European Journal of Operational Research

Size: px
Start display at page:

Download "European Journal of Operational Research"

Transcription

1 European Journal of Operaonal Research Conens lss avalable a ScenceDrec European Journal of Operaonal Research journal homepage: Producon, Manufacurng and Logscs Vald nequales for wo-perod relaxaons of bg-bucke lo-szng problems: Zero seup case Mahd Doosmohammad a, Kerem Akarunalı b, a School of Mahemacs, Unvers of Ednburgh, Ednburgh EH9 3FD, UK b Deparmen of Managemen Scence, Unvers of Srahclde, Glasgow G4 0GE, UK a r c l e n f o a b s r a c Arcle hsor: Receved 27 Sepember 2016 Acceped 8 November 2017 Avalable onlne 16 November 2017 Kewords: Producon Lo-szng Ineger programmng Polhedral analss Vald nequales In hs paper, we nvesgae wo-perod subproblems for bg-bucke lo-szng problems, whch have shown a grea poenal for obanng srong bounds. In parcular, we nvesgae he specal case of zero seup mes and denf wo mporan mxed neger ses represenng relaxaons of hese subproblems. We analze he polhedral srucure of hese ses, dervng several famles of vald nequales and presenng her face-defnng condons. We hen exend hese nequales n a novel fashon o he orgnal space of wo-perod subproblems, and also propose a new faml of vald nequales n he orgnal space. In order o nvesgae he rue srengh of he proposed nequales, we propose and mplemen exac separaon algorhms, whch are compuaonall esed over a broad range of es problems. In addon, we develop a heursc framework for separaon, n order o exend compuaonal ess o larger nsances. These compuaonal expermens ndcae he proposed nequales can be ndeed ver effecve m provng lower bounds subsanall The Auhors. Publshed b Elsever B.V. Ths s an open access arcle under he CC BY lcense. hp://creavecommons.org/lcenses/b/4.0/ 1. Inroducon The lo-szng problem ams o deermne an opmal producon plan dealng how much o produce and sock n each me perod of he plannng horzon, gven manufacurng ssem lmaons such as machne capaces and cusomer orders/forecased demand. Due o s srong mpac on manufacurng companes performance n erms of cusomer servce qual and operang coss, lo-szng has been a ver acve research area for man decades wh sgnfcan aenon from researchers as well as praconers. Due o s praccal mporance and lmed knowledge n he leraure, we focus n hs paper on he mul-em lo-szng problem wh bg-bucke capaces, where each resource s shared b mulple ems and more han one pe of em can be produced n an me perod. We sud hs problem from a heorecal perspecve, where we analze a wo-perod relaxaon of hs problem and characerze s mporan properes. Our man conrbuons are several famles of new vald nequales and her The research presened n hs paper s suppored b he EPSRC gran EP/L /1, enled Mul-Iem Producon Plannng: Theor, Compuaon and Pracce. Correspondng auhor. E-mal addresses: mahd.doosmohammad@ed.ac.uk M. Doosmohammad, kerem.akarunal@srah.ac.uk K. Akarunalı. face-defnng properes for he relaxaons of he wo-perod subproblem, novel exensons of hese nequales no he orgnal space of he wo-perod relaxaon, as well as new vald nequales for he orgnal space, and exac separaon algorhms desgned o es he praccal srengh of he proposed nequales. We also develop a smple bu effecve heursc approach n order o exend compuaonal expermens. Our compuaonal resuls show ha he proposed nequales have grea poenal o srenghen he lower bounds sgnfcanl Leraure revew Mos lo-szng problems are nherenl dffcul problems: from he heorecal complex perspecve, even he mul-em problem wh a sngle jon capac and whou seup mes s srongl N P -hard Chen & Thz, From a compuaonal perspecve, problems wh mulple ems and capaces, n parcular of ndusral scale, reman noorousl dffcul o solve o opmal, ofen resulng n hgh dual gaps Buschkühl, Sahlng, Helber, & Tempelmeer, Therefore, here s a wde specrum of research on lo-szng problems, rangng from praccall effcen heurscs e.g., Federgruen, Messner, & Tzur, 2007 and mea-heurscs e.g., Jans & Degraeve, 2007 o mahemacal programmng echnques, whch we dscuss nex n more deal due o her relevance o our sud. hps://do.org/ /j.ejor / 2017 The Auhors. Publshed b Elsever B.V. Ths s an open access arcle under he CC BY lcense. hp://creavecommons.org/lcenses/b/4.0/

2 M. Doosmohammad, K. Akarunalı / European Journal of Operaonal Research Because of her complex, mos researchers n he mahemacal programmng commun suded specal cases of lo-szng problems, whch sll provde valuable nsghs on some nheren srucures of more general problems and hence suppor he soluon mehodologes proposed. The exac approaches mos ofen emploed eher defnng vald nequales e.g., Baran, Van Ro, & Wolse, 1984; Küçükavuz & Poche, 2009; Poche & Wolse, 1994 or exended reformulaons e.g., Eppen & Marn, 1987; Krarup & Blde, 1977; Poche & Wolse, 2010 for varans of sngle-em problem, some of whch were also exended o mul-em problems e.g., Belvaux & Wolse, There are also few sudes usng oher echnques such as Lagrangan relaxaon e.g., Bllngon, McClan, & Thomas, 1986 and Danzg Wolfe decomposon e.g., Degraeve & Jans, Poche and Wolse 2006 provde a horough revew of dfferen varans of lo-szng problems, her complexes and soluon mehods used. Mos recenl, here have been nsghful polhedral resuls on mul-level problems, such as he vald nequales of Zhang, Küçükavuz, and Yaman 2012, and he compac formulaons of Van Vve, Wolse, and Yaman 2014 for small-bucke capaces,.e., ems do no share resources. Despe hs exensve leraure, he research explcl nvesgang complcaons arsng from mulple ems compeng for he lmed capaces nheren n bg-bucke problems s raher lmed, and onl few excepons exs o he bes of our knowledge. The polhedral analss of a sngle-perod relaxaon b Mller, Nemhauser, and Savelsbergh , provded us some nsghful properes of hs polhedron ncludng new vald nequales. The sud of Jans and Degraeve 2004 presened varous decomposons and ndcaed ha perod decomposons provde sronger bounds, whch s recenl nvesgaed furher b he branch-and-cu framework of de Araujo, Reck, Degraeve, Fragkos, and Jans 2015 resulng n promsng compuaonal resuls wh regards o gaps. The work of Van Vve and Wolse 2005 obaned srong lower bounds, mos ofen sronger han an prevous resuls, b applng approxmae exended reformulaons onl for a small number of perods. The exensve compuaonal sud of Akarunalı and Mller 2012 noed he boleneck n bg-bucke problems as he lack of a good undersandng of he convex hull of sngle-machne, mul-perod problems. Ths movaed he novel framework of Akarunalı, Fragkos, Mller, and Wu 2016, where he smalles such problem, a wo-perod relaxaon, s used o separae all volaed nequales b generang he exreme pons of s convex hull, whou pre-defnng famles of nequales. The compuaonal resuls of hs sud have shown grea promse o sgnfcanl close dual gaps for bg-bucke problems n general, whch movaed us o sud such a wo-perod relaxaon from a polhedral perspecve. In hs paper, we presen our work nvesgang he specal case of zero seup mes. Ths does no onl enable us o analcall sud nheren srucures and hence provde useful nsghs ha can be poenall exended o more complcaed problems, bu also mprove our undersandng abou hs mul-em problem wh zero seup mes ha has been acvel suded for man ears n he lo-szng leraure, e.g., he earler work of Dxon and Slver 1981 proposed heursc approaches for hs problem, and essenall he movaon for he semnal work of Padberg, van Ro, and Wolse 1985 semmed from hs problem. From a praccal perspecve, s worh nong ha seup mes are no necessarl zero, however, usng zero seup mes has been a ver effecve modellng approach n case of neglgble seup mes,.e., ver low seup mes compared o processng mes, n order o reduce he problem complex, see, e.g., Kuk, Salomon, & van Wassenhove, 1994 for a dscusson. Neglgble seup mes can be observed n varous manufacurng sengs, e.g., assembl operaons n auomoble ndusr and packng operaons n food ndusr. Moreover, as noed b Olaan and Geragh 2013 echnologcal advances such as agle oolng and maeral handlng make possble o produce dfferen producs on he same se of machnes and herefore enable producon lne desgners reduce seup coss sgnfcanl. We fnall refer he neresed reader o he revew of Jans and Degraeve 2008 for a horough dscusson abou hs specal case, achevemens and open challenges. In he nex secon, we presen he problem formulaon and he wo-perod relaxaon X 2 PL, orgnall proposed b Akarunalı e al. 2016, and sud some of s polhedral properes, ncludng he specal case wh no seup mes. In Secon 3, we presen wo relaxaons of X 2 PL, propose a number of vald nequales for hese relaxaons and dscuss her face-defnng properes. Secon 4 presens novel exensons of hese nequales no he orgnal space of X 2 PL, as well as a new faml of vald nequales. We presen exac separaon algorhms n Secon 5, and compuaonall es he srengh of he nequales n Secon 6, whch show promsng resuls for her effecveness. We also develop a smple bu effecve heursc separaon approach n Secon 7 enablng expermenaon wh larger nsances, where furher encouragng resuls are obaned. We conclude he paper wh a dscusson of possble exensons and generalzaons. We noe ha all essenal proofs are provded n he Onlne Supplemen due o her lengh and nvolved naure. 2. A wo-perod relaxaon for bg-bucke lo-szng problem Before we defne and sud he wo-perod relaxaon of neres, we frs provde he mahemacal formulaon of he mulem lo-szng problem wh bg-bucke capaces. We le T, I and R denoe he ses of me perods, ems, and machne resource pes, respecvel. We represen he producon, seup, and nvenor varables for em n perod b x,, and s, respecvel. mn T I f + T I h s 1 s.. x + s 1 s = d T, I 2 I a r x + ST r C r T, r R 3 x M T, I 4 { 0, 1 } T x I ; x, s 0 5 The objecve funcon 1 mnmzes oal cos, where f and h ndcae he seup and nvenor cos coeffcens, respecvel. The flow balance consrans 2 ensure ha he demand for each em n perod, denoed b d, s sasfed. We noe ha he model can be generalzed o nvolve mulple levels see, e.g., Akarunalı & Mller, 2012, however, we om hs for he sake of smplc. The bg-bucke capac consrans 3 ensure ha he capac C r of machne r s no exceeded n me perod, where a r and ST r represen he per un producon me and seup me for em, respecvel. Consrans 4 guaranee ha he seup varable s equal o 1 f producon occurs, where M represens he maxmum number of em ha can be produced n perod, whch s he mnmum of eher he remanng cumulave demand or he capac avalable. Fnall, he negral and non-negav consrans are gven b A wo-perod relaxaon: X 2 PL Nex, we presen he feasble regon of a wo-perod, sngle-machne relaxaon, as orgnall proposed b

3 88 M. Doosmohammad, K. Akarunalı / European Journal of Operaonal Research Akarunalı e al. 2016, of he mul-em producon plannng problem wh bg-bucke capaces, denoed b X 2 PL. x M I, = 1, 2 x d + s I, = 1, 2 x 1 + x 2 d d s I x 1 + x 2 d 1 + s I a x + ST C = 1, 2 I 2 x, s 0, { 0, 1 } I As we consder a sngle machne r R n hs relaxaon, we dropped he ndex r. We noe ha, for a gven me perod, he choce for he horzon of hs wo-perod subproblem wll be + α wh α = 1,..., NT. An obvous choce for α would be 1,.e., = 1, 2 relae o he perods of, + 1. The parameers can be assocaed wh he orgnal problem parameers usng he relaons M = M + 1 α, C = C r + 1 α, and d = d + 1 α, + α for all and = 1, 2. We refer he neresed reader o Akarunalı e al for a dealed dscusson of srucurng wo-perod subproblems. Nex, we noe some polhedral properes of X 2 PL. Proposon 1 Akarunalı e al W.l.o.g., we assume 0 < M and ST < C hold I, = 1, 2. Then, con v X 2 PL s fulldmensonal. Proposon 2. The rval face-defnng nequales for con v X 2 PL and her face-defnng condons f an are: 1. x 0, I, = 1, , I, = 1, s 0, I. 4. x M, I, = 1, x d + s, I, = 1, 2 f d < M. 6. x 1 + x 2 d d s, I f d < M, { 1, 2 }. 7. I a x + ST C, = 1, 2 f for {1, 2}, I a M + ST C + a k M k + ST k, k I. We om he proof for he sake of smplc of he presenaon. Nex, we presen some non-rval faces of con v X 2 PL. W.l.o.g., we assume a = 1, I n he remander of he paper snce varables can be scaled as needed. Proposon 3. For I, 1. The followng nequal s vald for X 2 PL : x 1 + x 2 d d 2 + s If d + ST C, { 1, 2 }, defnes a face of con v X 2 PL. 2. If ST C 2, hen he followng nequal s vald for X 2 PL : x 1 + x 2 d 1 C 2 ST 1 + C 2 ST + s If d 1 + ST C 1 and d 2 + ST > C 2, s face-defnng for con v X 2 PL. 3. If d 1 + ST C 1, hen he followng nequal s vald for X 2 PL : x 1 + x 2 d d 1 C 1 ST 2 + C 1 ST d 1 + s If d 1 + ST > C 1 and d 2 + ST C 2, defnes a face of con v X 2 PL. 4. If d ST C 1 + C 2, hen he followng nequal s vald for X 2 PL : x 1 + x 2 d 1 C 2 ST 1 + d 1 C 1 ST 2 + C 1 ST + C 2 ST d + s If d 1 + ST > C 1 and d 2 + ST > C 2, defnes a face of con v X 2 PL. 1 The proof s omed for he sake of brev. In hs paper, we nvesgae he specal case of zero seups,.e., ST = 0, I. We noe ha a companon paper Akarunalı, Doosmohammad, & Fragkos, 2017 sudes he polhedral properes of he general case of non-zero seups as well as srucurng an effecve compuaonal framework. In he nex secon, we esablsh wo relaxaons of X 2 PL and sud her polhedral srucures. We frs presen he known face-defnng nequales, and hen derve several classes of vald nequales and esablsh her face-defnng condons. 3. Polhedral analss of he relaxaons of X 2 PL Frs, we make necessar defnons for he remander of he paper. Defnon 1. For a gven : A cover of I for perod s a se S such ha λ = S d C > 0. For gven non-emp ses S I and T I\ S, we defne he paron slack as ξ = T M + S d C. We defne he se S + of srcl posve cover/paron elemens as follows: {{ S S + = d > λ } f S s a cover. { S d > ξ } f S s par of a paron. The posve maxmum funcon as b + = max { b, 0 }. Frs, for a gven, we defne he followng relaxaon, denoed b PIR 0, for X 2 PL, snce s suded n he leraure b varous researchers. x M, x C I I x 0, { 0, 1 } I We dropped here all he ndces as well as for he sake of smplc. We noe ha Defnon 1 remans vald n he same fashon ha we use same defnons for hs relaxaon wh all he ndces as well as dropped. Nex, we presen known face-defnng nequales for PIR 0. Proposon 4 Flow cover nequales Padberg e al., Le S be a cover, and S M = C + λ. Assume ha M = max S M > λ. Then, x M C M 6 S S S s vald and defnes a face of con v P IR 0. Moreover, for L I \ S and M = max M, M, he nequal x M M λ C M S L S L s vald and defnes a face of con v P IR 0 f 0 < M λ < M M, L. In addon o hs known relaxaon and s face-defnng nequales, we presen a second relaxaon of X 2 PL for a gven. We call hs as PIR 1 and sud mporan properes of n he remander of hs secon: x M, x d + s, x C I I I S 7

4 M. Doosmohammad, K. Akarunalı / European Journal of Operaonal Research x, s 0, { 0, 1 } I Smlar o he prevous relaxaon, we dropped here all he ndces as well as, and Defnon 1 remans vald n he same fashon. Snce he srucure of he se X 2 PL s que complex from a polhedral perspecve, hs relaxaon of X 2 PL wh a smpler polhedral srucure enables us o poenall derve valuable nsghs on he nheren srucure of X 2 PL. Nex, we noe some obvous properes of hs polhedron, ncludng he full dmensonal and rval faces of con v P IR 1. These proposons can be easl proven and herefore, we om dealed proofs here for he sake of brev. Proposon 5. dm con v P IR 1 = 3 I. Proposon 6. The followng nequales are he rval faces of PIR 1 : 1. x 0, I. 2. 1, I. 3. s 0, I. 4. x M, I. 5. x d + s f d < M, I 6. I x C f I M C + M k, k I. Nex, we dscuss famles of vald nequales and esablsh her face-defnng condons for PIR 1. Proposon 7. Le S be a cover of I. Then he followng nequal called cover nequal s vald for PIR 1 : x d s + C d 8 S S S Moreover, f K I \ S such ha M d holds K, where d = max S d λ and d = max d, d, K, hen he followng nequal called em-exended cover nequal s vald for PIR 1 : x d d λ s + C d S K S K The vald of 8 and 9 can be shown b consderng he mxed-neger se { I x s C, x s d, { 0, 1 } I }, whch s a sngle-node flow se whou he nonnegav consrans as well as a specal relaxaon of PIR 1, and hen dervng flow covers and exended flow covers from. Inuvel speakng, snce he ems n he cover have a oal demand ha s srcl exceedng he capac b λ, he cover nequal consders onl ems from hs se ha have an ndvdual demand srcl hgher han hs excess of λ, as s no possble o produce a leas one such em o s full demand. Proposon 8. If d < M S and S + 2 hold, 8 defnes a face of con v P IR 1. If, n addon, 0 < d λ < d d holds K, hen 9 defnes a face of con v P IR 1. We provde a dealed proof n he Onlne Supplemen. Proposon 9. Le S = be a subse of I, T = I\ S, and T, T be a paron of T such ha T = and ξ 0. Then he followng nequal called paron nequal s vald for PIR 1 : x + d ξ M ξ + 1 s + C S T S T S 10 We noe ha f eher S = or T =, hen hs reduces o eher flow cover nequales of Proposon 4 or cover nequales of Proposon 7, respecvel. The vald can be shown b consderng he mxed-neger se { S x s + I\ S x C, x s S S S 9 d, x M, { 0, 1 } I }, whch s a sngle-node flow se whou he non-negav consrans as well as a specal relaxaon of PIR 1, and hen dervng flow covers from. Proposon 10. Le ξ > 0, and assume ha for T + = { T M > ξ}, T + 1 holds. Moreover, assume ha d < M holds S. Then, 10 defnes a face of con v P IR 1. For he proof, we refer he reader o he Onlne Supplemen. Proposon 11. Le S be a non-emp subse of I, T = I\ S, T, T be a paron of T, K T and ξ 0. We defne { d p = : S M : T and p = max S T p ξ. We also defne p = max M, p, K. Then he followng nequal called em-exended paron nequal s vald for PIR 1 : x d ξ + M ξ + p ξ S T K S T K s + C d ξ + M ξ + 11 S S T The vald follows b dervng generalzed flow covers for he mxed-neger se { S x s + I\ S x C, x s d, x M, { 0, 1 } I }, whch s a sngle-node flow se whou he nonnegav consrans as well as a specal relaxaon of PIR 1. Proposon 12. Assume ha he condons presened n Proposon 10 hold. Moreover, le 0 < p ξ < M p hold K. Then, 11 defnes a face of con v P IR 1. For he proof, we refer he reader o he Onlne Supplemen. Example. Le I = { 1, 2, 3 }, and PIR 1 defned b: x , x , x x s 1, x s 2, x s 3 x 1 + x 2 + x 3 14 Consder S = { 1 } and T = { 3 }. Hence ξ = = 7. Then, we can generae a face-defnng paron nequal as follows: x 1 + x s x 1 + x s Wh S = { 1 } and T = { 3 }, we noe p 1 = 10, p 3 = 11. Hence, p = max S T p = 11 > ξ = 7. Le K = { 2 } and hence p 2 = max 11, 10 = 11. Then, we can derve he face-defnng emexended paron nequal: x 1 + x 2 + x s where bold elemens ndcae all erms ha are addonal compared o he prevous nequal. Usng PORTA Chrsof & Lobel, 2009, we can denf 6 face-defnng paron nequales and 3 face-defnng em-exended paron nequales for hs se. In Secon 5, we descrbe separaon algorhms for hese nequales. Nex, we dscuss how we can exend he resuls of hs secon o he space of he wo-perod relaxaon of X 2 PL. 4. Vald nequales n he orgnal space of X 2 PL We recall he orgnal space defned earler as X 2 PL. We frs map he nequales developed for PIR 1 n he prevous secon o he orgnal space of X 2 PL, and also defne new vald nequales for X 2 PL. Noe ha all he ndces as well as are nroduced here agan, whch were dropped n he prevous secon, snce he wll be par of he dscusson here.

5 90 M. Doosmohammad, K. Akarunalı / European Journal of Operaonal Research Corollar 1. Le {1, 2}, and S be a cover of I n perod. Then he nequal mapped from he cover nequal n PIR 1 s vald for X 2 PL : S x + S d + λ 1 S We can also exend hs nequal as follows. s + C 12 Proposon 13. Le, {1, 2}, =, and S be a cover of I for perod. In addon, assume L S. Then he followng nequal called perod-exended cover nequal s vald for X 2 PL : S x + L x + S d + λ 1 L d S s + C 13 We om he proof here, as Proposon 14 covers a more general case and hs s he specal case when K =. Nex, we dscuss em-exended cover nequales and how he can be represened n he orgnal space. Corollar 2. Le {1, 2} and S be a cover of I for perod. Le K I \ S such ha M d holds K, where d = max S d λ and d = max d, d. Then he followng nequal mapped from he em-exended cover nequal n PIR 1 s vald for X 2 PL : S K x + S d + λ 1 K d λ S s + C 14 The proof s sraghforward as follows he same logc as Proposon 7. We can also exend hs nequal as follows. Proposon 14. In addon o he assumpons and defnons of Corollar 2, le, {1, 2}, =, and L S. Then he followng nequal called em-and-perod-exended cover nequal s vald for X 2 PL : S K x L + L x + S S d + λ 1 K d λ d s + C 15 For he proof, we refer he reader o he Onlne Supplemen. Nex, we dscuss he exenson of paron nequales o orgnal space. Corollar 3. Le {1, 2} and S be a non-emp subse of I n perod. Le T = I\ S, and T, T be a paron of T such ha ξ 0. Then he followng nequal mapped from he paron nequal n PIR 1 s vald for X 2 PL : S T x + S d + ξ 1 S + T M + ξ 1 s + C 16 The proof s sraghforward as follows a smlar logc o he proof of Proposon 9. We can also exend hs nequal as follows. Proposon 15. In addon o he assumpons and defnons of Corollar 3, le L S, where, {1, 2} and =. Then he followng nequal called perod-exended paron nequal s vald for X 2 PL : S T L x + L x d + S S d + ξ 1 + T M + ξ 1 s + C 17 We om he proof as follows a smlar logc o he proof of Proposons 9 and 14. Nex, we dscuss em-exended paron nequales. Corollar 4. Le {1, 2}, S = be a subse of I for perod, T = I\ S and T, T be a paron of T such ha ξ 0 and K T. Le { d = : S M : T p p = max S T p ξ, and p = max M, p, K. Then he nequal mapped from he em-exended paron nequal n PIR 1 s vald for X 2 PL : + d + ξ 1 + M + ξ 1 x S T K K S p ξ S T s + C 18 The proof s omed as follows a smlar logc o he proof of Proposon 11. We can also exend hs nequal furher as follows. Proposon 16. In addon o he assumpons and defnons of Corollar 4, le L S,, {1, 2} and =. Then, he followng nequal called em-and-perod-exended paron nequal s vald for X 2 PL : x S T K + T + L x + M + ξ 1 S d + ξ 1 K p ξ L d S s + C 19 We om he proof here snce s ver smlar o he proof of Proposon 15. Fnall, we nroduce n he nex proposon a new faml of vald nequales ha s no derved from an of he nequales based on he sngle-perod relaxaon. Proposon 17. For, {1, 2} and =, le S I and T I\ S such ha θ = S d + T M C 0, and S Iand T I\ S such ha θ = S d + T M C 0. Then, he followng nequal called wo-perod paron nequal s vald for X 2 PL : S T x + S T x + S d + θ 1 + M θ T + T S M θ + 1 C + C + d θ + 1 S S s 20 We provde a dealed proof n Onlne Supplemen. We noe ha hs nequal s non-domnaed f S S =.

6 M. Doosmohammad, K. Akarunalı / European Journal of Operaonal Research Separaon algorhms for relaxaons and orgnal space The purpose of hs secon s o descrbe n deal he separaon problems assocaed wh all he famles of nequales defned n he prevous secons. Snce he man purpose of hs paper s o nvesgae he rue srengh of he cus generaed, we focus on defnng exac separaon algorhms raher han her compuaonal effcenc. Here, we follow he same srucure and order of he prevous wo secons: we frsl defne separaon problems assocaed wh famles of nequales defned for he relaxaons of he problems, and hen for hose assocaed wh he orgnal space. In he remander of hs secon, we le x, ȳ, s o represen a fraconal soluon vecor n he assocaed space ha s o be cu off. W.l.o.g., we assume all problem parameers o be neger valued Separaon n he relaxaon space We sar frs wh he faml of cover nequales as defned b Eq. 8. Frs, we noe ha we can rewre hese nequales as follows: x + d λ + 1 s C S Snce S mus be a cover, for a gven value of λ > 0, we can fnd he mos volaed nequal f an b solvng he followng knapsack problem: { } f λ = max τ j λ w j d w = C + λ, w { 0, 1 } I, j I I where τ j λ = x j + d j λ + 1 ȳ j s j. If f λ > C, hen a volaed cover nequal s denfed for he specfed λ. We noe ha snce λ Z +, one can solve hs separaon problem for an value of λ [1, I d C]. Nex, we dscuss he separaon procedure for he faml of em-exended cover nequales 9. We frs rewre hese nequales as follows: S x + d λ + 1 s + K x d λ C For a gven cover S, f d λ, hen we can defne he se K as follows: { K = I\ S x d λ ȳ > 0, M d } Therefore, one can generae covers usng he procedure defned for he cover nequales, and hen heurscall generae he se K. We noe ha hs s a smlar approach o he one proposed b Padberg e al p.854 for flow cover nequales. Fnall, we noe ha such a procedure s appled for he separaon of he known nequales 6 and 7. Nex, we dscuss he separaon procedure for he faml of paron nequales 10. Frs noe ha we can rewre hese nequales as follows: S x + d ξ + 1 s + T x + M ξ + 1 C For a gven value of ξ, we defne he followng IP for he separaon: f ξ = max x + d ξ + 1 ȳ s u I I + x + M ξ + 1 ȳ s.. d u + M v = C + ξ I I v C > d u I u + v 1, I u, v { 0, 1 }, I Here, u and v varables ndcae wheher an em belongs o se S or T, respecvel. The frs wo consrans ensure ha I v 1,.e., T =. A volaed nequal s found f f ξ > C. Sm- lar o he process for cover nequales, snce ξ Z +, one can solve hs separaon problem for an value of ξ [1, I max { d, M } C]. The separaon procedure for em-exended paron nequales 11 s smlar o he procedure descrbed for em-exended cover nequales Separaon n he orgnal space We sar hs secon wh he separaon of he perodexended cover nequales n he orgnal space as defned b Eq. 13. Frs, we rewre hese nequales as follows, where S I and L S, = : S x + d λ + 1 s + L x d C For a gven and fxed λ > 0, we can solve he separaon problem: max x + d λ + 1 ȳ s u + x d ȳ v I d I s.. u = C + λ ; v u I; u, v { 0, 1 } I If he opmal value of he problem s srcl greaer han C, hen a volaed nequal s denfed. Nex, we noe ha he separaon procedures n he orgnal space for he em-and-perod-exended cover nequales 15, for he perod-exended paron nequales 17, and for he emand-perod-exended paron nequales 19 follow a ver smlar logc o he separaon procedures of he perod-exended cover nequales n he orgnal space 13. Therefore, we om a dealed descrpon here for he sake of brev. Fnall, we nroduce he separaon procedure for wo-perod paron nequales defned b Eq. 20. Frs, we rewre as follows: x + d + ξ 1 s S + x + M + ξ 1 T + S T x + d ξ + 1 s + x + M ξ S S I s C + C For a gven par of, as well as fxed ξ > 0 and ξ > 0 values, we can defne he followng separaon problem 2PPI: 2PPI f ξ,ξ = max x + d + ξ 1 ȳ s u I I + x + M + ξ 1 ȳ v I + x + d ξ + 1 ȳ s I + x + M ξ + 1 ȳ v u + s z I

7 92 M. Doosmohammad, K. Akarunalı / European Journal of Operaonal Research s.. d u + M v = C + ξ I d u I I + M I v u + v 1, I u + v 1, I z u, I z u, I = C + ξ u, v, z, u, v { 0, 1 }, I If f ξ,ξ > C + C holds, hen he nequal s volaed and hence he cung plane s added o he formulaon. 6. Compuaonal resuls In hs secon, we presen numercal resuls ndcang he srengh of he varous cus proposed earler. We noe ha our prmar am here s no necessarl o buld a praccall effcen compuaonal framework, whch s addressed n a companon paper Akarunalı e al., 2017, bu nsead o exhausvel generae all volaed nequales b exac separaon o measure her praccal srengh and effecveness. All he separaon algorhms and mahemacal models are mplemened and execued usng he Mosel language of FICO Xpress Opmzaon Sue Mosel verson 3.6.0, Xpress-MP v7.7 on a PC wh Inel Core ggaherz processor and 4 ggabe RAM, where all possble wo-perod relaxaons, boh consecuve and non-consecuve, were consdered. In order o es he effecveness of he cus proposed, we have generaed 240 random es nsances n oal, whch we descrbe n deal nex. Frs of all, we noe ha exac separaon s compuaonall expensve, causng ssues wh avalable memor or prohbvel long mes when he problem sze became bgger han T = 12 and I = 10, so ha we se he maxmum sze o hese values. We also noe ha even wh hs maxmum sze, compuaonal mes can be exensve. On he oher hand, we have se he mnmum sze o T = 2 and I = 3, n order o capure he smples problem wh he wo-perod, mul-em srucure. We have vared T and I values wh nervals growng exponenall, n order o capure he vare creaed b he fac ha he problem complex grows exponenall raher han usng equal lengh nervals, resulng n 16 dfferen T, I combnaons. On he oher hand, we have consdered low, medum and hgh demand varabl for a good mx of problems, randoml generang d parame- ers n he nervals of [10, 20], [10, 40] and [10, 60], respecvel. Ths resuls n 48 combnaons, where for each combnaon, we have generaed 5 es nsances. The capaces n each perod are generaed as a random varable from he nerval [0.8 I md, 1.2 I md ], where md ndcaes he medan demand n ha nerval. Fnall, we noe ha he holdng coss h are randoml generaed from he nerval [0.1, 1] and he seup coss f akes a value of {1, 10, 50}, each wh probabl of 1 3, n order o generae a good mx of low and hgh seup cos ems and n beween. Nex, we presen he compuaonal resuls for low, medum and hgh demand varabl, n Tables 1, 2 and 3, respecvel. Snce he exac separaon procedures are sgnfcanl meconsumng for our proposed nequales, we generaed onl one round of volaed cus and added o he formulaon. In each able, he frs column ndcaes he combnaon T, I, followed b he columns ndcang average I nal G ap, G ap C losed b F low C over nequales onl.e., onl exended flow cover nequales are generaed, and he percenage G ap C losed for 5 nsances wh all he volaed cus generaed. Noe ha he nal gap s based on he srenghened LP relaxaon wh all volaed l, S nequales added a pror, whch are known o be ver effecve n pracce for mul-em problems, see, e.g., Akarunalı & Mller, In he remander of he ables, he columns ndcae he oal number of cus generaed of each pe for 5 nsances, n he followng order: C over 8, I em-exended C over 9, P erod-exended C over 13, I em-and- P erod-exended C over 15, P aron 10, I em-exended P aron 11, P erodexended P aron 17, I em-and- P erod-exended P aron 19, T wo- P erod P aron 20, and F low C over 7. We noe ha for bgges nsances of d [10, 40], [10, 60], he separaon procedure for wo-perod paron nequales ook prohbve mes and herefore he were removed from he framework, whch are ndcaed b n he ables. As he resuls n Tables 1 3 ndcae, he cus can close on average more han 25% of he nal gap. As one could naurall expec, he average gap closed b he cus deeroraes when eher he number of ems or he number of perods s ncreased, where hs deeroraon seems more sensve o he ncrease n he numbers of ems han o he ncrease n he numbers of perods. When he number of ems ncreases, he problem resembles more he srucure of an uncapacaed problem, he convex hull of whch can be effecvel descrbed b he l, S nequales and hence here s lle room for mprovemen b oher cus. Ths can be ndeed conssenl observed from he average nal gaps for all he problems wh 10 ems. On he oher hand, as he number of perods ncreases, he problem becomes furher awa from he deal wo-perod problem, for whch hese cus are orgnall derved. However, we noe ha when all nsances wh 10 ems are aken ou, even he average gap closed for he nsances wh 12 perods s 23.05%, whch s a subsanal mprovemen wh onl one round of added cus, and also onl slghl lower han 24.48%, he average gap closed for he nsances wh 6 perods and 3/4/6 ems. As we wll dscuss n Secon 7, he expermenaon wh 24 perod problems, albe usng a heursc approach, ndcae smlar gap mprovemens compared o nsances wh 12 perods. The resuls also ndcae whch pes of nequales are more nheren for dfferen szes of problems. A small number of cover nequales seem o close subsanal gaps for he 2-perod problems almos onl on her own, and he number of hese nequales do no var much as he problem sze ges bgger. On he oher hand, he number of paron nequales and wo-perod paron nequales generaed ncreases sgnfcanl as he number of ems and perods ncrease, makng boh pes of nequales mos ofen generaed nequales n our framework, hence also ponng o where an compuaonall effcen framework can focus on. We also noe our observaon from he compuaonal ess ha addng wo-perod paron nequales on op of cover and paron nequales and her varans s no ver effecve n closng he negral gap, alhough hs mgh also be he consequence of a sngle round of separaon. Anoher neresng aspec he resuls pon a s he fac ha he number of em- and perod-exended versons of he cus reman small compared o he smple versons of hese cus. Fnall, we make a remark on he effec of he proposed cus when he demand varabl changes. As he ables clearl ndcae, he cus n parcular paron nequales and s varans are more ofen volaed when he demand varabl ncreases: hese are also he nsances when our cus make more of an mpac for he amoun of he gap closed. Ths makes nuve sense ha paron nequales are more flexble han covers and hence a hgher demand varabl wll be able o generae more volaed nequales of hs pe. 7. A Heursc separaon approach As he exac separaon procedures dscussed n Secon 5 are sgnfcanl me-consumng and even prohbve wh respec

8 M. Doosmohammad, K. Akarunalı / European Journal of Operaonal Research Table 1 Average closed gaps and number of cus generaed of each pe for es problems wh d [10, 20]. # Cus generaed T, I IG GCFC GC C IC PC IPC P IP PP IPP TPP FC 2, , , , , , , , , , , , , , , , Ave = Table 2 Average closed gaps and number of cus generaed of each pe for es problems wh d [10, 40]. # Cus generaed T, I IG GCFC GC C IC PC IPC P IP PP IPP TPP FC 2, , , , , , , , , , , , , , , , Ave = o wo-perod paron nequales for larges nsances esed n Secon 6, we canno expec o generae all cus whn accepable compuaonal mes, n parcular for large nsances. Ths movaed us o furher analze he naure of he volaed nequales denfed b he exac separaon framework, n order o develop a smple bu effecve heursc approach for subsanall reducng he compuaonal mes requred b he separaon process. Frs of all, we conduced an exensve compuaonal expermen focusng prmarl on he cover nequales 8 and paron nequales 10, as her numbers and effecveness plaed a sgnfcan role n he gaps closed for he nsances esed n he prevous secon. In order o acheve unbased resuls, we generaed 80 new random nsances wh d [10, 60] and analzed he volaed nequales denfed. Ths analss ndcaed ha major of he volaed nequales of cover and paron nequales are generaed from he wo-perod subproblems conssng of wo consecuve perods raher han wo non-consecuve perods. Ths provdes a sgnfcan poenal for reducon n he number of separaon problems solved e.g., n a 12 perod problem, here are 66 possble combnaons of wo perods, whereas he consecuve wo perods are onl lmed o 11 combnaons, and unsurprsngl, for hese 80 nsances, we observed on average 70% mprovemen n compuaonal mes b generang consecuve wo-perod subproblems compared o all wo-perod subproblems. Mos mporanl, usng consecuve wo-perod subproblems, he closed gaps obaned were on average 95% of he closed gaps obaned b generang cus from all wo-perod subproblems. These observaons movaed us o generae he volaed nequales from he consecuve wo-perod subproblems onl. In addon, we performed an analss on he wo-perod paron nequales denfed for a randoml seleced se of 12 perod nsances from Secon 6. As dscussed earler, he exac separaon of hese nequales requres subsanall more compuaonal me compared o oher pes of nequales due o he sheer number of possble combnaons of ξ and ξ values n he separaon problem and hence he were no generaed for he larges nsances esed n Secon 6, and n addon, her effecveness for closng he dual gaps were observed o be raher lmed. Therefore, we lmed our analss here o prevousl used nsances raher han newl generaed nsances. Frs of all, smlar o our observaon for oher pes of nequales, we observed ha mos of he volaed wo-perod paron nequales were generaed from consecuve wo-perod subproblems. We also observed ha mos of he volaed wo-perod paron nequales occurred n he cenre of he plannng horzon, raher han earler or laer perods. Our prelmnar analss suggesed o focus on he woperod subproblems from he cenral 1/3 of he horzon, so, e.g., for a 12-perod problem, perods 5 o 8 o be used for he n he heursc separaon. Alhough we could no observe an parcular paern for he poenall more effecve value ranges of ξ,

9 94 M. Doosmohammad, K. Akarunalı / European Journal of Operaonal Research Table 3 Average closed gaps and number of cus generaed of each pe for es problems wh d [10, 60]. # Cus generaed T, I IG GCFC GC C IC PC IPC P IP PP IPP TPP FC 2, , , , , , , , , , , , , , , , Ave = we made he observaon ha ξ values ofen concde wh he ξ values of he paron nequales 10 from perod. Ths movaed us for he smple bu sgnfcanl me savng approach as presened n Algorhm 1. Algorhm 1: Heursc separaon algorhm for wo-perod paron nequales 20. Inpu : A fraconal soluon x, ȳ, s ; a wo-perod subproblem X 2 PL for, n he cenre of he plannng horzon Oupu : A volaed wo-perod paron nequal for ξ > 0 do f ξ { ξ here exss volaed paron nequal for ξ } hen for ξ > 0 do Solve he maxmzaon problem 2PPI f f ξ,ξ > C + C and S S hen Add he volaed cu 20 end end end end I s noeworh o remark ha even hough we are sll solvng he exac separaon problems of Secon 5, hs heursc algorhm sgnfcanl reduces he number of problems o be solved, and herefore, a subsanal reducon n compuaonal mes requred can be acheved. Thanks o hs advanage and based on our prelmnar ess, we decded o run 3 rounds of separaon wh hs heursc approach, as opposed o he sngle round we performed wh he exac separaon approach. The resuls are presened n Table 4, whch follows he same noaon and smlar srucure used n Secon 6, wh he addon of specfng he gaps closed wh he EXAc and HEUrsc approaches for he sake of comparson. Frs of all, when we compare he average performance of he heursc approach wh he performance of he exac approach for he nsances wh 12 perods, we noe ha s ver compeve, and even ofen superor, hanks o he advanage of 3 rounds of separaon and herefore beng able o denf furher nequales whch could no be denfed b he sngle round of he exac separaon approach. As expeced, compuaonal mes are sgnfcanl mproved, for example, nsances wh N T = 12, N I = 10 and d [10, 60] ook on average 7 das usng he exac approach whereas he heursc approach ook on average onl 3.5 hours for he same nsances. In order o furher es our heursc approach on larger nsances, we have also generaed nsances wh 24 perods, usng he same random generaon procedure descrbed n Secon 6. As compuaonal mes sgnfcanl ncrease even n case of our heursc, we noe ha he larges and mos challengng nsances n hs se.e., nsances wh N T = 24, N I = 10 and d [10, 60] ook on average 1.5 das, whch s sll superor o he mes aken b he exac separaon for nsances wh 12 perods. We presen our resuls n Table 4, where, for a gven number of ems, he average gap closure of he heursc for hese nsances s no sgnfcanl dfferen from s average performance on he nsances wh 12 perods. I s also noeworh o remark ha on average over all hese 120 nsances wh 12 and 24 perods, gaps are closed around 18%. We conclude ha n addon o he subsanal gans n compuaonal mes and abl o expermen wh larger nsances, he smple heursc framework acheves successful closure of gaps. 8. Conclusons In hs paper, we nvesgaed a wo-perod subproblem of he bg-bucke lo-szng problem from a heorecal perspecve. In parcular, we have denfed varous famles of vald nequales for a relaxaon of hs subproblem n he specal case of zero seup mes, descrbed her face-defnng properes, and we have also mapped and exended hese nequales o he orgnal space of he wo-perod subproblem. The compuaonal resuls ndcaed sgnfcan poenal for mprovng lower bounds, and we are currenl nvesgang hs horoughl n a companon sud n wo mmedae drecons: undersandng polhedral characerscs of he general case wh non-zero seup mes and denfng furher vald nequales, and desgnng a branch-and-cu framework wh rounes generang cung planes of boh zero and non-zero seup me sengs n realsc mes for mul-em loszng problems. The heorecal resuls we presened n hs paper can be exended o oher MIP problems hanks o he commonal of he mxed neger ses nheren n dfferen problems. We have alread noed varous sudes on he polhedron of he PIR 0, he sngle node fxed charge se, whch s a common mxed neger se n varous MIP problems. On he oher hand, he srucure of PIR 1 poses dfferen challenges and opporunes, and s worh nvesgang furher s lnk o oher mxed neger ses. Fnall, here s also mmedae neres n nvesgang f and how our undersand-

10 M. Doosmohammad, K. Akarunalı / European Journal of Operaonal Research Table 4 Average closed gaps usng he heursc procedure HEU vs. usng he exac separaon EXA. d [10, 20] d [10, 40] d [10, 60] GC GC GC T, I IG EXA HEU IG EXA HEU IG EXA HEU 12, , , , , , , , Ave = ng of he wo-perod subproblems can be furher exended o more sophscaed subproblems, e.g., a k -perod subproblem. As Van Vve and Wolse 2005 observed n her framework, even lmng o he values of k = 3 and k = 4, here s subsanal poenal o develop a horough undersandng of he complex loszng problems, whch we plan o sud n he fuure. Supplemenar maeral Supplemenar maeral assocaed wh hs arcle can be found, n he onlne verson, a /j.ejor References Akarunalı, K., Doosmohammad, M., & Fragkos, I A branch-and-cu algorhm usng wo-perod relaxaons for bg-bucke lo-szng problems wh nonzero seup mes. Workng paper, Unvers of Srahclde Busness School. Akarunalı, K., Fragkos, I., Mller, A., & Wu, T Local cus and wo-perod convex hull closures for bg-bucke lo-szng problems. INFORMS Journal on Compung, 28 4, Akarunalı, K., & Mller, A A compuaonal analss of lower bounds for bg bucke producon plannng problems. Compuaonal Opmzaon and Applcaons, 53 3, de Araujo, S., Reck, B. D., Degraeve, Z., Fragkos, I., & Jans, R Perod decomposons for he capacaed lo szng problem wh seup mes. INFORMS Journal on Compung, 27 3, Baran, I., Van Ro, T., & Wolse, L Srong formulaons for mul-em capacaed lo-szng. Managemen Scence, 30 10, Belvaux, G., & Wolse, L bc-prod: A specalzed branch-and-cu ssem for lo-szng problems. Managemen Scence, 46 5, Bllngon, P., McClan, J., & Thomas, L Heurscs for mullevel lo-szng wh a boleneck. Managemen Scence, 32, Buschkühl, L., Sahlng, F., Helber, S., & Tempelmeer, H Dnamc capacaed lo szng problems: A classfcaon and revew of soluon approaches. OR Specrum, 32, Chen, W.-H., & Thz, J.-M Analss of relaxaons for he mul-em capacaed lo-szng problem. Annals of Operaons Research, 26 1, Chrsof, T., & Lobel, A PORTA Polhedron represenaon ransformaon algorhm. hp://po.zb.de/op-long _ projecs/sofware/pora/, Accessed on 24 Januar Degraeve, Z., & Jans, R A new Danzg Wolfe reformulaon and branch-and-prce algorhm for he capacaed lo-szng problem wh seup mes. Operaons Research, 55 5, Dxon, P., & Slver, E A heursc soluon procedure for he mul-em, sngle-level, lmed capac, lo-szng problem. Journal of Operaons Managemen, 2 1, Eppen, G., & Marn, R Solvng mul-em capacaed lo-szng problems usng varable redefnon. Operaons Research, 35 6, Federgruen, A., Messner, J., & Tzur, M Progressve nerval heurscs for mul-em capacaed lo szng problem. Operaons Research, 55 3, Jans, R., & Degraeve, Z Improved lower bounds for he capacaed lo szng problem wh seup mes. Operaons Research Leers, 32, Jans, R., & Degraeve, Z Mea-heurscs for dnamc lo szng: A revew and comparson of soluon approaches. European Journal of Operaonal Research, 177 3, Jans, R., & Degraeve, Z Modelng ndusral lo szng problems: A revew. Inernaonal Journal of Producon Research, 46 6, Krarup, J., & Blde, O Plan locaon, se coverng and economc loszes: An omn algorhm for srucured problems. Opmerung be Graphenheoreschen und Ganzzahlgen Probleme pp Brkhauser Verlag. Küçükavuz, S., & Poche, Y Uncapacaed lo-szng wh backloggng: The convex hull. Mahemacal Programmng, 118 1, Kuk, R., Salomon, M., & van Wassenhove, L Bachng decsons: Srucure and models. European Journal of Operaonal Research, 75 2, Mller, A., Nemhauser, G., & Savelsbergh, M Solvng he mul-em capacaed lo-szng problem wh seup mes b branch-and-cu. Techncal repor CORE DP /39, Unversé Caholque de Louvan, Louvan-la-Neuve. Mller, A., Nemhauser, G., & Savelsbergh, M On he polhedral srucure of a mul-em producon plannng model wh seup mes. Mahemacal Programmng, , Olaan, O., & Geragh, J Evaluaon of producon conrol sraeges for neglgble-seup, mul-produc, seral lnes wh consderaon for robusness. Journal of Manufacurng Technolog Managemen, 24 3, Padberg, M., van Ro, T., & Wolse, L Vald lnear nequales for fxed charge problems. Operaons Research, 33 4, Poche, Y., & Wolse, L Polhedra for lo-szng wh Wagner Whn coss. Mahemacal Programmng, , Poche, Y., & Wolse, L Producon plannng b mxed neger programmng. Sprnger. Poche, Y., & Wolse, L Sngle em lo-szng wh non-decreasng capaces. Mahemacal Programmng, 121 1, Van Vve, M., & Wolse, L Approxmae exended formulaons. Mahemacal Programmng, 105 3, Van Vve, M., Wolse, L., & Yaman, H Relaxaons for wo-level mul-em lo-szng problems. Mahemacal Programmng, , Zhang, M., Küçükavuz, S., & Yaman, H A polhedral sud of mulechelon lo szng wh nermedae demands. Operaons Research, 60 4,

V.Abramov - FURTHER ANALYSIS OF CONFIDENCE INTERVALS FOR LARGE CLIENT/SERVER COMPUTER NETWORKS

V.Abramov - FURTHER ANALYSIS OF CONFIDENCE INTERVALS FOR LARGE CLIENT/SERVER COMPUTER NETWORKS R&RATA # Vol.) 8, March FURTHER AALYSIS OF COFIDECE ITERVALS FOR LARGE CLIET/SERVER COMPUTER ETWORKS Vyacheslav Abramov School of Mahemacal Scences, Monash Unversy, Buldng 8, Level 4, Clayon Campus, Wellngon

More information

THE PREDICTION OF COMPETITIVE ENVIRONMENT IN BUSINESS

THE PREDICTION OF COMPETITIVE ENVIRONMENT IN BUSINESS THE PREICTION OF COMPETITIVE ENVIRONMENT IN BUSINESS INTROUCTION The wo dmensonal paral dfferenal equaons of second order can be used for he smulaon of compeve envronmen n busness The arcle presens he

More information

Dynamic Team Decision Theory. EECS 558 Project Shrutivandana Sharma and David Shuman December 10, 2005

Dynamic Team Decision Theory. EECS 558 Project Shrutivandana Sharma and David Shuman December 10, 2005 Dynamc Team Decson Theory EECS 558 Proec Shruvandana Sharma and Davd Shuman December 0, 005 Oulne Inroducon o Team Decson Theory Decomposon of he Dynamc Team Decson Problem Equvalence of Sac and Dynamc

More information

Solution in semi infinite diffusion couples (error function analysis)

Solution in semi infinite diffusion couples (error function analysis) Soluon n sem nfne dffuson couples (error funcon analyss) Le us consder now he sem nfne dffuson couple of wo blocks wh concenraon of and I means ha, n a A- bnary sysem, s bondng beween wo blocks made of

More information

A Principled Approach to MILP Modeling

A Principled Approach to MILP Modeling A Prncpled Approach o MILP Modelng John Hooer Carnege Mellon Unvers Augus 008 Slde Proposal MILP modelng s an ar, bu need no be unprncpled. Slde Proposal MILP modelng s an ar, bu need no be unprncpled.

More information

Dual Approximate Dynamic Programming for Large Scale Hydro Valleys

Dual Approximate Dynamic Programming for Large Scale Hydro Valleys Dual Approxmae Dynamc Programmng for Large Scale Hydro Valleys Perre Carpener and Jean-Phlppe Chanceler 1 ENSTA ParsTech and ENPC ParsTech CMM Workshop, January 2016 1 Jon work wh J.-C. Alas, suppored

More information

Variants of Pegasos. December 11, 2009

Variants of Pegasos. December 11, 2009 Inroducon Varans of Pegasos SooWoong Ryu bshboy@sanford.edu December, 009 Youngsoo Cho yc344@sanford.edu Developng a new SVM algorhm s ongong research opc. Among many exng SVM algorhms, we wll focus on

More information

Reactive Methods to Solve the Berth AllocationProblem with Stochastic Arrival and Handling Times

Reactive Methods to Solve the Berth AllocationProblem with Stochastic Arrival and Handling Times Reacve Mehods o Solve he Berh AllocaonProblem wh Sochasc Arrval and Handlng Tmes Nsh Umang* Mchel Berlare* * TRANSP-OR, Ecole Polyechnque Fédérale de Lausanne Frs Workshop on Large Scale Opmzaon November

More information

CHAPTER 10: LINEAR DISCRIMINATION

CHAPTER 10: LINEAR DISCRIMINATION CHAPER : LINEAR DISCRIMINAION Dscrmnan-based Classfcaon 3 In classfcaon h K classes (C,C,, C k ) We defned dscrmnan funcon g j (), j=,,,k hen gven an es eample, e chose (predced) s class label as C f g

More information

Online Supplement for Dynamic Multi-Technology. Production-Inventory Problem with Emissions Trading

Online Supplement for Dynamic Multi-Technology. Production-Inventory Problem with Emissions Trading Onlne Supplemen for Dynamc Mul-Technology Producon-Invenory Problem wh Emssons Tradng by We Zhang Zhongsheng Hua Yu Xa and Baofeng Huo Proof of Lemma For any ( qr ) Θ s easy o verfy ha he lnear programmng

More information

GENERATING CERTAIN QUINTIC IRREDUCIBLE POLYNOMIALS OVER FINITE FIELDS. Youngwoo Ahn and Kitae Kim

GENERATING CERTAIN QUINTIC IRREDUCIBLE POLYNOMIALS OVER FINITE FIELDS. Youngwoo Ahn and Kitae Kim Korean J. Mah. 19 (2011), No. 3, pp. 263 272 GENERATING CERTAIN QUINTIC IRREDUCIBLE POLYNOMIALS OVER FINITE FIELDS Youngwoo Ahn and Kae Km Absrac. In he paper [1], an explc correspondence beween ceran

More information

A Tour of Modeling Techniques

A Tour of Modeling Techniques A Tour of Modelng Technques John Hooker Carnege Mellon Unversy EWO Semnar February 8 Slde Oulne Med neger lnear (MILP) modelng Dsuncve modelng Knapsack modelng Consran programmng models Inegraed Models

More information

Modeling and Solving of Multi-Product Inventory Lot-Sizing with Supplier Selection under Quantity Discounts

Modeling and Solving of Multi-Product Inventory Lot-Sizing with Supplier Selection under Quantity Discounts nernaonal ournal of Appled Engneerng Research SSN 0973-4562 Volume 13, Number 10 (2018) pp. 8708-8713 Modelng and Solvng of Mul-Produc nvenory Lo-Szng wh Suppler Selecon under Quany Dscouns Naapa anchanaruangrong

More information

Robustness Experiments with Two Variance Components

Robustness Experiments with Two Variance Components Naonal Insue of Sandards and Technology (NIST) Informaon Technology Laboraory (ITL) Sascal Engneerng Dvson (SED) Robusness Expermens wh Two Varance Componens by Ana Ivelsse Avlés avles@ns.gov Conference

More information

Solving the multi-period fixed cost transportation problem using LINGO solver

Solving the multi-period fixed cost transportation problem using LINGO solver Inernaonal Journal of Pure and Appled Mahemacs Volume 119 No. 12 2018, 2151-2157 ISSN: 1314-3395 (on-lne verson) url: hp://www.pam.eu Specal Issue pam.eu Solvng he mul-perod fxed cos ransporaon problem

More information

Comb Filters. Comb Filters

Comb Filters. Comb Filters The smple flers dscussed so far are characered eher by a sngle passband and/or a sngle sopband There are applcaons where flers wh mulple passbands and sopbands are requred Thecomb fler s an example of

More information

UNIVERSITAT AUTÒNOMA DE BARCELONA MARCH 2017 EXAMINATION

UNIVERSITAT AUTÒNOMA DE BARCELONA MARCH 2017 EXAMINATION INTERNATIONAL TRADE T. J. KEHOE UNIVERSITAT AUTÒNOMA DE BARCELONA MARCH 27 EXAMINATION Please answer wo of he hree quesons. You can consul class noes, workng papers, and arcles whle you are workng on he

More information

CS286.2 Lecture 14: Quantum de Finetti Theorems II

CS286.2 Lecture 14: Quantum de Finetti Theorems II CS286.2 Lecure 14: Quanum de Fne Theorems II Scrbe: Mara Okounkova 1 Saemen of he heorem Recall he las saemen of he quanum de Fne heorem from he prevous lecure. Theorem 1 Quanum de Fne). Le ρ Dens C 2

More information

EEL 6266 Power System Operation and Control. Chapter 5 Unit Commitment

EEL 6266 Power System Operation and Control. Chapter 5 Unit Commitment EEL 6266 Power Sysem Operaon and Conrol Chaper 5 Un Commmen Dynamc programmng chef advanage over enumeraon schemes s he reducon n he dmensonaly of he problem n a src prory order scheme, here are only N

More information

In spite of the remarkable improvements in the quality of general purpose mixed-integer

In spite of the remarkable improvements in the quality of general purpose mixed-integer Modellng Praccal Lo-Szng Problems as Mxed-Ineger Programs Gaean Belvaux Laurence A. Wolsey Cockerll-Sambre, Flémalle Work carred ou a CORE, Unversé Caholque de Louvan CORE and INMA, Unversé Caholque de

More information

On One Analytic Method of. Constructing Program Controls

On One Analytic Method of. Constructing Program Controls Appled Mahemacal Scences, Vol. 9, 05, no. 8, 409-407 HIKARI Ld, www.m-hkar.com hp://dx.do.org/0.988/ams.05.54349 On One Analyc Mehod of Consrucng Program Conrols A. N. Kvko, S. V. Chsyakov and Yu. E. Balyna

More information

The Dynamic Programming Models for Inventory Control System with Time-varying Demand

The Dynamic Programming Models for Inventory Control System with Time-varying Demand The Dynamc Programmng Models for Invenory Conrol Sysem wh Tme-varyng Demand Truong Hong Trnh (Correspondng auhor) The Unversy of Danang, Unversy of Economcs, Venam Tel: 84-236-352-5459 E-mal: rnh.h@due.edu.vn

More information

Notes on the stability of dynamic systems and the use of Eigen Values.

Notes on the stability of dynamic systems and the use of Eigen Values. Noes on he sabl of dnamc ssems and he use of Egen Values. Source: Macro II course noes, Dr. Davd Bessler s Tme Seres course noes, zarads (999) Ineremporal Macroeconomcs chaper 4 & Techncal ppend, and Hamlon

More information

Lecture 6: Learning for Control (Generalised Linear Regression)

Lecture 6: Learning for Control (Generalised Linear Regression) Lecure 6: Learnng for Conrol (Generalsed Lnear Regresson) Conens: Lnear Mehods for Regresson Leas Squares, Gauss Markov heorem Recursve Leas Squares Lecure 6: RLSC - Prof. Sehu Vjayakumar Lnear Regresson

More information

On computing differential transform of nonlinear non-autonomous functions and its applications

On computing differential transform of nonlinear non-autonomous functions and its applications On compung dfferenal ransform of nonlnear non-auonomous funcons and s applcaons Essam. R. El-Zahar, and Abdelhalm Ebad Deparmen of Mahemacs, Faculy of Scences and Humanes, Prnce Saam Bn Abdulazz Unversy,

More information

An introduction to Support Vector Machine

An introduction to Support Vector Machine An nroducon o Suppor Vecor Machne 報告者 : 黃立德 References: Smon Haykn, "Neural Neworks: a comprehensve foundaon, second edon, 999, Chaper 2,6 Nello Chrsann, John Shawe-Tayer, An Inroducon o Suppor Vecor Machnes,

More information

In the complete model, these slopes are ANALYSIS OF VARIANCE FOR THE COMPLETE TWO-WAY MODEL. (! i+1 -! i ) + [(!") i+1,q - [(!

In the complete model, these slopes are ANALYSIS OF VARIANCE FOR THE COMPLETE TWO-WAY MODEL. (! i+1 -! i ) + [(!) i+1,q - [(! ANALYSIS OF VARIANCE FOR THE COMPLETE TWO-WAY MODEL The frs hng o es n wo-way ANOVA: Is here neracon? "No neracon" means: The man effecs model would f. Ths n urn means: In he neracon plo (wh A on he horzonal

More information

Lecture VI Regression

Lecture VI Regression Lecure VI Regresson (Lnear Mehods for Regresson) Conens: Lnear Mehods for Regresson Leas Squares, Gauss Markov heorem Recursve Leas Squares Lecure VI: MLSC - Dr. Sehu Vjayakumar Lnear Regresson Model M

More information

CS434a/541a: Pattern Recognition Prof. Olga Veksler. Lecture 4

CS434a/541a: Pattern Recognition Prof. Olga Veksler. Lecture 4 CS434a/54a: Paern Recognon Prof. Olga Veksler Lecure 4 Oulne Normal Random Varable Properes Dscrmnan funcons Why Normal Random Varables? Analycally racable Works well when observaon comes form a corruped

More information

Cubic Bezier Homotopy Function for Solving Exponential Equations

Cubic Bezier Homotopy Function for Solving Exponential Equations Penerb Journal of Advanced Research n Compung and Applcaons ISSN (onlne: 46-97 Vol. 4, No.. Pages -8, 6 omoopy Funcon for Solvng Eponenal Equaons S. S. Raml *,,. Mohamad Nor,a, N. S. Saharzan,b and M.

More information

Tight results for Next Fit and Worst Fit with resource augmentation

Tight results for Next Fit and Worst Fit with resource augmentation Tgh resuls for Nex F and Wors F wh resource augmenaon Joan Boyar Leah Epsen Asaf Levn Asrac I s well known ha he wo smple algorhms for he classc n packng prolem, NF and WF oh have an approxmaon rao of

More information

Linear Response Theory: The connection between QFT and experiments

Linear Response Theory: The connection between QFT and experiments Phys540.nb 39 3 Lnear Response Theory: The connecon beween QFT and expermens 3.1. Basc conceps and deas Q: ow do we measure he conducvy of a meal? A: we frs nroduce a weak elecrc feld E, and hen measure

More information

Lecture 18: The Laplace Transform (See Sections and 14.7 in Boas)

Lecture 18: The Laplace Transform (See Sections and 14.7 in Boas) Lecure 8: The Lalace Transform (See Secons 88- and 47 n Boas) Recall ha our bg-cure goal s he analyss of he dfferenal equaon, ax bx cx F, where we emloy varous exansons for he drvng funcon F deendng on

More information

( ) () we define the interaction representation by the unitary transformation () = ()

( ) () we define the interaction representation by the unitary transformation () = () Hgher Order Perurbaon Theory Mchael Fowler 3/7/6 The neracon Represenaon Recall ha n he frs par of hs course sequence, we dscussed he chrödnger and Hesenberg represenaons of quanum mechancs here n he chrödnger

More information

Advanced time-series analysis (University of Lund, Economic History Department)

Advanced time-series analysis (University of Lund, Economic History Department) Advanced me-seres analss (Unvers of Lund, Economc Hsor Dearmen) 3 Jan-3 Februar and 6-3 March Lecure 4 Economerc echnues for saonar seres : Unvarae sochasc models wh Box- Jenns mehodolog, smle forecasng

More information

e-journal Reliability: Theory& Applications No 2 (Vol.2) Vyacheslav Abramov

e-journal Reliability: Theory& Applications No 2 (Vol.2) Vyacheslav Abramov June 7 e-ournal Relably: Theory& Applcaons No (Vol. CONFIDENCE INTERVALS ASSOCIATED WITH PERFORMANCE ANALYSIS OF SYMMETRIC LARGE CLOSED CLIENT/SERVER COMPUTER NETWORKS Absrac Vyacheslav Abramov School

More information

Combining Optimization and Constraint Programming

Combining Optimization and Constraint Programming Combnng Opmzaon and Consran Programmng J. N. Hooker Carnege Mellon Unvers 4 h Inernaonal Conference on Operaons Research March Havana Some work s jon wh Ignaco Grossmann Carnege Mellon Unvers Hak-Jn Km

More information

Testing a new idea to solve the P = NP problem with mathematical induction

Testing a new idea to solve the P = NP problem with mathematical induction Tesng a new dea o solve he P = NP problem wh mahemacal nducon Bacground P and NP are wo classes (ses) of languages n Compuer Scence An open problem s wheher P = NP Ths paper ess a new dea o compare he

More information

Department of Economics University of Toronto

Department of Economics University of Toronto Deparmen of Economcs Unversy of Torono ECO408F M.A. Economercs Lecure Noes on Heeroskedascy Heeroskedascy o Ths lecure nvolves lookng a modfcaons we need o make o deal wh he regresson model when some of

More information

( t) Outline of program: BGC1: Survival and event history analysis Oslo, March-May Recapitulation. The additive regression model

( t) Outline of program: BGC1: Survival and event history analysis Oslo, March-May Recapitulation. The additive regression model BGC1: Survval and even hsory analyss Oslo, March-May 212 Monday May 7h and Tuesday May 8h The addve regresson model Ørnulf Borgan Deparmen of Mahemacs Unversy of Oslo Oulne of program: Recapulaon Counng

More information

TSS = SST + SSE An orthogonal partition of the total SS

TSS = SST + SSE An orthogonal partition of the total SS ANOVA: Topc 4. Orhogonal conrass [ST&D p. 183] H 0 : µ 1 = µ =... = µ H 1 : The mean of a leas one reamen group s dfferen To es hs hypohess, a basc ANOVA allocaes he varaon among reamen means (SST) equally

More information

Econ107 Applied Econometrics Topic 5: Specification: Choosing Independent Variables (Studenmund, Chapter 6)

Econ107 Applied Econometrics Topic 5: Specification: Choosing Independent Variables (Studenmund, Chapter 6) Econ7 Appled Economercs Topc 5: Specfcaon: Choosng Independen Varables (Sudenmund, Chaper 6 Specfcaon errors ha we wll deal wh: wrong ndependen varable; wrong funconal form. Ths lecure deals wh wrong ndependen

More information

. The geometric multiplicity is dim[ker( λi. number of linearly independent eigenvectors associated with this eigenvalue.

. The geometric multiplicity is dim[ker( λi. number of linearly independent eigenvectors associated with this eigenvalue. Lnear Algebra Lecure # Noes We connue wh he dscusson of egenvalues, egenvecors, and dagonalzably of marces We wan o know, n parcular wha condons wll assure ha a marx can be dagonalzed and wha he obsrucons

More information

Graduate Macroeconomics 2 Problem set 5. - Solutions

Graduate Macroeconomics 2 Problem set 5. - Solutions Graduae Macroeconomcs 2 Problem se. - Soluons Queson 1 To answer hs queson we need he frms frs order condons and he equaon ha deermnes he number of frms n equlbrum. The frms frs order condons are: F K

More information

Lecture 11 SVM cont

Lecture 11 SVM cont Lecure SVM con. 0 008 Wha we have done so far We have esalshed ha we wan o fnd a lnear decson oundary whose margn s he larges We know how o measure he margn of a lnear decson oundary Tha s: he mnmum geomerc

More information

Mechanics Physics 151

Mechanics Physics 151 Mechancs Physcs 5 Lecure 9 Hamlonan Equaons of Moon (Chaper 8) Wha We Dd Las Tme Consruced Hamlonan formalsm H ( q, p, ) = q p L( q, q, ) H p = q H q = p H = L Equvalen o Lagrangan formalsm Smpler, bu

More information

Mechanics Physics 151

Mechanics Physics 151 Mechancs Physcs 5 Lecure 0 Canoncal Transformaons (Chaper 9) Wha We Dd Las Tme Hamlon s Prncple n he Hamlonan formalsm Dervaon was smple δi δ Addonal end-pon consrans pq H( q, p, ) d 0 δ q ( ) δq ( ) δ

More information

Mechanics Physics 151

Mechanics Physics 151 Mechancs Physcs 5 Lecure 9 Hamlonan Equaons of Moon (Chaper 8) Wha We Dd Las Tme Consruced Hamlonan formalsm Hqp (,,) = qp Lqq (,,) H p = q H q = p H L = Equvalen o Lagrangan formalsm Smpler, bu wce as

More information

Epistemic Game Theory: Online Appendix

Epistemic Game Theory: Online Appendix Epsemc Game Theory: Onlne Appendx Edde Dekel Lucano Pomao Marcano Snscalch July 18, 2014 Prelmnares Fx a fne ype srucure T I, S, T, β I and a probably µ S T. Le T µ I, S, T µ, βµ I be a ype srucure ha

More information

The preemptive resource-constrained project scheduling problem subject to due dates and preemption penalties: An integer programming approach

The preemptive resource-constrained project scheduling problem subject to due dates and preemption penalties: An integer programming approach Journal of Indusral Engneerng 1 (008) 35-39 The preempve resource-consraned projec schedulng problem subjec o due daes and preempon penales An neger programmng approach B. Afshar Nadjaf Deparmen of Indusral

More information

Robust and Accurate Cancer Classification with Gene Expression Profiling

Robust and Accurate Cancer Classification with Gene Expression Profiling Robus and Accurae Cancer Classfcaon wh Gene Expresson Proflng (Compuaonal ysems Bology, 2005) Auhor: Hafeng L, Keshu Zhang, ao Jang Oulne Background LDA (lnear dscrmnan analyss) and small sample sze problem

More information

Comparison of Differences between Power Means 1

Comparison of Differences between Power Means 1 In. Journal of Mah. Analyss, Vol. 7, 203, no., 5-55 Comparson of Dfferences beween Power Means Chang-An Tan, Guanghua Sh and Fe Zuo College of Mahemacs and Informaon Scence Henan Normal Unversy, 453007,

More information

New M-Estimator Objective Function. in Simultaneous Equations Model. (A Comparative Study)

New M-Estimator Objective Function. in Simultaneous Equations Model. (A Comparative Study) Inernaonal Mahemacal Forum, Vol. 8, 3, no., 7 - HIKARI Ld, www.m-hkar.com hp://dx.do.org/.988/mf.3.3488 New M-Esmaor Objecve Funcon n Smulaneous Equaons Model (A Comparave Sudy) Ahmed H. Youssef Professor

More information

ABSTRACT. KEYWORDS Hybrid, Genetic Algorithm, Shipping, Dispatching, Vehicle, Time Windows INTRODUCTION

ABSTRACT. KEYWORDS Hybrid, Genetic Algorithm, Shipping, Dispatching, Vehicle, Time Windows INTRODUCTION A HYBRID GENETIC ALGORITH FOR A DYNAIC INBOUND ORDERING AND SHIPPING AND OUTBOUND DISPATCHING PROBLE WITH HETEROGENEOUS VEHICLE TYPES AND DELIVERY TIE WINDOWS by Byung Soo Km, Woon-Seek Lee, and Young-Seok

More information

Approximate Analytic Solution of (2+1) - Dimensional Zakharov-Kuznetsov(Zk) Equations Using Homotopy

Approximate Analytic Solution of (2+1) - Dimensional Zakharov-Kuznetsov(Zk) Equations Using Homotopy Arcle Inernaonal Journal of Modern Mahemacal Scences, 4, (): - Inernaonal Journal of Modern Mahemacal Scences Journal homepage: www.modernscenfcpress.com/journals/jmms.aspx ISSN: 66-86X Florda, USA Approxmae

More information

. The geometric multiplicity is dim[ker( λi. A )], i.e. the number of linearly independent eigenvectors associated with this eigenvalue.

. The geometric multiplicity is dim[ker( λi. A )], i.e. the number of linearly independent eigenvectors associated with this eigenvalue. Mah E-b Lecure #0 Noes We connue wh he dscusson of egenvalues, egenvecors, and dagonalzably of marces We wan o know, n parcular wha condons wll assure ha a marx can be dagonalzed and wha he obsrucons are

More information

Survival Analysis and Reliability. A Note on the Mean Residual Life Function of a Parallel System

Survival Analysis and Reliability. A Note on the Mean Residual Life Function of a Parallel System Communcaons n Sascs Theory and Mehods, 34: 475 484, 2005 Copyrgh Taylor & Francs, Inc. ISSN: 0361-0926 prn/1532-415x onlne DOI: 10.1081/STA-200047430 Survval Analyss and Relably A Noe on he Mean Resdual

More information

[ ] 2. [ ]3 + (Δx i + Δx i 1 ) / 2. Δx i-1 Δx i Δx i+1. TPG4160 Reservoir Simulation 2018 Lecture note 3. page 1 of 5

[ ] 2. [ ]3 + (Δx i + Δx i 1 ) / 2. Δx i-1 Δx i Δx i+1. TPG4160 Reservoir Simulation 2018 Lecture note 3. page 1 of 5 TPG460 Reservor Smulaon 08 page of 5 DISCRETIZATIO OF THE FOW EQUATIOS As we already have seen, fne dfference appromaons of he paral dervaves appearng n he flow equaons may be obaned from Taylor seres

More information

HIERARCHICAL DECISIONS FOR LINEAR/NON-LINEAR DISJUNCTIVE PROBLEMS

HIERARCHICAL DECISIONS FOR LINEAR/NON-LINEAR DISJUNCTIVE PROBLEMS 2 nd Mercosur Congress on Chemcal Engneerng 4 h Mercosur Congress on Process Sysems Engneerng HIERARCHICAL DECISIONS FOR LINEAR/NON-LINEAR DISJUNCTIVE PROLEMS Jorge M. Monagna and Aldo R. Vecche * INGAR

More information

Chapter 6: AC Circuits

Chapter 6: AC Circuits Chaper 6: AC Crcus Chaper 6: Oulne Phasors and he AC Seady Sae AC Crcus A sable, lnear crcu operang n he seady sae wh snusodal excaon (.e., snusodal seady sae. Complee response forced response naural response.

More information

John Geweke a and Gianni Amisano b a Departments of Economics and Statistics, University of Iowa, USA b European Central Bank, Frankfurt, Germany

John Geweke a and Gianni Amisano b a Departments of Economics and Statistics, University of Iowa, USA b European Central Bank, Frankfurt, Germany Herarchcal Markov Normal Mxure models wh Applcaons o Fnancal Asse Reurns Appendx: Proofs of Theorems and Condonal Poseror Dsrbuons John Geweke a and Gann Amsano b a Deparmens of Economcs and Sascs, Unversy

More information

A NEW TECHNIQUE FOR SOLVING THE 1-D BURGERS EQUATION

A NEW TECHNIQUE FOR SOLVING THE 1-D BURGERS EQUATION S19 A NEW TECHNIQUE FOR SOLVING THE 1-D BURGERS EQUATION by Xaojun YANG a,b, Yugu YANG a*, Carlo CATTANI c, and Mngzheng ZHU b a Sae Key Laboraory for Geomechancs and Deep Underground Engneerng, Chna Unversy

More information

Computational results on new staff scheduling benchmark instances

Computational results on new staff scheduling benchmark instances TECHNICAL REPORT Compuaonal resuls on new saff shedulng enhmark nsanes Tm Curos Rong Qu ASAP Researh Group Shool of Compuer Sene Unersy of Nongham NG8 1BB Nongham UK Frs pulshed onlne: 19-Sep-2014 las

More information

Math 128b Project. Jude Yuen

Math 128b Project. Jude Yuen Mah 8b Proec Jude Yuen . Inroducon Le { Z } be a sequence of observed ndependen vecor varables. If he elemens of Z have a on normal dsrbuon hen { Z } has a mean vecor Z and a varancecovarance marx z. Geomercally

More information

ON THE WEAK LIMITS OF SMOOTH MAPS FOR THE DIRICHLET ENERGY BETWEEN MANIFOLDS

ON THE WEAK LIMITS OF SMOOTH MAPS FOR THE DIRICHLET ENERGY BETWEEN MANIFOLDS ON THE WEA LIMITS OF SMOOTH MAPS FOR THE DIRICHLET ENERGY BETWEEN MANIFOLDS FENGBO HANG Absrac. We denfy all he weak sequenal lms of smooh maps n W (M N). In parcular, hs mples a necessary su cen opologcal

More information

P R = P 0. The system is shown on the next figure:

P R = P 0. The system is shown on the next figure: TPG460 Reservor Smulaon 08 page of INTRODUCTION TO RESERVOIR SIMULATION Analycal and numercal soluons of smple one-dmensonal, one-phase flow equaons As an nroducon o reservor smulaon, we wll revew he smples

More information

New compact formulations for choice network revenue management

New compact formulations for choice network revenue management New compac formulaons for choce nework revenue managemen Sum Kunnumkal Kalyan Tallur Augus 13, 2015 Absrac The choce nework revenue managemen model ncorporaes cusomer purchase behavor as a funcon of he

More information

GORDON AND NEWELL QUEUEING NETWORKS AND COPULAS

GORDON AND NEWELL QUEUEING NETWORKS AND COPULAS Yugoslav Journal of Operaons Research Vol 9 (009) Number 0- DOI:0.98/YUJOR0900C GORDON AND NEWELL QUEUEING NETWORKS AND COPULAS Danel CIUIU Facul of Cvl Indusral and Agrculural Buldngs Techncal Unvers

More information

Lecture 2 L n i e n a e r a M od o e d l e s

Lecture 2 L n i e n a e r a M od o e d l e s Lecure Lnear Models Las lecure You have learned abou ha s machne learnng Supervsed learnng Unsupervsed learnng Renforcemen learnng You have seen an eample learnng problem and he general process ha one

More information

Ordinary Differential Equations in Neuroscience with Matlab examples. Aim 1- Gain understanding of how to set up and solve ODE s

Ordinary Differential Equations in Neuroscience with Matlab examples. Aim 1- Gain understanding of how to set up and solve ODE s Ordnary Dfferenal Equaons n Neuroscence wh Malab eamples. Am - Gan undersandng of how o se up and solve ODE s Am Undersand how o se up an solve a smple eample of he Hebb rule n D Our goal a end of class

More information

HEAT CONDUCTION PROBLEM IN A TWO-LAYERED HOLLOW CYLINDER BY USING THE GREEN S FUNCTION METHOD

HEAT CONDUCTION PROBLEM IN A TWO-LAYERED HOLLOW CYLINDER BY USING THE GREEN S FUNCTION METHOD Journal of Appled Mahemacs and Compuaonal Mechancs 3, (), 45-5 HEAT CONDUCTION PROBLEM IN A TWO-LAYERED HOLLOW CYLINDER BY USING THE GREEN S FUNCTION METHOD Sansław Kukla, Urszula Sedlecka Insue of Mahemacs,

More information

OPTIMIZATION OF A PRODUCTION LOT SIZING PROBLEM WITH QUANTITY DISCOUNT

OPTIMIZATION OF A PRODUCTION LOT SIZING PROBLEM WITH QUANTITY DISCOUNT NERNAONAL JOURNAL OF OPMZAON N CVL ENGNEERNG n. J. Opm. Cvl Eng., 26; 6(2:87-29 OPMZAON OF A PRODUCON LO SZNG PROBLEM WH QUANY DSCOUN S. hosrav and S.H. Mrmohammad *, Deparmen of ndusral and sysems Engneerng,

More information

Panel Data Regression Models

Panel Data Regression Models Panel Daa Regresson Models Wha s Panel Daa? () Mulple dmensoned Dmensons, e.g., cross-secon and me node-o-node (c) Pongsa Pornchawseskul, Faculy of Economcs, Chulalongkorn Unversy (c) Pongsa Pornchawseskul,

More information

Relative controllability of nonlinear systems with delays in control

Relative controllability of nonlinear systems with delays in control Relave conrollably o nonlnear sysems wh delays n conrol Jerzy Klamka Insue o Conrol Engneerng, Slesan Techncal Unversy, 44- Glwce, Poland. phone/ax : 48 32 37227, {jklamka}@a.polsl.glwce.pl Keywor: Conrollably.

More information

Sampling Procedure of the Sum of two Binary Markov Process Realizations

Sampling Procedure of the Sum of two Binary Markov Process Realizations Samplng Procedure of he Sum of wo Bnary Markov Process Realzaons YURY GORITSKIY Dep. of Mahemacal Modelng of Moscow Power Insue (Techncal Unversy), Moscow, RUSSIA, E-mal: gorsky@yandex.ru VLADIMIR KAZAKOV

More information

Multi-priority Online Scheduling with Cancellations

Multi-priority Online Scheduling with Cancellations Submed o Operaons Research manuscrp (Please, provde he manuscrp number!) Auhors are encouraged o subm new papers o INFORMS journals by means of a syle fle emplae, whch ncludes he journal le. However, use

More information

Outline. Probabilistic Model Learning. Probabilistic Model Learning. Probabilistic Model for Time-series Data: Hidden Markov Model

Outline. Probabilistic Model Learning. Probabilistic Model Learning. Probabilistic Model for Time-series Data: Hidden Markov Model Probablsc Model for Tme-seres Daa: Hdden Markov Model Hrosh Mamsuka Bonformacs Cener Kyoo Unversy Oulne Three Problems for probablsc models n machne learnng. Compung lkelhood 2. Learnng 3. Parsng (predcon

More information

FTCS Solution to the Heat Equation

FTCS Solution to the Heat Equation FTCS Soluon o he Hea Equaon ME 448/548 Noes Gerald Reckenwald Porland Sae Unversy Deparmen of Mechancal Engneerng gerry@pdxedu ME 448/548: FTCS Soluon o he Hea Equaon Overvew Use he forward fne d erence

More information

DEEP UNFOLDING FOR MULTICHANNEL SOURCE SEPARATION SUPPLEMENTARY MATERIAL

DEEP UNFOLDING FOR MULTICHANNEL SOURCE SEPARATION SUPPLEMENTARY MATERIAL DEEP UNFOLDING FOR MULTICHANNEL SOURCE SEPARATION SUPPLEMENTARY MATERIAL Sco Wsdom, John Hershey 2, Jonahan Le Roux 2, and Shnj Waanabe 2 Deparmen o Elecrcal Engneerng, Unversy o Washngon, Seale, WA, USA

More information

Single-loop System Reliability-Based Design & Topology Optimization (SRBDO/SRBTO): A Matrix-based System Reliability (MSR) Method

Single-loop System Reliability-Based Design & Topology Optimization (SRBDO/SRBTO): A Matrix-based System Reliability (MSR) Method 10 h US Naonal Congress on Compuaonal Mechancs Columbus, Oho 16-19, 2009 Sngle-loop Sysem Relably-Based Desgn & Topology Opmzaon (SRBDO/SRBTO): A Marx-based Sysem Relably (MSR) Mehod Tam Nguyen, Junho

More information

A fuzzy approach to capacity constrained MRP systems *

A fuzzy approach to capacity constrained MRP systems * A fuzzy approach o capacy consraned MRP sysems * J. Mula CGP (Research Cenre on Producon Managemen and Engneerng) Polyechnc Unversy of Valenca Plaza Ferrándz y Carbonell, 080 Alcoy (Alcane) fmula@cgp.upv.es

More information

Analysis And Evaluation of Econometric Time Series Models: Dynamic Transfer Function Approach

Analysis And Evaluation of Econometric Time Series Models: Dynamic Transfer Function Approach 1 Appeared n Proceedng of he 62 h Annual Sesson of he SLAAS (2006) pp 96. Analyss And Evaluaon of Economerc Tme Seres Models: Dynamc Transfer Funcon Approach T.M.J.A.COORAY Deparmen of Mahemacs Unversy

More information

12d Model. Civil and Surveying Software. Drainage Analysis Module Detention/Retention Basins. Owen Thornton BE (Mech), 12d Model Programmer

12d Model. Civil and Surveying Software. Drainage Analysis Module Detention/Retention Basins. Owen Thornton BE (Mech), 12d Model Programmer d Model Cvl and Surveyng Soware Dranage Analyss Module Deenon/Reenon Basns Owen Thornon BE (Mech), d Model Programmer owen.hornon@d.com 4 January 007 Revsed: 04 Aprl 007 9 February 008 (8Cp) Ths documen

More information

Part II CONTINUOUS TIME STOCHASTIC PROCESSES

Part II CONTINUOUS TIME STOCHASTIC PROCESSES Par II CONTINUOUS TIME STOCHASTIC PROCESSES 4 Chaper 4 For an advanced analyss of he properes of he Wener process, see: Revus D and Yor M: Connuous marngales and Brownan Moon Karazas I and Shreve S E:

More information

College of William & Mary Department of Computer Science

College of William & Mary Department of Computer Science 1 Techncal Repor WM-CS-2012-01 College of Wllam & Mary Deparmen of Compuer Scence WM-CS-2012-01 Onlne Vecor Schedulng and Generalzed Load Balancng Xaojun Zhu, Qun L, Wezhen Mao, Guha Chen March 5, 2012

More information

Existence and Uniqueness Results for Random Impulsive Integro-Differential Equation

Existence and Uniqueness Results for Random Impulsive Integro-Differential Equation Global Journal of Pure and Appled Mahemacs. ISSN 973-768 Volume 4, Number 6 (8), pp. 89-87 Research Inda Publcaons hp://www.rpublcaon.com Exsence and Unqueness Resuls for Random Impulsve Inegro-Dfferenal

More information

ON THE LINEAR PROGRAMMING AND GROUP RELAXATIONS OF THE UNCAPACITATED FACILITY LOCATION PROBLEM

ON THE LINEAR PROGRAMMING AND GROUP RELAXATIONS OF THE UNCAPACITATED FACILITY LOCATION PROBLEM ON THE LINEAR PROGRAMMING AND GROUP RELAXATIONS O THE UNAPAITATED AILITY LOATION PROLEM MOHAMMAD KHALIL A THESIS PRESENTED TO THE GRADUATE SHOOL O THE UNIVERSITY O LORIDA IN PARTIAL ULILLMENT O THE REQUIREMENTS

More information

Efficient Asynchronous Channel Hopping Design for Cognitive Radio Networks

Efficient Asynchronous Channel Hopping Design for Cognitive Radio Networks Effcen Asynchronous Channel Hoppng Desgn for Cognve Rado Neworks Chh-Mn Chao, Chen-Yu Hsu, and Yun-ng Lng Absrac In a cognve rado nework (CRN), a necessary condon for nodes o communcae wh each oher s ha

More information

ELASTIC MODULUS ESTIMATION OF CHOPPED CARBON FIBER TAPE REINFORCED THERMOPLASTICS USING THE MONTE CARLO SIMULATION

ELASTIC MODULUS ESTIMATION OF CHOPPED CARBON FIBER TAPE REINFORCED THERMOPLASTICS USING THE MONTE CARLO SIMULATION THE 19 TH INTERNATIONAL ONFERENE ON OMPOSITE MATERIALS ELASTI MODULUS ESTIMATION OF HOPPED ARBON FIBER TAPE REINFORED THERMOPLASTIS USING THE MONTE ARLO SIMULATION Y. Sao 1*, J. Takahash 1, T. Masuo 1,

More information

ABSTRACT KEYWORDS. Bonus-malus systems, frequency component, severity component. 1. INTRODUCTION

ABSTRACT KEYWORDS. Bonus-malus systems, frequency component, severity component. 1. INTRODUCTION EERAIED BU-MAU YTEM ITH A FREQUECY AD A EVERITY CMET A IDIVIDUA BAI I AUTMBIE IURACE* BY RAHIM MAHMUDVAD AD HEI HAAI ABTRACT Frangos and Vronos (2001) proposed an opmal bonus-malus sysems wh a frequency

More information

Operations Research Letters

Operations Research Letters Operaons Research Leers 39 (2011) 323 328 Conens lss avalable a ScVerse ScenceDrec Operaons Research Leers journal homepage: www.elsever.com/locae/orl Rank of Handelman herarchy for Max-Cu Myoung-Ju Park,

More information

Clustering (Bishop ch 9)

Clustering (Bishop ch 9) Cluserng (Bshop ch 9) Reference: Daa Mnng by Margare Dunham (a slde source) 1 Cluserng Cluserng s unsupervsed learnng, here are no class labels Wan o fnd groups of smlar nsances Ofen use a dsance measure

More information

APOC #232 Capacity Planning for Fault-Tolerant All-Optical Network

APOC #232 Capacity Planning for Fault-Tolerant All-Optical Network APOC #232 Capacy Plannng for Faul-Toleran All-Opcal Nework Mchael Kwok-Shng Ho and Kwok-wa Cheung Deparmen of Informaon ngneerng The Chnese Unversy of Hong Kong Shan, N.T., Hong Kong SAR, Chna -mal: kwcheung@e.cuhk.edu.hk

More information

Performance Analysis for a Network having Standby Redundant Unit with Waiting in Repair

Performance Analysis for a Network having Standby Redundant Unit with Waiting in Repair TECHNI Inernaonal Journal of Compung Scence Communcaon Technologes VOL.5 NO. July 22 (ISSN 974-3375 erformance nalyss for a Nework havng Sby edundan Un wh ang n epar Jendra Sngh 2 abns orwal 2 Deparmen

More information

RELATIONSHIP BETWEEN VOLATILITY AND TRADING VOLUME: THE CASE OF HSI STOCK RETURNS DATA

RELATIONSHIP BETWEEN VOLATILITY AND TRADING VOLUME: THE CASE OF HSI STOCK RETURNS DATA RELATIONSHIP BETWEEN VOLATILITY AND TRADING VOLUME: THE CASE OF HSI STOCK RETURNS DATA Mchaela Chocholaá Unversy of Economcs Braslava, Slovaka Inroducon (1) one of he characersc feaures of sock reurns

More information

SOME NOISELESS CODING THEOREMS OF INACCURACY MEASURE OF ORDER α AND TYPE β

SOME NOISELESS CODING THEOREMS OF INACCURACY MEASURE OF ORDER α AND TYPE β SARAJEVO JOURNAL OF MATHEMATICS Vol.3 (15) (2007), 137 143 SOME NOISELESS CODING THEOREMS OF INACCURACY MEASURE OF ORDER α AND TYPE β M. A. K. BAIG AND RAYEES AHMAD DAR Absrac. In hs paper, we propose

More information

CS 268: Packet Scheduling

CS 268: Packet Scheduling Pace Schedulng Decde when and wha pace o send on oupu ln - Usually mplemened a oupu nerface CS 68: Pace Schedulng flow Ion Soca March 9, 004 Classfer flow flow n Buffer managemen Scheduler soca@cs.bereley.edu

More information

Bernoulli process with 282 ky periodicity is detected in the R-N reversals of the earth s magnetic field

Bernoulli process with 282 ky periodicity is detected in the R-N reversals of the earth s magnetic field Submed o: Suden Essay Awards n Magnecs Bernoull process wh 8 ky perodcy s deeced n he R-N reversals of he earh s magnec feld Jozsef Gara Deparmen of Earh Scences Florda Inernaonal Unversy Unversy Park,

More information

Online Appendix for. Strategic safety stocks in supply chains with evolving forecasts

Online Appendix for. Strategic safety stocks in supply chains with evolving forecasts Onlne Appendx for Sraegc safey socs n supply chans wh evolvng forecass Tor Schoenmeyr Sephen C. Graves Opsolar, Inc. 332 Hunwood Avenue Hayward, CA 94544 A. P. Sloan School of Managemen Massachuses Insue

More information

( ) [ ] MAP Decision Rule

( ) [ ] MAP Decision Rule Announcemens Bayes Decson Theory wh Normal Dsrbuons HW0 due oday HW o be assgned soon Proec descrpon posed Bomercs CSE 90 Lecure 4 CSE90, Sprng 04 CSE90, Sprng 04 Key Probables 4 ω class label X feaure

More information