Chapter Introduction

Size: px
Start display at page:

Download "Chapter Introduction"

Transcription

1 Unversty of Pretora etd Wle, D N (5) Chapter Objectvty of the PSOA. Introducton In scence, most physcal phenomena are nvarant. It s fundamental that mathematcal representaton of these phenomena reflects ths nvarance. Ths fundamental requrement s nown as objectvty, frame-ndfference or observer ndependence, and s well nown n classcal mechancs [5]. For objectvty, the descrpton of some quantty has to be nvarant under pure translatons, as well as pure rotatons. Objectvty or observer ndependence s also hghly desrable (almost essental) n optmzaton procedures, to reflect the nvarance of the physcal processes that are optmzed. Robust optmzaton procedures and algorthms should defntely be nvarant. In classcal gradent based optmzaton, the gradent vector (or some conjugate drecton to the gradent), ndcates some drecton of mprovement, even f ths drecton s not optmal. Ths accounts for the reference frame; classcal optmzaton s (usually) frame nvarant. In modern (stochastc) optmzaton procedures, the requrement of observer ndependence s equally essental. These algorthms nclude genetc programmng [7], genetc algorthms [8], evolutonary strateges [9], dfferental evoluton [] and the partcle swarm optmzaton algorthm (PSOA) [, ]. For the genetc algorthm (GA), Salomon [5, 55] demonstrated the lac of rotatonal nvarance of the algorthm. He showed that the GA s performance at low mutaton rates s sgnfcantly nfluenced by the frame of reference used to pose a problem. The PSOA was ntroduced by Kennedy and Eberhart [] as a gradent free stochastc optmzaton algorthm. The fundamental prncple behnd the PSOA s the evolutonary advantages that the sharng of nformaton offers. Ths s often nown as collaboratve searchng. The PSOA s qute smple: At frst, a swarm of p partcles s randomly deployed n an n-dmensonal desgn doman. The partcles then update ther postons n the desgn doman over unt tme ncrements usng a smple stochastc rule, nown as the velocty rule. The qualty of each partcle s poston at each teraton s then evaluated usng the objectve or cost functon. Each partcle s cogntve memory allows t to remember t s own best cost functon 7

2 Unversty of Pretora etd Wle, D N (5) CHAPTER. OBJECTIVITY OF THE PSOA 8 value, wth assocated poston, over tme. Importantly, the socal nteracton and awareness of the partcles allow them to also remember the best cost functon value the swarm tself found over tme. In Chapter t was shown that mplementaton subtletes due to ambguous notaton have resulted n two dstnctly dfferent mplementatons of the PSOA. Whle ths does not repute negatvely on the ngenuty of the dea of Kennedy and Eberhart, dscernng between these two mplementatons s of crucal mportance. The behavor of the respectve mplementatons s maredly dfferent, although they only dffer n the formulaton of the velocty updatng rule. In fact, the dfferences are merely due to subtle dfferences n the ntroducton of randomness nto the algorthm. In ths chapter, the objectvty of the PSOA s nvestgated. It s shown that the frst formulaton PSOAF s objectve, combned wth the dsadvantage that the partcle trajectores collapse to lne searches. It s then show that the second formulaton PSOAF s not objectve, although t has the advantage that the partcle trajectores are n-dmensonal space fllng. A novel formulaton that s both objectve and dverse,.e. the algorthm generates partcle trajectores that are space fllng, s then presented.. Notes on the nvestgaton The nvestgaton nto the objectvty of the PSOA s started by defnng the nstantaneous search doman of a partcle, vz. the doman to whch the search of a partcle at teraton s restrcted as dscussed n Chapter. From Eqs. (.) and (.), t s observed that the nstantaneous search doman s composed from a determnstc contrbuton gven by (x + wv ), and a stochastc contrbuton due to ν. The stochastc doman s bounded, and has an assocated probablty dstrbuton, due to the random scalng wth fnte scalars. In order to nvestgate the objectvty of the stochastc contrbuton ν of the nstantaneous search doman, Monte Carlo smulatons [56] are used. These are conducted for dfferent values of p, pg and x. Scatter plots are constructed to defne the doman of possble stochastc vectors ν by generatng nstances of ν. In all nvestgatons c = c =.. Formulaton (PSOAF) For PSOAF, the stochastc vector ν s gven by ν = c r (p x ) + c r (pg x ), (.) where r and r represent two unform real random scalars between and, whch are updated at every teraton, and for each partcle n the swarm. The random numbers r and r ndependently scale only the magntudes of the cogntve and socal vectors, respectvely gven by c (p x ) and c (p g x ). The cogntve vector c (p x ) and the socal vector c (p g x ) can be anythng from normal to parallel w.r.t. each other. When the cogntve and socal vectors are not parallel, Eq. (.) may be nterpreted as the vector equaton of a bounded plane P n n-dmensonal space. The bounded plane s then translated n

3 Unversty of Pretora etd Wle, D N (5) CHAPTER. OBJECTIVITY OF THE PSOA 9 x + () P (p x ) (p g x ) x + wv x + () Fgure.: Parttonng the poston vector x + nto a determnstc contrbuton (x + wv ), and a stochastc contrbuton (ν P ), for c = c =. x + () L (p x ) (p g x ) x + wv x + () Fgure.: Parttonng the poston vector x + nto a determnstc contrbuton (x + wv ), and a stochastc contrbuton (ν L ), for c = c =. n-dmensonal space by the addton of x and wv, as depcted n Fgure.. Whenever the cogntve and socal vectors c (p x ) and c (p g x ) are parallel, Eq. (.) may be nterpreted as the vector equaton of a bounded lne L n n-dmensonal space. Agan, the bounded lne s translated n n-dmensonal space by the addton of x and wv, as depcted n Fgure.. The ntrnsc propertes of a vector are ts magntude and drecton; these exst ndependent of a reference frame [57]. In PSOAF, only the vector magntudes (whch are nvarant) are randomly scaled. Also, snce the vectors c (p x ) and c (p g x ) are constructed through the subtracton of two vectors, they are also translatonally nvarant. Both crtera for observer ndependence are met, hence PSOAF s objectve... PSOAF: Investgaton of the nstantaneous search doman Objectvty of PSOAF s now llustrated by conductng Monte Carlo smulatons. Smlar smulatons wll be conducted for the algorthmc formulatons n sectons to come.

4 Unversty of Pretora etd Wle, D N (5) CHAPTER. OBJECTIVITY OF THE PSOA () ν ν+().5 (pg x).5 (pg x ).5.5 (p x) ν+().5 (p x) ν+().5.5 Fgure.: PSOAF: Scatter plot of stochastc vectors ν, generated usng Monte possble Carlo smulatons, wth a) p x = [ ] and pg x = [ ] and b) p x = [ ] and pg x = [ ]. Each pont represents the end pont of a stochastc vector wth orgn at [ ]. Frst, a study s conducted for non-parallel cogntve and socal vectors c (p x ) and c (pg x ). In Fgure., the vectors p x and pg x are respectvely gven by [ ] and [ ]. A scatter plot yelds the plane P, wth c and c merely scalng P. A scatter plot s then constructed after rotatngthe vectors pg x and p x 5 clocwse, as depcted n Fgure.. Hence p x and pg x are respectvely gven by [ ] and [ ]. From Fgure., t s clear that the doman remans a bounded plane P, whch s merely rotated 5 clocwse. It also follows from random varable theory that the probablty dstrbuton over the doman P s unform [5], as llustrated n Fgures. and.. Secondly, a smlar study s conducted for parallel cogntve and socal vectors c (p x ) and c (pg x ), as depcted n Fgure.. In Fgure., the parallel vectors p x and pg x are respectvely gven by [ ] and [ ]. The doman s a bounded lne L wth c and c merely scalng the length of L. Agan a scatter plot s constructed after rotatng pg x and p x 5 clocwse, as depcted n Fgure.. Now, p x and pg x are respectvely gven by [ ] and [ ]. As shown n Fgure., the bounded lne L s merely rotated. It follows from random varable theory that the probablty dstrbuton over the bounded lne s tr-lnear [5], for dfferent vector lengths p x and pg x. As dscussed earler and graphcally demonstrated here, PSOAF s objectve. A rotaton of the vectors p, pg and x merely results n a rotaton of the stochastc domans, P and L. Ths

5 Unversty of Pretora etd Wle, D N (5) CHAPTER. OBJECTIVITY OF THE PSOA ν+().5 () ν.5 (pg x ).5.5 (p x ).5 (p x ) (pg x ) ν+() ν+() 5 6 Fgure.: PSOAF: Scatter plot of possble stochastc vectors ν, generated usng Monte g Carlo smulatons, wth a) p x = [ ] and p x = [ ] and b) p x = [ ] and pg x = [ ]. follows snce only the magntude of the cogntve and socal vectors are scaled n PSOAF.. Formulaton (PSOAF) The stochastc vector ν of PSOAF s gven by ν = c r (p x ) + c r (pg x ), (.) where the operator ndcates component by component multplcaton between two vectors. Hence the random vectors r m are gven by r m = (ρ, ρ,, ρn ), m =,, (.) wth ρl, l =,,, n unform random numbers between and. Eq. (.) s no longer a vector equaton of a bounded plane P, snce every non-zero component of (p x ) and (pg x ) s ndependently scaled. As a result, the doman of possble stochastc vectors s generalzed to n-dmensonal space S. However, snce the components of a vector are gven wth respect to a specfc reference frame, PSOAF s rotatonally varant. (Although PSOAF s of course translatonally nvarant.) Nevertheless, PSOAF s observer dependent, snce only one of the two crtera of objectvty s met... PSOAF: Investgaton of the nstantaneous search doman The observer dependence of PSOAF s now quantfed, usng Monte Carlo smulatons, smlar to those n Secton...

6 Unversty of Pretora etd Wle, D N (5) CHAPTER. OBJECTIVITY OF THE PSOA g (p x) ().5 ν ν+() ν+().5.5 (p x).5 (p x).5 (pg x ) ν+().5.5 Fgure.5: PSOAF: Scatter plot of possble stochastc vectors ν, generated usng Monte Carlo smulatons wth a) p x = [ ] and pg x = [ ] and b) p x = [ ] and pg x = [ ]. As before, the study s conducted for non-parallel cogntve and socal vectors c (p x ) and c (pg x ). Fgure.5 depcts that the doman s an n-dmensonal space S (wth n = n ths case), wth c and c merely scalng S. It s also clear that the probablty dstrbuton over S s non-unform. The scatter plot after rotatng the vectors pg x and p x 5 clocwse, s depcted n Fgure.5. It s clear that the doman changes after rotaton of the vectors. However, the doman remans an n-dmensonal space S, but the sze of, and the probablty dstrbuton over, the doman depends on the orentaton w.r.t. the Cartesan coordnate axs. The study s repeated for parallel cogntve and socal vectors c (p x ) and c (pg x ), as depcted n Fgure.6. The doman s stll generalzed to n-dmensonal space S wth c and c merely scalng the sze of S. The scatter plot after rotatng pg x and p x 5 clocwse, s depcted n Fgure.6. It s clear that the doman changes sgnfcantly after rotaton of the vectors. In fact, the doman collapses to a bounded lne L, snce both vectors are parallel to one of the Cartesan bass vectors. As dscussed earler and graphcally demonstrated here, PSOAF s observer dependent. A rotaton of the vectors (p x ) and (pg x ) results n the sze of, and the probablty dstrbuton over, the stochastc doman to change. Ths follows snce PSOAF scales the components of the cogntve and socal vectors. Snce the components of a vector are observer dependent, PSOAF s also observer dependent. However, the advantage of PSOAF s that the partcle trajectores reman space fllng n ndmensonal space as shown n Chapter. The result s that dversty n partcle trajectores are mantaned.

7 Unversty of Pretora etd Wle, D N (5) CHAPTER. OBJECTIVITY OF THE PSOA ν+().5 () ν.5 g (p x).5.5 (p x ).5 ν+() 5 (p x ) (pg x ) 6.5 ν+() 5 6 Fgure.6: PSOAF: Scatter plot of possble stochastc vectors ν, generated usng Monte g Carlo smulatons wth a) p x = [ ] and p x = [ ] and b) p x = [ ] and pg x = [ ]..5 Novel Formulaton: PSOAF* As dscussed n Secton., PSOAF s objectve, although the partcle trajectores collapse to lnes. (The advantage of dverse (n-dmensonal) partcle search trajectores are quantfed n Chapter.) On the other hand, PSOAF allows for partcles to have dverse search trajectores, but unfortunately ths comes at the cost of sacrfcng objectvty. An mplementaton of the PSOA s now presented that allows for dverse partcle search trajectores, whle retanng objectvty. Based on PSOAF, the novel, dverse mplementaton, s denoted PSOAF*. In PSOAF*, the vector magntudes are scaled, and the vector drectons of (p x ) and (pg x ) perturbed, by multplyng each of the above vectors wth an ndependent random rotaton matrx. The random rotaton matrces are constructed anew for each partcle and for every teraton, hence ν = c r Q (p x ) + c r Q (pg x ), (.) wth each Ql, l =,, a random rotaton matrx of dmenson n n. Q s a proper orthogonal matrx (wth determnant ). Numerous methods are avalable to construct rotaton matrces (e.g. see the approach of Salomon [5]. Constructng n n matrces usng Salomon s routne s however computatonally expensve, snce (n )(n ) matrx-matrx multplcatons are requred.) As a computatonally vable alternatve, the exponental map s used [58]. There are agan numerous ways to construct exponental maps. The smple seres method s selected [58]. The general

8 Unversty of Pretora etd Wle, D N (5) CHAPTER. OBJECTIVITY OF THE PSOA seres expanson of an exponental map s gven by Q = I + W + W W + 6 W W W +, (.5) where I s the dentty matrx and W s a sew matrx. The random sew matrx W s constructed as follows: W = απ 8 (A AT ), (.6) wth A an n n random matrx wth each entry a unform random number between.5 and.5. α s a real scalng factor and superscrpt T denotes the matrx transpose. The author selects to construct the exponental map Q for small perturbatons, usng only the frst two terms of a truncated seres method,.e. Q = I + W. (.7) Ths s the lnear approxmaton to a rotaton matrx, and s vald for small perturbatons, snce the entres of the hgher order terms are close to zero. The advantage of the smplfcaton s that the number of matrx-matrx multplcatons s zero. (It s mportant to note that the varable bounds defnng D should be normalzed, such that the boundary ranges are equal.).5. PSOAF*: Investgaton of the nstantaneous search doman As before, the objectvty of PSOAF* s quantfed usng Monte Carlo smulatons. In dmensons, α = 5 s selected. (Although ths s not small, ths serves to clearly llustrate the proposed concept). Agan, the study for non-parallel cogntve and socal vectors s conducted, as depcted n Fgure.7. The doman generalzes to n-dmensonal space S, wth c and c scalng S. The scatter plot after rotatng the vectors ( p g ( ) x and p x) 5 clocwse s depcted n Fgure.7. Clearly, the doman remans generalzed to n-dmensonal space S, rotated 5 clocwse. The probablty dstrbuton over the doman S s non-unform. Secondly, the study for parallel cogntve and socal vectors c (p x ) and c (p g x ) s conducted. Agan the doman generalzes to n-dmensonal space S, wth c and c merely scalng the doman. A scatter plot after rotatng ( p g x ) and ( p x ) 5 clocwse s constructed, as depcted n Fgure.8. Evdently, the n-dmensonal space S s merely rotated, and the probablty dstrbuton over the doman s non-unform. As dscussed earler and graphcally demonstrated here, PSOAF* s an objectve formulaton. A rotaton of the vectors (p x ) and (pg x ) merely results n a rotaton of the stochastc doman S. The drawbac of PSOAF s overcome n PSOAF*, where n addton to scalng the vector magntudes, the vectors are drectonally perturbed. The magntudes and drectons of (p x ) and

9 Unversty of Pretora etd Wle, D N (5) ν+().5 ν+() CHAPTER. OBJECTIVITY OF THE PSOA.5.5 g ( p x) ( pg x).5 ( p x).5 ν+() ( p x) ν () () ν ν+() Fgure.7: PSOAF*: Scatter plot of possble stochastc vectors ν, generated usng Monte g Carlo smulatons, wth a) p x = [ ] and p x = [ ] and b) p x = [ ] and pg x = [ ]..5.5 ( p.5 ( p x) ( pg x ) ( pg x).5 x) ν+() 5 ν+() 5 generated usng Fgure.8: PSOAF*: Scatter plot of nstances of the stochastc vectors ν, g Monte Carlo smulatons, wth a) p x = [ ] and p x = [ ] and b) p x = [ ] and pg x = [ ].

10 Unversty of Pretora etd Wle, D N (5) CHAPTER. OBJECTIVITY OF THE PSOA 6 (p g x ) are used to only ndcate potental mprovement, thereby placng only some fath n both drecton and step sze. Ths s n contrast to PSOAF, where absolute fath s placed n the drectons prescrbed by (p x ) and (pg x ), whle only some fath s placed on step sze. (Incdentally, termnaton occurs when p, pg and x converge on the same pont n n-dmensonal space, combned wth wv.).6 Numercal experments An emprcal study to quantfy the (lac of) objectvty of the three dscussed mplementatons of the PSOA s now performed. A synchronous updatng method s used []. Real varables are mplemented usng double-precson floatng-pont arthmetc. For ths study the algorthm parameters are c = c =, the swarm sze s p = partcles and the computatons are performed for varous constant nerta factors w. Intal veloctes are assumed to equal. In PSOAF*, α = s smply selected. (The author does not see an optmal value for α, but merely wshes to llustrate the effects of perturbng the vector drectons.) Furthermore, no boundary or velocty restrctons are mplemented. Each run conssts of functon evaluatons ( teratons). All results presented are averaged over runs. In the study the followng fve test functons are used: ) The Rosenbroc functon (unmodal, f ): f (x) = n ( ( ) x x ( ) ) + x. = ) The Quadrc functon (unmodal, f ): f (x) = ) The Acley functon (multmodal, f ): f (x) = exp ( exp ( ( n ) x j. = j= ). n n = x ) n n = cos(πx ) + + e. v) The generalzed Rastrgn functon (multmodal, f ): n ( ) f (x) = x cos(πx ) +. v) Fnally, the generalzed Grewan functon (multmodal, f ): f (x) = n n ( x x ) cos +. = = =

11 Unversty of Pretora etd Wle, D N (5) CHAPTER. OBJECTIVITY OF THE PSOA 7 Table.: Test functon parameters Functon n doman f ±.8 f ±. f ±. f ± 5. f ± 6. The parameters used n the study are gven n Table.. The doman column represents the range of each dmenson of the desgn varables; the test functon domans are symmetrcal about. n all dmensons. The multmodal functons f and f are decomposable [5], vz. the desgn varables are uncoupled. Ths mples that once an optmal value for a gven desgn varable s obtaned, t remans optmal, ndependent of the other desgn varables. Ths s smlar to optmzng n -dmensonal optmzaton problems, nstead of n-dmensonal coupled optmzaton problem. The test set s therefore studed n the unrotated or decomposable reference frame f(x), as well as n an arbtrary rotated reference frame f(rx), n whch the desgn varables are coupled [55]. Here, R s a random, proper orthogonal transformaton matrx, constructed as n [5]. The transformaton matrx results n a pure rotaton of each test functon. For each of the ndependent runs, a new random rotaton matrx R s constructed, to ensure that there s no bas toward any partcular reference frame..7 Dscusson of Results Depcted n Fgures.9,.,.,. and. are the mean objectve functon values after 5 functon evaluatons (or teratons) averaged over runs for both the unrotated and rotated functons. The rotatonal nvarance of PSOAF and PSOAF* are evdent from Fgures.9,.,.,. and.. The poor performance of PSOAF drectly results from the partcle trajectores collapsng to lnes as shown n Chapter. There s a sgnfcant mproved performance for all the test functons wth PSOAF*, due to the scalng of the vector magntudes and perturbaton of the vector drectons. The rotatonal varance of PSOAF s evdent from Fgures.9,.,.,. and.. There s a severe performance loss for some of the rotated functons compared to the unrotated functons. Two functons result n smlar performance for the rotated and unrotated functons, namely the Quadrc functon f, and the Grewan functon, f. The Quadrc and Grewan functons are almost nsenstve to rotaton. (The Grewan functon s a sphercal functon on whch snusodal nose s mposed. Hence ths functon s artfcally ndfferent to rotaton, snce many local mnma appear, rrespectve of rotaton.)

12 Unversty of Pretora etd Wle, D N (5) CHAPTER. OBJECTIVITY OF THE PSOA 8 Average objectve functon value PSOAF unrotated PSOAF rotated PSOAF* (α=) unrotated PSOAF* (α=) rotated Average objectve functon value PSOAF unrotated PSOAF rotated Inerta constant ω Inerta constant ω Fgure.9: Average functon value obtaned wth a) PSOAF and PSOAF*, and b) PSOAF after 5 functon evaluatons ( teratons) averaged over runs on the rotated and unrotated Rosenbroc test functon f. Average objectve functon value PSOAF unrotated PSOAF rotated PSOAF* (α=) unrotated PSOAF* (α=) rotated Average objectve functon value PSOAF unrotated PSOAF rotated Inerta constant ω Inerta constant ω Fgure.: Average functon value obtaned wth a) PSOAF and PSOAF*, and b) PSOAF after 5 functon evaluatons ( teratons) averaged over runs on the rotated and unrotated Quadrc test functon f.

13 Unversty of Pretora etd Wle, D N (5) CHAPTER. OBJECTIVITY OF THE PSOA 9 Average objectve functon value PSOAF unrotated PSOAF rotated PSOAF* (α=) unrotated PSOAF* (α=) rotated Inerta constant ω Average objectve functon value PSOAF unrotated PSOAF rotated Inerta constant ω Fgure.: Average functon value obtaned wth a) PSOAF and PSOAF*, and b) PSOAF after 5 functon evaluatons ( teratons) averaged over runs on the rotated and unrotated Acley test functon f. Average objectve functon value PSOAF unrotated PSOAF rotated PSOAF* (α=) unrotated PSOAF* (α=) rotated Average objectve functon value PSOAF unrotated PSOAF rotated Inerta constant ω Inerta constant ω Fgure.: Average functon value obtaned wth a) PSOAF and PSOAF*, and b) PSOAF after 5 functon evaluatons ( teratons) averaged over runs on the rotated and unrotated Rastrgn test functon f.

14 Unversty of Pretora etd Wle, D N (5) CHAPTER. OBJECTIVITY OF THE PSOA Average objectve functon value PSOAF unrotated PSOAF rotated PSOAF* (α=) unrotated PSOAF* (α=) rotated Average objectve functon value PSOAF unrotated PSOAF rotated Inerta constant ω Inerta constant ω Fgure.: Average functon value obtaned wth a) PSOAF and PSOAF*, and b) PSOAF after 5 functon evaluatons ( teratons) averaged over runs on the rotated and unrotated Grewan test functon f. Inadvertently, ths also suggest that non-sphercal unmodal test functons should be used to evaluate objectvty. The two unmodal functons, namely the Rosenbroc functon f and the Quadrc functon f, are of some nterest, snce they ndcate the ablty of an algorthm to search wthn a local basn. The performance of PSOAF* s sgnfcantly better than PSOAF for both functons, for both the rotated and unrotated test functons. PSOAF demonstrates a severe performance loss for the Rosenbroc functon, for the rotated functon compared to the unrotated functon. (Note the scale of the graphs n Fgure.9.) The performance dfference between PSOAF and PSOAF* for the unmodal Quadrc test functon f, s depcted n Fgure.. Fgure. depcts the mean functon value convergence hstory of PSOAF (wth w =.8), PSOAF (wth w =.) and PSOAF* (wth w =.5 and α = ) over 5 teratons. The values for w are optmal for each algorthm, but no attempt was made to optmze α. Of the three formulatons, t s clear that PSOAF* s computatonally the most effectve on the Quadrc test functon. For the multmodal functons, PSOAF demonstrates notable performance loss. See for example the Acley functon f, and the Rastrgn functon f. In contrast, the performance of PSOAF* s comparable to the best obtaned wth PSOAF, wth no performance loss due to rotaton. For the sae of clarty, an overvew of the performances of PSOAF, PSOAF and PSOAF* s gven n Table.. The table summarzes the best functon values obtaned, together wth the nerta factor at whch the best functon value s obtaned after 5 functon evaluatons ( teratons). The results for both the unrotated and rotated test functons are gven.

15 Unversty of Pretora etd Wle, D N (5) CHAPTER. OBJECTIVITY OF THE PSOA Table.: Constant nerta factor at whch the best average objectve functon value s obtaned for the unrotated test functons. The accompanyng average objectve functon value for rotated test functons s also presented. PSOAF w f unrotated f best avg f rotated f best avg f f f f f w PSOAF f unrotated f best avg f rotated f best avg f f f f f.6.5. w PSOAF* (α = ) f unrotated f best avg f rotated f best avg f f.5..8 f f f.6..

16 Unversty of Pretora etd Wle, D N (5) CHAPTER. OBJECTIVITY OF THE PSOA Average objectve functon value 5 5 PSOAF unrotated PSOAF rotated PSOAF unrotated PSOAF rotated PSOAF* (α=) unrotated PSOAF* (α=) rotated Number of teratons Fgure.: Mean functon value hstory plot averaged over runs on the rotated and unrotated Quadrc test functon f wth PSOAF (wth w =.8), PSOAF (wth w =.) and PSOAF* (wth w =.5 and α = )..8 Comments on PSOAF*.8. On nvarance It s now shown that PSOAF* s not strctly rotatonally nvarant, but only n a stochastc sense. Consder an arbtrary vector, expressed n two dfferent reference frames, by respectvely y and y. The two reference frames are related by a pure rotaton M, hence where Orth + ndcates the space of proper orthogonal matrces. y = My, M Orth +, (.8) Now apply two ndependent drectonal perturbatons (rotatons) Q Orth + and Q Orth + to y and y respectvely. The vectors ŷ and ŷ then obtaned, are respectvely gven by and ŷ = Qy, (.9) ŷ = Q y. (.) Strct determnstc rotatonal nvarance requres a one-to-one mappng of the perturbed vectors n ether reference frame. Hence ŷ = Mŷ, M Orth + (.) By substtutng Eqs. (.8), (.9) and (.) nto Eq. (.), the followng s obtaned Eq. (.) s rewrtten as Q My = MQy, M Orth +. (.) (Q M MQ)y =, M Orth +. (.)

17 Unversty of Pretora etd Wle, D N (5) CHAPTER. OBJECTIVITY OF THE PSOA Snce Eq. (.) has to hold for any arbtrary vector y, t follows that Q M = MQ, M Orth +. (.) The unque soluton to Eq. (.) s that both Q and Q are the second-order sotropc tensor,.e. Q = Q = I. (.5) The foregong mples that a strct enforcement of rotatonal nvarance results n Q l = I, l =,. In other words, PSOAF* reduces to PSOAF. However, snce the PSOA s a stochastc algorthm, t s adequate to satsfy Eq. (.5) n an average sense only. In order to satsfy Q = Q = I n a stochastc sense, t s suffcent to requre that mean(q ) = mean(q) = I, f the probablty dstrbutons of Q and Q are chosen equal over dentcal domans..8. Implementatonal ssues of PSOAF* Further to the mplementaton n Secton.5, numerous strateges exst to acheve ndependent drectonal perturbaton. An obvous, computatonally nexpensve possblty s to randomly perturb each component of the unt vectors (p x )/ (p x ) and (pg x )/ (pg x ) ; the vectors (p x ) and (p g x ) are then reconstructed from the normalzaton of the perturbed vectors. (Although ths maes a rgorous mathematcal analyss of the algorthm dffcult.) In the mplementaton, n updatng Q, strateges to lmt the computatonal expense assocated wth matrx multplcatons and the generaton of random numbers may also be mplemented. For example, multplyng the sum of c (p x ) and c (p g x ) by a sngle random rotaton matrx, reduces the number of matrx multplcatons by half. However, depcted n Fgure.5 s the dfference n nstantaneous search doman that results when ndependent rotaton matrces Q Q are used, as opposed to dentcal rotaton matrces Q = Q. To reduce the computatonal effort even further, the vectors c (p x ) and c (p g x ) of all the partcles can be drectonally perturbed by the same ndependent rotaton matrces, vz. Q l = Q l, l =, and =,,, p..8. Alternatves to PSOAF* Fnally, there are of course numerous methods to ntroduce dversty nto PSOAF, as opposed to the proposed opton of ndependent drectonal perturbaton. Only a sngle alternatve s mentoned here, namely an ncrease n the socal awareness of the partcles. In turn, ths may for example be effected by ncreasng the number of partcles that contrbute to Eq. (.) [5, ]. (One may of course acheve n-dmensonal searches, f the number of partcles p n, unless the partcle trajectores are parallel.) Addtonal nformaton about the objectve functon s then also used n the searches of any partcle.

18 Unversty of Pretora etd Wle, D N (5) ().5 + ν ν + () CHAPTER. OBJECTIVITY OF THE PSOA ( p x ) g ( p x).5 ( p x) g ( p x).5.5 ν+() ν+() Fgure.5: Scatter plot of possble stochastc vectors ν, generated usng Monte Carlo smulatons, wth (p x ) = [ ] and (p x ) = [ ] usng a) dentcal rotaton matrces Q = Q and b) ndependent rotaton matrces Q 6= Q..9 Closure It s shown that PSOAF s objectve, but t demonstrates an overall poor performance, due to the partcle trajectores collapsng to lnes. Ths s a drect result of only scalng the magntude of the cogntve and socal vectors c (p x ) and c (pg x ). In turn, PSOAF s not objectve, whch results n severe performance loss for rotated functons. Nevertheless, PSOAF stll outperforms PSOAF for both rotated and unrotated test functons, snce the algorthm s dverse,.e. the partcle trajectores do not collapse to lnes. A novel mplementaton denoted PSOAF* s proposed, whch s both objectve and dverse. In PSOAF*, the magntudes are scaled, and the drectons perturbed ndependently, of both the cogntve and socal vectors c (p x ) and c (pg x ). (Ths however comes at the cost of an addtonal scalng factor.) PSOAF* outperforms PSOAF for the unmodal functons used, for both the rotated and unrotated test functons. In addton, ts performance s comparable to PSOAF for the multmodal functons, wth the added advantage of beng ndependent of the reference frame n whch the objectve functon s formulated.

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

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

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

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

CSci 6974 and ECSE 6966 Math. Tech. for Vision, Graphics and Robotics Lecture 21, April 17, 2006 Estimating A Plane Homography

CSci 6974 and ECSE 6966 Math. Tech. for Vision, Graphics and Robotics Lecture 21, April 17, 2006 Estimating A Plane Homography CSc 6974 and ECSE 6966 Math. Tech. for Vson, Graphcs and Robotcs Lecture 21, Aprl 17, 2006 Estmatng A Plane Homography Overvew We contnue wth a dscusson of the major ssues, usng estmaton of plane projectve

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

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

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

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

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

ELASTIC WAVE PROPAGATION IN A CONTINUOUS MEDIUM

ELASTIC WAVE PROPAGATION IN A CONTINUOUS MEDIUM ELASTIC WAVE PROPAGATION IN A CONTINUOUS MEDIUM An elastc wave s a deformaton of the body that travels throughout the body n all drectons. We can examne the deformaton over a perod of tme by fxng our look

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

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

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

CHAPTER 14 GENERAL PERTURBATION THEORY

CHAPTER 14 GENERAL PERTURBATION THEORY CHAPTER 4 GENERAL PERTURBATION THEORY 4 Introducton A partcle n orbt around a pont mass or a sphercally symmetrc mass dstrbuton s movng n a gravtatonal potental of the form GM / r In ths potental t moves

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

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

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

LINEAR REGRESSION ANALYSIS. MODULE IX Lecture Multicollinearity

LINEAR REGRESSION ANALYSIS. MODULE IX Lecture Multicollinearity LINEAR REGRESSION ANALYSIS MODULE IX Lecture - 30 Multcollnearty Dr. Shalabh Department of Mathematcs and Statstcs Indan Insttute of Technology Kanpur 2 Remedes for multcollnearty Varous technques have

More information

FREQUENCY DISTRIBUTIONS Page 1 of The idea of a frequency distribution for sets of observations will be introduced,

FREQUENCY DISTRIBUTIONS Page 1 of The idea of a frequency distribution for sets of observations will be introduced, FREQUENCY DISTRIBUTIONS Page 1 of 6 I. Introducton 1. The dea of a frequency dstrbuton for sets of observatons wll be ntroduced, together wth some of the mechancs for constructng dstrbutons of data. Then

More information

Rigid body simulation

Rigid body simulation Rgd bod smulaton Rgd bod smulaton Once we consder an object wth spacal etent, partcle sstem smulaton s no longer suffcent Problems Problems Unconstraned sstem rotatonal moton torques and angular momentum

More information

Econ107 Applied Econometrics Topic 3: Classical Model (Studenmund, Chapter 4)

Econ107 Applied Econometrics Topic 3: Classical Model (Studenmund, Chapter 4) I. Classcal Assumptons Econ7 Appled Econometrcs Topc 3: Classcal Model (Studenmund, Chapter 4) We have defned OLS and studed some algebrac propertes of OLS. In ths topc we wll study statstcal propertes

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

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 particle in a state of uniform motion remain in that state of motion unless acted upon by external force.

A particle in a state of uniform motion remain in that state of motion unless acted upon by external force. The fundamental prncples of classcal mechancs were lad down by Galleo and Newton n the 16th and 17th centures. In 1686, Newton wrote the Prncpa where he gave us three laws of moton, one law of gravty,

More information

Spin-rotation coupling of the angularly accelerated rigid body

Spin-rotation coupling of the angularly accelerated rigid body Spn-rotaton couplng of the angularly accelerated rgd body Loua Hassan Elzen Basher Khartoum, Sudan. Postal code:11123 E-mal: louaelzen@gmal.com November 1, 2017 All Rghts Reserved. Abstract Ths paper s

More information

3.1 Expectation of Functions of Several Random Variables. )' be a k-dimensional discrete or continuous random vector, with joint PMF p (, E X E X1 E X

3.1 Expectation of Functions of Several Random Variables. )' be a k-dimensional discrete or continuous random vector, with joint PMF p (, E X E X1 E X Statstcs 1: Probablty Theory II 37 3 EPECTATION OF SEVERAL RANDOM VARIABLES As n Probablty Theory I, the nterest n most stuatons les not on the actual dstrbuton of a random vector, but rather on a number

More information

MMA and GCMMA two methods for nonlinear optimization

MMA and GCMMA two methods for nonlinear optimization MMA and GCMMA two methods for nonlnear optmzaton Krster Svanberg Optmzaton and Systems Theory, KTH, Stockholm, Sweden. krlle@math.kth.se Ths note descrbes the algorthms used n the author s 2007 mplementatons

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

Composite Hypotheses testing

Composite Hypotheses testing Composte ypotheses testng In many hypothess testng problems there are many possble dstrbutons that can occur under each of the hypotheses. The output of the source s a set of parameters (ponts n a parameter

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

Simulated Power of the Discrete Cramér-von Mises Goodness-of-Fit Tests

Simulated Power of the Discrete Cramér-von Mises Goodness-of-Fit Tests Smulated of the Cramér-von Mses Goodness-of-Ft Tests Steele, M., Chaselng, J. and 3 Hurst, C. School of Mathematcal and Physcal Scences, James Cook Unversty, Australan School of Envronmental Studes, Grffth

More information

12. The Hamilton-Jacobi Equation Michael Fowler

12. The Hamilton-Jacobi Equation Michael Fowler 1. The Hamlton-Jacob Equaton Mchael Fowler Back to Confguraton Space We ve establshed that the acton, regarded as a functon of ts coordnate endponts and tme, satsfes ( ) ( ) S q, t / t+ H qpt,, = 0, and

More information

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

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

Uncertainty and auto-correlation in. Measurement

Uncertainty and auto-correlation in. Measurement Uncertanty and auto-correlaton n arxv:1707.03276v2 [physcs.data-an] 30 Dec 2017 Measurement Markus Schebl Federal Offce of Metrology and Surveyng (BEV), 1160 Venna, Austra E-mal: markus.schebl@bev.gv.at

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

U.C. Berkeley CS294: Spectral Methods and Expanders Handout 8 Luca Trevisan February 17, 2016

U.C. Berkeley CS294: Spectral Methods and Expanders Handout 8 Luca Trevisan February 17, 2016 U.C. Berkeley CS94: Spectral Methods and Expanders Handout 8 Luca Trevsan February 7, 06 Lecture 8: Spectral Algorthms Wrap-up In whch we talk about even more generalzatons of Cheeger s nequaltes, and

More information

1 Matrix representations of canonical matrices

1 Matrix representations of canonical matrices 1 Matrx representatons of canoncal matrces 2-d rotaton around the orgn: ( ) cos θ sn θ R 0 = sn θ cos θ 3-d rotaton around the x-axs: R x = 1 0 0 0 cos θ sn θ 0 sn θ cos θ 3-d rotaton around the y-axs:

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

U.C. Berkeley CS294: Beyond Worst-Case Analysis Luca Trevisan September 5, 2017

U.C. Berkeley CS294: Beyond Worst-Case Analysis Luca Trevisan September 5, 2017 U.C. Berkeley CS94: Beyond Worst-Case Analyss Handout 4s Luca Trevsan September 5, 07 Summary of Lecture 4 In whch we ntroduce semdefnte programmng and apply t to Max Cut. Semdefnte Programmng Recall that

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

Singular Value Decomposition: Theory and Applications

Singular Value Decomposition: Theory and Applications Sngular Value Decomposton: Theory and Applcatons Danel Khashab Sprng 2015 Last Update: March 2, 2015 1 Introducton A = UDV where columns of U and V are orthonormal and matrx D s dagonal wth postve real

More information

2 More examples with details

2 More examples with details Physcs 129b Lecture 3 Caltech, 01/15/19 2 More examples wth detals 2.3 The permutaton group n = 4 S 4 contans 4! = 24 elements. One s the dentty e. Sx of them are exchange of two objects (, j) ( to j and

More information

Week 9 Chapter 10 Section 1-5

Week 9 Chapter 10 Section 1-5 Week 9 Chapter 10 Secton 1-5 Rotaton Rgd Object A rgd object s one that s nondeformable The relatve locatons of all partcles makng up the object reman constant All real objects are deformable to some extent,

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

Physics 181. Particle Systems

Physics 181. Particle Systems Physcs 181 Partcle Systems Overvew In these notes we dscuss the varables approprate to the descrpton of systems of partcles, ther defntons, ther relatons, and ther conservatons laws. We consder a system

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

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

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

Approximate Smallest Enclosing Balls

Approximate Smallest Enclosing Balls Chapter 5 Approxmate Smallest Enclosng Balls 5. Boundng Volumes A boundng volume for a set S R d s a superset of S wth a smple shape, for example a box, a ball, or an ellpsod. Fgure 5.: Boundng boxes Q(P

More information

Chapter 8 Indicator Variables

Chapter 8 Indicator Variables Chapter 8 Indcator Varables In general, e explanatory varables n any regresson analyss are assumed to be quanttatve n nature. For example, e varables lke temperature, dstance, age etc. are quanttatve n

More information

Salmon: Lectures on partial differential equations. Consider the general linear, second-order PDE in the form. ,x 2

Salmon: Lectures on partial differential equations. Consider the general linear, second-order PDE in the form. ,x 2 Salmon: Lectures on partal dfferental equatons 5. Classfcaton of second-order equatons There are general methods for classfyng hgher-order partal dfferental equatons. One s very general (applyng even to

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

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

Iterative General Dynamic Model for Serial-Link Manipulators

Iterative General Dynamic Model for Serial-Link Manipulators EEL6667: Knematcs, Dynamcs and Control of Robot Manpulators 1. Introducton Iteratve General Dynamc Model for Seral-Lnk Manpulators In ths set of notes, we are gong to develop a method for computng a general

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

Physics 5153 Classical Mechanics. Principle of Virtual Work-1

Physics 5153 Classical Mechanics. Principle of Virtual Work-1 P. Guterrez 1 Introducton Physcs 5153 Classcal Mechancs Prncple of Vrtual Work The frst varatonal prncple we encounter n mechancs s the prncple of vrtual work. It establshes the equlbrum condton of a mechancal

More information

The Feynman path integral

The Feynman path integral The Feynman path ntegral Aprl 3, 205 Hesenberg and Schrödnger pctures The Schrödnger wave functon places the tme dependence of a physcal system n the state, ψ, t, where the state s a vector n Hlbert space

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

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

VQ widely used in coding speech, image, and video

VQ widely used in coding speech, image, and video at Scalar quantzers are specal cases of vector quantzers (VQ): they are constraned to look at one sample at a tme (memoryless) VQ does not have such constrant better RD perfomance expected Source codng

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

Some Comments on Accelerating Convergence of Iterative Sequences Using Direct Inversion of the Iterative Subspace (DIIS)

Some Comments on Accelerating Convergence of Iterative Sequences Using Direct Inversion of the Iterative Subspace (DIIS) Some Comments on Acceleratng Convergence of Iteratve Sequences Usng Drect Inverson of the Iteratve Subspace (DIIS) C. Davd Sherrll School of Chemstry and Bochemstry Georga Insttute of Technology May 1998

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

Canonical transformations

Canonical transformations Canoncal transformatons November 23, 2014 Recall that we have defned a symplectc transformaton to be any lnear transformaton M A B leavng the symplectc form nvarant, Ω AB M A CM B DΩ CD Coordnate transformatons,

More information

Transfer Functions. Convenient representation of a linear, dynamic model. A transfer function (TF) relates one input and one output: ( ) system

Transfer Functions. Convenient representation of a linear, dynamic model. A transfer function (TF) relates one input and one output: ( ) system Transfer Functons Convenent representaton of a lnear, dynamc model. A transfer functon (TF) relates one nput and one output: x t X s y t system Y s The followng termnology s used: x y nput output forcng

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

STAT 309: MATHEMATICAL COMPUTATIONS I FALL 2018 LECTURE 16

STAT 309: MATHEMATICAL COMPUTATIONS I FALL 2018 LECTURE 16 STAT 39: MATHEMATICAL COMPUTATIONS I FALL 218 LECTURE 16 1 why teratve methods f we have a lnear system Ax = b where A s very, very large but s ether sparse or structured (eg, banded, Toepltz, banded plus

More information

THEOREMS OF QUANTUM MECHANICS

THEOREMS OF QUANTUM MECHANICS THEOREMS OF QUANTUM MECHANICS In order to develop methods to treat many-electron systems (atoms & molecules), many of the theorems of quantum mechancs are useful. Useful Notaton The matrx element A mn

More information

arxiv: v1 [math.ho] 18 May 2008

arxiv: v1 [math.ho] 18 May 2008 Recurrence Formulas for Fbonacc Sums Adlson J. V. Brandão, João L. Martns 2 arxv:0805.2707v [math.ho] 8 May 2008 Abstract. In ths artcle we present a new recurrence formula for a fnte sum nvolvng the Fbonacc

More information

Markov Chain Monte Carlo (MCMC), Gibbs Sampling, Metropolis Algorithms, and Simulated Annealing Bioinformatics Course Supplement

Markov Chain Monte Carlo (MCMC), Gibbs Sampling, Metropolis Algorithms, and Simulated Annealing Bioinformatics Course Supplement Markov Chan Monte Carlo MCMC, Gbbs Samplng, Metropols Algorthms, and Smulated Annealng 2001 Bonformatcs Course Supplement SNU Bontellgence Lab http://bsnuackr/ Outlne! Markov Chan Monte Carlo MCMC! Metropols-Hastngs

More information

COMPLEX NUMBERS AND QUADRATIC EQUATIONS

COMPLEX NUMBERS AND QUADRATIC EQUATIONS COMPLEX NUMBERS AND QUADRATIC EQUATIONS INTRODUCTION We know that x 0 for all x R e the square of a real number (whether postve, negatve or ero) s non-negatve Hence the equatons x, x, x + 7 0 etc are not

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

χ x B E (c) Figure 2.1.1: (a) a material particle in a body, (b) a place in space, (c) a configuration of the body

χ x B E (c) Figure 2.1.1: (a) a material particle in a body, (b) a place in space, (c) a configuration of the body Secton.. Moton.. The Materal Body and Moton hyscal materals n the real world are modeled usng an abstract mathematcal entty called a body. Ths body conssts of an nfnte number of materal partcles. Shown

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

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

In this section is given an overview of the common elasticity models.

In this section is given an overview of the common elasticity models. Secton 4.1 4.1 Elastc Solds In ths secton s gven an overvew of the common elastcty models. 4.1.1 The Lnear Elastc Sold The classcal Lnear Elastc model, or Hooean model, has the followng lnear relatonshp

More information

The Multiple Classical Linear Regression Model (CLRM): Specification and Assumptions. 1. Introduction

The Multiple Classical Linear Regression Model (CLRM): Specification and Assumptions. 1. Introduction ECONOMICS 5* -- NOTE (Summary) ECON 5* -- NOTE The Multple Classcal Lnear Regresson Model (CLRM): Specfcaton and Assumptons. Introducton CLRM stands for the Classcal Lnear Regresson Model. The CLRM s also

More information

Tensor Analysis. For orthogonal curvilinear coordinates, ˆ ˆ (98) Expanding the derivative, we have, ˆ. h q. . h q h q

Tensor Analysis. For orthogonal curvilinear coordinates, ˆ ˆ (98) Expanding the derivative, we have, ˆ. h q. . h q h q For orthogonal curvlnear coordnates, eˆ grad a a= ( aˆ ˆ e). h q (98) Expandng the dervatve, we have, eˆ aˆ ˆ e a= ˆ ˆ a h e + q q 1 aˆ ˆ ˆ a e = ee ˆˆ ˆ + e. h q h q Now expandng eˆ / q (some of the detals

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

Lecture 5.8 Flux Vector Splitting

Lecture 5.8 Flux Vector Splitting Lecture 5.8 Flux Vector Splttng 1 Flux Vector Splttng The vector E n (5.7.) can be rewrtten as E = AU (5.8.1) (wth A as gven n (5.7.4) or (5.7.6) ) whenever, the equaton of state s of the separable form

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

The Geometry of Logit and Probit

The Geometry of Logit and Probit The Geometry of Logt and Probt Ths short note s meant as a supplement to Chapters and 3 of Spatal Models of Parlamentary Votng and the notaton and reference to fgures n the text below s to those two chapters.

More information

Relaxation Methods for Iterative Solution to Linear Systems of Equations

Relaxation Methods for Iterative Solution to Linear Systems of Equations Relaxaton Methods for Iteratve Soluton to Lnear Systems of Equatons Gerald Recktenwald Portland State Unversty Mechancal Engneerng Department gerry@pdx.edu Overvew Techncal topcs Basc Concepts Statonary

More information

A PROCEDURE FOR SIMULATING THE NONLINEAR CONDUCTION HEAT TRANSFER IN A BODY WITH TEMPERATURE DEPENDENT THERMAL CONDUCTIVITY.

A PROCEDURE FOR SIMULATING THE NONLINEAR CONDUCTION HEAT TRANSFER IN A BODY WITH TEMPERATURE DEPENDENT THERMAL CONDUCTIVITY. Proceedngs of the th Brazlan Congress of Thermal Scences and Engneerng -- ENCIT 006 Braz. Soc. of Mechancal Scences and Engneerng -- ABCM, Curtba, Brazl,- Dec. 5-8, 006 A PROCEDURE FOR SIMULATING THE NONLINEAR

More information

SCALARS AND VECTORS All physical quantities in engineering mechanics are measured using either scalars or vectors.

SCALARS AND VECTORS All physical quantities in engineering mechanics are measured using either scalars or vectors. SCALARS AND ECTORS All phscal uanttes n engneerng mechancs are measured usng ether scalars or vectors. Scalar. A scalar s an postve or negatve phscal uantt that can be completel specfed b ts magntude.

More information

Foundations of Arithmetic

Foundations of Arithmetic Foundatons of Arthmetc Notaton We shall denote the sum and product of numbers n the usual notaton as a 2 + a 2 + a 3 + + a = a, a 1 a 2 a 3 a = a The notaton a b means a dvdes b,.e. ac = b where c s an

More information

CME 302: NUMERICAL LINEAR ALGEBRA FALL 2005/06 LECTURE 13

CME 302: NUMERICAL LINEAR ALGEBRA FALL 2005/06 LECTURE 13 CME 30: NUMERICAL LINEAR ALGEBRA FALL 005/06 LECTURE 13 GENE H GOLUB 1 Iteratve Methods Very large problems (naturally sparse, from applcatons): teratve methods Structured matrces (even sometmes dense,

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

Structure and Drive Paul A. Jensen Copyright July 20, 2003

Structure and Drive Paul A. Jensen Copyright July 20, 2003 Structure and Drve Paul A. Jensen Copyrght July 20, 2003 A system s made up of several operatons wth flow passng between them. The structure of the system descrbes the flow paths from nputs to outputs.

More information

8.592J: Solutions for Assignment 7 Spring 2005

8.592J: Solutions for Assignment 7 Spring 2005 8.59J: Solutons for Assgnment 7 Sprng 5 Problem 1 (a) A flament of length l can be created by addton of a monomer to one of length l 1 (at rate a) or removal of a monomer from a flament of length l + 1

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

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

Physics 53. Rotational Motion 3. Sir, I have found you an argument, but I am not obliged to find you an understanding.

Physics 53. Rotational Motion 3. Sir, I have found you an argument, but I am not obliged to find you an understanding. Physcs 53 Rotatonal Moton 3 Sr, I have found you an argument, but I am not oblged to fnd you an understandng. Samuel Johnson Angular momentum Wth respect to rotatonal moton of a body, moment of nerta plays

More information

ECE559VV Project Report

ECE559VV Project Report ECE559VV Project Report (Supplementary Notes Loc Xuan Bu I. MAX SUM-RATE SCHEDULING: THE UPLINK CASE We have seen (n the presentaton that, for downlnk (broadcast channels, the strategy maxmzng the sum-rate

More information

Inductance Calculation for Conductors of Arbitrary Shape

Inductance Calculation for Conductors of Arbitrary Shape CRYO/02/028 Aprl 5, 2002 Inductance Calculaton for Conductors of Arbtrary Shape L. Bottura Dstrbuton: Internal Summary In ths note we descrbe a method for the numercal calculaton of nductances among conductors

More information

Advanced Quantum Mechanics

Advanced Quantum Mechanics Advanced Quantum Mechancs Rajdeep Sensarma! sensarma@theory.tfr.res.n ecture #9 QM of Relatvstc Partcles Recap of ast Class Scalar Felds and orentz nvarant actons Complex Scalar Feld and Charge conjugaton

More information

Analytical Gradient Evaluation of Cost Functions in. General Field Solvers: A Novel Approach for. Optimization of Microwave Structures

Analytical Gradient Evaluation of Cost Functions in. General Field Solvers: A Novel Approach for. Optimization of Microwave Structures IMS 2 Workshop Analytcal Gradent Evaluaton of Cost Functons n General Feld Solvers: A Novel Approach for Optmzaton of Mcrowave Structures P. Harscher, S. Amar* and R. Vahldeck and J. Bornemann* Swss Federal

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