A Fast On-Chip Decoupling Capacitance Budgeting Algorithm Using Macromodeling and Linear Programming

Size: px
Start display at page:

Download "A Fast On-Chip Decoupling Capacitance Budgeting Algorithm Using Macromodeling and Linear Programming"

Transcription

1 1. A Fas On-Chp Decouplng Capacance Budgeng Algorhm Usng Macromodelng and Lnear Programmng Mn Zhao, Rajendran Panda, Savhr Sundareswaran, Shu Yan and Yuhong Fu Freescale Semconducor, Inc. ABSTRACT We propose a novel and effcen charge-based decouplng capacance budgeng algorhm. Our mehod uses he macromodelng echnque and effecve radus of decouplng capacance o reduce he sze of he problem. We formulae he nonlnear opmzaon no a lnear program (LP) by negrang he nodal equaons across a me perod of neres and hrough ceran approxmaons. To reduce he error caused by lnearzaon, we do mulple eraons of he lnear program. Expermenal resuls demonsrae ha, wh he proposed algorhm, even very large power neworks (eg. 5 mllon nodes) can be opmzed n a couple of hours wh 1- ransen analyses. Comparson of our algorhm wh anoher heursc mehod shows area effcency and run me advanage of our mehod. Caegores and Subjec Descrpors B.. [Hardware]: INTEGRATED CIRCUITS layou General Terms Algorhms, Performance, Relably, Verfcaon Keywords Decouplng capacance, budgeng, macromodelng, sequence of lnear programmng 1. INTRODUCTION Wh he ncrease n VLSI crcu frequency and supply volage scalng, desgnng a robus power dsrbuon nework has become a challengng ask. Sac IR drop s usually addressed hrough ncreased mealzaon (reducon of ressance), pad placemen, opology opmzaon and power densy aware floorplannng. However, he chef echnque for lmng dynamc volage flucuaons s o place decouplng capacors (known as decap, nanabbrevaed manner) close o problem spo. As he fabrca- Permsson o make dgal or hard copes of all or par of hs work for personal or classroom use s graned whou fee provded ha copes are no made or dsrbued for prof or commercal advanage and ha copes bear hs noce and he full caon on he frs page. To copy oherwse, o republsh, o pos on servers or o redsrbue o lss, requres pror specfc permsson and/or a fee. DAC 00, July 8, 00, San Francsco, Calforna, USA. Copyrgh 00 ACM /0/000...$5.00. on process moves up he echnology nodes ladder, he ncreased wre ressance s aggravang he supply nose problem. However, he hgh leakage curren n hese echnology nodes dscourages addon of abundan decouplng capacance. Decouplng capacance also uses up de area and affecs de yeld adversely. For hese reasons, dynamc supply nose needs o be addressed wh a mnmum amoun of decap whch needs o be placed opmally. In hs paper, we address hs mporan problem. We formulae he problem o mnmze he oal decap subjec o volage consrans on he nework nodes, and consrans on he decap amoun ha can be added a varous places. The soluon ams a realzng dynamc volages beer han a user-specfed hreshold level a all mes. Decap budgeng and placemen s a non-lnear opmzaon problem [1]. Recenly, several sensvy-based echnques were proposed. The auhors of [1] proposed a sensvy-based nonlnear program, whch s solved by quadrac programmng and usng a compressed pecewse lnear form o sore adjon sensvy. [] used he conjugae graden(cg) mehod, wh a merged adjon sensvy heursc o speed up he sensvy calculaon. [] reduced he problem sze usng he geomerc mulgrd concep and hen used a sequenal quadrac program on he reduced grd. Is applcaon s hus lmed o regular mesh srucures. In [], L, e.al. used dvde-and-conquer o reduce he sze of he sensvy-based opmzaon. [5] solved he nonlnear opmzaon hrough a sequence of lnear programmng mehod. Sensves o decaps were used as lnear consrans n he opmzaon. All hese mehods requre calculaon of sensves wh respec o decouplng capacance locaon. The adjon mehod of sensvy calculaon [ 8] needs o sore waveforms a every node n boh he orgnal and he adjon neworks, whch may exhaus he memory resource for large neworks. Moreover, he foresad mehods add decap and recompue sensvy n an erave procedure. As he complexy of one adjon sensvy compuaon s he same as one ransen analyss, he erave nonlnear opmzaon procedure becomes que expensve. Charge-based decap esmaon echnques have also been proposed [9, 10] n he conex of power supply nose-aware floorplannng. An approxmae lumped decouplng capacance s esmaed for each floorplan module, assumng full VDD volage as he nal volage of decouplng capacance. Our approach dffers from he prevous works n many respecs: (1) The charge bass of consrans, formed hrough negraon of he nodal equaons over a me perod of n- 1

2 eres, smplfes compuaons subsanally. Our approach frs lnearzes he charge-based consrans hrough ceran approxmaons (o be explaned furher) and hen performs a small number of eraons o accoun for he non-lneary. A small lnear program s solved n each eraon. () The conceps of macromodelng [11] and effecve radus of capacance are used for problem sze reducon. A small macromodel wh few pors, whch are he poenal ses for decap connecon, s creaed by reducng he nework. () I handles he decap pre-exsng n he nework and he decap o be added n dfferen ways. Inrnsc decap of devces and nerconnec and pre-placed decap cells are modeled nsde he macromodel hrough her companon models placed n he nework before he macromodelng procedure. Addonal decap ha s opmzed s exernal o he macromodel, and s conneced a he model s pors.. BACKGROUND.1 Overvew of Power Grd Smulaon A chp s power dsrbuon sysem s modeled as a lnear RLC nework wh ndependen me-varyng curren sources modelng he swchng currens of he ranssors. Smulang he nework requres solvng he followng sysem of dfferenal equaons, whch s formed by a Modfed Nodal Analyss (MNA) [1] approach: G x()+c x () =b(), (1) where G s a conducance marx, C s an admance marx resulng from capacve and nducve elemens, x() s he me-varyng vecor of volages a he nodes, and currens hrough nducors and volage sources, and b() s he vecor of ndependen me-varyng currens and volages from ndependen sources and companon models. Ths dfferenal sysem s very effcenly solved by reducng o a lnear algebrac sysem (G + C/h) x() =b()+c/h x( h), () usng Backward Euler (BE) echnque. A fxed me sep, h, s used o make he LHS marx saonary so ha s facors can be reused n a ransen smulaon.. Macromodelng A par of he power grd or he enre power dsrbuon nework can be modeled as a mul-por lnear elemen wh curren ransfer characerscs gven by I = A V + S, I, V, S R m,a R m m () where m s he number of pors n he model, A s he por admance marx, V s he vecor of por volages, I s he vecor of currens hrough he pors from an exernal nework no he model, and S s a vecor of curren sources conneced beween each por and he reference node. S essenally brngs he effec of movng all nernal curren sources o he pors. Macromodelng s he procedure of dervng Equaon () from he modfed nodal equaons () of a power grd, ncludng he nrnsc and already placed decouplng capacors modeled hrough her companon models usng backward Euler echnque.» G11 G 1 G T 1 G» U V =» J 1 J + I () where I s he vecor of currens hrough he nerface, and U and V are volages a he nernal nodes and pors respecvely. G 1, G 11, G are pars of he admance marx whch ncludes he companon conducance C/h of equaon (). J 1 and J are curren sources a he nernal nodes and pors respecvely, and nclude he companon currens C/h x( h) of equaon (). The dealed macromodelng procedure s descrbed n [11].. DECAP BUDGETING.1 Problem Formulaon We se ou o mnmze he oal capacance o be added a he nework nodes, subjec o he consrans ha () he dynamc node volages are beer han a specfed hreshold, () he capacance ha can be added a a node s bounded, and () he node volages sasfy he MNA equaons. mnmze subjec o P C () V k V hre, () C C max,, () V k should sasfy MNA a me pon k, for all power grd nodes, and me pons k Here, V hre s specfed by he user and C max, s se dependng on local congeson and he decap amoun realzable per un area.. Flow Oulne Overall flow of he proposed echnque s oulned below. Sep 1 Run ransen analyss of he power nework, wh nrnsc capacances and decouplng capacances ha have already been placed. Sep Inspec he ransen volage waveforms and deermne () he regons where he volage consrans are volaed (called volaon regons), () he me wndows durng whch volaons occur (called he volaon me wndows), and () a se of nodes o be used for macromodelng and opmzaon (called he samplng nodes). Ths sep s dealed ou n Secon.. If no volaon regon s found, hen opmzaon s compleed. Oherwse, connue wh sep. Sep For each volaon regon, deermne he opmal amoun of decap o be added a he samplng nodes. Ths s done by frs macromodelng he nework usng he samplng nodes n he volaon regon as he pors, and hen runnng few eraons of a lnear program descrbed n Secon and 5 on hs model. For volaon regons ha overlap each oher, he opmzaon for hese regons s done smulaneously. Sep Dsrbue he decap of he samplng nodes evenly no he respecve samplng regons and go o sep 1.. Volaon Regon, Samplng Nodes, and Volaon Tme Wndow A volaon regon s llusraed n Fgure 1. Frs, he de s paroned no unformly szed les usng pre-defned x and y pches. A se of conguous les (he cenral crcular regon plus he darkly shaded regon) wheren one or more nodes volae he volage consran defnes he core volaon regon. The core volaon regon s hen expanded n all drecons by a pre-deermned dsance (known as he 18

3 A le The core volaon regon The volaon regon Fgure 1: An llusraon of he volaon regon effecve radus) o nclude addonal les (lghly shaded les surroundng he darkly shaded ones). Ths expanded regon s called he volaon regon. The expanson s based on he fac ha decap added ousde he expanded regon has lle or no mpac on he volages n he core regon, and vce versa [1, ]. Moreover, he expanded regon provdes more space for decap addon. The large chunk of nework ousde he expanded regon can be absraced away by reducng hose nodes n he macromodel. There can be more han one volaon regon, hough only one s shown n he Fgure 1. By runnng he opmzaon ndependenly on dfferen volaon regons, we are hus able o reduce he problem sze sgnfcanly. Ths also brngs n an opporuny for parallelzaon. If wo or more volaon regons overlap each oher, hen hese regons are opmzed ogeher. Opmzng hem separaely n such cases wll resul n sub-opmal resuls as we wll hen gnore he srong neracon of capacances and volages across hese regons. The effecve radus for he decap depends on he effecve ressance whch s deermned by he echnology node, he curren varaon, and he mealzaon used n each layer. I s very easy o come up wh a conservave esmaon of how far he core regon should be expanded based on one-me expermens wh represenave desgns n a gven echnology, or hrough approxmae calculaons. Samplng nodes are represenave nodes sampled from each le havng a volage volaon. A few (less han 10) nodes per le wh wors volage volaon are sampled o represen he behavor of all nodes n hose les.we keep he le small enough for hs assumpon o be vald. The res of he nodes are absraced away by macromodelng. The LP formulaon assumes ha all he decap s added a he samplng nodes. Afer solvng he LP, however, we dsrbue he decap a he samplng nodes evenly among he nodes n he le regons ha he samplng nodes belong o. Vdd Vhre Wors Volage 0 1 Fgure : A volaon me wndow [, 1] The volaon me wndows for a samplng node are deermned from he volage waveform a ha node obaned hrough he dynamc analyss of he power nework. A volaon me wndow s marked by me nsances:,he me nsance of maxmum volage ha occurred before a volaon, and 1, he me nsance when he volage recovers back o V hre afer a volaon. An example of a volaon me wndow s shown n Fgure. The volage a s assumed o be he nal volage of he decap before dscharge. If a volaon regon has mulple volaon me wndows, he opmzaon s done sasfyng he charge ransfer consrans (dscussed n Secon.1) for each of hese wndows.. LP-BASED DECAP BUDGETING.1 Charge Transfer Equaons of a Macromodel The charge ransfer equaons correspondng o a volaon me wndow are derved by negrang equaon () over he volaon me wndow, [, 1]:» " R 1 # " R 1 # G11 G 1 Ud R G T 1 G 1 = 0 J R 1d 1 Vd J d + R 1 Id Now he macromodelng procedure as saed n Secon. can be appled o oban a charge-based macromodel: where Q s R 1 and B s (G T 1G 1 11 Q = A W + B (5) R Id, A s G G T 1G 1 11 G1, W s 1 Vd, R R 1 J 1d 1 J d). Noe ha Q represens he oal charge flowng hrough he pors from he decap o he nework durng he me perod [, 1]andW represens he average volages of he pors mulpled by he me duraon [, 1]. Now our sysem consss of he nework wh pre-exsng capacors and nducors as a macromodel, and he decap o be added as exernal elemens conneced o he pors of he macromodel, as shown n Fgure. The problem hus reduces o opmzng he decap added a he pors whle provdng suffcen charge o pull up he por volages above V hre durng he volaon me wndows. decaps Q sample nodes macromodel Q = A W + B Fgure : Charge macromodel wh decaps. Volage Based Consrans Fgure depcs he volage waveform a a node before and afer addng decap. The objecve of addng decap s o pull up he volage above he V hre level as shown n ha fgure. Vdd Vhre Volage Vo 0 W The volage curve afer addng decap The volage curve before addng decap 1 Fgure : Volage waveforms wh & whou decap Applyng he charge equaons of he macromodel (5) o he case where decap has been R added (.e. he upper volage waveform n Fgure, W s 1 Vd, and represens he shaded area shown n ha fgure. Approxmang he upper volage waveform o a sragh lne beween and 1,we can wre he volage consrans as: W (V o, + V hre ) ( 1 )/, SP, () 19

4 R where SP s he se of samplng nodes (pors), W s 1 V d for he samplng node, andv o, s he volage of he node a he sar of he wndow. The above volage based consrans make anoher approxmaon, vz. he V o obaned from he ransen analyss before addon of decap has no changed apprecably wh he addon of decap. In he charge equaons of he macromodel (5), Q s he charge flow from he decap node o he macromodel. In order o keep he supply nework volage above V hre, enough charge should be released from he decap. To keep he decap node self above V hre, he maxmum charge ha can be released from he decap s C (V o V hre ). Therefore, we have he consrans: where M = V o,1 V hre V o, V hre V o,m V hre M C A W + B () 5, C =[C1,C,.., Cm]T where C s he decap value a he node, V o, s he orgnal volage of he node. The operaon represens Hadamard s produc,.e, he enry-wse produc of vecors M and C.. Consrans on Capacance The decouplng capacance has o sasfy he capacance consrans, whch are specfed for he le regons. Afer we deermne he decap a he samplng nodes, we evenly dsrbue hem whn he le regons ha he samplng nodes belong o. Therefore, f C max, s he maxmum amoun of decap allowed n he le regon, we wll have X C k C max, (8) k le C max, s decded by he capacance densy possble for a gven echnology and he amoun of whe space avalable n he le. When a block consss of mulple les and he decap placemen s used for floorplan purpose, a more complex se of consrans on capacance can be derved from he floorplan consrans.. The Complee LP Formulaon Combnng he charge ransfer consrans (), volagebased consrans (), consrans on capacance (8), he decap budgeng problem for a volaon regon can be formulaed no a lnear programmng formulaon mnmze subjec o P SP C (9) M C A W + B W L P k le C k C max,, le volaon regon, where m = SP, C =[C 1,C,.., C m] T, M = V o,1 V 1 V o, V V o,m V m 5, L = (V hre + V o,1) ( 1 )/ (V hre + V o,) ( 1 )/ (V hre + V o,m) ( 1 )/ where SP s he se of samplng nodes. If we assume ha V reaches he hreshold volage a 1,henV s V hre and M s he same as M gven n (). Noe ha, n he above 5, formulaon, he elemens of vecor C are he only varables of opmzaon. 5. ENHANCEMENTS 5.1 Sequence of Lnear Programmng In Secon., we assumed ha he maxmum charge ha could be dscharged from a decap s C (V o, V hre ). Ths s rue only f he node s wors volage a 1, denoed as V, s V hre. Acually, V self s dependen on C. Therefore, consrans () should acually be rewren as (V o, V ) C X 1 k m A k W k + B, SP (10) Snce boh V and C are varables, consran (10) s no lnear anymore. To handle hs nonlneary, we use he followng erave procedure, wheren he lnear programmng opmzaon s solved n each eraon. 1. Se p = 0 and he nal value V p based on he ransen analyss resuls. If V p <V hre,sev p = V hre.. Se V p V, and deermne he decap budge C p by solvng he LP problem (9).. Se C p C, and ge he new node volage V by subsung W by (V + V o,1) ( 1 )/ andhen solvng he lnear equaons (V o,1 V 1) C 1 (V o, V ) C (V o,n V m) C m 5 = A (V 1 + V o,1) T (V + V o,) T (V m + V o,m) T 5 + B where ( 1 )/ st. Noe ha he lef hand sde of he above equaons mply Backward Euler formulaon of he decap, C. Here, V C s equvalen o C/h x() of equaon and V o, C s equvalen o he C/h x( h) ofequaon.. Updae V p+1. Here, we se a sep lengh σ o lm he soluon space o he neghborhood of V p.ifv s below V hre,weseobev hre. The procedure can be descrbed as follows. f V V p s sasfacorly small, sop; else f V <V hre, V p+1 else f V else f V else V p+1 = V p = p +1 = V hre V p σ, V p+1 V p + σ, V p+1 = V p = V p σ + σ Here, we could also use he frs order Tayler seres approxmaons of he nonlnear consrans (10) o replace he formulaon (9), as done n he radonal successve lnear programmng (SLP) [1]. Our comparson of he wo mehods showed he resuls o be smlar. Therefore, we use he above descrbed nuve, erave, procedure. 5. Relaxaon of Capacance Consrans Durng he erave lnear programmng procedure, he volage of he samplng nodes may bounce up and down before selng down. Bu, f he consrans on capacance are gh, hs may lead he opmzaon no he nfeasble 0

5 regon whou a chance for recovery. Therefore, we added a relaxaon facor o relax he capacance consrans. Wh hs, he objecve of LP formulaon (9) becomes mnmze X SP C + β C relax and he consrans on capacance become X C k C max, + C relax, le volaon regon k le where C relax s he relaxaon facor and β s a wegh facor. Generally, β s se o a very large number o force he capacance consrans o be sasfed. By seng he β, he relaxaon facor could also provde a rade-off beween he capacance consrans and he oal capacance requred. 5. Nonunform Tlng When here are mulple volaon regons n a nework, unform sze lng may cause problem. Typcally, some volaon regons occupy a large area wh a slow volage graden across he regon and oher volaon regons occupy very lle area wh a large volage graden across he regon. Gven wo nodes n1 andn, he volage graden s defned as V (n1) V (n) dvded by he dsance of wo nodes. When graden s large, he volage and curren drawn a wo closely locaed nodes could be que dfferen. The wo nodes hen can no be represened well by he samplng node anymore. Therefore, a very fne gran lng may become necessary. However, he fne gran lng wll cause large volaon regons o generae a very large LP problem. In vew of hs, we use non-unformed lng. Fne gran lng s appled o he small volaon regons wh hgh volage graden, whle coarse gran lng s used for large volaon regons wh small volage graden.. COMPLEXITY ANALYSIS AND EXPERIMENTAL RESULTS The proposed echnques were mplemened as par of an exsng n-house power grd analyss ool [1]. An effcen drec lnear solver based on Cholesky facorzaon was used for macromodelng and a publc doman lnear programmng solver, lp solve [15], was ulzed o solve he lnear programmng problem. All he expermens were performed on Lnux machnes wh.ghz CPU and 1GB-GB memory. Table 1: Benchmark Deals Chp #nodes Supply Wors #me Analyss vol(v) Vol(v) pons CPU(s) Chp-1 198, Chp-,88, The proposed mehod was benchmarked usng global power neworks from real desgns. The number of nodes, supply volage, wors volage, number of me seps smulaed and CPU me (n Seconds) for he dynamc analyss are lsed n Table 1. Ideally, we would lke o compare our resuls wh he exsng sensvy-based mehods ha could handle very large power neworks wh non-unform meshes. However, for example, [] requres abou 000s on a GHz Lnux machne o solve a non-unform power nework of abou 1 mllon nodes. Due o he absence of common benchmarks and he effor nvolved n mplemenng he complcaed sensvy-based algorhms whn he framework of our ool, we wll compare he expermenal resuls wh a smpler heursc mehod, and also presen a complexy analyss of our mehod..1 Complexy Analyss The varous approxmaons dscussed before usually cause he decap o be under/over-esmaed slghly. So, as saed n Secon., we erae beween he opmzaon and ransen analyss verfcaon (seps -) unl he volage consrans are sasfed. From column 11 of Table, we can see ha consrans are sasfed ypcally n 1- eraons. For each such (ouer) eraon, one sequence of lnear programmng opmzaon and one ransen analyss are execued. Suppose n, m, l, N and h l are he number of nodes, samplng nodes, me pons, volaon regons and lnear program eraons respecvely. Le O lp (m) beheme complexy of LP. Le n 1.5 l be he me complexy of he ransen analyss, as assumed n []. Then he me spen per ouer eraon s n 1.5 l + n 1.5 N + N O lp (m) h 1,where n 1.5 N s me spen on macromodelng of volaon regons. Generally he sze of samples m s several orders smaller han he orgnal nework sze, n. Mos mporan of all, by adjusng he lng granulary and usng non-unform lng, he sze of m can be kep whn conrol, ndependen of he sze of he power grd and he number of he volaon nodes. In our mplemenaon, we adjus he lng such ha m 000. Also, we se h l 10. If he volages of he samplng nodes do no converge whn 10 LP runs, we connue wh he nex sep of flow. Overall, he opmzaon me s lmed compared o ha of one ransen analyss of he enre nework, especally for large neworks. Ths can be seen from columns 10 and 11 of Table. For example, n case of Chp- wh 1.85V specfcaon, he me spen on ransen analyss runs (one for sep 1 and one for sep of Secon.) s 90s, whle he me spen on opmzaon s only 8s.. Comparson wh a Bnary Search based Greedy Mehod To benchmark he qualy of our LP based decap opmzaon, we compared he resuls wh anoher nuve mehod whch s a greedy mehod based on bnary search. The bnary search mehod places decap unformly across he core volaon regon. For each core volaon regon, he amoun of decap placed s decded hrough bnary search. The search begns by placng 0.1ff/um area capacance n he volaon regons. The capacance s ncreased by 10 mes n each sep unl he frs successful soluon s found. From hen on, he bnary search s used o deermne a mnmum capacance ha sasfes he volage consrans. Table compares he resuls of our mehod wh ha of he bnary search mehod. The volage specfcaon and number of volaon nodes are shown n columns and respecvely. The wors volage afer opmzaon, amoun of decap added, he CPU me, and he number of eraons are gven n columns - for he Bnary Search mehod and n columns 8-11 for our mehod. The CPU me ncludes he enre decap budgeng flow as gven n Secon., ncludng he pre-opmzaon ransen analyss, sequence of LP, and he pos-opmzaon ransen analyss verfcaon. The number of eraons refers o he number of opmzaonverfcaon eraons. The oal number of ransen analyss runs s hus #er +1. 1

6 Table : Comparson wh he Bnary Search Mehod Chp Spec #Vo he Bnary Search Mehod Our Mehod Vol(v) Nodes Vol(v) Decap CPU(s) #er Vol(v) Decap CPU(s) #er Chp nf nf nf nf nf nf nf nf nf nf 9.9 Chp pf pf pf pf pf pf pf pf 11. From Table, we can see ha our mehod always uses less decap han he bnary search based mehod. The larger he volaon regons, he more gans our mehod shows. Ths s because he proposed opmzaon places he decap a he mos needed locaon. In addon, our mehod always consumes much less CPU me. Noe ha, n order o speed up he greedy mehod, we dd he bnary search of all he volaon regons smulaneously. Oherwse, he CPU me per eraon wll be #voregon mes he CPU me per eraon lsed n Table. The drawback wh he smulaanous bnary search s ha a capacance value ha orgnally pulled he volages up successfully above he hreshold level may fal because of he capacance reducon of he neghborng volaon regons. In hs case, he bnary search has o be rese. Moreover, s dffcul o apply he greedy echnque o floorplan or placemen purpose due o he nflexbly n placng dfferen decap a dfferen locaons of one volaon regon based on floorplan/placemen consrans. To llusrae how he erave LP procedure mproves he qualy of opmzaon, Table shows he resuls of he frs eraons of 5 volaon regons of Chp-1 a.0v specfcaon. Column s he rao of he volaon regon area o he oal area, and column s he number of samplng nodes. The decap added n each of he eraons are repored under columns -. We can see ha only one eraon of LP may sgnfcanly overesmae he decap value. Afer a couple of eraons, he decap value becomes sable. In our mplemenaon, we se change n volage dv 0.00 as he condon of convergence. Wh only LP eraons, he volaon regons,, and 5 converged. Table : The frs four LP eraons Vo Area #Sample 1s nd rd h regon rao nodes (nf) (nf) (nf) (nf) Vo Vo Vo Vo Vo CONCLUSION In hs paper, we have proposed a novel and effcen charge-based decouplng capacor budgeng algorhm. We proposed several echnques o reduce he problem sze and also o have he problem sze scale well w.r.o he nework sze and he number of volaon nodes. We used macromodelng, and he localy effecs of he decouplng capacance, along wh non-unform lng. By negrang he nodal equaons over he volaon me perod, we replaced he me consumng sensvy calculaons and nonlnear programmng opmzaon wh a much cheaper sequence of lnear programmng procedure. Acually, each ndvdual echnque proposed here can be combned wh oher exsng approaches. Expermenal resuls were gven o demonsrae ha he proposed algorhm s capable of opmzng power neworks wh 5 mllon nodes usng 1- ransen analyss runs, and n a couple of hours. Comparson of our algorhm wh a greedy heursc shows he advanage of area effcency and run me effcency of our mehod. By modfyng he capacance consrans, he mehod can easly be ncorporaed no physcal desgn flows. 8. REFERENCES [1] H. Su, S. S. Sapanekar, and S. R. Nassf, An algorhm for opmal decouplng capacor szng and placemen for sandard cell layous, n ISPD, pp. 8, 00. [] J.Fu,Z.Luo,X.Hong,Y.Ca,S.X.D.Tan,andZ.Pan, A fas decouplng capacor budgeng algorhm for robus on-chp power delvery, n ASP-DAC, pp , 00. [] K. Wang and M. M. Sadowska, On-chp power supply nework opmzaon usng mulgrd-based echnque, n DAC, pp , 00. [] H.L,Z.Q,S.X.D.Tan,L.Wu,Y.Ca,andX.Hong, Paronng-based approach o fas on-chp decouplng capacor budgeng and mnmzaon, n DAC, pp , 005. [5] Z.Q,H.L,J.Fan,S.X.D.Tan,Y.Ca,andX.Hong, On-chp decouplng capacor budgeng by sequence of lnear programmng, n Proceedngs of h Inernaonal Conference on ASIC, 005. [] S. W. Drecor and R. A. Rohrer, The generalzed adjon nework and nework sensves, IEEE Transacons on Crcu Theory, vol. 1, pp. 18, Aug [] P. Feldmann, T. V. Nguyen, S. W. Drecor, and R. A. Rohrer, Sensvy compuaon n pecewse approxmae crcu smulaon, TCAD, vol. 10, pp , Feb [8] L. T. Pllage, R. A. Rohrer, and C. Vsweswarah, Elecronc and Sysem Smulaon Mehods. New York: McGraw-Hll, [9] S. Zhao, K. Roy, and C. K. Koh, Decouplng capacance allocaon for power supply nose suppresson, n ISPD, 001. [10] H. H. Chen and D. D. Lng, Power supply nose analyss mehodology for deep-submcron VLSI chp desgn, n DAC, pp. 8, 199. [11] M. Zhao, R. V. Panda, S. S. Sapanekar, and D. T. Blaauw, Herarchcal analayss of power dsrbuon neworks, TCAD, vol. 1, pp , Feb. 00. [1] C. Ho, A. Ruehl, and P. Brennan, The modfed nodal approach o nework analyss, IEEE Trans. Crcus and Sysems, vol. CAS-, no., pp , 195. [1] M. S. Bazaraa, H. D. Sheral, and C. M. Shey, Nonlnear Programmng. John Wley and Sons, Inc., 199. [1] A. Dharchoudhury, R. Panda, D. Blaauw, R. Vadyanahan, B. Tuuanu, and D. Bearden, Desgn and analyss of power dsrbuon neworks n PowerPC mcroprocessors, n DAC, pp. 8, [15] M. R. C. M. Berkelaar and e. al., LP SOLVE 5.5 Users Manual, 005.

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

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

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

First-order piecewise-linear dynamic circuits

First-order piecewise-linear dynamic circuits Frs-order pecewse-lnear dynamc crcus. Fndng he soluon We wll sudy rs-order dynamc crcus composed o a nonlnear resse one-por, ermnaed eher by a lnear capacor or a lnear nducor (see Fg.. Nonlnear resse one-por

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

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

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

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

( ) () 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

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

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

10. A.C CIRCUITS. Theoretically current grows to maximum value after infinite time. But practically it grows to maximum after 5τ. Decay of current :

10. A.C CIRCUITS. Theoretically current grows to maximum value after infinite time. But practically it grows to maximum after 5τ. Decay of current : . A. IUITS Synopss : GOWTH OF UNT IN IUIT : d. When swch S s closed a =; = d. A me, curren = e 3. The consan / has dmensons of me and s called he nducve me consan ( τ ) of he crcu. 4. = τ; =.63, n one

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

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

[ ] 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

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

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

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

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

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

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

J i-1 i. J i i+1. Numerical integration of the diffusion equation (I) Finite difference method. Spatial Discretization. Internal nodes.

J i-1 i. J i i+1. Numerical integration of the diffusion equation (I) Finite difference method. Spatial Discretization. Internal nodes. umercal negraon of he dffuson equaon (I) Fne dfference mehod. Spaal screaon. Inernal nodes. R L V For hermal conducon le s dscree he spaal doman no small fne spans, =,,: Balance of parcles for an nernal

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

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

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

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

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

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

Volatility Interpolation

Volatility Interpolation Volaly Inerpolaon Prelmnary Verson March 00 Jesper Andreasen and Bran Huge Danse Mares, Copenhagen wan.daddy@danseban.com brno@danseban.com Elecronc copy avalable a: hp://ssrn.com/absrac=69497 Inro Local

More information

ISSN MIT Publications

ISSN MIT Publications MIT Inernaonal Journal of Elecrcal and Insrumenaon Engneerng Vol. 1, No. 2, Aug 2011, pp 93-98 93 ISSN 2230-7656 MIT Publcaons A New Approach for Solvng Economc Load Dspach Problem Ansh Ahmad Dep. of Elecrcal

More information

The Finite Element Method for the Analysis of Non-Linear and Dynamic Systems

The Finite Element Method for the Analysis of Non-Linear and Dynamic Systems Swss Federal Insue of Page 1 The Fne Elemen Mehod for he Analyss of Non-Lnear and Dynamc Sysems Prof. Dr. Mchael Havbro Faber Dr. Nebojsa Mojslovc Swss Federal Insue of ETH Zurch, Swzerland Mehod of Fne

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

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

2/20/2013. EE 101 Midterm 2 Review

2/20/2013. EE 101 Midterm 2 Review //3 EE Mderm eew //3 Volage-mplfer Model The npu ressance s he equalen ressance see when lookng no he npu ermnals of he amplfer. o s he oupu ressance. I causes he oupu olage o decrease as he load ressance

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

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

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

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

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

Time-interval analysis of β decay. V. Horvat and J. C. Hardy

Time-interval analysis of β decay. V. Horvat and J. C. Hardy Tme-nerval analyss of β decay V. Horva and J. C. Hardy Work on he even analyss of β decay [1] connued and resuled n he developmen of a novel mehod of bea-decay me-nerval analyss ha produces hghly accurae

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

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

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

Moving Least Square Method for Reliability-Based Design Optimization

Moving Least Square Method for Reliability-Based Design Optimization Movng Leas Square Mehod for Relably-Based Desgn Opmzaon K.K. Cho, Byeng D. Youn, and Ren-Jye Yang* Cener for Compuer-Aded Desgn and Deparmen of Mechancal Engneerng, he Unversy of Iowa Iowa Cy, IA 52242

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

Chapter 2 Linear dynamic analysis of a structural system

Chapter 2 Linear dynamic analysis of a structural system Chaper Lnear dynamc analyss of a srucural sysem. Dynamc equlbrum he dynamc equlbrum analyss of a srucure s he mos general case ha can be suded as akes no accoun all he forces acng on. When he exernal loads

More information

This document is downloaded from DR-NTU, Nanyang Technological University Library, Singapore.

This document is downloaded from DR-NTU, Nanyang Technological University Library, Singapore. Ths documen s downloaded from DR-NTU, Nanyang Technologcal Unversy Lbrary, Sngapore. Tle A smplfed verb machng algorhm for word paron n vsual speech processng( Acceped verson ) Auhor(s) Foo, Say We; Yong,

More information

3. OVERVIEW OF NUMERICAL METHODS

3. OVERVIEW OF NUMERICAL METHODS 3 OVERVIEW OF NUMERICAL METHODS 3 Inroducory remarks Ths chaper summarzes hose numercal echnques whose knowledge s ndspensable for he undersandng of he dfferen dscree elemen mehods: he Newon-Raphson-mehod,

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

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

Bayes rule for a classification problem INF Discriminant functions for the normal density. Euclidean distance. Mahalanobis distance

Bayes rule for a classification problem INF Discriminant functions for the normal density. Euclidean distance. Mahalanobis distance INF 43 3.. Repeon Anne Solberg (anne@f.uo.no Bayes rule for a classfcaon problem Suppose we have J, =,...J classes. s he class label for a pxel, and x s he observed feaure vecor. We can use Bayes rule

More information

PARTICLE SWARM OPTIMIZATION FOR INTERACTIVE FUZZY MULTIOBJECTIVE NONLINEAR PROGRAMMING. T. Matsui, M. Sakawa, K. Kato, T. Uno and K.

PARTICLE SWARM OPTIMIZATION FOR INTERACTIVE FUZZY MULTIOBJECTIVE NONLINEAR PROGRAMMING. T. Matsui, M. Sakawa, K. Kato, T. Uno and K. Scenae Mahemacae Japoncae Onlne, e-2008, 1 13 1 PARTICLE SWARM OPTIMIZATION FOR INTERACTIVE FUZZY MULTIOBJECTIVE NONLINEAR PROGRAMMING T. Masu, M. Sakawa, K. Kao, T. Uno and K. Tamada Receved February

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

Single-loop system reliability-based topology optimization considering statistical dependence between limit-states

Single-loop system reliability-based topology optimization considering statistical dependence between limit-states Sruc Muldsc Opm 2011) 44:593 611 DOI 10.1007/s00158-011-0669-0 RESEARCH PAPER Sngle-loop sysem relably-based opology opmzaon consderng sascal dependence beween lm-saes Tam H. Nguyen Junho Song Glauco H.

More information

Introduction to Compact Dynamical Modeling. III.1 Reducing Linear Time Invariant Systems. Luca Daniel Massachusetts Institute of Technology

Introduction to Compact Dynamical Modeling. III.1 Reducing Linear Time Invariant Systems. Luca Daniel Massachusetts Institute of Technology SF & IH Inroducon o Compac Dynamcal Modelng III. Reducng Lnear me Invaran Sysems Luca Danel Massachuses Insue of echnology Course Oulne Quck Sneak Prevew I. Assemblng Models from Physcal Problems II. Smulang

More information

Robustness of DEWMA versus EWMA Control Charts to Non-Normal Processes

Robustness of DEWMA versus EWMA Control Charts to Non-Normal Processes Journal of Modern Appled Sascal Mehods Volume Issue Arcle 8 5--3 Robusness of D versus Conrol Chars o Non- Processes Saad Saeed Alkahan Performance Measuremen Cener of Governmen Agences, Insue of Publc

More information

( ) lamp power. dx dt T. Introduction to Compact Dynamical Modeling. III.1 Reducing Linear Time Invariant Systems

( ) lamp power. dx dt T. Introduction to Compact Dynamical Modeling. III.1 Reducing Linear Time Invariant Systems SF & IH Inroducon o Compac Dynamcal Modelng III. Reducng Lnear me Invaran Sysems Luca Danel Massachuses Insue of echnology Movaons dx A x( + b u( y( c x( Suppose: we are jus neresed n ermnal.e. npu/oupu

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

Motion in Two Dimensions

Motion in Two Dimensions Phys 1 Chaper 4 Moon n Two Dmensons adzyubenko@csub.edu hp://www.csub.edu/~adzyubenko 005, 014 A. Dzyubenko 004 Brooks/Cole 1 Dsplacemen as a Vecor The poson of an objec s descrbed by s poson ecor, r The

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

Dynamic Team Decision Theory

Dynamic Team Decision Theory Dynamc Team Decson Theory EECS 558 Proec Repor Shruvandana Sharma and Davd Shuman December, 005 I. Inroducon Whle he sochasc conrol problem feaures one decson maker acng over me, many complex conrolled

More information

Chapter 5. Circuit Theorems

Chapter 5. Circuit Theorems Chaper 5 Crcu Theorems Source Transformaons eplace a olage source and seres ressor by a curren and parallel ressor Fgure 5.-1 (a) A nondeal olage source. (b) A nondeal curren source. (c) Crcu B-conneced

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

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

Including the ordinary differential of distance with time as velocity makes a system of ordinary differential equations.

Including the ordinary differential of distance with time as velocity makes a system of ordinary differential equations. Soluons o Ordnary Derenal Equaons An ordnary derenal equaon has only one ndependen varable. A sysem o ordnary derenal equaons consss o several derenal equaons each wh he same ndependen varable. An eample

More information

Li An-Ping. Beijing , P.R.China

Li An-Ping. Beijing , P.R.China A New Type of Cpher: DICING_csb L An-Png Bejng 100085, P.R.Chna apl0001@sna.com Absrac: In hs paper, we wll propose a new ype of cpher named DICING_csb, whch s derved from our prevous sream cpher DICING.

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

. 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

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

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

Diffusion of Heptane in Polyethylene Vinyl Acetate: Modelisation and Experimentation

Diffusion of Heptane in Polyethylene Vinyl Acetate: Modelisation and Experimentation IOSR Journal of Appled hemsry (IOSR-JA) e-issn: 78-5736.Volume 7, Issue 6 Ver. I. (Jun. 4), PP 8-86 Dffuson of Hepane n Polyehylene Vnyl Aceae: odelsaon and Expermenaon Rachd Aman *, Façal oubarak, hammed

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

Chapter Lagrangian Interpolation

Chapter Lagrangian Interpolation Chaper 5.4 agrangan Inerpolaon Afer readng hs chaper you should be able o:. dere agrangan mehod of nerpolaon. sole problems usng agrangan mehod of nerpolaon and. use agrangan nerpolans o fnd deraes and

More information

Learning Objectives. Self Organization Map. Hamming Distance(1/5) Introduction. Hamming Distance(3/5) Hamming Distance(2/5) 15/04/2015

Learning Objectives. Self Organization Map. Hamming Distance(1/5) Introduction. Hamming Distance(3/5) Hamming Distance(2/5) 15/04/2015 /4/ Learnng Objecves Self Organzaon Map Learnng whou Exaples. Inroducon. MAXNET 3. Cluserng 4. Feaure Map. Self-organzng Feaure Map 6. Concluson 38 Inroducon. Learnng whou exaples. Daa are npu o he syse

More information

Computing Relevance, Similarity: The Vector Space Model

Computing Relevance, Similarity: The Vector Space Model Compung Relevance, Smlary: The Vecor Space Model Based on Larson and Hears s sldes a UC-Bereley hp://.sms.bereley.edu/courses/s0/f00/ aabase Managemen Sysems, R. Ramarshnan ocumen Vecors v ocumens are

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

Advanced Macroeconomics II: Exchange economy

Advanced Macroeconomics II: Exchange economy Advanced Macroeconomcs II: Exchange economy Krzyszof Makarsk 1 Smple deermnsc dynamc model. 1.1 Inroducon Inroducon Smple deermnsc dynamc model. Defnons of equlbrum: Arrow-Debreu Sequenal Recursve Equvalence

More information

, t 1. Transitions - this one was easy, but in general the hardest part is choosing the which variables are state and control variables

, t 1. Transitions - this one was easy, but in general the hardest part is choosing the which variables are state and control variables Opmal Conrol Why Use I - verss calcls of varaons, opmal conrol More generaly More convenen wh consrans (e.g., can p consrans on he dervaves More nsghs no problem (a leas more apparen han hrogh calcls of

More information

January Examinations 2012

January Examinations 2012 Page of 5 EC79 January Examnaons No. of Pages: 5 No. of Quesons: 8 Subjec ECONOMICS (POSTGRADUATE) Tle of Paper EC79 QUANTITATIVE METHODS FOR BUSINESS AND FINANCE Tme Allowed Two Hours ( hours) Insrucons

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

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

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

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

Solving Equation [5.61], the helical fiber thickness required to contain the internal pressure is:

Solving Equation [5.61], the helical fiber thickness required to contain the internal pressure is: 5.4.3 eng Analyss of Cylndrcal Pressure Vessels S. T. Peers 001 Ths sofware s provded free for your use wh no guaranee as o s effecveness. I s copyrghed by Process-Research and may no be duplcaed, gven

More information

Planar truss bridge optimization by dynamic programming and linear programming

Planar truss bridge optimization by dynamic programming and linear programming IABSE-JSCE Jon Conference on Advances n Brdge Engneerng-III, Augus 1-, 015, Dhaka, Bangladesh. ISBN: 978-984-33-9313-5 Amn, Oku, Bhuyan, Ueda (eds.) www.abse-bd.org Planar russ brdge opmzaon by dynamc

More information

Fall 2010 Graduate Course on Dynamic Learning

Fall 2010 Graduate Course on Dynamic Learning Fall 200 Graduae Course on Dynamc Learnng Chaper 4: Parcle Flers Sepember 27, 200 Byoung-Tak Zhang School of Compuer Scence and Engneerng & Cognve Scence and Bran Scence Programs Seoul aonal Unversy hp://b.snu.ac.kr/~bzhang/

More information

Equalization on Graphs: Linear Programming and Message Passing

Equalization on Graphs: Linear Programming and Message Passing Equalzaon on Graphs: Lnear Programmng and Message Passng Mohammad H. Taghav and Paul H. Segel Cener for Magnec Recordng Research Unversy of Calforna, San Dego La Jolla, CA 92093-0401, USA Emal: (maghav,

More information

Digital Variable Frequency Control for Zero Voltage Switching and Interleaving of Synchronous Buck Converters

Digital Variable Frequency Control for Zero Voltage Switching and Interleaving of Synchronous Buck Converters Dgal Varable Frequency Conrol for Zero Volage Swchng and Inerleavng of Synchronous Buck Converers Pål Andreassen, Guseppe Gud, Tore M. Undeland Norwegan Unversy of Scence and Technology, Trondhem, Norway

More information

Sampling Coordination of Business Surveys Conducted by Insee

Sampling Coordination of Business Surveys Conducted by Insee Samplng Coordnaon of Busness Surveys Conduced by Insee Faben Guggemos 1, Olver Sauory 1 1 Insee, Busness Sascs Drecorae 18 boulevard Adolphe Pnard, 75675 Pars cedex 14, France Absrac The mehod presenly

More information

Study on Distribution Network Reconfiguration with Various DGs

Study on Distribution Network Reconfiguration with Various DGs Inernaonal Conference on Maerals Engneerng and Informaon Technology Applcaons (MEITA 205) Sudy on Dsrbuon ework Reconfguraon wh Varous DGs Shengsuo u a, Y Dng b and Zhru Lang c School of Elecrcal Engneerng,

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

. 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

A Simulation Based Optimal Control System For Water Resources

A Simulation Based Optimal Control System For Water Resources Cy Unversy of New York (CUNY) CUNY Academc Works Inernaonal Conference on Hydronformacs 8--4 A Smulaon Based Opmal Conrol Sysem For Waer Resources Aser acasa Maro Morales-Hernández Plar Brufau Plar García-Navarro

More information

Structural Optimization Using Metamodels

Structural Optimization Using Metamodels Srucural Opmzaon Usng Meamodels 30 Mar. 007 Dep. o Mechancal Engneerng Dong-A Unvers Korea Kwon-Hee Lee Conens. Numercal Opmzaon. Opmzaon Usng Meamodels Impac beam desgn WB Door desgn 3. Robus Opmzaon

More information

A NOVEL NETWORK METHOD DESIGNING MULTIRATE FILTER BANKS AND WAVELETS

A NOVEL NETWORK METHOD DESIGNING MULTIRATE FILTER BANKS AND WAVELETS A NOVEL NEWORK MEHOD DESIGNING MULIRAE FILER BANKS AND WAVELES Yng an Deparmen of Elecronc Engneerng and Informaon Scence Unversy of Scence and echnology of Chna Hefe 37, P. R. Chna E-mal: yan@usc.edu.cn

More information

A Systematic Framework for Dynamically Optimizing Multi-User Wireless Video Transmission

A Systematic Framework for Dynamically Optimizing Multi-User Wireless Video Transmission A Sysemac Framework for Dynamcally Opmzng ul-user Wreless Vdeo Transmsson Fangwen Fu, haela van der Schaar Elecrcal Engneerng Deparmen, UCLA {fwfu, mhaela}@ee.ucla.edu Absrac In hs paper, we formulae he

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

Energy Storage Devices

Energy Storage Devices Energy Sorage Deces Objece of Lecure Descrbe he consrucon of a capacor and how charge s sored. Inroduce seeral ypes of capacors Dscuss he elecrcal properes of a capacor The relaonshp beween charge, olage,

More information