Discrete Particle Swarm Optimization for TSP: Theoretical Results and Experimental Evaluations

Size: px
Start display at page:

Download "Discrete Particle Swarm Optimization for TSP: Theoretical Results and Experimental Evaluations"

Transcription

1 Dscrete Partcle Swarm Optmzaton for TSP: Theoretcal Results and Expermental Evaluatons Matthas Hoffmann, Mortz Mühlenthaler, Sabne Helwg, Rolf Wanka Department of Computer Scence, Unversty of Erlangen-Nuremberg, Germany Abstract. Partcle swarm optmzaton (PSO) s a nature-nspred technque orgnally desgned for solvng contnuous optmzaton problems. There already exst several approaches that use PSO also as bass for solvng dscrete optmzaton problems, n partcular the Travelng Salesperson Problem (TSP). In ths paper, () we present the frst theoretcal analyss of a dscrete PSO algorthm for TSP whch also provdes nsght nto the convergence behavor of the swarm. In partcular, we prove that the popular choce of usng sequences of transpostons as the dfference between tours tends to decrease the convergence rate. () In the lght of ths observaton, we present a new noton of dfference between tours based on edge exchanges and a new method to combne dfferences by computng ther centrod. Ths leads to a more PSO-lke behavor of the algorthm and avods the observed slow down effect. () Then, we nvestgate mplementatons of our methods and compare them wth prevous mplementatons showng the compettveness of our new approaches. 1 Introducton The problem. Partcle Swarm Optmzaton (PSO) s a popular metaheurstc desgned for solvng optmzaton problems on contnuous domans. It was ntroduced by Kennedy and Eberhard [11, 5] and has snce then been appled successfully to a wde range of optmzaton problems. Snce the structure of the PSO algorthm s relatvely smple, PSO has to some extent been open for theoretcal studes of the swarm behavor. Clerk and Kennedy [4], Trelea [17], and Jang et al. [9] provde analyses of the convergence behavor of partcle swarms, whch offer some nsghts on how to select the swarm parameters, and the ntal behavor of a swarm has been analyzed n [8]. Inspred by the performance of PSO on contnuous optmzaton problems, several approaches have also been proposed for applyng PSO to dscrete problems, such as functon optmzaton on bnary domans [12], schedulng problems [1], and the Travelng Salesperson Problem (TSP) [2, 18, 6, 14, 15, 19].

2 The TSP s one of the classcal problems n dscrete optmzaton. A wealth of methods specfcally talored for solvng TSP has been developed and mathematcally and expermentally nvestgated. A comprehensve overvew of ths lne of research can be found n [7]. But the TSP s also well suted to be approached by (meta-)heurstc methods lke PSO. For dscrete PSO, new nterpretatons of movement and velocty are necessary. The frst approach to adaptng the PSO scheme to TSP s due to Clerc [2, 3]. However, t turns out that ths dscrete PSO (DPSO) by tself s not as successful as the orgnal PSO for contnuous problems. Consequently, subsequent approaches to solvng TSP by PSO typcally rely on downstream optmzaton technques such as k-opt [15, 14] and Ln-Kernghan [6] appled after one PSO teraton to mprove the qualty of the soluton obtaned by PSO. Unfortunately, whereas these hybrd algorthms are evaluated expermentally by beng run on benchmark nstances, they are hard to analyze mathematcally, and so far, no theoretcal nsghts were ganed about the partcles behavor n dscrete PSO at all. In fact, the downstream optmzaton even conceals the performance of plan DPSO. Our contrbuton. In ths paper, we present the frst theoretcal analyss of the dscrete PSO algorthms of Clerc [2] and Wang et al. [18], whch, to some extent, also apples to the approach of Sh et al. [14]. In partcular, we provde for the frst tme theoretcal evdence for why the convergence behavor of these DPSO algorthms for the TSP s qute dfferent from what we would expect from the classcal PSO for contnuous problems. The key nsght s that n later stages of the optmzaton process, the partcles are not lkely to converge towards the best soluton found so far. In fact, we prove that the dstance to the best soluton even remans more or less constant. In the lght of the theoretcal fndngs, we then propose a novel nterpretaton of partcle moton avodng the convergence problem mentoned above. Our method s smlar to computng the mdpont of a dscrete lne. Addtonally, we ntroduce a new representaton of the velocty of a partcle, whch s based on exchangng edges n a potental soluton of a TSP nstance. We evaluate our proposed DPSO wth respect to seven nstances from the TSPlb [13] wth 52 to 105 ctes. In these expermental evaluatons, our focus s on the DPSO performance of dfferent velocty representatons because we are n ths context manly nterested n the performance of the plan DPSO approaches wthout subsequent local mprovement phases. Our results ndcate that the combnaton of the mdpont-based partcle moton wth the edge-exchange operator outperforms the other operators as well as those methods whch suffer from the dentfed convergence problem. In order to also compare our DPSO to the prevous approaches whch use n the PSO teratons addtonal local mprovement heurstcs, we hybrdze our DPSO teraton wth a 2-OPT local optmzaton appled to the global attractor. Here, we make two observatons: The frst one s that better performance of the plan DPSO results n a better performance of the hybrdzed DPSO. And second, for the frst tme t s clearly documented that the huge performance gans (that

3 s acheved when usng local optmzaton n comparson to the plan DPSO) ndcate that the qualty of the solutons found by the DPSO algorthms wth local optmzaton s almost completely determned by the qualty of the local optmzaton. In the prevous work [14, 18, 19], t s not dfferentated between the contrbuton of the PSO and the addtonal local mprovement methods. The remander of ths paper s organzed as follows. In Sec. 2, we provde relevant background nformaton on TSP and PSO. Sec. 3 contans our theoretcal analyss of the partcle convergence behavor of dscrete PSOs for permutaton problems. In Sec. 4, we propose the new dscrete PSO for the TSP whch uses a centrod-based approach for the partcle movement. Sec. 5 provdes the expermental results whch ndcate that our proposed approach outperforms other dscrete PSOs for the TSP. 2 Prelmnares 2.1 The Travelng Salesperson Problem The Travelng Salesperson Problem (TSP) s a classcal combnatoral optmzaton problem. An nstance I = (n, dst) of the TSP wth n ctes {1,..., n} conssts of the dstances between each par of ctes, gven as an n n-nteger matrx dst. The task s to fnd a tour wth mnmum length vstng each cty exactly once, ncludng the way back to the ntal cty. A tour s gven as a permutaton π of the ctes {1,..., n}, where π() denotes the th vsted cty. Hence the set of optmal solutons of an nstance I = (n, dst) s ( argmn dst π(n),π(1) + ) dst π(),π(+1), π S n 1 <n where S n s the symmetrc group on {1,..., n} wth the usual composton as group operaton. The operaton s used for explorng the search space. Note that n ths formulaton the search space conssts of all elements of S n, so one specfc cycle of length n s represented by many permutatons. The advantage of usng all elements of S n for the proposed dscrete PSO s that the partcles can move around freely n the search space wthout the danger of encounterng permutatons whch do not correspond to a vald tour. The decson varant of TSP ( gven an nteger L, s there a tour n I of length at most L? ) s NP-hard and hence, TSP can presumably not be solved exactly n polynomal tme. 2.2 (Dscrete) Partcle Swarm Optmzaton Introduced by Kennedy and Eberhard [11, 5], partcle swarm optmzaton (PSO) s a populaton-based metaheurstc that uses a swarm of potental solutons called partcles to cooperatvely solve optmzaton problems. Typcally, the search space of a problem nstance s an n-dmensonal rectangle B R n, and the objectve functon (often also called ftness functon) s of the form f : R n R. PSO works n teratons. In teraton t, each partcle has a poston x (t) B

4 and a velocty v (t) R n. Whle movng through the search space, the partcles evaluate f at x (t). Each partcle remembers ts best poston p so far (called local attractor) and the best poston p glob of all partcles n the swarm so far (called global attractor). In teraton t, the poston and velocty of each partcle s updated accordng to the followng movement equatons: v (t+1) x (t+1) = a v (t) = x (t) + r loc b loc (p x (t) ) + r glob b glob (p glob x (t) ) (1) + v (t+1). (2) The parameters a, b loc, b glob R are constant weghts whch can be selected by the user. The nerta a adjusts the relatve mportance of the nerta of the partcles, and the so-called acceleraton coeffcents b loc and b glob determne the nfluence of the local and the global attractor, resp. In every teraton, r loc and r glob are drawn unformly at random from [0, 1]. Partcles exchange nformaton about the search space exclusvely va the global attractor p glob. p x (t) s the local attracton exerted by the local attractor on partcle, and p glob x (t) s the global attracton exerted by the global attractor. The PSO algorthm was orgnally desgned for solvng optmzaton problems on contnuous domans. Inspred by the success and conceptual smplcty of PSO, several approaches have been proposed to adapt the PSO dynamcs to dscrete problems ncludng the TSP [2, 6, 15, 14]. In order to adapt PSO s movement equatons (1) and (2) to the dscrete doman of the TSP, Clerk suggests n [2] the followng modfcatons (or new nterpretatons) of the terms nvolved: The partcle s poston x (t) s a permutaton π of the ctes,. e., π = (c 1 c 2... c n ) whch corresponds to the tour c 1 c 2 c n c 1. The dfference x y between two postons x and y (also called the attracton of x to y) s represented by a shortest sequence of transpostons T = t 1,..., t k such that y T = x. Transposton t = (c m c r ) exchanges the two ctes c m and c r n a gven round-trp. The length of a dfference s the length of the sequence T of transpostons. The multplcaton s T of a dfference T = t 1,..., t k wth a scalar s, 0 < s 1, s defned as t 1,..., t s k. For s = 0, s T =. Here, we omt the cases s > 1 and s < 0 snce they do not occur n our proposed PSO. The addton T 1 + T 2 of dfferences T 1 = t 1 1,..., t 1 k and T 1 = t 2 1,..., t 2 l s defned as T 1 + T 2 = t 1 1,..., t 1 k, t2 1,..., t 2 l. The addton of a dfference and a poston s defned as applyng the transpostons of the dfference to the poston. A small example s presented after the proof of Theorem 1. In [2], [18] and [14], the dfference between two postons x and y n the search space s represented as a lst of transpostons that transform round-trp x nto round-trp y. In [6], ths representaton s restrcted to adjacent transpostons. In our new approach n Sec. 4, we replace the transposton by a representaton whch successvely exchanges two edges n a round-trp. Hence, the dfference of two postons x and y s a sequence of edge exchanges of mnmal length whch transforms x nto y.

5 3 Theoretcal Analyss In ths secton, we prove that under certan condtons the prevously developed varants of DPSO mentoned n Sec. 2 behave counterntutvely when compared to the classcal contnuous PSO snce the convergence rate n DPSO s slowed down. More specfcally, we show that transpostons whch occur both n the local attracton and n the global attracton cancel each other and prevent the partcle from movng closer to the global and local attractor. Ths s qute dfferent from the behavor observed n contnuous PSO where common components n these attractons even result n an amplfed attractve force. The phenomenon that the local and the global attracton n prevous approaches have a lot of transpostons n common n the later stages of the optmzaton process can be observed expermentally. Evaluatng the two attractons p x (t) and p glob x (t) for sample runs (see Fg. 1), we see that (n ths example) on average about 30% of the transpostons occur n both attractons. Summng the partcles n the rght half of the bns n Fg. 1, we can conclude that for roughly 20% of the partcles, more than a half of the transpostons are shared by the two attractons. We analyzed the DPSO methods from [2, 18] that fracton of partcles (%) % 10-20% 20-30% 30-40% 40-50% fracton of common transpostons n the local and global attractors Fg. 1. Smlarty of local and global attracton on Clerc s DPSO [2], averaged over 100 runs on the TSP nstance berln52, consderng all partcles n teratons 990 to % 60-70% 70-80% 80-90% % use transpostons for representng dstances between partcles n what we call the Long Term Dscrete PSO model (LTD). In ths model, we assume that the followng four condtons hold: Dfferences between postons are represented by sequences of transpostons. p = p glob =: p, for all partcles. a = 0 r loc and r glob are unformly dstrbuted. When the full swarm converges to a common best soluton p, all local and global attractors are dentcal. If p = p glob for a certan partcle, then t has vsted the

6 global attractor at least once. We assume the nerta a of the partcles beng 0 snce n our experments, the performance of the PSO algorthm even becomes worse f the nerta weght s set to a hgher value. r loc and r glob are qute often unformly dstrbuted n practce. Ths assumpton s also made n the mathematcal analyss n [17]. For Theorem 1, we assume b loc = b glob, whch allows for a closed and smple representaton. After ts proof, we deal wth the more general case whch can be analyzed analogously and present a small example. Theorem 1. Let s [0, 1], and let b loc = b glob = b. The probablty that n the LTD model a certan partcle reduces ts dstance to p n an teraton by a factor of at least b s, s (1 s) 2. Proof. As a = 0, the two movement equatons (1) and (2) can be reduced to one: x (t+1) = x (t) + r loc b (p x (t) ) + r glob b (p x (t) ) Let d be the number of transpostons n the attracton (p x (t) ). Snce we multply the dfference wth r loc b and r glob b, resp., we apply the frst r loc b d and then the frst r glob b d transpostons to x (t). Both dfferences have a common part consstng of the frst mn(r loc, r glob ) b d transpostons. By applyng the frst r loc b d transpostons, for each transposton an element of x (t) reaches the place that t also has n p. However, when applyng the transpostons of the second dfference, the common part of both dfferences s appled twce and the elements of the permutaton that were already at the rght place move now to another place. To brng the elements back to the orgnal place we have to apply the nverse of the common part. Snce the nverse of the common part has exactly the same number of transpostons as the common part, the dstance to p s only reduced by the transpostons that are not common n both dfferences and so are only appled once. The number of the transpostons that are appled only once s r loc r glob b d. Only these transpostons contrbute to the convergence towards p because the other transpostons move the partcle further away from p when they are appled a second tme. Therefore, we call transpostons that are appled only once effectve transpostons. The probablty that the fracton of effectve transposton s at least b s s, s gven by ( rloc r glob b d ) P b s = P( r loc r glob s). d Snce r loc and r glob are unformly dstrbuted, we may conclude (see also Fg. 2 choosng b loc = b glob = b): P( r loc r glob s) = (1 s) 2

7 If b loc b glob, we analogously get the followng expresson for the probablty q s of the fracton of effectve transpostons beng larger than s: q s = P( r loc b loc r glob b glob s) ( = P r glob b loc r loc s ) ( + P r glob b loc r loc + s ) b glob b glob 1 ( { { }} { { }}) = mn 1, max 0, b loc r loc s b glob + mn 1, max 0, b loc r loc +s b glob dr loc 0 r glob 1 s b glob s b loc b glob s b loc b glob +s b loc Fg. 2. The shaded area denotes q s the probablty that the partcle reduces ts dstance to p by at least 25% s ( ) Our analyss drectly apples to Clerc s DPSO [2]. The algorthm proposed by Wang et al. [18] works a bt dfferent wth respect to the scalng of the attractons. In [18], Wang et al. proposed to scale the attractons by b loc, b glob [0, 1] keepng each transposton wth probablty b loc and b glob, resp., n the attracton. So n the LTD model, the movement equatons (1) and (2) reduce to x (t+1) = x (t) + b glob (p x (t) ) + b loc (p x (t) ). A transposton becomes an effectve transposton f t s kept n exactly one of the two attractons. Therefore, effectve transpostons occur wth probablty b loc (1 b glob ) + b glob (1 b loc ) = b loc + b glob 2b glob b loc. Ths s also the expected value of the fracton of effectve transpostons. The coeffcents b loc and b glob are ntended to adjust the weght of the local and the global attractor. Intutvely, f the attractors should exert a large nfluence on the partcles, b loc and b glob are set to 1. Ths works fne n the classcal PSO for contnuous problems. In the dscrete case however, whenever the LTD model apples, the local and global attractons do not pull the partcles closer to the attractors at all. 1 The probablty q s can be vsual- zed lke shown n Fg. 2, where q s amounts to the shaded area. Consder the followng small example. Let x (t) = ( ) and p = ( ). Then p x (t) = ((2 3) (3 5) (4 7) (6 8) (8 9)). Wth b = 0.8, we have b (p x (t) ) = ((2 3) (3 5) (4 7) (6 8)). Wth r loc = 0.75 and r glob = 0.5, we get r loc b (p x (t) ) = ((2 3)(3 5)(4 7)) and r glob b (p x (t) ) = ((2 3) (3 5)), and fnally x (t+1) = x (t) + r loc b (p x (t) ) + r glob b (p x (t) ) = ( ). The transposton (4 7) s the only effectve transposton. By Theorem 1, r loc

8 4 A new DPSO for the TSP 4.1 Centrod-based partcle movement: a new nterpretaton of addton As descrbed n Sec. 2.2, concatenaton s used n [2] and [18] as addton of dfferences,. e., attractons. In Sec. 3, we showed that ths approach has the dsadvantage that after some tme the expected progress becomes consderably slow. Now we propose a new method of combnng the attractons that avods ths dsadvantage. Instead of composng two attractons to a long lst of operators, we look at the destnatons, whch are the ponts the dfferent weghted attractons lead to, and compute the centrod of those destnatons. In our approach, we use no nerta (. e., we set a = 0), but only the attracton to the local and to the global attractor, each weghted n accordance wth equaton (1). Snce we have only two attractors, the centrod can be calculated easly by computng the dfference between the destnatons, scalng them by one half and addng the result to the frst destnaton. The PSO movement equatons can now be expressed wth the destnaton ponts of the attracton to the local attractor, to the global attractor and a random velocty: d loc = x (t) d glob = x (t) + r loc b loc (p x (t) ) + r glob b glob (p glob x (t) ) v rand = r rand b rand (p rand x (t) ) x (t+1) = d glob (d loc d glob ) + v rand The random movement n the end ensures that the swarm does not converge too fast. A graphcal, contnuous representaton s depcted n Fg. 3. x (t+1) p glob v rand d glob p d loc x (t) Fg. 3. Centrod-based partcle movement The advantage of ths model s that t takes the spatal structure of the search space nto account. Snce the centrod s the mean of the destnatons, our factors b loc and b glob can be transferred easly to the classcal PSO by dvdng them by 2.

9 4.2 Edge recombnaton: a new nterpretaton of velocty In [2], [18] and [14], the dfference between two postons b and a n the search space,. e., the velocty or the attracton of b to a s expressed as a lst of transpostons that transforms one sequence of ctes nto the other. Here, we propose a new method that s based on edge exchanges. Edge exchanges are a common technque used n local search methods for the TSP [7]. The dea s to mprove a soluton by exchangng crossng edges, whch results n a shorter tour. For an example, see the transformatons from Fgures 4(a) to (c). A generalzaton of ths operaton s the edge recombnaton operator. Gven a lst l = (c 1 c 2... c n ) of ctes representng a round-trp, the edge recombnaton operator edger(, j) nverts the sequence of ctes between ndces and j: l edger(, j) = (c 1... c 1 c j c j 1 c j 2... c +2 c +1 c c j+1... c n ) In our approach, we use ths operator to express the dfference between two partcles b and a. Instead of a lst of transpostons, the dfference (or velocty, or attracton) s now a lst of edge recombnaton operators that yelds b f the operators are appled to a. For example, the dfference between a = ( ) and b = ( ) s b a = (edger(5, 6) edger(3, 6)) (see Fg. 4). v 1 v 1 v 1 v 6 v 6 v 6 v 2 v 2 v 2 v 5 v 5 v 5 v 3 v 3 v 3 v 4 (a) ( ) edger(5,6) v 4 (b) ( ) edger(3,6) v 4 (c) ( ) Fg. 4. Vsualzaton of two edge recombnatons The problem of fndng the mnmum number of edge exchanges that s needed to transform one permutaton nto the other s NP-complete [16]. Therefore, we use n our experments the smple and fast approxmaton algorthm GetLstOfEdgeExchanges from [10] that s smlar to the algorthm that fnds the mnmum number of transpostons. The soluton found by the approxmaton algorthm GetLstOfEdgeExchanges has n the worst case 1 2 (n 1) tmes more edge exchanges than the optmal soluton [10]. 5 Expermental Results In our experments, we have compared our new approaches to the prevous exstng ones. We have on purpose ntally not done any local optmzaton between

10 Table 1. Comparson of dfferent dstance representatons and movement types wthout local optmzaton (left) and wth an addtonal 2-OPT-based local optmzaton of the global attractor (rght). Move type Compostoston Centrod CentrodCentrod Compo- Centrod CentrodCentrod Transposton Dstance Repr. Transposton Adj. Trans- edger Transposton poston Problem berln % 194.6% 70.5% 22.5% (7542) ± ±531.6 ±863.0 ± pr % 317.7% 156.5% 88.9% (108159) gr % 430.4% 220.8% 128.5% (55209) ± ± ± ± kroa % 529.2% 238.0% 111.2% (21282) ± ± ± ± kroc % 537.4% 256.2% 133.9% (20749) ± ± ± ± krod % 503.1% 239.0% 127.7% (21294) ± ± ± ± ln % 575.8% 305.3% 188.5% (14379) ± ± ± ± Adj. Transposton edger Transposton 24.2% 186,2% 8,2% 7.0% ±362.6 ± ±241.5 ± % 229.1% 5.8% 4.7% ± ± ± ± ± ± ± ± % 368.1% 9.7% 6.3% ± ± ± % 401.9% 7.4% 5.5% ± ± ± ±620.1 ± % 435.9% 8.2% 7.1% ± ± ±728.4 ± % 368.9% 7.9% 7.1% ± ± ±589.7 ± % 475.5% 18.0% 7.1% ± ± ±626.2 ± two teratons to see the clear mpact of exchangng the exstng approaches wth ours. The swarm we use conssts of 100 partcles and we use 1000 teratons to optmze the functon. Each confguraton s run 100 tmes to compute the mean error and the standard devaton. The entres n Table 1 provde data n the followng format: Problem name (optmal value) relatve error maxmal value found by the algorthm mean value ± standard devaton best soluton found by the algorthm In Table 1, the left four result columns present our results obtaned wth the proposed DPSO varants wthout local optmzaton. In order to make our results also comparable to other approaches, we have added the local optmzaton method from Sh et al. n [14]. These results are shown n the rght four columns

11 of Table 1. Smlarly to our proposed approach, the method of Sh also avods the convergence problems analyzed n Sec. 3, but seems to result n a smaller relatve error. In every four columns block, the frst column shows the results of the method representng dfferences as transpostons and usng a smple composton to combne the dfferences. The other three columns show the results obtaned by the centrod-based method from Sec The centrod-based approach s combned wth varous representatons of dfferences, namely adjacent transpostons, transpostons and the edge recombnatons ntroduced n Sec Our centrod-based approach avods the counter-ntutve convergence behavor explaned by the theoretcal analyss n Sec. 3. The experments show that ths method s a better choce than the smple composton of dfferences. Another crucal factor s the choce of the representaton of partcle veloctes. Our expermental results show that transpostons are better than adjacent transpostons and that the proposed edge recombnaton method performs best. Fnally, a comparson between dfferent approaches of dscrete PSO can only be sgnfcant, f the actual contrbuton of the PSO algorthm s not obfuscated by an addtonal local search procedure. Ths s why we show the results of the pure PSO wthout local optmzaton, whch can serve as a reference for future DPSO varants for the TSP. 6 Conclusons In our theoretcal analyss of dscrete PSO for the TSP we showed that the convergence behavor the convergence behavor dffers sgnfcantly from what we expect from a PSO for contnuous problems. Our analyss can be appled manly to the DPSO varants of Clerc n [2] and Wang n [18] but can also be extended to the approaches of Sh et al. [14]. The convergence behavor can be observed whenever the local and the global attractor of a partcle nearly concde. If ths s the case the transpostons occurrng n respectve veloctes cancel each other out. Ths slows down the partcles and prevents convergence. We proposed a new model for partcle moton whch from a theoretcal pont of vew does not suffer from the aforementoned convergence problem. Ths s backed by our experments, whch show a clear mprovement of the DPSO performance wth ths model. Addtonally we ntroduced a representaton for partcle veloctes, whch s based on edge exchanges n a tour. Our evaluaton shows that the edge exchange-based representaton produces better results than tradtonal approaches from the lterature. References 1. D. Anghnolf and M. Paolucc. A new dscrete partcle swarm optmzaton approach for the sngle-machne total weghted tardness schedulng problem wth sequence-dependent setup tmes. European Journal of Operatonal Research, 193:73 85, 2009.

12 2. M. Clerc. Dscrete Partcle Swarm Optmzaton, llustrated by the Travelng Salesman Problem. Webste, tsp/dscrete PSO TSP.zp. 3. M. Clerc. Dscrete partcle swarm optmzaton, llustrated by the travelng salesman problem. In G. C. Onwubolu and B. V. Babu, edtors, New Optmzaton Technques n Engneerng, Studes n Fuzzness and Soft Computng, chapter 8, pages Sprnger, M. Clerc and J. Kennedy. The partcle swarm Exploson, stablty, and convergence n a multdmensonal complex space. IEEE Transactons on Evolutonary Computaton, 6:58 73, R. C. Eberhart and J. Kennedy. A new optmzer usng partcle swarm theory. In Proc. 6th Internatonal Symposum on Mcro Machne and Human Scence, pages 39 43, E. F. G. Goldbarg, G. R. de Souza, and M. C. Goldbarg. Partcle swarm for the travelng salesman problem. In Proc. 6th European Conference on Evolutonary Computaton n Combnatoral Optmzaton (EvoCOP), pages , G. Gutn and A. P. Punnen, edtors. The Travelng Salesman Problem and Its Varatons, volume 12 of Combnatoral Optmzaton. Sprnger, S. Helwg and R. Wanka. Theoretcal analyss of ntal partcle swarm behavor. In Proc. 10th Int. Conf. on Parallel Problem Solvng from Nature (PPSN), pages , M. Jang, Y. P. Luo, and S. Y. Yang. Stochastc convergence analyss and parameter selecton of the standard partcle swarm optmzaton algorthm. Inf. Process. Lett., 102:8 16, J. Kececoglu and D. Sankoff. Exact and approxmaton algorthms for sortng by reversals, wth applcaton to genome rearrangement. Algorthmca, 13: , J. Kennedy and R. C. Eberhart. Partcle swarm optmzaton. In Proc. IEEE Internatonal Conference on Neural Networks, volume 4, pages , J. Kennedy and R. C. Eberhart. A dscrete bnary verson of the partcle swarm algorthm. In Proc. IEEE Int. Conf. on Systems, Man, and Cybernetcs, volume 5, pages , G. Renelt. TSPLIB A travelng salesman problem lbrary. ORSA Journal on Computng, 3(4): , X. H. Sh, Y. C. Lang, H. P. Lee, C. Lu, and Q. X. Wang. Partcle swarm optmzaton-based algorthms for TSP and generalzed TSP. Inf. Process. Lett., 103: , X. H. Sh, Y. Zhou, L. M. Wang, Q. Wang, and Y. C. Lang. A dscrete partcle swarm optmzaton algorthm for travellng salesman problem. In Proc. 1st Int. Conf. on Computaton Methods (ICCM), volume 2, pages , A. Solomon, P. Sutclffe, and R. Lster. Sortng crcular permutatons by reversal. In Proc. 8th Int. W shop on Algorthms and Data Structures (WADS), pages , I. C. Trelea. The partcle swarm optmzaton algorthm: Convergence analyss and parameter selecton. Inf. Process. Lett., 85: , K. P. Wang, L. Huang, C. G. Zhou, and W. Pang. Partcle swarm optmzaton for travelng salesman problem. In Proc. 2nd Int. Conf. on Machne Learnng and Cybernetcs, volume 3, pages , W. Zhong, J. Zhang, and W. Chen. A novel dscrete partcle swarm optmzaton to solve travelng salesman problem. In Proc. IEEE Congress on Evolutonary Computaton (CEC), pages , 2007.

The Study of Teaching-learning-based Optimization Algorithm

The Study of Teaching-learning-based Optimization Algorithm Advanced Scence and Technology Letters Vol. (AST 06), pp.05- http://dx.do.org/0.57/astl.06. The Study of Teachng-learnng-based Optmzaton Algorthm u Sun, Yan fu, Lele Kong, Haolang Q,, Helongang Insttute

More information

Markov Chain Monte Carlo Lecture 6

Markov Chain Monte Carlo Lecture 6 where (x 1,..., x N ) X N, N s called the populaton sze, f(x) f (x) for at least one {1, 2,..., N}, and those dfferent from f(x) are called the tral dstrbutons n terms of mportance samplng. Dfferent ways

More information

Speeding up Computation of Scalar Multiplication in Elliptic Curve Cryptosystem

Speeding up Computation of Scalar Multiplication in Elliptic Curve Cryptosystem H.K. Pathak et. al. / (IJCSE) Internatonal Journal on Computer Scence and Engneerng Speedng up Computaton of Scalar Multplcaton n Ellptc Curve Cryptosystem H. K. Pathak Manju Sangh S.o.S n Computer scence

More information

Appendix B: Resampling Algorithms

Appendix B: Resampling Algorithms 407 Appendx B: Resamplng Algorthms A common problem of all partcle flters s the degeneracy of weghts, whch conssts of the unbounded ncrease of the varance of the mportance weghts ω [ ] of the partcles

More information

Week3, Chapter 4. Position and Displacement. Motion in Two Dimensions. Instantaneous Velocity. Average Velocity

Week3, Chapter 4. Position and Displacement. Motion in Two Dimensions. Instantaneous Velocity. Average Velocity Week3, Chapter 4 Moton n Two Dmensons Lecture Quz A partcle confned to moton along the x axs moves wth constant acceleraton from x =.0 m to x = 8.0 m durng a 1-s tme nterval. The velocty of the partcle

More information

Errors for Linear Systems

Errors for Linear Systems Errors for Lnear Systems When we solve a lnear system Ax b we often do not know A and b exactly, but have only approxmatons  and ˆb avalable. Then the best thng we can do s to solve ˆx ˆb exactly whch

More information

MODIFIED PARTICLE SWARM OPTIMIZATION FOR OPTIMIZATION PROBLEMS

MODIFIED PARTICLE SWARM OPTIMIZATION FOR OPTIMIZATION PROBLEMS Journal of Theoretcal and Appled Informaton Technology 3 st ecember 0. Vol. No. 005 0 JATIT & LLS. All rghts reserved. ISSN: 9985 www.jatt.org EISSN: 87395 MIFIE PARTICLE SARM PTIMIZATIN FR PTIMIZATIN

More information

Module 3 LOSSY IMAGE COMPRESSION SYSTEMS. Version 2 ECE IIT, Kharagpur

Module 3 LOSSY IMAGE COMPRESSION SYSTEMS. Version 2 ECE IIT, Kharagpur Module 3 LOSSY IMAGE COMPRESSION SYSTEMS Verson ECE IIT, Kharagpur Lesson 6 Theory of Quantzaton Verson ECE IIT, Kharagpur Instructonal Objectves At the end of ths lesson, the students should be able to:

More information

Research Article Green s Theorem for Sign Data

Research Article Green s Theorem for Sign Data Internatonal Scholarly Research Network ISRN Appled Mathematcs Volume 2012, Artcle ID 539359, 10 pages do:10.5402/2012/539359 Research Artcle Green s Theorem for Sgn Data Lous M. Houston The Unversty of

More information

Chapter - 2. Distribution System Power Flow Analysis

Chapter - 2. Distribution System Power Flow Analysis Chapter - 2 Dstrbuton System Power Flow Analyss CHAPTER - 2 Radal Dstrbuton System Load Flow 2.1 Introducton Load flow s an mportant tool [66] for analyzng electrcal power system network performance. Load

More information

Kernel Methods and SVMs Extension

Kernel Methods and SVMs Extension Kernel Methods and SVMs Extenson The purpose of ths document s to revew materal covered n Machne Learnng 1 Supervsed Learnng regardng support vector machnes (SVMs). Ths document also provdes a general

More information

Comparison of the Population Variance Estimators. of 2-Parameter Exponential Distribution Based on. Multiple Criteria Decision Making Method

Comparison of the Population Variance Estimators. of 2-Parameter Exponential Distribution Based on. Multiple Criteria Decision Making Method Appled Mathematcal Scences, Vol. 7, 0, no. 47, 07-0 HIARI Ltd, www.m-hkar.com Comparson of the Populaton Varance Estmators of -Parameter Exponental Dstrbuton Based on Multple Crtera Decson Makng Method

More information

College of Computer & Information Science Fall 2009 Northeastern University 20 October 2009

College of Computer & Information Science Fall 2009 Northeastern University 20 October 2009 College of Computer & Informaton Scence Fall 2009 Northeastern Unversty 20 October 2009 CS7880: Algorthmc Power Tools Scrbe: Jan Wen and Laura Poplawsk Lecture Outlne: Prmal-dual schema Network Desgn:

More information

Annexes. EC.1. Cycle-base move illustration. EC.2. Problem Instances

Annexes. EC.1. Cycle-base move illustration. EC.2. Problem Instances ec Annexes Ths Annex frst llustrates a cycle-based move n the dynamc-block generaton tabu search. It then dsplays the characterstcs of the nstance sets, followed by detaled results of the parametercalbraton

More information

Problem Set 9 Solutions

Problem Set 9 Solutions Desgn and Analyss of Algorthms May 4, 2015 Massachusetts Insttute of Technology 6.046J/18.410J Profs. Erk Demane, Srn Devadas, and Nancy Lynch Problem Set 9 Solutons Problem Set 9 Solutons Ths problem

More information

Generalized Linear Methods

Generalized Linear Methods Generalzed Lnear Methods 1 Introducton In the Ensemble Methods the general dea s that usng a combnaton of several weak learner one could make a better learner. More formally, assume that we have a set

More information

Global Sensitivity. Tuesday 20 th February, 2018

Global Sensitivity. Tuesday 20 th February, 2018 Global Senstvty Tuesday 2 th February, 28 ) Local Senstvty Most senstvty analyses [] are based on local estmates of senstvty, typcally by expandng the response n a Taylor seres about some specfc values

More information

An Adaptive Learning Particle Swarm Optimizer for Function Optimization

An Adaptive Learning Particle Swarm Optimizer for Function Optimization An Adaptve Learnng Partcle Swarm Optmzer for Functon Optmzaton Changhe L and Shengxang Yang Abstract Tradtonal partcle swarm optmzaton (PSO) suffers from the premature convergence problem, whch usually

More information

An Interactive Optimisation Tool for Allocation Problems

An Interactive Optimisation Tool for Allocation Problems An Interactve Optmsaton ool for Allocaton Problems Fredr Bonäs, Joam Westerlund and apo Westerlund Process Desgn Laboratory, Faculty of echnology, Åbo Aadem Unversty, uru 20500, Fnland hs paper presents

More information

A New Scrambling Evaluation Scheme based on Spatial Distribution Entropy and Centroid Difference of Bit-plane

A New Scrambling Evaluation Scheme based on Spatial Distribution Entropy and Centroid Difference of Bit-plane A New Scramblng Evaluaton Scheme based on Spatal Dstrbuton Entropy and Centrod Dfference of Bt-plane Lang Zhao *, Avshek Adhkar Kouch Sakura * * Graduate School of Informaton Scence and Electrcal Engneerng,

More information

Psychology 282 Lecture #24 Outline Regression Diagnostics: Outliers

Psychology 282 Lecture #24 Outline Regression Diagnostics: Outliers Psychology 282 Lecture #24 Outlne Regresson Dagnostcs: Outlers In an earler lecture we studed the statstcal assumptons underlyng the regresson model, ncludng the followng ponts: Formal statement of assumptons.

More information

A PROBABILITY-DRIVEN SEARCH ALGORITHM FOR SOLVING MULTI-OBJECTIVE OPTIMIZATION PROBLEMS

A PROBABILITY-DRIVEN SEARCH ALGORITHM FOR SOLVING MULTI-OBJECTIVE OPTIMIZATION PROBLEMS HCMC Unversty of Pedagogy Thong Nguyen Huu et al. A PROBABILITY-DRIVEN SEARCH ALGORITHM FOR SOLVING MULTI-OBJECTIVE OPTIMIZATION PROBLEMS Thong Nguyen Huu and Hao Tran Van Department of mathematcs-nformaton,

More information

Assortment Optimization under MNL

Assortment Optimization under MNL Assortment Optmzaton under MNL Haotan Song Aprl 30, 2017 1 Introducton The assortment optmzaton problem ams to fnd the revenue-maxmzng assortment of products to offer when the prces of products are fxed.

More information

CS : Algorithms and Uncertainty Lecture 17 Date: October 26, 2016

CS : Algorithms and Uncertainty Lecture 17 Date: October 26, 2016 CS 29-128: Algorthms and Uncertanty Lecture 17 Date: October 26, 2016 Instructor: Nkhl Bansal Scrbe: Mchael Denns 1 Introducton In ths lecture we wll be lookng nto the secretary problem, and an nterestng

More information

COMPARISON OF SOME RELIABILITY CHARACTERISTICS BETWEEN REDUNDANT SYSTEMS REQUIRING SUPPORTING UNITS FOR THEIR OPERATIONS

COMPARISON OF SOME RELIABILITY CHARACTERISTICS BETWEEN REDUNDANT SYSTEMS REQUIRING SUPPORTING UNITS FOR THEIR OPERATIONS Avalable onlne at http://sck.org J. Math. Comput. Sc. 3 (3), No., 6-3 ISSN: 97-537 COMPARISON OF SOME RELIABILITY CHARACTERISTICS BETWEEN REDUNDANT SYSTEMS REQUIRING SUPPORTING UNITS FOR THEIR OPERATIONS

More information

Lecture 12: Discrete Laplacian

Lecture 12: Discrete Laplacian Lecture 12: Dscrete Laplacan Scrbe: Tanye Lu Our goal s to come up wth a dscrete verson of Laplacan operator for trangulated surfaces, so that we can use t n practce to solve related problems We are mostly

More information

Lecture Notes on Linear Regression

Lecture Notes on Linear Regression Lecture Notes on Lnear Regresson Feng L fl@sdueducn Shandong Unversty, Chna Lnear Regresson Problem In regresson problem, we am at predct a contnuous target value gven an nput feature vector We assume

More information

DUE: WEDS FEB 21ST 2018

DUE: WEDS FEB 21ST 2018 HOMEWORK # 1: FINITE DIFFERENCES IN ONE DIMENSION DUE: WEDS FEB 21ST 2018 1. Theory Beam bendng s a classcal engneerng analyss. The tradtonal soluton technque makes smplfyng assumptons such as a constant

More information

Design and Optimization of Fuzzy Controller for Inverse Pendulum System Using Genetic Algorithm

Design and Optimization of Fuzzy Controller for Inverse Pendulum System Using Genetic Algorithm Desgn and Optmzaton of Fuzzy Controller for Inverse Pendulum System Usng Genetc Algorthm H. Mehraban A. Ashoor Unversty of Tehran Unversty of Tehran h.mehraban@ece.ut.ac.r a.ashoor@ece.ut.ac.r Abstract:

More information

Uncertainty as the Overlap of Alternate Conditional Distributions

Uncertainty as the Overlap of Alternate Conditional Distributions Uncertanty as the Overlap of Alternate Condtonal Dstrbutons Olena Babak and Clayton V. Deutsch Centre for Computatonal Geostatstcs Department of Cvl & Envronmental Engneerng Unversty of Alberta An mportant

More information

On the Multicriteria Integer Network Flow Problem

On the Multicriteria Integer Network Flow Problem BULGARIAN ACADEMY OF SCIENCES CYBERNETICS AND INFORMATION TECHNOLOGIES Volume 5, No 2 Sofa 2005 On the Multcrtera Integer Network Flow Problem Vassl Vasslev, Marana Nkolova, Maryana Vassleva Insttute of

More information

Lectures - Week 4 Matrix norms, Conditioning, Vector Spaces, Linear Independence, Spanning sets and Basis, Null space and Range of a Matrix

Lectures - Week 4 Matrix norms, Conditioning, Vector Spaces, Linear Independence, Spanning sets and Basis, Null space and Range of a Matrix Lectures - Week 4 Matrx norms, Condtonng, Vector Spaces, Lnear Independence, Spannng sets and Bass, Null space and Range of a Matrx Matrx Norms Now we turn to assocatng a number to each matrx. We could

More information

CHAPTER 5 NUMERICAL EVALUATION OF DYNAMIC RESPONSE

CHAPTER 5 NUMERICAL EVALUATION OF DYNAMIC RESPONSE CHAPTER 5 NUMERICAL EVALUATION OF DYNAMIC RESPONSE Analytcal soluton s usually not possble when exctaton vares arbtrarly wth tme or f the system s nonlnear. Such problems can be solved by numercal tmesteppng

More information

Suppose that there s a measured wndow of data fff k () ; :::; ff k g of a sze w, measured dscretely wth varable dscretzaton step. It s convenent to pl

Suppose that there s a measured wndow of data fff k () ; :::; ff k g of a sze w, measured dscretely wth varable dscretzaton step. It s convenent to pl RECURSIVE SPLINE INTERPOLATION METHOD FOR REAL TIME ENGINE CONTROL APPLICATIONS A. Stotsky Volvo Car Corporaton Engne Desgn and Development Dept. 97542, HA1N, SE- 405 31 Gothenburg Sweden. Emal: astotsky@volvocars.com

More information

Particle Swarm Optimization with Adaptive Mutation in Local Best of Particles

Particle Swarm Optimization with Adaptive Mutation in Local Best of Particles 1 Internatonal Congress on Informatcs, Envronment, Energy and Applcatons-IEEA 1 IPCSIT vol.38 (1) (1) IACSIT Press, Sngapore Partcle Swarm Optmzaton wth Adaptve Mutaton n Local Best of Partcles Nanda ulal

More information

The Minimum Universal Cost Flow in an Infeasible Flow Network

The Minimum Universal Cost Flow in an Infeasible Flow Network Journal of Scences, Islamc Republc of Iran 17(2): 175-180 (2006) Unversty of Tehran, ISSN 1016-1104 http://jscencesutacr The Mnmum Unversal Cost Flow n an Infeasble Flow Network H Saleh Fathabad * M Bagheran

More information

V is the velocity of the i th

V is the velocity of the i th Proceedngs of the 007 IEEE Swarm Intellgence Symposum (SIS 007) Probablstcally rven Partcle Swarms for Optmzaton of Mult Valued screte Problems : esgn and Analyss Kalyan Veeramachanen, Lsa Osadcw, Ganapath

More information

The Order Relation and Trace Inequalities for. Hermitian Operators

The Order Relation and Trace Inequalities for. Hermitian Operators Internatonal Mathematcal Forum, Vol 3, 08, no, 507-57 HIKARI Ltd, wwwm-hkarcom https://doorg/0988/mf088055 The Order Relaton and Trace Inequaltes for Hermtan Operators Y Huang School of Informaton Scence

More information

829. An adaptive method for inertia force identification in cantilever under moving mass

829. An adaptive method for inertia force identification in cantilever under moving mass 89. An adaptve method for nerta force dentfcaton n cantlever under movng mass Qang Chen 1, Mnzhuo Wang, Hao Yan 3, Haonan Ye 4, Guola Yang 5 1,, 3, 4 Department of Control and System Engneerng, Nanng Unversty,

More information

Numerical Heat and Mass Transfer

Numerical Heat and Mass Transfer Master degree n Mechancal Engneerng Numercal Heat and Mass Transfer 06-Fnte-Dfference Method (One-dmensonal, steady state heat conducton) Fausto Arpno f.arpno@uncas.t Introducton Why we use models and

More information

Physics 5153 Classical Mechanics. D Alembert s Principle and The Lagrangian-1

Physics 5153 Classical Mechanics. D Alembert s Principle and The Lagrangian-1 P. Guterrez Physcs 5153 Classcal Mechancs D Alembert s Prncple and The Lagrangan 1 Introducton The prncple of vrtual work provdes a method of solvng problems of statc equlbrum wthout havng to consder the

More information

NUMERICAL DIFFERENTIATION

NUMERICAL DIFFERENTIATION NUMERICAL DIFFERENTIATION 1 Introducton Dfferentaton s a method to compute the rate at whch a dependent output y changes wth respect to the change n the ndependent nput x. Ths rate of change s called the

More information

Chapter 5. Solution of System of Linear Equations. Module No. 6. Solution of Inconsistent and Ill Conditioned Systems

Chapter 5. Solution of System of Linear Equations. Module No. 6. Solution of Inconsistent and Ill Conditioned Systems Numercal Analyss by Dr. Anta Pal Assstant Professor Department of Mathematcs Natonal Insttute of Technology Durgapur Durgapur-713209 emal: anta.bue@gmal.com 1 . Chapter 5 Soluton of System of Lnear Equatons

More information

Chapter 6. Supplemental Text Material

Chapter 6. Supplemental Text Material Chapter 6. Supplemental Text Materal S6-. actor Effect Estmates are Least Squares Estmates We have gven heurstc or ntutve explanatons of how the estmates of the factor effects are obtaned n the textboo.

More information

A Hybrid Variational Iteration Method for Blasius Equation

A Hybrid Variational Iteration Method for Blasius Equation Avalable at http://pvamu.edu/aam Appl. Appl. Math. ISSN: 1932-9466 Vol. 10, Issue 1 (June 2015), pp. 223-229 Applcatons and Appled Mathematcs: An Internatonal Journal (AAM) A Hybrd Varatonal Iteraton Method

More information

8.4 COMPLEX VECTOR SPACES AND INNER PRODUCTS

8.4 COMPLEX VECTOR SPACES AND INNER PRODUCTS SECTION 8.4 COMPLEX VECTOR SPACES AND INNER PRODUCTS 493 8.4 COMPLEX VECTOR SPACES AND INNER PRODUCTS All the vector spaces you have studed thus far n the text are real vector spaces because the scalars

More information

Homework Assignment 3 Due in class, Thursday October 15

Homework Assignment 3 Due in class, Thursday October 15 Homework Assgnment 3 Due n class, Thursday October 15 SDS 383C Statstcal Modelng I 1 Rdge regresson and Lasso 1. Get the Prostrate cancer data from http://statweb.stanford.edu/~tbs/elemstatlearn/ datasets/prostate.data.

More information

Difference Equations

Difference Equations Dfference Equatons c Jan Vrbk 1 Bascs Suppose a sequence of numbers, say a 0,a 1,a,a 3,... s defned by a certan general relatonshp between, say, three consecutve values of the sequence, e.g. a + +3a +1

More information

Air Age Equation Parameterized by Ventilation Grouped Time WU Wen-zhong

Air Age Equation Parameterized by Ventilation Grouped Time WU Wen-zhong Appled Mechancs and Materals Submtted: 2014-05-07 ISSN: 1662-7482, Vols. 587-589, pp 449-452 Accepted: 2014-05-10 do:10.4028/www.scentfc.net/amm.587-589.449 Onlne: 2014-07-04 2014 Trans Tech Publcatons,

More information

Chapter 13: Multiple Regression

Chapter 13: Multiple Regression Chapter 13: Multple Regresson 13.1 Developng the multple-regresson Model The general model can be descrbed as: It smplfes for two ndependent varables: The sample ft parameter b 0, b 1, and b are used to

More information

Supporting Information

Supporting Information Supportng Informaton The neural network f n Eq. 1 s gven by: f x l = ReLU W atom x l + b atom, 2 where ReLU s the element-wse rectfed lnear unt, 21.e., ReLUx = max0, x, W atom R d d s the weght matrx to

More information

The equation of motion of a dynamical system is given by a set of differential equations. That is (1)

The equation of motion of a dynamical system is given by a set of differential equations. That is (1) Dynamcal Systems Many engneerng and natural systems are dynamcal systems. For example a pendulum s a dynamcal system. State l The state of the dynamcal system specfes t condtons. For a pendulum n the absence

More information

ANSWERS. Problem 1. and the moment generating function (mgf) by. defined for any real t. Use this to show that E( U) var( U)

ANSWERS. Problem 1. and the moment generating function (mgf) by. defined for any real t. Use this to show that E( U) var( U) Econ 413 Exam 13 H ANSWERS Settet er nndelt 9 deloppgaver, A,B,C, som alle anbefales å telle lkt for å gøre det ltt lettere å stå. Svar er gtt . Unfortunately, there s a prntng error n the hnt of

More information

Simultaneous Optimization of Berth Allocation, Quay Crane Assignment and Quay Crane Scheduling Problems in Container Terminals

Simultaneous Optimization of Berth Allocation, Quay Crane Assignment and Quay Crane Scheduling Problems in Container Terminals Smultaneous Optmzaton of Berth Allocaton, Quay Crane Assgnment and Quay Crane Schedulng Problems n Contaner Termnals Necat Aras, Yavuz Türkoğulları, Z. Caner Taşkın, Kuban Altınel Abstract In ths work,

More information

A Robust Method for Calculating the Correlation Coefficient

A Robust Method for Calculating the Correlation Coefficient A Robust Method for Calculatng the Correlaton Coeffcent E.B. Nven and C. V. Deutsch Relatonshps between prmary and secondary data are frequently quantfed usng the correlaton coeffcent; however, the tradtonal

More information

Mathematical Preparations

Mathematical Preparations 1 Introducton Mathematcal Preparatons The theory of relatvty was developed to explan experments whch studed the propagaton of electromagnetc radaton n movng coordnate systems. Wthn expermental error the

More information

VARIATION OF CONSTANT SUM CONSTRAINT FOR INTEGER MODEL WITH NON UNIFORM VARIABLES

VARIATION OF CONSTANT SUM CONSTRAINT FOR INTEGER MODEL WITH NON UNIFORM VARIABLES VARIATION OF CONSTANT SUM CONSTRAINT FOR INTEGER MODEL WITH NON UNIFORM VARIABLES BÂRZĂ, Slvu Faculty of Mathematcs-Informatcs Spru Haret Unversty barza_slvu@yahoo.com Abstract Ths paper wants to contnue

More information

Feature Selection: Part 1

Feature Selection: Part 1 CSE 546: Machne Learnng Lecture 5 Feature Selecton: Part 1 Instructor: Sham Kakade 1 Regresson n the hgh dmensonal settng How do we learn when the number of features d s greater than the sample sze n?

More information

EEE 241: Linear Systems

EEE 241: Linear Systems EEE : Lnear Systems Summary #: Backpropagaton BACKPROPAGATION The perceptron rule as well as the Wdrow Hoff learnng were desgned to tran sngle layer networks. They suffer from the same dsadvantage: they

More information

Lecture 5 Decoding Binary BCH Codes

Lecture 5 Decoding Binary BCH Codes Lecture 5 Decodng Bnary BCH Codes In ths class, we wll ntroduce dfferent methods for decodng BCH codes 51 Decodng the [15, 7, 5] 2 -BCH Code Consder the [15, 7, 5] 2 -code C we ntroduced n the last lecture

More information

Time-Varying Systems and Computations Lecture 6

Time-Varying Systems and Computations Lecture 6 Tme-Varyng Systems and Computatons Lecture 6 Klaus Depold 14. Januar 2014 The Kalman Flter The Kalman estmaton flter attempts to estmate the actual state of an unknown dscrete dynamcal system, gven nosy

More information

Inner Product. Euclidean Space. Orthonormal Basis. Orthogonal

Inner Product. Euclidean Space. Orthonormal Basis. Orthogonal Inner Product Defnton 1 () A Eucldean space s a fnte-dmensonal vector space over the reals R, wth an nner product,. Defnton 2 (Inner Product) An nner product, on a real vector space X s a symmetrc, blnear,

More information

Some modelling aspects for the Matlab implementation of MMA

Some modelling aspects for the Matlab implementation of MMA Some modellng aspects for the Matlab mplementaton of MMA Krster Svanberg krlle@math.kth.se Optmzaton and Systems Theory Department of Mathematcs KTH, SE 10044 Stockholm September 2004 1. Consdered optmzaton

More information

Introduction to Vapor/Liquid Equilibrium, part 2. Raoult s Law:

Introduction to Vapor/Liquid Equilibrium, part 2. Raoult s Law: CE304, Sprng 2004 Lecture 4 Introducton to Vapor/Lqud Equlbrum, part 2 Raoult s Law: The smplest model that allows us do VLE calculatons s obtaned when we assume that the vapor phase s an deal gas, and

More information

2E Pattern Recognition Solutions to Introduction to Pattern Recognition, Chapter 2: Bayesian pattern classification

2E Pattern Recognition Solutions to Introduction to Pattern Recognition, Chapter 2: Bayesian pattern classification E395 - Pattern Recognton Solutons to Introducton to Pattern Recognton, Chapter : Bayesan pattern classfcaton Preface Ths document s a soluton manual for selected exercses from Introducton to Pattern Recognton

More information

Boostrapaggregating (Bagging)

Boostrapaggregating (Bagging) Boostrapaggregatng (Baggng) An ensemble meta-algorthm desgned to mprove the stablty and accuracy of machne learnng algorthms Can be used n both regresson and classfcaton Reduces varance and helps to avod

More information

Linear Regression Analysis: Terminology and Notation

Linear Regression Analysis: Terminology and Notation ECON 35* -- Secton : Basc Concepts of Regresson Analyss (Page ) Lnear Regresson Analyss: Termnology and Notaton Consder the generc verson of the smple (two-varable) lnear regresson model. It s represented

More information

Physics 207: Lecture 20. Today s Agenda Homework for Monday

Physics 207: Lecture 20. Today s Agenda Homework for Monday Physcs 207: Lecture 20 Today s Agenda Homework for Monday Recap: Systems of Partcles Center of mass Velocty and acceleraton of the center of mass Dynamcs of the center of mass Lnear Momentum Example problems

More information

Single-Facility Scheduling over Long Time Horizons by Logic-based Benders Decomposition

Single-Facility Scheduling over Long Time Horizons by Logic-based Benders Decomposition Sngle-Faclty Schedulng over Long Tme Horzons by Logc-based Benders Decomposton Elvn Coban and J. N. Hooker Tepper School of Busness, Carnege Mellon Unversty ecoban@andrew.cmu.edu, john@hooker.tepper.cmu.edu

More information

Module 9. Lecture 6. Duality in Assignment Problems

Module 9. Lecture 6. Duality in Assignment Problems Module 9 1 Lecture 6 Dualty n Assgnment Problems In ths lecture we attempt to answer few other mportant questons posed n earler lecture for (AP) and see how some of them can be explaned through the concept

More information

THE SUMMATION NOTATION Ʃ

THE SUMMATION NOTATION Ʃ Sngle Subscrpt otaton THE SUMMATIO OTATIO Ʃ Most of the calculatons we perform n statstcs are repettve operatons on lsts of numbers. For example, we compute the sum of a set of numbers, or the sum of the

More information

Formulas for the Determinant

Formulas for the Determinant page 224 224 CHAPTER 3 Determnants e t te t e 2t 38 A = e t 2te t e 2t e t te t 2e 2t 39 If 123 A = 345, 456 compute the matrx product A adj(a) What can you conclude about det(a)? For Problems 40 43, use

More information

Notes on Frequency Estimation in Data Streams

Notes on Frequency Estimation in Data Streams Notes on Frequency Estmaton n Data Streams In (one of) the data streamng model(s), the data s a sequence of arrvals a 1, a 2,..., a m of the form a j = (, v) where s the dentty of the tem and belongs to

More information

BOOTSTRAP METHOD FOR TESTING OF EQUALITY OF SEVERAL MEANS. M. Krishna Reddy, B. Naveen Kumar and Y. Ramu

BOOTSTRAP METHOD FOR TESTING OF EQUALITY OF SEVERAL MEANS. M. Krishna Reddy, B. Naveen Kumar and Y. Ramu BOOTSTRAP METHOD FOR TESTING OF EQUALITY OF SEVERAL MEANS M. Krshna Reddy, B. Naveen Kumar and Y. Ramu Department of Statstcs, Osmana Unversty, Hyderabad -500 007, Inda. nanbyrozu@gmal.com, ramu0@gmal.com

More information

NON-CENTRAL 7-POINT FORMULA IN THE METHOD OF LINES FOR PARABOLIC AND BURGERS' EQUATIONS

NON-CENTRAL 7-POINT FORMULA IN THE METHOD OF LINES FOR PARABOLIC AND BURGERS' EQUATIONS IJRRAS 8 (3 September 011 www.arpapress.com/volumes/vol8issue3/ijrras_8_3_08.pdf NON-CENTRAL 7-POINT FORMULA IN THE METHOD OF LINES FOR PARABOLIC AND BURGERS' EQUATIONS H.O. Bakodah Dept. of Mathematc

More information

Riccardo Poli, James Kennedy, Tim Blackwell: Particle swarm optimization. Swarm Intelligence 1(1): (2007)

Riccardo Poli, James Kennedy, Tim Blackwell: Particle swarm optimization. Swarm Intelligence 1(1): (2007) Sldes largely based on: Rccardo Pol, James Kennedy, Tm Blackwell: Partcle swarm optmzaton. Swarm Intellgence 1(1): 33-57 (2007) Partcle Swarm Optmzaton Sldes largely based on: Rccardo Pol, James Kennedy,

More information

Fixed point method and its improvement for the system of Volterra-Fredholm integral equations of the second kind

Fixed point method and its improvement for the system of Volterra-Fredholm integral equations of the second kind MATEMATIKA, 217, Volume 33, Number 2, 191 26 c Penerbt UTM Press. All rghts reserved Fxed pont method and ts mprovement for the system of Volterra-Fredholm ntegral equatons of the second knd 1 Talaat I.

More information

Lecture 4. Instructor: Haipeng Luo

Lecture 4. Instructor: Haipeng Luo Lecture 4 Instructor: Hapeng Luo In the followng lectures, we focus on the expert problem and study more adaptve algorthms. Although Hedge s proven to be worst-case optmal, one may wonder how well t would

More information

Irregular vibrations in multi-mass discrete-continuous systems torsionally deformed

Irregular vibrations in multi-mass discrete-continuous systems torsionally deformed (2) 4 48 Irregular vbratons n mult-mass dscrete-contnuous systems torsonally deformed Abstract In the paper rregular vbratons of dscrete-contnuous systems consstng of an arbtrary number rgd bodes connected

More information

HMMT February 2016 February 20, 2016

HMMT February 2016 February 20, 2016 HMMT February 016 February 0, 016 Combnatorcs 1. For postve ntegers n, let S n be the set of ntegers x such that n dstnct lnes, no three concurrent, can dvde a plane nto x regons (for example, S = {3,

More information

Real-Time Systems. Multiprocessor scheduling. Multiprocessor scheduling. Multiprocessor scheduling

Real-Time Systems. Multiprocessor scheduling. Multiprocessor scheduling. Multiprocessor scheduling Real-Tme Systems Multprocessor schedulng Specfcaton Implementaton Verfcaton Multprocessor schedulng -- -- Global schedulng How are tasks assgned to processors? Statc assgnment The processor(s) used for

More information

A Hybrid Differential Evolution Algorithm Game Theory for the Berth Allocation Problem

A Hybrid Differential Evolution Algorithm Game Theory for the Berth Allocation Problem A Hybrd Dfferental Evoluton Algorthm ame Theory for the Berth Allocaton Problem Nasser R. Sabar, Sang Yew Chong, and raham Kendall The Unversty of Nottngham Malaysa Campus, Jalan Broga, 43500 Semenyh,

More information

SPECTRAL ANALYSIS USING EVOLUTION STRATEGIES

SPECTRAL ANALYSIS USING EVOLUTION STRATEGIES SPECTRAL ANALYSIS USING EVOLUTION STRATEGIES J. FEDERICO RAMÍREZ AND OLAC FUENTES Insttuto Naconal de Astrofísca, Óptca y Electrónca Lus Enrque Erro # 1 Santa María Tonanzntla, Puebla, 784, Méxco framrez@cseg.naoep.mx,

More information

Appendix B. The Finite Difference Scheme

Appendix B. The Finite Difference Scheme 140 APPENDIXES Appendx B. The Fnte Dfference Scheme In ths appendx we present numercal technques whch are used to approxmate solutons of system 3.1 3.3. A comprehensve treatment of theoretcal and mplementaton

More information

ADVANCED MACHINE LEARNING ADVANCED MACHINE LEARNING

ADVANCED MACHINE LEARNING ADVANCED MACHINE LEARNING 1 ADVANCED ACHINE LEARNING ADVANCED ACHINE LEARNING Non-lnear regresson technques 2 ADVANCED ACHINE LEARNING Regresson: Prncple N ap N-dm. nput x to a contnuous output y. Learn a functon of the type: N

More information

4 Analysis of Variance (ANOVA) 5 ANOVA. 5.1 Introduction. 5.2 Fixed Effects ANOVA

4 Analysis of Variance (ANOVA) 5 ANOVA. 5.1 Introduction. 5.2 Fixed Effects ANOVA 4 Analyss of Varance (ANOVA) 5 ANOVA 51 Introducton ANOVA ANOVA s a way to estmate and test the means of multple populatons We wll start wth one-way ANOVA If the populatons ncluded n the study are selected

More information

FUZZY GOAL PROGRAMMING VS ORDINARY FUZZY PROGRAMMING APPROACH FOR MULTI OBJECTIVE PROGRAMMING PROBLEM

FUZZY GOAL PROGRAMMING VS ORDINARY FUZZY PROGRAMMING APPROACH FOR MULTI OBJECTIVE PROGRAMMING PROBLEM Internatonal Conference on Ceramcs, Bkaner, Inda Internatonal Journal of Modern Physcs: Conference Seres Vol. 22 (2013) 757 761 World Scentfc Publshng Company DOI: 10.1142/S2010194513010982 FUZZY GOAL

More information

Linear Feature Engineering 11

Linear Feature Engineering 11 Lnear Feature Engneerng 11 2 Least-Squares 2.1 Smple least-squares Consder the followng dataset. We have a bunch of nputs x and correspondng outputs y. The partcular values n ths dataset are x y 0.23 0.19

More information

arxiv:cs.cv/ Jun 2000

arxiv:cs.cv/ Jun 2000 Correlaton over Decomposed Sgnals: A Non-Lnear Approach to Fast and Effectve Sequences Comparson Lucano da Fontoura Costa arxv:cs.cv/0006040 28 Jun 2000 Cybernetc Vson Research Group IFSC Unversty of São

More information

EPR Paradox and the Physical Meaning of an Experiment in Quantum Mechanics. Vesselin C. Noninski

EPR Paradox and the Physical Meaning of an Experiment in Quantum Mechanics. Vesselin C. Noninski EPR Paradox and the Physcal Meanng of an Experment n Quantum Mechancs Vesseln C Nonnsk vesselnnonnsk@verzonnet Abstract It s shown that there s one purely determnstc outcome when measurement s made on

More information

Report on Image warping

Report on Image warping Report on Image warpng Xuan Ne, Dec. 20, 2004 Ths document summarzed the algorthms of our mage warpng soluton for further study, and there s a detaled descrpton about the mplementaton of these algorthms.

More information

Semi-supervised Classification with Active Query Selection

Semi-supervised Classification with Active Query Selection Sem-supervsed Classfcaton wth Actve Query Selecton Jao Wang and Swe Luo School of Computer and Informaton Technology, Beng Jaotong Unversty, Beng 00044, Chna Wangjao088@63.com Abstract. Labeled samples

More information

Calculation of time complexity (3%)

Calculation of time complexity (3%) Problem 1. (30%) Calculaton of tme complexty (3%) Gven n ctes, usng exhaust search to see every result takes O(n!). Calculaton of tme needed to solve the problem (2%) 40 ctes:40! dfferent tours 40 add

More information

A new construction of 3-separable matrices via an improved decoding of Macula s construction

A new construction of 3-separable matrices via an improved decoding of Macula s construction Dscrete Optmzaton 5 008 700 704 Contents lsts avalable at ScenceDrect Dscrete Optmzaton journal homepage: wwwelsevercom/locate/dsopt A new constructon of 3-separable matrces va an mproved decodng of Macula

More information

Workshop: Approximating energies and wave functions Quantum aspects of physical chemistry

Workshop: Approximating energies and wave functions Quantum aspects of physical chemistry Workshop: Approxmatng energes and wave functons Quantum aspects of physcal chemstry http://quantum.bu.edu/pltl/6/6.pdf Last updated Thursday, November 7, 25 7:9:5-5: Copyrght 25 Dan Dll (dan@bu.edu) Department

More information

Note on EM-training of IBM-model 1

Note on EM-training of IBM-model 1 Note on EM-tranng of IBM-model INF58 Language Technologcal Applcatons, Fall The sldes on ths subject (nf58 6.pdf) ncludng the example seem nsuffcent to gve a good grasp of what s gong on. Hence here are

More information

PARTICIPATION FACTOR IN MODAL ANALYSIS OF POWER SYSTEMS STABILITY

PARTICIPATION FACTOR IN MODAL ANALYSIS OF POWER SYSTEMS STABILITY POZNAN UNIVE RSITY OF TE CHNOLOGY ACADE MIC JOURNALS No 86 Electrcal Engneerng 6 Volodymyr KONOVAL* Roman PRYTULA** PARTICIPATION FACTOR IN MODAL ANALYSIS OF POWER SYSTEMS STABILITY Ths paper provdes a

More information

On the correction of the h-index for career length

On the correction of the h-index for career length 1 On the correcton of the h-ndex for career length by L. Egghe Unverstet Hasselt (UHasselt), Campus Depenbeek, Agoralaan, B-3590 Depenbeek, Belgum 1 and Unverstet Antwerpen (UA), IBW, Stadscampus, Venusstraat

More information

Grover s Algorithm + Quantum Zeno Effect + Vaidman

Grover s Algorithm + Quantum Zeno Effect + Vaidman Grover s Algorthm + Quantum Zeno Effect + Vadman CS 294-2 Bomb 10/12/04 Fall 2004 Lecture 11 Grover s algorthm Recall that Grover s algorthm for searchng over a space of sze wors as follows: consder the

More information

Beyond Zudilin s Conjectured q-analog of Schmidt s problem

Beyond Zudilin s Conjectured q-analog of Schmidt s problem Beyond Zudln s Conectured q-analog of Schmdt s problem Thotsaporn Ae Thanatpanonda thotsaporn@gmalcom Mathematcs Subect Classfcaton: 11B65 33B99 Abstract Usng the methodology of (rgorous expermental mathematcs

More information