, 111 Church Sree S.E., Minneapolis, MN 55455 billcoop@me.umn.edu

More information

adaptive control; stochastic systems; certainty equivalence principle; long-term

adaptive control; stochastic systems; certainty equivalence principle; long-term COMMUICATIOS I IFORMATIO AD SYSTEMS c 2006 Inernaional Press Vol. 6, o. 4, pp. 299-320, 2006 003 ADAPTIVE COTROL OF LIEAR TIME IVARIAT SYSTEMS: THE BET O THE BEST PRICIPLE S. BITTATI AD M. C. CAMPI Absrac.

More information

PROFIT TEST MODELLING IN LIFE ASSURANCE USING SPREADSHEETS PART ONE

PROFIT TEST MODELLING IN LIFE ASSURANCE USING SPREADSHEETS PART ONE Profi Tes Modelling in Life Assurance Using Spreadshees PROFIT TEST MODELLING IN LIFE ASSURANCE USING SPREADSHEETS PART ONE Erik Alm Peer Millingon 2004 Profi Tes Modelling in Life Assurance Using Spreadshees

More information

Random Walk in 1-D. 3 possible paths x vs n. -5 For our random walk, we assume the probabilities p,q do not depend on time (n) - stationary

Random Walk in 1-D. 3 possible paths x vs n. -5 For our random walk, we assume the probabilities p,q do not depend on time (n) - stationary Random Walk in -D Random walks appear in many cones: diffusion is a random walk process undersanding buffering, waiing imes, queuing more generally he heory of sochasic processes gambling choosing he bes

More information

MTH6121 Introduction to Mathematical Finance Lesson 5

MTH6121 Introduction to Mathematical Finance Lesson 5 26 MTH6121 Inroducion o Mahemaical Finance Lesson 5 Conens 2.3 Brownian moion wih drif........................... 27 2.4 Geomeric Brownian moion........................... 28 2.5 Convergence of random

More information

STABILITY OF LOAD BALANCING ALGORITHMS IN DYNAMIC ADVERSARIAL SYSTEMS

STABILITY OF LOAD BALANCING ALGORITHMS IN DYNAMIC ADVERSARIAL SYSTEMS STABILITY OF LOAD BALANCING ALGORITHMS IN DYNAMIC ADVERSARIAL SYSTEMS ELLIOT ANSHELEVICH, DAVID KEMPE, AND JON KLEINBERG Absrac. In he dynamic load balancing problem, we seek o keep he job load roughly

More information

Efficient Risk Sharing with Limited Commitment and Hidden Storage

Efficient Risk Sharing with Limited Commitment and Hidden Storage Efficien Risk Sharing wih Limied Commimen and Hidden Sorage Árpád Ábrahám and Sarola Laczó March 30, 2012 Absrac We exend he model of risk sharing wih limied commimen e.g. Kocherlakoa, 1996) by inroducing

More information

Principal components of stock market dynamics. Methodology and applications in brief (to be updated ) Andrei Bouzaev, bouzaev@ya.

Principal components of stock market dynamics. Methodology and applications in brief (to be updated ) Andrei Bouzaev, bouzaev@ya. Principal componens of sock marke dynamics Mehodology and applicaions in brief o be updaed Andrei Bouzaev, bouzaev@ya.ru Why principal componens are needed Objecives undersand he evidence of more han one

More information

Stock Trading with Recurrent Reinforcement Learning (RRL) CS229 Application Project Gabriel Molina, SUID 5055783

Stock Trading with Recurrent Reinforcement Learning (RRL) CS229 Application Project Gabriel Molina, SUID 5055783 Sock raing wih Recurren Reinforcemen Learning (RRL) CS9 Applicaion Projec Gabriel Molina, SUID 555783 I. INRODUCION One relaively new approach o financial raing is o use machine learning algorihms o preic

More information

Dynamic programming models and algorithms for the mutual fund cash balance problem

Dynamic programming models and algorithms for the mutual fund cash balance problem Submied o Managemen Science manuscrip Dynamic programming models and algorihms for he muual fund cash balance problem Juliana Nascimeno Deparmen of Operaions Research and Financial Engineering, Princeon

More information

Morningstar Investor Return

Morningstar Investor Return Morningsar Invesor Reurn Morningsar Mehodology Paper Augus 31, 2010 2010 Morningsar, Inc. All righs reserved. The informaion in his documen is he propery of Morningsar, Inc. Reproducion or ranscripion

More information

Task-Execution Scheduling Schemes for Network Measurement and Monitoring

Task-Execution Scheduling Schemes for Network Measurement and Monitoring Task-Execuion Scheduling Schemes for Nework Measuremen and Monioring Zhen Qin, Robero Rojas-Cessa, and Nirwan Ansari Deparmen of Elecrical and Compuer Engineering New Jersey Insiue of Technology Universiy

More information

Signal Processing and Linear Systems I

Signal Processing and Linear Systems I Sanford Universiy Summer 214-215 Signal Processing and Linear Sysems I Lecure 5: Time Domain Analysis of Coninuous Time Sysems June 3, 215 EE12A:Signal Processing and Linear Sysems I; Summer 14-15, Gibbons

More information

Analogue and Digital Signal Processing. First Term Third Year CS Engineering By Dr Mukhtiar Ali Unar

Analogue and Digital Signal Processing. First Term Third Year CS Engineering By Dr Mukhtiar Ali Unar Analogue and Digial Signal Processing Firs Term Third Year CS Engineering By Dr Mukhiar Ali Unar Recommended Books Haykin S. and Van Veen B.; Signals and Sysems, John Wiley& Sons Inc. ISBN: 0-7-380-7 Ifeachor

More information

Economics Honors Exam 2008 Solutions Question 5

Economics Honors Exam 2008 Solutions Question 5 Economics Honors Exam 2008 Soluions Quesion 5 (a) (2 poins) Oupu can be decomposed as Y = C + I + G. And we can solve for i by subsiuing in equaions given in he quesion, Y = C + I + G = c 0 + c Y D + I

More information

Analysis of Tailored Base-Surge Policies in Dual Sourcing Inventory Systems

Analysis of Tailored Base-Surge Policies in Dual Sourcing Inventory Systems Analysis of Tailored Base-Surge Policies in Dual Sourcing Invenory Sysems Ganesh Janakiraman, 1 Sridhar Seshadri, 2, Anshul Sheopuri. 3 Absrac We sudy a model of a firm managing is invenory of a single

More information

Analysis of Pricing and Efficiency Control Strategy between Internet Retailer and Conventional Retailer

Analysis of Pricing and Efficiency Control Strategy between Internet Retailer and Conventional Retailer Recen Advances in Business Managemen and Markeing Analysis of Pricing and Efficiency Conrol Sraegy beween Inerne Reailer and Convenional Reailer HYUG RAE CHO 1, SUG MOO BAE and JOG HU PARK 3 Deparmen of

More information

Heuristics for dimensioning large-scale MPLS networks

Heuristics for dimensioning large-scale MPLS networks Heurisics for dimensioning large-scale MPLS newors Carlos Borges 1, Amaro de Sousa 1, Rui Valadas 1 Depar. of Elecronics and Telecommunicaions Universiy of Aveiro, Insiue of Telecommunicaions pole of Aveiro

More information

The naive method discussed in Lecture 1 uses the most recent observations to forecast future values. That is, Y ˆ t + 1

The naive method discussed in Lecture 1 uses the most recent observations to forecast future values. That is, Y ˆ t + 1 Business Condiions & Forecasing Exponenial Smoohing LECTURE 2 MOVING AVERAGES AND EXPONENTIAL SMOOTHING OVERVIEW This lecure inroduces ime-series smoohing forecasing mehods. Various models are discussed,

More information

Term Structure of Prices of Asian Options

Term Structure of Prices of Asian Options Term Srucure of Prices of Asian Opions Jirô Akahori, Tsuomu Mikami, Kenji Yasuomi and Teruo Yokoa Dep. of Mahemaical Sciences, Risumeikan Universiy 1-1-1 Nojihigashi, Kusasu, Shiga 525-8577, Japan E-mail:

More information

SELF-EVALUATION FOR VIDEO TRACKING SYSTEMS

SELF-EVALUATION FOR VIDEO TRACKING SYSTEMS SELF-EVALUATION FOR VIDEO TRACKING SYSTEMS Hao Wu and Qinfen Zheng Cenre for Auomaion Research Dep. of Elecrical and Compuer Engineering Universiy of Maryland, College Park, MD-20742 {wh2003, qinfen}@cfar.umd.edu

More information

Making a Faster Cryptanalytic Time-Memory Trade-Off

Making a Faster Cryptanalytic Time-Memory Trade-Off Making a Faser Crypanalyic Time-Memory Trade-Off Philippe Oechslin Laboraoire de Securié e de Crypographie (LASEC) Ecole Polyechnique Fédérale de Lausanne Faculé I&C, 1015 Lausanne, Swizerland philippe.oechslin@epfl.ch

More information

TEMPORAL PATTERN IDENTIFICATION OF TIME SERIES DATA USING PATTERN WAVELETS AND GENETIC ALGORITHMS

TEMPORAL PATTERN IDENTIFICATION OF TIME SERIES DATA USING PATTERN WAVELETS AND GENETIC ALGORITHMS TEMPORAL PATTERN IDENTIFICATION OF TIME SERIES DATA USING PATTERN WAVELETS AND GENETIC ALGORITHMS RICHARD J. POVINELLI AND XIN FENG Deparmen of Elecrical and Compuer Engineering Marquee Universiy, P.O.

More information

The Grantor Retained Annuity Trust (GRAT)

The Grantor Retained Annuity Trust (GRAT) WEALTH ADVISORY Esae Planning Sraegies for closely-held, family businesses The Granor Reained Annuiy Trus (GRAT) An efficien wealh ransfer sraegy, paricularly in a low ineres rae environmen Family business

More information

Constant Data Length Retrieval for Video Servers with Variable Bit Rate Streams

Constant Data Length Retrieval for Video Servers with Variable Bit Rate Streams IEEE Inernaional Conference on Mulimedia Compuing & Sysems, June 17-3, 1996, in Hiroshima, Japan, p. 151-155 Consan Lengh Rerieval for Video Servers wih Variable Bi Rae Sreams Erns Biersack, Frédéric Thiesse,

More information

USE OF EDUCATION TECHNOLOGY IN ENGLISH CLASSES

USE OF EDUCATION TECHNOLOGY IN ENGLISH CLASSES USE OF EDUCATION TECHNOLOGY IN ENGLISH CLASSES Mehme Nuri GÖMLEKSİZ Absrac Using educaion echnology in classes helps eachers realize a beer and more effecive learning. In his sudy 150 English eachers were

More information

Appendix A: Area. 1 Find the radius of a circle that has circumference 12 inches.

Appendix A: Area. 1 Find the radius of a circle that has circumference 12 inches. Appendi A: Area worked-ou s o Odd-Numbered Eercises Do no read hese worked-ou s before aemping o do he eercises ourself. Oherwise ou ma mimic he echniques shown here wihou undersanding he ideas. Bes wa

More information

A Note on Using the Svensson procedure to estimate the risk free rate in corporate valuation

A Note on Using the Svensson procedure to estimate the risk free rate in corporate valuation A Noe on Using he Svensson procedure o esimae he risk free rae in corporae valuaion By Sven Arnold, Alexander Lahmann and Bernhard Schwezler Ocober 2011 1. The risk free ineres rae in corporae valuaion

More information

On the degrees of irreducible factors of higher order Bernoulli polynomials

On the degrees of irreducible factors of higher order Bernoulli polynomials ACTA ARITHMETICA LXII.4 (1992 On he degrees of irreducible facors of higher order Bernoulli polynomials by Arnold Adelberg (Grinnell, Ia. 1. Inroducion. In his paper, we generalize he curren resuls on

More information

INTEREST RATE FUTURES AND THEIR OPTIONS: SOME PRICING APPROACHES

INTEREST RATE FUTURES AND THEIR OPTIONS: SOME PRICING APPROACHES INTEREST RATE FUTURES AND THEIR OPTIONS: SOME PRICING APPROACHES OPENGAMMA QUANTITATIVE RESEARCH Absrac. Exchange-raded ineres rae fuures and heir opions are described. The fuure opions include hose paying

More information

Simultaneous Perturbation Stochastic Approximation in Decentralized Load Balancing Problem

Simultaneous Perturbation Stochastic Approximation in Decentralized Load Balancing Problem Preprins, 1s IFAC Conference on Modelling, Idenificaion and Conrol of Nonlinear Sysems June 24-26, 2015. Sain Peersburg, Russia Simulaneous Perurbaion Sochasic Approximaion in Decenralized Load Balancing

More information

Large Scale Online Learning.

Large Scale Online Learning. Large Scale Online Learning. Léon Boou NEC Labs America Princeon NJ 08540 leon@boou.org Yann Le Cun NEC Labs America Princeon NJ 08540 yann@lecun.com Absrac We consider siuaions where raining daa is abundan

More information

Chapter 8: Regression with Lagged Explanatory Variables

Chapter 8: Regression with Lagged Explanatory Variables Chaper 8: Regression wih Lagged Explanaory Variables Time series daa: Y for =1,..,T End goal: Regression model relaing a dependen variable o explanaory variables. Wih ime series new issues arise: 1. One

More information

Chapter 7. Response of First-Order RL and RC Circuits

Chapter 7. Response of First-Order RL and RC Circuits Chaper 7. esponse of Firs-Order L and C Circuis 7.1. The Naural esponse of an L Circui 7.2. The Naural esponse of an C Circui 7.3. The ep esponse of L and C Circuis 7.4. A General oluion for ep and Naural

More information

Keldysh Formalism: Non-equilibrium Green s Function

Keldysh Formalism: Non-equilibrium Green s Function Keldysh Formalism: Non-equilibrium Green s Funcion Jinshan Wu Deparmen of Physics & Asronomy, Universiy of Briish Columbia, Vancouver, B.C. Canada, V6T 1Z1 (Daed: November 28, 2005) A review of Non-equilibrium

More information

Relationships between Stock Prices and Accounting Information: A Review of the Residual Income and Ohlson Models. Scott Pirie* and Malcolm Smith**

Relationships between Stock Prices and Accounting Information: A Review of the Residual Income and Ohlson Models. Scott Pirie* and Malcolm Smith** Relaionships beween Sock Prices and Accouning Informaion: A Review of he Residual Income and Ohlson Models Sco Pirie* and Malcolm Smih** * Inernaional Graduae School of Managemen, Universiy of Souh Ausralia

More information

A Generalized Bivariate Ornstein-Uhlenbeck Model for Financial Assets

A Generalized Bivariate Ornstein-Uhlenbeck Model for Financial Assets A Generalized Bivariae Ornsein-Uhlenbeck Model for Financial Asses Romy Krämer, Mahias Richer Technische Universiä Chemniz, Fakulä für Mahemaik, 917 Chemniz, Germany Absrac In his paper, we sudy mahemaical

More information

Present Value Methodology

Present Value Methodology Presen Value Mehodology Econ 422 Invesmen, Capial & Finance Universiy of Washingon Eric Zivo Las updaed: April 11, 2010 Presen Value Concep Wealh in Fisher Model: W = Y 0 + Y 1 /(1+r) The consumer/producer

More information

cooking trajectory boiling water B (t) microwave 0 2 4 6 8 101214161820 time t (mins)

cooking trajectory boiling water B (t) microwave 0 2 4 6 8 101214161820 time t (mins) Alligaor egg wih calculus We have a large alligaor egg jus ou of he fridge (1 ) which we need o hea o 9. Now here are wo accepable mehods for heaing alligaor eggs, one is o immerse hem in boiling waer

More information

Energy and Performance Management of Green Data Centers: A Profit Maximization Approach

Energy and Performance Management of Green Data Centers: A Profit Maximization Approach Energy and Performance Managemen of Green Daa Ceners: A Profi Maximizaion Approach Mahdi Ghamkhari, Suden Member, IEEE, and Hamed Mohsenian-Rad, Member, IEEE Absrac While a large body of work has recenly

More information

STABILITY OF LOAD BALANCING ALGORITHMS IN DYNAMIC ADVERSARIAL SYSTEMS

STABILITY OF LOAD BALANCING ALGORITHMS IN DYNAMIC ADVERSARIAL SYSTEMS SIAM J. COMPUT. Vol. 37, No. 5, pp. 1656 1673 c 2008 Sociey for Indusrial and Applied Mahemaics STABILITY OF LOAD BALANCING ALGORITHMS IN DYNAMIC ADVERSARIAL SYSTEMS ELLIOT ANSHELEVICH, DAVID KEMPE, AND

More information

We consider a decentralized assembly system in which a buyer purchases components from several first-tier

We consider a decentralized assembly system in which a buyer purchases components from several first-tier MANAGEMENT SCIENCE Vol. 55, No. 4, April 2009, pp. 552 567 issn 0025-1909 eissn 1526-5501 09 5504 0552 informs doi 10.1287/mnsc.1080.0961 2009 INFORMS Dynamic Cos Reducion Through Process Improvemen in

More information

AP Calculus AB 2010 Scoring Guidelines

AP Calculus AB 2010 Scoring Guidelines AP Calculus AB 1 Scoring Guidelines The College Board The College Board is a no-for-profi membership associaion whose mission is o connec sudens o college success and opporuniy. Founded in 1, he College

More information

Improvement of a TCP Incast Avoidance Method for Data Center Networks

Improvement of a TCP Incast Avoidance Method for Data Center Networks Improvemen of a Incas Avoidance Mehod for Daa Cener Neworks Kazuoshi Kajia, Shigeyuki Osada, Yukinobu Fukushima and Tokumi Yokohira The Graduae School of Naural Science and Technology, Okayama Universiy

More information

OPTIMAL PORTFOLIO MANAGEMENT WITH TRANSACTIONS COSTS AND CAPITAL GAINS TAXES

OPTIMAL PORTFOLIO MANAGEMENT WITH TRANSACTIONS COSTS AND CAPITAL GAINS TAXES OPTIMAL PORTFOLIO MANAGEMENT WITH TRANSACTIONS COSTS AND CAPITAL GAINS TAXES Hayne E. Leland Haas School of Business Universiy of California, Berkeley Curren Version: December, 1999 Absrac We examine he

More information

Stochastic Recruitment: A Limited-Feedback Control Policy for Large Ensemble Systems

Stochastic Recruitment: A Limited-Feedback Control Policy for Large Ensemble Systems Sochasic Recruimen: A Limied-Feedback Conrol Policy for Large Ensemble Sysems Lael Odhner and Harry Asada Absrac This paper is abou sochasic recruimen, a conrol archiecure for cenrally conrolling he ensemble

More information

Spectrum-Aware Data Replication in Intermittently Connected Cognitive Radio Networks

Spectrum-Aware Data Replication in Intermittently Connected Cognitive Radio Networks Specrum-Aware Daa Replicaion in Inermienly Conneced Cogniive Radio Neworks Absrac The opening of under-uilized specrum creaes an opporuniy for unlicensed users o achieve subsanial performance improvemen

More information

Cointegration: The Engle and Granger approach

Cointegration: The Engle and Granger approach Coinegraion: The Engle and Granger approach Inroducion Generally one would find mos of he economic variables o be non-saionary I(1) variables. Hence, any equilibrium heories ha involve hese variables require

More information

Impact of Human Mobility on Opportunistic Forwarding Algorithms

Impact of Human Mobility on Opportunistic Forwarding Algorithms Impac of Human Mobiliy on Opporunisic Forwarding Algorihms Augusin Chainreau, Pan Hui, Jon Crowcrof, Chrisophe Dio, Richard Gass, and James Sco *, Universiy of Cambridge Microsof Research Thomson Research

More information

Time Series Analysis Using SAS R Part I The Augmented Dickey-Fuller (ADF) Test

Time Series Analysis Using SAS R Part I The Augmented Dickey-Fuller (ADF) Test ABSTRACT Time Series Analysis Using SAS R Par I The Augmened Dickey-Fuller (ADF) Tes By Ismail E. Mohamed The purpose of his series of aricles is o discuss SAS programming echniques specifically designed

More information

Individual Health Insurance April 30, 2008 Pages 167-170

Individual Health Insurance April 30, 2008 Pages 167-170 Individual Healh Insurance April 30, 2008 Pages 167-170 We have received feedback ha his secion of he e is confusing because some of he defined noaion is inconsisen wih comparable life insurance reserve

More information

A UNIFIED APPROACH TO MATHEMATICAL OPTIMIZATION AND LAGRANGE MULTIPLIER THEORY FOR SCIENTISTS AND ENGINEERS

A UNIFIED APPROACH TO MATHEMATICAL OPTIMIZATION AND LAGRANGE MULTIPLIER THEORY FOR SCIENTISTS AND ENGINEERS A UNIFIED APPROACH TO MATHEMATICAL OPTIMIZATION AND LAGRANGE MULTIPLIER THEORY FOR SCIENTISTS AND ENGINEERS RICHARD A. TAPIA Appendix E: Differeniaion in Absrac Spaces I should be no surprise ha he differeniaion

More information

Pricing Fixed-Income Derivaives wih he Forward-Risk Adjused Measure Jesper Lund Deparmen of Finance he Aarhus School of Business DK-8 Aarhus V, Denmark E-mail: jel@hha.dk Homepage: www.hha.dk/~jel/ Firs

More information

Working Paper On the timing option in a futures contract. SSE/EFI Working Paper Series in Economics and Finance, No. 619

Working Paper On the timing option in a futures contract. SSE/EFI Working Paper Series in Economics and Finance, No. 619 econsor www.econsor.eu Der Open-Access-Publikaionsserver der ZBW Leibniz-Informaionszenrum Wirschaf The Open Access Publicaion Server of he ZBW Leibniz Informaion Cenre for Economics Biagini, Francesca;

More information

Direc Manipulaion Inerface and EGN algorithms

Direc Manipulaion Inerface and EGN algorithms A Direc Manipulaion Inerface for 3D Compuer Animaion Sco Sona Snibbe y Brown Universiy Deparmen of Compuer Science Providence, RI 02912, USA Absrac We presen a new se of inerface echniques for visualizing

More information

A Load Balancing Method in Downlink LTE Network based on Load Vector Minimization

A Load Balancing Method in Downlink LTE Network based on Load Vector Minimization A Load Balancing Mehod in Downlink LTE Nework based on Load Vecor Minimizaion Fanqin Zhou, Lei Feng, Peng Yu, and Wenjing Li Sae Key Laboraory of Neworking and Swiching Technology, Beijing Universiy of

More information

ARCH 2013.1 Proceedings

ARCH 2013.1 Proceedings Aricle from: ARCH 213.1 Proceedings Augus 1-4, 212 Ghislain Leveille, Emmanuel Hamel A renewal model for medical malpracice Ghislain Léveillé École d acuaria Universié Laval, Québec, Canada 47h ARC Conference

More information

Optimal Life Insurance, Consumption and Portfolio: A Dynamic Programming Approach

Optimal Life Insurance, Consumption and Portfolio: A Dynamic Programming Approach 28 American Conrol Conference Wesin Seale Hoel, Seale, Washingon, USA June 11-13, 28 WeA1.5 Opimal Life Insurance, Consumpion and Porfolio: A Dynamic Programming Approach Jinchun Ye (Pin: 584) Absrac A

More information

A Distributed Multiple-Target Identity Management Algorithm in Sensor Networks

A Distributed Multiple-Target Identity Management Algorithm in Sensor Networks A Disribued Muliple-Targe Ideniy Managemen Algorihm in Sensor Neworks Inseok Hwang, Kaushik Roy, Hamsa Balakrishnan, and Claire Tomlin Dep. of Aeronauics and Asronauics, Sanford Universiy, CA 94305 Elecrical

More information

Dispatch-and-Search: Dynamic Multi-Ferry Control in Partitioned Mobile Networks

Dispatch-and-Search: Dynamic Multi-Ferry Control in Partitioned Mobile Networks Dispach-and-Search: Dynamic Muli-Ferry Conrol in Pariioned Mobile Newors ing He, Ananhram Swami, and Kang-Won Lee IBM Research, Hawhorne, NY 1532 USA {he,angwon}@us.ibm.com Army Research Laboraory, Adelphi,

More information

Longevity 11 Lyon 7-9 September 2015

Longevity 11 Lyon 7-9 September 2015 Longeviy 11 Lyon 7-9 Sepember 2015 RISK SHARING IN LIFE INSURANCE AND PENSIONS wihin and across generaions Ragnar Norberg ISFA Universié Lyon 1/London School of Economics Email: ragnar.norberg@univ-lyon1.fr

More information

Dependent Interest and Transition Rates in Life Insurance

Dependent Interest and Transition Rates in Life Insurance Dependen Ineres and ransiion Raes in Life Insurance Krisian Buchard Universiy of Copenhagen and PFA Pension January 28, 2013 Absrac In order o find marke consisen bes esimaes of life insurance liabiliies

More information

Verification Theorems for Models of Optimal Consumption and Investment with Retirement and Constrained Borrowing

Verification Theorems for Models of Optimal Consumption and Investment with Retirement and Constrained Borrowing MATHEMATICS OF OPERATIONS RESEARCH Vol. 36, No. 4, November 2, pp. 62 635 issn 364-765X eissn 526-547 364 62 hp://dx.doi.org/.287/moor..57 2 INFORMS Verificaion Theorems for Models of Opimal Consumpion

More information

A PROPOSAL TO OBTAIN A LONG QUARTERLY CHILEAN GDP SERIES *

A PROPOSAL TO OBTAIN A LONG QUARTERLY CHILEAN GDP SERIES * CUADERNOS DE ECONOMÍA, VOL. 43 (NOVIEMBRE), PP. 285-299, 2006 A PROPOSAL TO OBTAIN A LONG QUARTERLY CHILEAN GDP SERIES * JUAN DE DIOS TENA Universidad de Concepción y Universidad Carlos III, España MIGUEL

More information

The Real Business Cycle paradigm. The RBC model emphasizes supply (technology) disturbances as the main source of

The Real Business Cycle paradigm. The RBC model emphasizes supply (technology) disturbances as the main source of Prof. Harris Dellas Advanced Macroeconomics Winer 2001/01 The Real Business Cycle paradigm The RBC model emphasizes supply (echnology) disurbances as he main source of macroeconomic flucuaions in a world

More information

Impact of Human Mobility on the Design of Opportunistic Forwarding Algorithms

Impact of Human Mobility on the Design of Opportunistic Forwarding Algorithms Impac of Human Mobiliy on he Design of Opporunisic Forwarding Algorihms Augusin Chainreau, Pan Hui *, Jon Crowcrof *, Chrisophe Dio, Richard Gass, and James Sco, Thomson Research 46 quai A Le Gallo 92648

More information

Making Use of Gate Charge Information in MOSFET and IGBT Data Sheets

Making Use of Gate Charge Information in MOSFET and IGBT Data Sheets Making Use of ae Charge Informaion in MOSFET and IBT Daa Shees Ralph McArhur Senior Applicaions Engineer Advanced Power Technology 405 S.W. Columbia Sree Bend, Oregon 97702 Power MOSFETs and IBTs have

More information

Quality-Of-Service Class Specific Traffic Matrices in IP/MPLS Networks

Quality-Of-Service Class Specific Traffic Matrices in IP/MPLS Networks ualiy-of-service Class Specific Traffic Marices in IP/MPLS Neworks Sefan Schnier Deusche Telekom, T-Sysems D-4 Darmsad +4 sefan.schnier@-sysems.com Franz Harleb Deusche Telekom, T-Sysems D-4 Darmsad +4

More information

MACROECONOMIC FORECASTS AT THE MOF A LOOK INTO THE REAR VIEW MIRROR

MACROECONOMIC FORECASTS AT THE MOF A LOOK INTO THE REAR VIEW MIRROR MACROECONOMIC FORECASTS AT THE MOF A LOOK INTO THE REAR VIEW MIRROR The firs experimenal publicaion, which summarised pas and expeced fuure developmen of basic economic indicaors, was published by he Minisry

More information

MODEL AND ALGORITHMS FOR THE REAL TIME MANAGEMENT OF RESIDENTIAL ELECTRICITY DEMAND. A. Barbato, G. Carpentieri

MODEL AND ALGORITHMS FOR THE REAL TIME MANAGEMENT OF RESIDENTIAL ELECTRICITY DEMAND. A. Barbato, G. Carpentieri MODEL AND ALGORITHMS FOR THE REAL TIME MANAGEMENT OF RESIDENTIAL ELECTRICITY DEMAND A. Barbao, G. Carpenieri Poliecnico di Milano, Diparimeno di Eleronica e Informazione, Email: barbao@ele.polimi.i, giuseppe.carpenieri@mail.polimi.i

More information

Module 4. Single-phase AC circuits. Version 2 EE IIT, Kharagpur

Module 4. Single-phase AC circuits. Version 2 EE IIT, Kharagpur Module 4 Single-phase A circuis ersion EE T, Kharagpur esson 5 Soluion of urren in A Series and Parallel ircuis ersion EE T, Kharagpur n he las lesson, wo poins were described:. How o solve for he impedance,

More information

Quality-Of-Service Class Specific Traffic Matrices in IP/MPLS Networks

Quality-Of-Service Class Specific Traffic Matrices in IP/MPLS Networks ualiy-of-service Class Specific Traffic Marices in IP/MPLS Neworks Sefan Schnier Deusche Telekom, T-Sysems D-4 Darmsad +4 sefan.schnier@-sysems.com Franz Harleb Deusche Telekom, T-Sysems D-4 Darmsad +4

More information

BALANCE OF PAYMENTS. First quarter 2008. Balance of payments

BALANCE OF PAYMENTS. First quarter 2008. Balance of payments BALANCE OF PAYMENTS DATE: 2008-05-30 PUBLISHER: Balance of Paymens and Financial Markes (BFM) Lena Finn + 46 8 506 944 09, lena.finn@scb.se Camilla Bergeling +46 8 506 942 06, camilla.bergeling@scb.se

More information