ETNA Kent State University
|
|
- Gladys Hart
- 5 years ago
- Views:
Transcription
1 Elecronic Tranacion on Numerical Analyi. Volume 37, pp. 7-86,. Copyrigh, Ken Sae Univeriy. ISSN ETNA Ken Sae Univeriy hp://ena.mah.ken.edu THE ANALYTIC SVD: ON THE NON-GENERIC POINTS ON THE PATH DÁŠA JANOVSKÁ AND VLADIMíR JANOVSKÝ Abrac. A new echnique for compuing he Analyic SVD i propoed. The idea i o follow a branch of ju one imple ingular value and he correponding lef/righ ingular vecor. Numerical compuaion may collape a non-generic poin; we will conider he cae when he coninuaion ge uck due o a nonzero muliple ingular value. We inerpre uch a poin a a ingulariy of he branch. We employ ingulariy heory in order o decribe and claify hi poin. Since i codimenion i one, we mee uch a poin rarely. Key word. SVD, ASVD, coninuaion, ingulariy heory. AMS ubjec claificaion. 65F5. Inroducion. A ingular value decompoiion SVD of a real marix A R m n, m n, i a facorizaion A = UΣV T, where U R m m and V R n n are orhogonal marice and Σ = diag,..., n R m n. The value i, i =,...,n, are called ingular value. They may be defined o be nonnegaive and o be arranged in nonincreaing order. Le A depend moohly on a parameer R, [a, b]. The aim i o conruc a pah of SVD. A = UΣV T, where U, Σ and V depend moohly on [a, b]. If A i a real analyic marix funcion on [a, b], hen here exi an Analyic Singular Value Decompoiion ASVD [], a facorizaion. ha inerpolae he claical SVD defined a = a, i.e. he facor U, V and Σ are real analyic on [a, b], for each [a, b], boh U R m m and V R n n are orhogonal marice, and Σ = diag,..., n R m n i a diagonal marix, a = a, he marice Ua, Σa and V a are he facor of he claical SVD of he marix Aa. The diagonal enrie i R of Σ are called ingular value. Due o he requiremen of moohne, ingular value may be negaive, and heir ordering may by arbirary. Under cerain aumpion, he ASVD may be uniquely deermined by he facor a = a. For heoreical background, ee [9]. A far a he compuaion i concerned, an incremenal echnique i propoed in []. Given a poin on he pah, one compue a claical SVD for a neighboring parameer value. Nex, one compue permuaion marice which link he claical SVD o he nex poin on he pah. The procedure i approximaive wih a local error of order Oh, where h i he ep ize. An alernaive echnique for compuing he ASVD i preened in [3]. A non-auonomou vecor field H : R R N R N of large dimenion N = n + n + m can be conruced in uch a way ha he oluion of he iniial value problem for he yem x = H, x i linked o he pah of he ASVD. Moreover, [3] conribue o he analyi of non-generic poin of he ASVD pah; ee []. Thee poin could be, in fac, inerpreed a ingulariie of he vecor field R N. In [], boh approache are compared. Received November 8, 6. Acceped November, 9. Publihed online April 7,. Recommended by Bruno Lang. The reearch of boh auhor wa parially uppored by he Gran Agency of he Czech Republic gran No. /6/356. The fir and he econd auhor acknowledge financial uppor by he reearch projec MSM and MSM 6839, repecively, of The Miniry of Educaion, Youh and Spor, Czech Republic. Iniue of Chemical Technology, Prague janovkd@vch.cz. Charle Univeriy, Faculy of Mahemaic and Phyic, Prague janovky@karlin.mff.cuni.cz. 7
2 Ken Sae Univeriy hp://ena.mah.ken.edu ASVD: NON-GENERIC POINTS 7 A coninuaion algorihm for compuing he ASVD i preened in [7]. I follow a pah of a few eleced ingular value and lef/righ ingular vecor. I i appropriae for large pare marice. The coninuaion algorihm i of predicor-correcor ype. The relevan predicor i baed on he Euler mehod, hence on an ODE olver. In hi repec, here i a link o [3]. Neverhele, he Newon-ype correcor guaranee he oluion wih precribed preciion. The coninuaion may ge uck a poin where a nonimple ingular value i appear for a paricular parameer and index i. In [, 3], uch poin are called non-generic poin of he pah. They are relaed o branching of he ingular value pah. The code in [7] incorporae exrapolaion raegie in order o jump over uch poin. In hi paper, we inveigae non-generic poin. In Secion, we give an example ha moivaed our reearch. Then we inroduce a pah-following mehod for coninuaion of a imple ingular value and he correponding lef/righ ingular vecor. In Secion 4, we define and analyze a ingulariy on he pah. Nex, we perurb hi ingulariy; ee Secion 5. We ummarize our concluion in Secion 6. Finally, in Appendix A, we provide deail of he expanion ued in our aympoic analyi.. Moivaion. Le A = ; + ee [, Example ]. The relevan ASVD, A = UΣV T,.5.5, can be compued explicily: U = V =, =, = +. Obviouly, = = i a nonimple muliple ingular value of A. We will ak he following queion: doe he ASVD-pah peri for an arbirary ufficienly mall perurbaion? Le. A = [ + ε + / /4 Conider he relevan ASVD. Thi ime, we compue i numerically uing he echnique decribed in [7]. We how he reul for he unperurbed and perurbed marice in Figure. and., repecively. Noice ha he branche in Figure. and in Figure. are qualiaively differen. We oberve a eniive dependence on he iniial condiion of he branche. 3. Coninuaion of a imple ingular value. 3.. Preliminarie. Le u recall he noion of a ingular value of a marix A R m n, m n. DEFINITION 3.. Le R. We ay ha i a ingular value of he marix A if here exi u R m and v R n uch ha 3. Av u =, A T u v =, u = v =. The vecor v and u are called he righ and he lef ingular vecor of he marix A. Noe ha i defined up o i ign: if he riple, u, v aifie 3. hen a lea hree more riple, u, v,, u, v,, u, v, ].
3 Ken Sae Univeriy hp://ena.mah.ken.edu 7 D. JANOVSKÁ AND V. JANOVSKÝ v v FIGURE.. Perurbaion ε =. Lef: Branche of ingular value and in red and blue a funcion of. Righ: The relevan righ ingular vecor in red and blue v v FIGURE.. Perurbaion ε =.. Lef: Branche of ingular value and in red and blue a funcion of. Righ: The relevan righ ingular vecor in red and blue. can be inerpreed a ingular value and lef and righ ingular vecor of A. DEFINITION 3.. Le R. We ay ha i a imple ingular value of a marix A if here exi u R m, u, and v R n, v, where, u, v,, u, v,, u, v,, u, v are he only oluion o 3.. A ingular value which i no imple i called a nonimple muliple ingular value. REMARK 3.3. Le. Then i a imple ingular value of A if and only if i a imple eigenvalue of A T A. In paricular, v R n and u R m uch ha A T Av = v, v =, u = Av are he relevan righ and lef ingular vecor of A. REMARK 3.4. = i a imple ingular value of A if and only if m = n and dimkera =. REMARK 3.5. Le i, j be imple ingular value of A wih i j. Then i j and i j. Le u recall he idea of [7]. The branche of eleced ingular value i and he correponding lef/righ ingular vecor U i R m, V i R n are conidered, i.e., AV i = i U i, A T U i = i V i, U i T U i = V i T V i =, for [a, b]. The naural orhogonaliy condiion U i T U j = V i T V j =, i j, [a, b], are added. For p n, he eleced ingular value S =,..., p R p and he correponding lef/righ ingular vecor U = [U,..., U p ] R m p and V = [V,..., V p ] R n p are followed for [a, b].
4 Ken Sae Univeriy hp://ena.mah.ken.edu ASVD: NON-GENERIC POINTS The pah of imple ingular value. In hi ecion, we conider he idea of pahfollowing for one ingular value and he correponding lef/righ ingular vecor. We expec he pah o be locally a branch of i, U i R m, V i R n, aifying condiion 3. and 3.3 for [a, b]. We conider he ih branch, i n, namely, he branch which i iniialized by i a, U i a R m, V i a R n, compued by he claical SVD [4]. Noe ha he SVD algorihm order all ingular value in decending order a i a n a. We aume ha i a i imple; ee Remark 3.3 and 3.4. DEFINITION 3.6. For a given [a, b] and R, le u e Im A 3.4 M, A T, I n where I m R m m and I n R n n are ideniie. REMARK 3.7. Le, [a, b].. i a ingular value of A if and only if dim Ker M,.. i a imple ingular value of A if and only if dim Ker M, =. REMARK 3.8. Le [, [a, b]. u. If M, = hen u v] T u = v T v. [ũ ]. If in addiion M, = hen u ṽ T ũ = v T ṽ. Noe ha if i hen due o Remark 3.8 one of he caling condiion 3.3 i redundan. Thi moivae he following definiion. DEFINITION 3.9. Conider he mapping f : R R +m+n R +m+n, R, x =, u, v R R m R n f, x R +m+n, where u + Av 3.5 f, x A T u v. v T v A an alernaive o 3.5 we will alo ue u + Av 3.6 f, x A T u v. u T u + v T v The equaion 3.7 f, x =, x =, u, v, may locally define a branch x =, u, v R +m+n of ingular value and lef/righ ingular vecor u and v. The branch i iniialized a, which play he role of = a. I i aumed ha here exi x R +m+n uch ha f, x =. The iniial condiion x =, u, v R +m+n play he role of already compued SVD facor i a R, U i a R m and V i a R n. We olve 3.7 on an open inerval J of parameer uch ha J. THEOREM 3.. Conider 3.5. Le, x J R +m+n, x =, u, v, be a roo of f, x =. Aume ha i a imple ingular value of A. Then
5 Ken Sae Univeriy hp://ena.mah.ken.edu 74 D. JANOVSKÁ AND V. JANOVSKÝ here exi an open ubinerval I J conaining and a unique funcion I x R +m+n uch ha f, x = for all I and ha x = x. Moreover, if A C k I, R m n, k, hen x C k I, R +m+n. If A C ω I, R m n hen x C ω I, R +m+n. Proof. We will how ha he aumpion imply ha he parial differenial f x a he poin, x i a regular + m + n + m + n marix. Le δx = δ, δu, δv R +m+n, 3.8 f x, x δx = u I m A v A T I n δ δu = R +m+n. T m v T δv Thi i equivalen o he yem 3.9 M, δu u = δ δv v, v T δv =. Projecing he fir equaion on u, v, and uing he ymmery of he marix M,, yield u T T 3. v M, δu δu = M δv δv, u v = δ u + v. u By definiion 3.5, M, v = R m+n. Therefore, δ =. δu u Due o Remark 3.7, here exi a conan c uch ha = c δv v. The econd condiion in 3.9 implie ha c =. Hence δx = δ, δu, δv =, which prove he claim. Auming ha A C k I, R m n, k, he aemen i a conequence of he Implici Funcion Theorem; ee, e.g., [6]. In cae ha A C ω I, R m n, i.e., when A i real analyic, he reul again follow from he Implici Funcion Theorem; ee []. REMARK 3.. The above aemen alo hold for he alernaive caling 3.6. The argumen i imilar. The pracical advanage of 3.7 i ha we can ue andard package for coninuaion of an implicily defined curve. In paricular, we ue he MATLAB oolbox MATCONT [3]. Pah-following of he oluion e of 3.7 via MATCONT i very robu. However, one ha o be careful when inerpreing he reul. In paricular, he lower bound for he ep ize, MinSepize, hould be choen ufficienly mall. We will commen on hi obervaion in Secion 6. In order o illurae he performance, we conider he ame problem a in [7], namely, he homoopy 3. A = A + A, [, ], where he marice A well33.mx, A illc33.mx are aken from he Marix Marke []. Noe ha A, A R 33 3 are pare and A, A are well- and illcondiioned, repecively. The aim i o coninue he en malle ingular value and correponding lef/righ ingular vecor of A. The coninuaion i iniialized a =. The iniial decompoiion of A i compued via vd; ee he MATLAB Funcion Reference. The reul of coninuaion are diplayed in Figure 3.. We run MATCONT en ime, once for each ingular value. The branche do no cro. The compuaion complie wih Theorem 3.. Each curve i compued a a equence of iolaed poin marked by circle; ee he zoom on he righ. The adapive epize conrol refine he epize individually for each branch.
6 Ken Sae Univeriy hp://ena.mah.ken.edu ASVD: NON-GENERIC POINTS 75. x FIGURE 3.. Branche of ingular value 3,..., 3 a funcion of. On he righ: The relevan zoom. 4. Singular poin on he pah. Le f, x =, x =, u, v,. Le u dicu he cae when he aumpion of Theorem 3. do no hold; namely, aume ha i a nonimple muliple ingular value of A. Then we conclude ha dimker M, ; ee Remark 3.7. In paricular, we will aume ha dimker M, =. Then, here exi δu R m, δu =, and δv R n, δv =, uch ha δu M, =, v δv T δv =. Noe ha hi implie u T δu =. Compuing Kerf x, x, ee , we conclude ha dimkerf x, = and 4. Kerf x, x = pan δu. δv 4.. Dimenional reducion. Our aim i o analyze he ingular roo, x of he parameer dependen equaion 3.7. A andard echnique i Lyapunov-Schmid reducion [6]. We apply a verion baed on bordered marice [5]. We aume ha dimkerf x, x =, i.e., he corank of he marix f x, x i one. Uing he proper erminology, we deal wih a corank = ingulariy. The algorihm of he reducion i a follow. Le u fix vecor B, C R +m+n. Find ξ R, τ R, x R +m+n and ϕ R uch ha f + τ, x + x + ϕb =, C T x = ξ. We define an operaor relaed o he above equaion: 4.4 Le u aume ha 4.5 de F : R R R +m+n R R +m+n R, f Fτ, ξ, x, ϕ + τ, x + x + ϕb C T. x ξ fx, x B C T.
7 Ken Sae Univeriy hp://ena.mah.ken.edu 76 D. JANOVSKÁ AND V. JANOVSKÝ I can be hown, ee [5], ha hi aumpion i aified for a generic choice of bordering vecor B and C. Neverhele, laer on we pecify B and C. Obviouly, ξ, τ, x, ϕ =,,, R R R +m+n R i a roo of F, i.e., F,,, =. The parial differenial of F wih repec o he variable x, ϕ, namely, he marix F x,ϕ,,, R +m+n +m+n, i regular a he origin,,, by he aumpion 4.5. Due o he Implici Funcion Theorem [6], he oluion manifold of Fξ, τ, x, ϕ = can be locally parameerized by τ and ξ; ha i, here exi funcion 4.6 ϕ = ϕτ, ξ R, x = xτ, ξ R +m+n, uch ha Fτ, ξ, xτ, ξ, ϕτ, ξ = for all τ and ξ being mall. From 4. and he fac ha B due o 4.5, we conclude ha 4.7 f + τ, x + xτ, ξ = if and only if 4.8 ϕτ, ξ =. The calar equaion 4.8 i called he bifurcaion equaion. There i a one-o-one link beween he oluion τ, ξ R of he bifurcaion equaion 4.8 and he oluion, x R R +m+n of he equaion 3.7: 4.9 = + τ, x = x + xτ, ξ. The aemen ha an obviou local meaning: i decribe all roo of 3.7 in a neighborhood of, x. A a rule, he oluion of he bifurcaion equaion can be approximaed only numerically. The uual echnique i. approximae he mapping τ, ξ ϕτ, ξ via i Taylor expanion a he origin,. olve a runcaed bifurcaion equaion, i.e., he equaion wih runcaed higher order erm. The Taylor expanion read 4. ϕτ, ξ = ϕ + ϕ τ τ + ϕ ξ ξ + ϕ ττ τ + ϕ ξτ ξτ + ϕ ξξ ξ + h.o.., where he parial derivaive of ϕ = ϕτ, ξ are underood o be evaluaed a he origin, e.g., ϕ ϕ, or ϕ ξτ ϕ ξτ,. Noe ha ϕ, =. The ymbol h.o.. denoe higher order erm. We alo need o expand 4. xτ, ξ = x+ x τ τ+ x ξ ξ+ x ττ τ + x ξτ ξτ+ x ξξ ξ +h.o... The parial derivaive of x = xτ, ξ are underood o be evaluaed a he origin. Thi expanion i needed o approximae he link 4.9 beween 4.8 and 3.7. Compuing he coefficien of boh expanion 4. and 4. i a rouine procedure; ee [5, Secion 6.]. For example, he coefficien x ξ, ϕ ξ aify a linear yem wih he marix from 4.5, [ fx, x B xξ C T =. ϕ ξ ]
8 Ken Sae Univeriy hp://ena.mah.ken.edu ASVD: NON-GENERIC POINTS 77 The coefficien x ξξ, ϕ ξξ are defined via a linear yem wih he ame marix, fx, x B xξξ C T = ϕ ξξ [ fxx, x x ξ x ξ ec. We conidered he paricular funcion f defined in 3.6. In our compuaion, we pecified B and C a 4. B = δu δv, C = δu. δv Noe ha condiion 4.5 i aified. Moreover, he compuaion are implified. In Appendix A, we li Taylor coefficien of 4. and 4. up o he econd order. The coefficien are compued by uing he pecific daa of he problem:, x, δu, δv, A and higher derivaive of A a. 4.. Bifurcaion analyi. Le u analyze he bifurcaion equaion 4.8 in a neighborhood of he origin. Le u ar wih a heuriic. Due o A., he bifurcaion equaion can be facored a 4.3 ϕτ, ξ = τ ητ, ξ =, ητ, ξ ϕ τ + ϕ ττ τ + ϕ ξτ ξ + h.o.., where h.o.. are of econd order. The oluion τ, ξ of 4.3 are linked o he oluion, x of 3.7 via he ranformaion 4.9, where he incremen xτ, ξ are expanded a ], τ, ξ = τ τ + ττ τ + ξτ ξτ + h.o.., uτ, ξ = u τ τ + δu ξ 8 u ξ + u ττ τ + u ξτ ξτ + h.o.., vτ, ξ = v τ τ + δv ξ 8 v ξ + v ττ τ + v ξτ ξτ + h.o... Noe ha we exploied A., A.6, and A.9. The h.o.. are of hird order. Obviouly, τ = i a rivial oluion of 4.3. In cae ϕ τ, hi i he only oluion locally available. In wha follow, le u conider he cae ϕ τ =. We olve 4.3 on an open inerval J of parameer auming ha J. THEOREM 4.. Le, x J R +m+n, x =, u, v be a roo of f, x =. Le dim Kerf x, x = ; ha i, le δu R m, δu =, δv R n, δv =, be uch ha 4. hold. Aume ha ϕ τ =, ϕ ττ. ϕ τ δut A v + u T A δv, For ϕ ττ, ee A.4. Le A = A be mooh, i.e., A C ω J, R +m+n. Then here exi an open ubinerval I J conaining and a unique funcion I x R +m+n, x C ω I, R +m+n, uch ha f, x = for all I wih x = x. Proof. By virue of he facorizaion 4.3 we have o olve ητ, ξ = for τ and ξ. The aumpion 4.8 make i poible o apply he Implici Funcion Theorem for real analyic funcion [].
9 Ken Sae Univeriy hp://ena.mah.ken.edu 78 D. JANOVSKÁ AND V. JANOVSKÝ In order o inroduce required erminology, le u briefly review Singulariy Theory [5, Chaper 6]. Le, x R R +m+n be a roo of f, i.e., f, x = ; ee 3.6 in hi paricular conex. If hi roo, x aifie he aumpion of Theorem 3., hen we ay ha i i a regular roo. If no, hen, x i aid o be a ingular roo. In oher word,, x i a ingular poin of he mapping f : R R +m+n R +m+n. In Secion 4. we have already menioned he codimenion of a ingular poin. Theorem 4. claifie codim = ingular poin. The codimenion i no he only apec of he claificaion. In Theorem 4. we require he equaliy 4.7 and he inequaliy 4.8. The former condiion i called he defining condiion and he laer one i he nondegeneracy condiion. The number of defining condiion i called he codimenion. Theorem 4. deal wih a ingular poin of codim =. The nex iem of he claificaion li, namely, a ingular poin of codim =, would be defined by he condiion ϕ τ = ϕ ττ = and he nondegeneracy condiion ϕ τττ, ec. Wih ome abue of language, we noe he following: REMARK 4.. Equaion 3.7 define a pah of parameer dependen ingular value and correponding lef/righ ingular vecor. There are ingular poin on hi pah. One of hem could be claified a corank =, codim = ingular poin. Thi paricular poin, x, x =, u, v, i relaed o a nonimple muliple ingular value. PROPOSITION 4.3. Le he aumpion of Theorem 4. be aified. Le, x be a ingular poin of f of corank = and codim =. Le δu R m, δu =, δv R n, δv =, pan he kernel 4.. Then he poin, y R R +m+n, y, δu, δv, i alo a ingular poin of f of corank = and codim =. Moreover, he Taylor expanion a, y can be obained from ha a he poin, x, i.e., from 4.3, More preciely, le he coefficien of he Taylor expanion a, y be marked by ilde, i.e., ϕ ξ, ϕ τ, ϕ ξτ, ϕ ξτ, j eϕ ξ,, and j f j ξ,. Then j 4.9 ϕ τ = ϕ τ, ϕ ξτ = ϕ ξτ, ϕ ττ = ϕ ττ and ϕ ξ = ϕ ξξ = = j ϕ, = =, ξj ξ = ξξ = = j, = = ξj for j =,,.... Proof. The aemen concerning he ingulariy a, y follow from he properie of he kernel of 3.4. The formulae can be readily verified. Hence, he claificaion of, y follow from he aumpion 4.7 and 4.8. A menioned in Secion, he paper [3] conain a deailed analyi of non-generic poin on he pah. I wa noed ha [3] aemp o rack all ingular value, i.e., o conruc he pah of Σ, U and V ; ee.. Neverhele, on he analyical level, one can peak abou pah of ingular value j and k. REMARK 4.4. In [3, Secion 3], he auhor inveigae imple cro over of he pah. He propoed a e of defining condiion for hi phenomenon. Neverhele, he doe no reolve hi yem explicily. I i inuiively clear ha hi imple cro over of he pah hould be he ingular poin reaed in Theorem 4.. In our analyi, we can poin ou compuable conan A. and A.4 o decide abou he cae.
10 Ken Sae Univeriy hp://ena.mah.ken.edu ASVD: NON-GENERIC POINTS Example. We conider [3, Example ]; ee alo [, Example ]. Le A USU, R, where S = diag.5 +,,,, c U = c c c c 3 3, 3 c 3 wih c = co, = in, c = co+, = in+, c 3 = co+, 3 = in +. There are five ingular poin of all pah a =.5,.5,.75,,.5. Thee are manifeed a imple cro over of he pah or imple cro over in erm of [3] relaed o paricular pair of pah; ee Figure 4. on he lef. Noe ha if =.5, hen = =.5. Due o Remark 3.5, eiher or i nonimple acually, boh of hem are nonimple. We are able o coninue eparae branche of ingular value and lef/righ ingular vecor. Le u e =., compue he claical SVD of A = UΣV T and iniialize he coninuaion a 3, U 3, V 3, which are he hird ingular value and he hird column of U and V a =.. We acually coninue a curve in R R In Figure 4. and Figure 4., and v 4 are depiced in green. On hi curve here are ju wo ingular poin, one a =.5 and he oher a =.75. Le u perform an aympoic analyi of he former poin. Uing he noaion of Theorem 4., =.5, =.75, u = [.9689;.474;.;.], v = [.9689;.78;.475;.87], δu = [.475;.5776;.98;.778], δv = [;.949;.98;.453]. The leading Taylor coefficien of ϕτ, ξ and τ, ξ are ϕ ττ =.687, ϕ ξτ =, τ =, ξτ =, ττ = Similarly, we can compue he leading coefficien of uτ, ξ R 4 and vτ, ξ R 4. Neglecing quadraic erm in 4.3, , we ge a local approximaion of he green branch being parameerized by τ. In paricular, we e τ =. :.5 :.5 regularly paced poin on he inerval [.,.5] wih incremen.5 and mark he reuling poin by black diamond; ee he zoom in Figure 4. and Figure 4.. Due o Propoiion 4.3, we can ge an aympoic expanion of he blue branch for free. 5. An unfolding. Le 5. f : R R +m+n R R +m+n, R, x =, u, v R R m R n, ε R f, x, ε R +m+n, u + A + εzv f, x, ε A + εz T u v. u T u + v T v The mapping 5. i an example of an unfolding of he mapping 3.6. I i required ha f, x, via 5. complie wih f, x via 3.6, which i obviou. For a fixed value of ε,
11 Ken Sae Univeriy hp://ena.mah.ken.edu 8 D. JANOVSKÁ AND V. JANOVSKÝ 4 3 "green".8.6 "yellow".4..8 "green".6.4. "yellow" FIGURE 4.. Branche of ingular value,..., 4 in red, blue, green, and yellow a funcion of. Zoomed: The ingular poin, marked a quare, on he green branch. The approximaion via aympoic analyi are marked by diamond..5 "yellow".8.6 "yellow".4 "green".5. v 4 v 4..5 "green" FIGURE 4.. Red, blue, green, and yellow branche of he ninh oluion componen, i.e., v 4 = v 4. Zoomed: The relevan aympoic analyi of he green branch. we conider he equaion 5. f, x, ε =, x =, u, v, for R and x R +m+n. The unfolding may alo model an imperfecion of he original mapping f, x. Le, x, be a ingular poin of he above f. Le hi poin aify he aumpion of Theorem 4.. Our aim i o udy oluion of 5. for a mall fixed ε. 5.. Dimenional reducion reviied. We adap he dimenional reducion from Secion 4. o he unfolding 5.. Le u fix vecor B R +m+n, C R +m+n. Find ξ R, τ R, ε R, x R +m+n and ϕ R uch ha f + τ, x + x, ε + ϕb =, C T x = ξ. Under he aumpion 4.5, he oluion ϕ and x of 5.3 and 5.4 can be locally parameerized by τ, ξ, and ε, i.e., 5.5 ϕ = ϕτ, ξ, ε R, x = xτ, ξ, ε R +m+n. Uing he ame argumen a in Secion 4., we conclude ha 5.6 f + τ, x + xτ, ξ, ε, ε = if and only if 5.7 ϕτ, ξ, ε =
12 Ken Sae Univeriy hp://ena.mah.ken.edu ASVD: NON-GENERIC POINTS 8 for mall τ, ξ and ε. The aim i o compue Taylor expanion of he funcion ϕτ, ξ, eε and xτ, ξ, ε a he origin in order o approximae oluion of 5.7 and 5.6. In Appendix A, we li he relevan Taylor coefficien up o econd order. The bordering i choen a in Imperfec bifurcaion. Conider he expanion ϕτ, ξ, ε =τ ϕ τ ϕ ττ τ + ϕ ξτ ξ + ε ϕ ε + ϕ ξε ξ + ϕ τε τ + ϕ εε ε + h.o.. and τ, ξ, ε =τ τ + ττ τ + x ξτ ξ + ε ε + ξε ξ + τε τ + εε ε + h.o.., uτ, ξ, ε = u τ τ + δu ξ 8 u ξ + u ττ τ + u ξτ ξτ + ε u ε + u ξε ξ + u τε τ + u εε ε + h.o.., vτ, ξ, ε = v τ τ + δv ξ 8 v ξ + v ττ τ + v ξτ ξτ + ε v ε + v ξε ξ + v τε τ + v εε ε + h.o... The h.o.. are of hird order. Inead of 5.7, we olve he runcaed bifurcaion equaion 5. τ ϕ τ + ϕ ττ τ + ϕ ξτ ξ + ε ϕ ε + ϕ ξε ξ + ϕ τε τ + ϕ εε ε = for ξ and τ and fixed ε. If ϕ ξε, he oluion o 5. can be parameerized by τ. Hence, given a mall value of τ we compue ξ = ξτ a a oluion of he runcaed bifurcaion equaion 5.. Then we ubiue hi pair τ, ξτ ino he runcaed verion of We ge an approximaion of he roo of 5.6. Le u conider he funcion. from Secion and e ε =.. In Figure., in fac, he oluion e of 5. are depiced, namely, he oluion componen x =, x 4 = v and x 5 = v a.5.5, for boh red and blue branche. In Figure 5., here are zoom of boh red and blue branche howing he oluion componen, on he lef, and he componen v, on he righ. Thee oluion were compued numerically via pah-following; ee Secion 3.. Approximaion via 5. and he runcaed verion of are marked by red and blue diamond. They are reaonably accurae for mall Example coninued. Le u conider A a in Example 4.3. We model an imperfecion 5., n = m = 4. In paricular, we e ε =., Z
13 Ken Sae Univeriy hp://ena.mah.ken.edu 8 D. JANOVSKÁ AND V. JANOVSKÝ v FIGURE 5.. Analyi of moivaing example from Secion, ε =.: zoom of boh red and blue branche, namely, he oluion componen, v, compared wih reul of he runcaed approximaion marked by diamond "green" "yellow" FIGURE 5.. Example from Secion 5.3, ε =.: Four perurbed branche of he analyic SVD,. On he lef: An illuraion of he adapive ep lengh. We compue oluion of 5. for.. In Figure 5., here are four oluion branche, namely, he fir four componen of parameerized by,., colored by red, blue, green, and yellow. The iniial condiion of hee branche are he perurbaion of he iniial condiion from Example 4.3. Oberve ha he blue branch i cloe o he red one, and he green branch o he yellow one for.5 and.5, repecively. We hould have in mind our moivaion, a illuraed in Figure. and Figure.. The imple cro over branching degenerae o a ouching of differen branche like. A far a he lef/righ ingular vecor pah are concerned, a perurbaion implie wiing. Coming back o he perurbed Example 5.3, namely, o he coninuaion of he blue and he green branche, we oberved wi of lef/righ ingular vecor for.5 and.5, repecively. In Figure 5.3, on he lef, he u, u -oluion componen of he blue and red branche are hown near.5. The paricular componen wi. Similarly, in Figure 5.3, on he righ, we depic he u, u 3 -oluion componen of he green and yellow branche a.5. Again, here i a wi. Noe ha a imilar obervaion can be made a he blue and green branche nearly ouch for.5. Comparing Figure 4. and Figure 5., we conclude ha he global branching cenario may change dramaically under a perurbaion. 6. Concluion. We inroduced a new pah-following echnique o follow imple ingular value and he correponding lef/righ ingular vecor. We inveigaed ingulariie on he pah; namely, we claified a ingulariy wih corank = and codim =. Thi ingulariy i relaed o imple cro over in erm of [3]. We alo udied hi ingulariy ubjec o a perurbaion an imperfecion.
14 Ken Sae Univeriy hp://ena.mah.ken.edu ASVD: NON-GENERIC POINTS 83. u.5 u u u FIGURE 5.3. The oluion componen. Lef: The blue and red branche for.5. Righ: The green and yellow branche for FIGURE 6.. Branche of ingular value,..., a funcion of.4. The branche 6 and 7 are in magena and cyan. They cro on he lef while hey do no cro on he righ. For he zoom, ee Figure 6.. Thee inveigaion ry o uppor he claim ha he aforemenioned pah-following echnique work generically. Conider he homoopy 3. again. Le u compue he pah of he en large ingular value uing MATCONT [3]. The reul i ummarized in Figure 6.. The pah 6 and 7 inerec on he lef. The Coninuer wa run under a defaul parameer eing IniSepize =., MinSepize = 5, MaxSepize =.. If we increae he preciion eing IniSepize =., MinSepize = 6, MaxSepize =. he relevan branche do no inerec, a hown in Figure 6. on he righ. Figure 6. how zoom of he menioned branche compued wih defaul on he lef and he increaed on he righ preciion. The figure on he lef ugge ha he croing i a numerical arifac. We refer o he magena curve, i.e., he numerically compued branch 6. The cyan branch 7 no hown here exhibi imilar zig-zag ocillaion. The doed cyan line i a rimmed curve 7 wih removed ocillaion. Generically, parameer-dependen ingular value do no change he ordering excep if hey change he ign. In a forhcoming paper we will inveigae zero ingular value ubjec o perurbaion. Appendix A. Deail of Taylor expanion. We review leading erm of he expanion 4.3, , and he imperfec verion 5.8, Noe ha he compuaion of hee erm follow a rouine chain rule procedure indicaed a he end of Secion 4.. We ake advanage of he rucure of f x and higher parial differenial of f.
15 Ken Sae Univeriy hp://ena.mah.ken.edu 84 D. JANOVSKÁ AND V. JANOVSKÝ "magena" "magena" "cyan" FIGURE 6.. Zoom: The croing of he branche 6 and 7, hown in magena and cyan, on he lef of Figure 6. i acually a numerical arifac. If he preciion i increaed, hen he numerically compued branche do no cro. The Taylor coefficien depend on he following daa: u, v, δu, δv, A, A, A and higher derivaive of A, and on Z, Z and higher derivaive of Z. Noe ha Z i relaed o he unfolding 5.. Concerning 4.3: A. ϕ ξ = ϕ ξξ = = j ϕ, = =, j =,,..., ξj A. A.3 A.4 ϕ τ = δut A v u T A δv, ϕ ξτ = u T A v δu T A δv, ϕ ττ = δu T A v + u T A δv δu T A v τ + u τ T A δv, where A.5 uτ A = M v +, v τ A T u, M, i defined in 3.4. Concerning : A.6 ξ =, u ξ = δu, v ξ = δv, A.7 τ = u T A v, A.8 ξτ = ϕ τ, A.9 ξξ =, uτ = M v +, A v τ A T u, uξτ = A v ξτ M+, δv A T, δu uξξ = u v ξξ 4 v,
16 Ken Sae Univeriy hp://ena.mah.ken.edu ASVD: NON-GENERIC POINTS 85 A. ττ = u T A v + u T A v τ + u τ T A v, uττ v ττ A = M +, v [ A T u ] A M +, δv τ A T δu τ u T A v M +, [ A v A T u ], A. ξξξ =, uξξξ =. v ξξξ Moreover, A. ξ = ξξ = = j, =... =, j =,,.... ξj REMARK A.. The formulae A. and A. can be verified by inducion. Concerning : A.3 A.4 A.5 A.6 ϕ ε = δu T Z v + u T Z δv, ϕ ξε = u T Z v δu T Z δv, ϕ εε = δu T Z v ε u ε T Z δv, ϕ τε = δu T A v ε + δv T A T u ε δu T Z v + δv T Z T u δu T Z v τ + δv T Z T u τ, where u τ, v τ are defined by A.5, uε Z A.7 = M v +, v ε Z T u. A.8 ε = u T Z v, A.9 ξε = 4 u T Z δv + δu T Z v, A. A. τε = uξε = Z v ξε M+, δv Z T, δu u T A v ε + v T A T u ε + δu T Z v τ + v T Z T u τ + u T Z v,
17 Ken Sae Univeriy hp://ena.mah.ken.edu 86 D. JANOVSKÁ AND V. JANOVSKÝ A. uτε = M v +, A v ε τε A T u ε + τ M +, uε v ε [ M +, Z v τ Z T u τ M +, Z v Z T u ] v τ [ + ε M +, uτ ], A.3 εε = u T Z v ε + u ε T Z v, A.4 uεε = v ε M +, uε M εε v +, Z v ε ε Z T. u ε REFERENCES [] A. BUNSE-GERSTNER, R. BYERS, V. MEHRMANN, AND N. K. NICHOLS, Numerical compuaion of an analyic ingular value decompoiion of a marix valued funcion, Numer. Mah., 6 99, pp. 39. [] P. DEUFLHART AND A. HOHMANN, Numerical Analyi in Modern Scienific Compuing. An Inroducion, Springer, New York, 3. [3] A. DHOOGE, W. GOVAERTS, AND YU. A. KUZETSOV, MATCONT: A Malab package for numerical bifurcaion analyi of ODE, ACM Tran. Mah. Sofware, 3 3, pp [4] G. H. GOLUB AND C. F. VAN LOAN, Marix Compuaion, hird ed., The John Hopkin Univeriy Pre, Balimore, 996. [5] W. GOVAERTS, Numerical Mehod for Bifurcaion of Dynamical Equilibria, SIAM, Philadelphia,. [6] S. N. CHOW AND J. K. HALE, Mehod of Bifurcaion Theory, Springer, New York, 98. [7] V. JANOVSKÝ, D. JANOVSKÁ, AND K. TANABE, Compuing he Analyic Singular Value Decompoiion via a pahfollowing, in Proceeding of ENUMATH 5, A. B. de Caro, D. Gómez, P. Quinela, and P. Salgado, ed., Springer, New York, 6, pp [8] D. JANOVSKÁ AND V. JANOVSKÝ, On non-generic poin of he Analyic SVD, in Inernaional Conference on Numerical Analyi and Applied Mahemaic 6, T. E. Simo, G. Pihoyio, and Ch. Tioura, ed., WILEY-VCH, Weinheim, 6, pp [9] T. KATO, Perurbaion Theory for Linear Operaor, econd ed., Springer, New York, 976. [] S. KRANTZ AND H. PARKS, A Primer of Real Analyic Funcion. Birkhäuer, New York,. [] Marix Marke. Available a hp://mah.ni.gov/marixmarke/. [] V. MEHRMANN AND W. RATH, Numerical mehod for he compuaion of analyic ingular value decompoiion, Elecron. Tran. Numer. Anal., 993, pp hp://ena.mah.ken.edu/vol..993/pp7-88.dir/. [3] K. WRIGHT, Differenial equaion for he analyic ingular value decompoion of a marix. Numer. Mah., 63 99, pp
6.8 Laplace Transform: General Formulas
48 HAP. 6 Laplace Tranform 6.8 Laplace Tranform: General Formula Formula Name, ommen Sec. F() l{ f ()} e f () d f () l {F()} Definiion of Tranform Invere Tranform 6. l{af () bg()} al{f ()} bl{g()} Lineariy
More information, the. L and the L. x x. max. i n. It is easy to show that these two norms satisfy the following relation: x x n x = (17.3) max
ecure 8 7. Sabiliy Analyi For an n dimenional vecor R n, he and he vecor norm are defined a: = T = i n i (7.) I i eay o how ha hee wo norm aify he following relaion: n (7.) If a vecor i ime-dependen, hen
More informationSystems of nonlinear ODEs with a time singularity in the right-hand side
Syem of nonlinear ODE wih a ime ingulariy in he righ-hand ide Jana Burkoová a,, Irena Rachůnková a, Svaolav Saněk a, Ewa B. Weinmüller b, Sefan Wurm b a Deparmen of Mahemaical Analyi and Applicaion of
More informationClassification of 3-Dimensional Complex Diassociative Algebras
Malayian Journal of Mahemaical Science 4 () 41-54 (010) Claificaion of -Dimenional Complex Diaociaive Algebra 1 Irom M. Rihiboev, Iamiddin S. Rahimov and Wiriany Bari 1,, Iniue for Mahemaical Reearch,,
More informationFractional Ornstein-Uhlenbeck Bridge
WDS'1 Proceeding of Conribued Paper, Par I, 21 26, 21. ISBN 978-8-7378-139-2 MATFYZPRESS Fracional Ornein-Uhlenbeck Bridge J. Janák Charle Univeriy, Faculy of Mahemaic and Phyic, Prague, Czech Republic.
More informationEECE 301 Signals & Systems Prof. Mark Fowler
EECE 31 Signal & Syem Prof. Mark Fowler Noe Se #27 C-T Syem: Laplace Tranform Power Tool for yem analyi Reading Aignmen: Secion 6.1 6.3 of Kamen and Heck 1/18 Coure Flow Diagram The arrow here how concepual
More informationFIXED POINTS AND STABILITY IN NEUTRAL DIFFERENTIAL EQUATIONS WITH VARIABLE DELAYS
PROCEEDINGS OF THE AMERICAN MATHEMATICAL SOCIETY Volume 136, Number 3, March 28, Page 99 918 S 2-9939(7)989-2 Aricle elecronically publihed on November 3, 27 FIXED POINTS AND STABILITY IN NEUTRAL DIFFERENTIAL
More informationt is a basis for the solution space to this system, then the matrix having these solutions as columns, t x 1 t, x 2 t,... x n t x 2 t...
Mah 228- Fri Mar 24 5.6 Marix exponenials and linear sysems: The analogy beween firs order sysems of linear differenial equaions (Chaper 5) and scalar linear differenial equaions (Chaper ) is much sronger
More information18.03SC Unit 3 Practice Exam and Solutions
Sudy Guide on Sep, Dela, Convoluion, Laplace You can hink of he ep funcion u() a any nice mooh funcion which i for < a and for > a, where a i a poiive number which i much maller han any ime cale we care
More informationRandomized Perfect Bipartite Matching
Inenive Algorihm Lecure 24 Randomized Perfec Biparie Maching Lecurer: Daniel A. Spielman April 9, 208 24. Inroducion We explain a randomized algorihm by Ahih Goel, Michael Kapralov and Sanjeev Khanna for
More informationIntroduction to SLE Lecture Notes
Inroducion o SLE Lecure Noe May 13, 16 - The goal of hi ecion i o find a ufficien condiion of λ for he hull K o be generaed by a imple cure. I urn ou if λ 1 < 4 hen K i generaed by a imple curve. We will
More informationCHAPTER 7: SECOND-ORDER CIRCUITS
EEE5: CI RCUI T THEORY CHAPTER 7: SECOND-ORDER CIRCUITS 7. Inroducion Thi chaper conider circui wih wo orage elemen. Known a econd-order circui becaue heir repone are decribed by differenial equaion ha
More informationOn the Exponential Operator Functions on Time Scales
dvance in Dynamical Syem pplicaion ISSN 973-5321, Volume 7, Number 1, pp. 57 8 (212) hp://campu.m.edu/ada On he Exponenial Operaor Funcion on Time Scale laa E. Hamza Cairo Univeriy Deparmen of Mahemaic
More informationNECESSARY AND SUFFICIENT CONDITIONS FOR LATENT SEPARABILITY
NECESSARY AND SUFFICIENT CONDITIONS FOR LATENT SEPARABILITY Ian Crawford THE INSTITUTE FOR FISCAL STUDIES DEPARTMENT OF ECONOMICS, UCL cemmap working paper CWP02/04 Neceary and Sufficien Condiion for Laen
More informationAlgorithmic Discrete Mathematics 6. Exercise Sheet
Algorihmic Dicree Mahemaic. Exercie Shee Deparmen of Mahemaic SS 0 PD Dr. Ulf Lorenz 7. and 8. Juni 0 Dipl.-Mah. David Meffer Verion of June, 0 Groupwork Exercie G (Heap-Sor) Ue Heap-Sor wih a min-heap
More informationFLAT CYCLOTOMIC POLYNOMIALS OF ORDER FOUR AND HIGHER
#A30 INTEGERS 10 (010), 357-363 FLAT CYCLOTOMIC POLYNOMIALS OF ORDER FOUR AND HIGHER Nahan Kaplan Deparmen of Mahemaic, Harvard Univeriy, Cambridge, MA nkaplan@mah.harvard.edu Received: 7/15/09, Revied:
More informationChapter 7: Inverse-Response Systems
Chaper 7: Invere-Repone Syem Normal Syem Invere-Repone Syem Baic Sar ou in he wrong direcion End up in he original eady-ae gain value Two or more yem wih differen magniude and cale in parallel Main yem
More informationLaplace Transform. Inverse Laplace Transform. e st f(t)dt. (2)
Laplace Tranform Maoud Malek The Laplace ranform i an inegral ranform named in honor of mahemaician and aronomer Pierre-Simon Laplace, who ued he ranform in hi work on probabiliy heory. I i a powerful
More informationMathematische Annalen
Mah. Ann. 39, 33 339 (997) Mahemaiche Annalen c Springer-Verlag 997 Inegraion by par in loop pace Elon P. Hu Deparmen of Mahemaic, Norhweern Univeriy, Evanon, IL 628, USA (e-mail: elon@@mah.nwu.edu) Received:
More informationu(t) Figure 1. Open loop control system
Open loop conrol v cloed loop feedbac conrol The nex wo figure preen he rucure of open loop and feedbac conrol yem Figure how an open loop conrol yem whoe funcion i o caue he oupu y o follow he reference
More informationMatrix Versions of Some Refinements of the Arithmetic-Geometric Mean Inequality
Marix Versions of Some Refinemens of he Arihmeic-Geomeric Mean Inequaliy Bao Qi Feng and Andrew Tonge Absrac. We esablish marix versions of refinemens due o Alzer ], Carwrigh and Field 4], and Mercer 5]
More information1 Motivation and Basic Definitions
CSCE : Deign and Analyi of Algorihm Noe on Max Flow Fall 20 (Baed on he preenaion in Chaper 26 of Inroducion o Algorihm, 3rd Ed. by Cormen, Leieron, Rive and Sein.) Moivaion and Baic Definiion Conider
More information5.2 GRAPHICAL VELOCITY ANALYSIS Polygon Method
ME 352 GRHICL VELCITY NLYSIS 52 GRHICL VELCITY NLYSIS olygon Mehod Velociy analyi form he hear of kinemaic and dynamic of mechanical yem Velociy analyi i uually performed following a poiion analyi; ie,
More informationStability in Distribution for Backward Uncertain Differential Equation
Sabiliy in Diribuion for Backward Uncerain Differenial Equaion Yuhong Sheng 1, Dan A. Ralecu 2 1. College of Mahemaical and Syem Science, Xinjiang Univeriy, Urumqi 8346, China heng-yh12@mail.inghua.edu.cn
More informationChapter 2. First Order Scalar Equations
Chaper. Firs Order Scalar Equaions We sar our sudy of differenial equaions in he same way he pioneers in his field did. We show paricular echniques o solve paricular ypes of firs order differenial equaions.
More informationChapter 3 Boundary Value Problem
Chaper 3 Boundary Value Problem A boundary value problem (BVP) is a problem, ypically an ODE or a PDE, which has values assigned on he physical boundary of he domain in which he problem is specified. Le
More informationConvergence of the gradient algorithm for linear regression models in the continuous and discrete time cases
Convergence of he gradien algorihm for linear regreion model in he coninuou and dicree ime cae Lauren Praly To cie hi verion: Lauren Praly. Convergence of he gradien algorihm for linear regreion model
More informationt )? How would you have tried to solve this problem in Chapter 3?
Exercie 9) Ue Laplace ranform o wrie down he oluion o 2 x x = F in x = x x = v. wha phenomena do oluion o hi DE illurae (even hough we're forcing wih in co )? How would you have ried o olve hi problem
More informationMon Apr 2: Laplace transform and initial value problems like we studied in Chapter 5
Mah 225-4 Week 2 April 2-6 coninue.-.3; alo cover par of.4-.5, EP 7.6 Mon Apr 2:.-.3 Laplace ranform and iniial value problem like we udied in Chaper 5 Announcemen: Warm-up Exercie: Recall, The Laplace
More informationCHAPTER 7. Definition and Properties. of Laplace Transforms
SERIES OF CLSS NOTES FOR 5-6 TO INTRODUCE LINER ND NONLINER PROBLEMS TO ENGINEERS, SCIENTISTS, ND PPLIED MTHEMTICINS DE CLSS NOTES COLLECTION OF HNDOUTS ON SCLR LINER ORDINRY DIFFERENTIL EQUTIONS (ODE")
More informationProblem Set If all directed edges in a network have distinct capacities, then there is a unique maximum flow.
CSE 202: Deign and Analyi of Algorihm Winer 2013 Problem Se 3 Inrucor: Kamalika Chaudhuri Due on: Tue. Feb 26, 2013 Inrucion For your proof, you may ue any lower bound, algorihm or daa rucure from he ex
More information2. VECTORS. R Vectors are denoted by bold-face characters such as R, V, etc. The magnitude of a vector, such as R, is denoted as R, R, V
ME 352 VETS 2. VETS Vecor algebra form he mahemaical foundaion for kinemaic and dnamic. Geomer of moion i a he hear of boh he kinemaic and dnamic of mechanical em. Vecor anali i he imehonored ool for decribing
More informationAdditional Methods for Solving DSGE Models
Addiional Mehod for Solving DSGE Model Karel Meren, Cornell Univeriy Reference King, R. G., Ploer, C. I. & Rebelo, S. T. (1988), Producion, growh and buine cycle: I. he baic neoclaical model, Journal of
More informationLet. x y. denote a bivariate time series with zero mean.
Linear Filer Le x y : T denoe a bivariae ime erie wih zero mean. Suppoe ha he ime erie {y : T} i conruced a follow: y a x The ime erie {y : T} i aid o be conruced from {x : T} by mean of a Linear Filer.
More informationIntroduction to Congestion Games
Algorihmic Game Theory, Summer 2017 Inroducion o Congeion Game Lecure 1 (5 page) Inrucor: Thoma Keelheim In hi lecure, we ge o know congeion game, which will be our running example for many concep in game
More informationTo become more mathematically correct, Circuit equations are Algebraic Differential equations. from KVL, KCL from the constitutive relationship
Laplace Tranform (Lin & DeCarlo: Ch 3) ENSC30 Elecric Circui II The Laplace ranform i an inegral ranformaion. I ranform: f ( ) F( ) ime variable complex variable From Euler > Lagrange > Laplace. Hence,
More informationNetwork Flows: Introduction & Maximum Flow
CSC 373 - lgorihm Deign, nalyi, and Complexiy Summer 2016 Lalla Mouaadid Nework Flow: Inroducion & Maximum Flow We now urn our aenion o anoher powerful algorihmic echnique: Local Search. In a local earch
More informationFinish reading Chapter 2 of Spivak, rereading earlier sections as necessary. handout and fill in some missing details!
MAT 257, Handou 6: Ocober 7-2, 20. I. Assignmen. Finish reading Chaper 2 of Spiva, rereading earlier secions as necessary. handou and fill in some missing deails! II. Higher derivaives. Also, read his
More informationCONTROL SYSTEMS. Chapter 10 : State Space Response
CONTROL SYSTEMS Chaper : Sae Space Repone GATE Objecive & Numerical Type Soluion Queion 5 [GATE EE 99 IIT-Bombay : Mark] Conider a econd order yem whoe ae pace repreenaion i of he form A Bu. If () (),
More informationSection 3.5 Nonhomogeneous Equations; Method of Undetermined Coefficients
Secion 3.5 Nonhomogeneous Equaions; Mehod of Undeermined Coefficiens Key Terms/Ideas: Linear Differenial operaor Nonlinear operaor Second order homogeneous DE Second order nonhomogeneous DE Soluion o homogeneous
More information10. State Space Methods
. Sae Space Mehods. Inroducion Sae space modelling was briefly inroduced in chaper. Here more coverage is provided of sae space mehods before some of heir uses in conrol sysem design are covered in he
More informationCSC 364S Notes University of Toronto, Spring, The networks we will consider are directed graphs, where each edge has associated with it
CSC 36S Noe Univeriy of Torono, Spring, 2003 Flow Algorihm The nework we will conider are direced graph, where each edge ha aociaed wih i a nonnegaive capaciy. The inuiion i ha if edge (u; v) ha capaciy
More informationARTIFICIAL INTELLIGENCE. Markov decision processes
INFOB2KI 2017-2018 Urech Univeriy The Neherland ARTIFICIAL INTELLIGENCE Markov deciion procee Lecurer: Silja Renooij Thee lide are par of he INFOB2KI Coure Noe available from www.c.uu.nl/doc/vakken/b2ki/chema.hml
More information23.2. Representing Periodic Functions by Fourier Series. Introduction. Prerequisites. Learning Outcomes
Represening Periodic Funcions by Fourier Series 3. Inroducion In his Secion we show how a periodic funcion can be expressed as a series of sines and cosines. We begin by obaining some sandard inegrals
More informationLecture 21: Bezier Approximation and de Casteljau s Algorithm. and thou shalt be near unto me Genesis 45:10
Lecure 2: Bezier Approximaion and de Caeljau Algorihm and hou hal be near uno me Genei 45:0. Inroducion In Lecure 20, we ued inerpolaion o pecify hape. Bu inerpolaion i no alway a good way o decribe he
More informationESTIMATES FOR THE DERIVATIVE OF DIFFUSION SEMIGROUPS
Elec. Comm. in Probab. 3 (998) 65 74 ELECTRONIC COMMUNICATIONS in PROBABILITY ESTIMATES FOR THE DERIVATIVE OF DIFFUSION SEMIGROUPS L.A. RINCON Deparmen of Mahemaic Univeriy of Wale Swanea Singleon Par
More informationGLOBAL ANALYTIC REGULARITY FOR NON-LINEAR SECOND ORDER OPERATORS ON THE TORUS
PROCEEDINGS OF THE AMERICAN MATHEMATICAL SOCIETY Volume 131, Number 12, Page 3783 3793 S 0002-9939(03)06940-5 Aricle elecronically publihed on February 28, 2003 GLOBAL ANALYTIC REGULARITY FOR NON-LINEAR
More informationExplicit form of global solution to stochastic logistic differential equation and related topics
SAISICS, OPIMIZAION AND INFOMAION COMPUING Sa., Opim. Inf. Compu., Vol. 5, March 17, pp 58 64. Publihed online in Inernaional Academic Pre (www.iapre.org) Explici form of global oluion o ochaic logiic
More informationThe structure of a set of positive solutions to Dirichlet BVPs with time and space singularities
The rucure of a e of poiive oluion o Dirichle BVP wih ime and pace ingulariie Irena Rachůnková a, Alexander Spielauer b, Svaolav Saněk a and Ewa B. Weinmüller b a Deparmen of Mahemaical Analyi, Faculy
More informationFUZZY n-inner PRODUCT SPACE
Bull. Korean Mah. Soc. 43 (2007), No. 3, pp. 447 459 FUZZY n-inner PRODUCT SPACE Srinivaan Vijayabalaji and Naean Thillaigovindan Reprined from he Bullein of he Korean Mahemaical Sociey Vol. 43, No. 3,
More informationMATH 128A, SUMMER 2009, FINAL EXAM SOLUTION
MATH 28A, SUMME 2009, FINAL EXAM SOLUTION BENJAMIN JOHNSON () (8 poins) [Lagrange Inerpolaion] (a) (4 poins) Le f be a funcion defined a some real numbers x 0,..., x n. Give a defining equaion for he Lagrange
More informationResearch Article An Upper Bound on the Critical Value β Involved in the Blasius Problem
Hindawi Publihing Corporaion Journal of Inequaliie and Applicaion Volume 2010, Aricle ID 960365, 6 page doi:10.1155/2010/960365 Reearch Aricle An Upper Bound on he Criical Value Involved in he Blaiu Problem
More informationThe Residual Graph. 11 Augmenting Path Algorithms. Augmenting Path Algorithm. Augmenting Path Algorithm
Augmening Pah Algorihm Greedy-algorihm: ar wih f (e) = everywhere find an - pah wih f (e) < c(e) on every edge augmen flow along he pah repea a long a poible The Reidual Graph From he graph G = (V, E,
More informationLower and Upper Approximation of Fuzzy Ideals in a Semiring
nernaional Journal of Scienific & Engineering eearch, Volume 3, ue, January-0 SSN 9-558 Lower and Upper Approximaion of Fuzzy deal in a Semiring G Senhil Kumar, V Selvan Abrac n hi paper, we inroduce he
More informationChapter 6. Systems of First Order Linear Differential Equations
Chaper 6 Sysems of Firs Order Linear Differenial Equaions We will only discuss firs order sysems However higher order sysems may be made ino firs order sysems by a rick shown below We will have a sligh
More information18 Extensions of Maximum Flow
Who are you?" aid Lunkwill, riing angrily from hi ea. Wha do you wan?" I am Majikhie!" announced he older one. And I demand ha I am Vroomfondel!" houed he younger one. Majikhie urned on Vroomfondel. I
More informationHamilton- J acobi Equation: Weak S olution We continue the study of the Hamilton-Jacobi equation:
M ah 5 7 Fall 9 L ecure O c. 4, 9 ) Hamilon- J acobi Equaion: Weak S oluion We coninue he sudy of he Hamilon-Jacobi equaion: We have shown ha u + H D u) = R n, ) ; u = g R n { = }. ). In general we canno
More informationNEUTRON DIFFUSION THEORY
NEUTRON DIFFUSION THEORY M. Ragheb 4//7. INTRODUCTION The diffuion heory model of neuron ranpor play a crucial role in reacor heory ince i i imple enough o allow cienific inigh, and i i ufficienly realiic
More informationMath Week 14 April 16-20: sections first order systems of linear differential equations; 7.4 mass-spring systems.
Mah 2250-004 Week 4 April 6-20 secions 7.-7.3 firs order sysems of linear differenial equaions; 7.4 mass-spring sysems. Mon Apr 6 7.-7.2 Sysems of differenial equaions (7.), and he vecor Calculus we need
More informationOn the Benney Lin and Kawahara Equations
JOURNAL OF MATEMATICAL ANALYSIS AND APPLICATIONS 11, 13115 1997 ARTICLE NO AY975438 On he BenneyLin and Kawahara Equaion A Biagioni* Deparmen of Mahemaic, UNICAMP, 1381-97, Campina, Brazil and F Linare
More informationResearch Article On Double Summability of Double Conjugate Fourier Series
Inernaional Journal of Mahemaic and Mahemaical Science Volume 22, Aricle ID 4592, 5 page doi:.55/22/4592 Reearch Aricle On Double Summabiliy of Double Conjugae Fourier Serie H. K. Nigam and Kuum Sharma
More informationClass Meeting # 10: Introduction to the Wave Equation
MATH 8.5 COURSE NOTES - CLASS MEETING # 0 8.5 Inroducion o PDEs, Fall 0 Professor: Jared Speck Class Meeing # 0: Inroducion o he Wave Equaion. Wha is he wave equaion? The sandard wave equaion for a funcion
More informationOnline Appendix to Solution Methods for Models with Rare Disasters
Online Appendix o Soluion Mehods for Models wih Rare Disasers Jesús Fernández-Villaverde and Oren Levinal In his Online Appendix, we presen he Euler condiions of he model, we develop he pricing Calvo block,
More informationResearch Article Existence and Uniqueness of Solutions for a Class of Nonlinear Stochastic Differential Equations
Hindawi Publihing Corporaion Abrac and Applied Analyi Volume 03, Aricle ID 56809, 7 page hp://dx.doi.org/0.55/03/56809 Reearch Aricle Exience and Uniquene of Soluion for a Cla of Nonlinear Sochaic Differenial
More informationFlow networks. Flow Networks. A flow on a network. Flow networks. The maximum-flow problem. Introduction to Algorithms, Lecture 22 December 5, 2001
CS 545 Flow Nework lon Efra Slide courey of Charle Leieron wih mall change by Carola Wenk Flow nework Definiion. flow nework i a direced graph G = (V, E) wih wo diinguihed verice: a ource and a ink. Each
More informationAlgorithms and Data Structures 2011/12 Week 9 Solutions (Tues 15th - Fri 18th Nov)
Algorihm and Daa Srucure 2011/ Week Soluion (Tue 15h - Fri 18h No) 1. Queion: e are gien 11/16 / 15/20 8/13 0/ 1/ / 11/1 / / To queion: (a) Find a pair of ube X, Y V uch ha f(x, Y) = f(v X, Y). (b) Find
More informationEE Control Systems LECTURE 2
Copyrigh F.L. Lewi 999 All righ reerved EE 434 - Conrol Syem LECTURE REVIEW OF LAPLACE TRANSFORM LAPLACE TRANSFORM The Laplace ranform i very ueful in analyi and deign for yem ha are linear and ime-invarian
More informationEchocardiography Project and Finite Fourier Series
Echocardiography Projec and Finie Fourier Series 1 U M An echocardiagram is a plo of how a porion of he hear moves as he funcion of ime over he one or more hearbea cycles If he hearbea repeas iself every
More informationThe multisubset sum problem for finite abelian groups
Alo available a hp://amc-journal.eu ISSN 1855-3966 (prined edn.), ISSN 1855-3974 (elecronic edn.) ARS MATHEMATICA CONTEMPORANEA 8 (2015) 417 423 The muliube um problem for finie abelian group Amela Muraović-Ribić
More informationChapter 6. Laplace Transforms
6- Chaper 6. Laplace Tranform 6.4 Shor Impule. Dirac Dela Funcion. Parial Fracion 6.5 Convoluion. Inegral Equaion 6.6 Differeniaion and Inegraion of Tranform 6.7 Syem of ODE 6.4 Shor Impule. Dirac Dela
More informationDiscussion Session 2 Constant Acceleration/Relative Motion Week 03
PHYS 100 Dicuion Seion Conan Acceleraion/Relaive Moion Week 03 The Plan Today you will work wih your group explore he idea of reference frame (i.e. relaive moion) and moion wih conan acceleraion. You ll
More informationLinear Algebra Primer
Linear Algebra rimer And a video dicuion of linear algebra from EE263 i here (lecure 3 and 4): hp://ee.anford.edu/coure/ee263 lide from Sanford CS3 Ouline Vecor and marice Baic Mari Operaion Deerminan,
More informationChapter 6. Laplace Transforms
Chaper 6. Laplace Tranform Kreyzig by YHLee;45; 6- An ODE i reduced o an algebraic problem by operaional calculu. The equaion i olved by algebraic manipulaion. The reul i ranformed back for he oluion of
More informationA Theoretical Model of a Voltage Controlled Oscillator
A Theoreical Model of a Volage Conrolled Ocillaor Yenming Chen Advior: Dr. Rober Scholz Communicaion Science Iniue Univeriy of Souhern California UWB Workhop, April 11-1, 6 Inroducion Moivaion The volage
More informationCS4445/9544 Analysis of Algorithms II Solution for Assignment 1
Conider he following flow nework CS444/944 Analyi of Algorihm II Soluion for Aignmen (0 mark) In he following nework a minimum cu ha capaciy 0 Eiher prove ha hi aemen i rue, or how ha i i fale Uing he
More informationMATH 4330/5330, Fourier Analysis Section 6, Proof of Fourier s Theorem for Pointwise Convergence
MATH 433/533, Fourier Analysis Secion 6, Proof of Fourier s Theorem for Poinwise Convergence Firs, some commens abou inegraing periodic funcions. If g is a periodic funcion, g(x + ) g(x) for all real x,
More informationMath 334 Fall 2011 Homework 11 Solutions
Dec. 2, 2 Mah 334 Fall 2 Homework Soluions Basic Problem. Transform he following iniial value problem ino an iniial value problem for a sysem: u + p()u + q() u g(), u() u, u () v. () Soluion. Le v u. Then
More informationCHAPTER. Forced Equations and Systems { } ( ) ( ) 8.1 The Laplace Transform and Its Inverse. Transforms from the Definition.
CHAPTER 8 Forced Equaion and Syem 8 The aplace Tranform and I Invere Tranform from he Definiion 5 5 = b b {} 5 = 5e d = lim5 e = ( ) b {} = e d = lim e + e d b = (inegraion by par) = = = = b b ( ) ( )
More informationNotes on cointegration of real interest rates and real exchange rates. ρ (2)
Noe on coinegraion of real inere rae and real exchange rae Charle ngel, Univeriy of Wiconin Le me ar wih he obervaion ha while he lieraure (mo prominenly Meee and Rogoff (988) and dion and Paul (993))
More informationModel Reduction for Dynamical Systems Lecture 6
Oo-von-Guericke Universiä Magdeburg Faculy of Mahemaics Summer erm 07 Model Reducion for Dynamical Sysems ecure 6 v eer enner and ihong Feng Max lanck Insiue for Dynamics of Complex echnical Sysems Compuaional
More informationNote on Matuzsewska-Orlich indices and Zygmund inequalities
ARMENIAN JOURNAL OF MATHEMATICS Volume 3, Number 1, 21, 22 31 Noe on Mauzewka-Orlic indice and Zygmund inequaliie N. G. Samko Univeridade do Algarve, Campu de Gambela, Faro,85 139, Porugal namko@gmail.com
More informationReminder: Flow Networks
0/0/204 Ma/CS 6a Cla 4: Variou (Flow) Execie Reminder: Flow Nework A flow nework i a digraph G = V, E, ogeher wih a ource verex V, a ink verex V, and a capaciy funcion c: E N. Capaciy Source 7 a b c d
More informationRough Paths and its Applications in Machine Learning
Pah ignaure Machine learning applicaion Rough Pah and i Applicaion in Machine Learning July 20, 2017 Rough Pah and i Applicaion in Machine Learning Pah ignaure Machine learning applicaion Hiory and moivaion
More informationThe Residual Graph. 12 Augmenting Path Algorithms. Augmenting Path Algorithm. Augmenting Path Algorithm
Augmening Pah Algorihm Greedy-algorihm: ar wih f (e) = everywhere find an - pah wih f (e) < c(e) on every edge augmen flow along he pah repea a long a poible The Reidual Graph From he graph G = (V, E,
More informationSyntactic Complexity of Suffix-Free Languages. Marek Szykuła
Inroducion Upper Bound on Synacic Complexiy of Suffix-Free Language Univeriy of Wrocław, Poland Join work wih Januz Brzozowki Univeriy of Waerloo, Canada DCFS, 25.06.2015 Abrac Inroducion Sae and ynacic
More informationdy dx = xey (a) y(0) = 2 (b) y(1) = 2.5 SOLUTION: See next page
Assignmen 1 MATH 2270 SOLUTION Please wrie ou complee soluions for each of he following 6 problems (one more will sill be added). You may, of course, consul wih your classmaes, he exbook or oher resources,
More informationPerformance Comparison of LCMV-based Space-time 2D Array and Ambiguity Problem
Inernaional journal of cience Commerce and umaniie Volume No 2 No 3 April 204 Performance Comparion of LCMV-baed pace-ime 2D Arra and Ambigui Problem 2 o uan Chang and Jin hinghia Deparmen of Communicaion
More informationExponential Sawtooth
ECPE 36 HOMEWORK 3: PROPERTIES OF THE FOURIER TRANSFORM SOLUTION. Exponenial Sawooh: The eaie way o do hi problem i o look a he Fourier ranform of a ingle exponenial funcion, () = exp( )u(). From he able
More informationGeneralized Orlicz Spaces and Wasserstein Distances for Convex-Concave Scale Functions
Generalized Orlicz Space and Waerein Diance for Convex-Concave Scale Funcion Karl-Theodor Surm Abrac Given a ricly increaing, coninuou funcion ϑ : R + R +, baed on he co funcional ϑ (d(x, y dq(x, y, we
More informationTime Varying Multiserver Queues. W. A. Massey. Murray Hill, NJ Abstract
Waiing Time Aympoic for Time Varying Mulierver ueue wih Abonmen Rerial A. Melbaum Technion Iniue Haifa, 3 ISRAEL avim@x.echnion.ac.il M. I. Reiman Bell Lab, Lucen Technologie Murray Hill, NJ 7974 U.S.A.
More informationMath 2142 Exam 1 Review Problems. x 2 + f (0) 3! for the 3rd Taylor polynomial at x = 0. To calculate the various quantities:
Mah 4 Eam Review Problems Problem. Calculae he 3rd Taylor polynomial for arcsin a =. Soluion. Le f() = arcsin. For his problem, we use he formula f() + f () + f ()! + f () 3! for he 3rd Taylor polynomial
More informationOrdinary Differential Equations
Ordinary Differenial Equaions 5. Examples of linear differenial equaions and heir applicaions We consider some examples of sysems of linear differenial equaions wih consan coefficiens y = a y +... + a
More informationdt = C exp (3 ln t 4 ). t 4 W = C exp ( ln(4 t) 3) = C(4 t) 3.
Mah Rahman Exam Review Soluions () Consider he IVP: ( 4)y 3y + 4y = ; y(3) = 0, y (3) =. (a) Please deermine he longes inerval for which he IVP is guaraneed o have a unique soluion. Soluion: The disconinuiies
More informationVehicle Arrival Models : Headway
Chaper 12 Vehicle Arrival Models : Headway 12.1 Inroducion Modelling arrival of vehicle a secion of road is an imporan sep in raffic flow modelling. I has imporan applicaion in raffic flow simulaion where
More informationEXERCISES FOR SECTION 1.5
1.5 Exisence and Uniqueness of Soluions 43 20. 1 v c 21. 1 v c 1 2 4 6 8 10 1 2 2 4 6 8 10 Graph of approximae soluion obained using Euler s mehod wih = 0.1. Graph of approximae soluion obained using Euler
More informationOrdinary dierential equations
Chaper 5 Ordinary dierenial equaions Conens 5.1 Iniial value problem........................... 31 5. Forward Euler's mehod......................... 3 5.3 Runge-Kua mehods.......................... 36
More informationExistence of positive solution for a third-order three-point BVP with sign-changing Green s function
Elecronic Journal of Qualiaive Theory of Differenial Equaions 13, No. 3, 1-11; hp://www.mah.u-szeged.hu/ejqde/ Exisence of posiive soluion for a hird-order hree-poin BVP wih sign-changing Green s funcion
More informationCERTAIN CLASSES OF SOLUTIONS OF LAGERSTROM EQUATIONS
SARAJEVO JOURNAL OF MATHEMATICS Vol.10 (22 (2014, 67 76 DOI: 10.5644/SJM.10.1.09 CERTAIN CLASSES OF SOLUTIONS OF LAGERSTROM EQUATIONS ALMA OMERSPAHIĆ AND VAHIDIN HADŽIABDIĆ Absrac. This paper presens sufficien
More informationUndetermined coefficients for local fractional differential equations
Available online a www.isr-publicaions.com/jmcs J. Mah. Compuer Sci. 16 (2016), 140 146 Research Aricle Undeermined coefficiens for local fracional differenial equaions Roshdi Khalil a,, Mohammed Al Horani
More informationBuckling of a structure means failure due to excessive displacements (loss of structural stiffness), and/or
Buckling Buckling of a rucure mean failure due o exceive diplacemen (lo of rucural iffne), and/or lo of abiliy of an equilibrium configuraion of he rucure The rule of humb i ha buckling i conidered a mode
More information