Natural hp-bem for the electric field integral equation with singular solutions

Size: px
Start display at page:

Download "Natural hp-bem for the electric field integral equation with singular solutions"

Transcription

1 Natural hp-bem for the electrc feld ntegral equaton wth sngular solutons Alexe Bespalov Norbert Heuer Abstract We apply the hp-verson of the boundary element method (BEM) for the numercal soluton of the electrc feld ntegral equaton (EFIE) on a Lpschtz polyhedral surface Γ. The underlyng meshes are supposed to be quas-unform trangulatons of Γ, and the approxmatons are based on ether Ravart-Thomas or Brezz-Douglas-Marn famles of surface elements. Non-smoothness of Γ leads to sngulartes n the soluton of the EFIE, severely affectng convergence rates of the BEM. However, the sngular behavour of the soluton can be explctly specfed usng a fnte set of power functons (vertex-, edge-, and vertex-edge sngulartes). In ths paper we use ths fact to perform an a pror error analyss of the hp- BEM on quas-unform meshes. We prove precse error estmates n terms of the polynomal degree p, the mesh sze h, and the sngularty exponents. Key words: hp-verson wth quas-unform meshes, boundary element method, electrc feld ntegral equaton, sngulartes, a pror error estmate AMS Subect Classfcaton: 65N38, 65N15, 78M15, 41A10 1 Introducton In ths paper we study numercal approxmatons of the electrc feld ntegral equaton (EFIE) on a surface Γ = Ω, where Ω R 3 s a Lpschtz polyhedron. The EFIE s a boundary ntegral formulaton of a boundary value problem for the tme-harmonc Maxwell equatons n the doman exteror to Ω. It models the scatterng of electromagnetc waves at a perfectly conductng body. For the numercal soluton of the EFIE we use the Galerkn boundary element method (BEM). It employs H(dv Γ )-conformng famles of surface elements, namely Ravart-Thomas (RT) and Brezz-Douglas-Marn (BDM) elements, to dscretse the varatonal formulaton of the EFIE (called Rumsey s prncple). Ths approach s referred to as the natural BEM for the EFIE. As n fnte element methods, the convergence may be acheved ether by keepng Supported by EPSRC under grant no. EP/E058094/1. School of Mathematcs, Unversty of Manchester, Manchester, M13 9PL, UK. Emal: albespalov@yahoo.com Facultad de Matemátcas, Pontfca Unversdad Católca de Chle, Avenda Vcuña Mackenna 4860, Santago, Chle. Emal: nheuer@mat.puc.cl 1

2 polynomal degrees fxed and refnng the mesh (h-verson), or by fxng the mesh and ncreasng polynomal degrees (p-verson), or by smultaneous h-refnement and p-enrchment (hp-verson). The Galerkn BEM s wdely used n the engneerng practce for the smulaton of electromagnetc scatterng. Moreover, t had been used long before a rgorous theoretcal analyss of the method became avalable. Error analyses of dfferent boundary element schemes for the eddy current problem on smooth obstacles are gven by MacCamy and Stephan n [24, 25]. Despte these early results the BEM-analyss for non-smooth obstacles started much later. In fact, the convergence and a pror error analyss of the h-bem for the EFIE on pecewse smooth (open or closed) surfaces has been developed wthn the last decade (see [13, 22, 16, 10, 15]), and the correspondng results for hgh-order methods (p- and hp-bem) are very recent (see [6, 4, 7]). In partcular, [6] and [7] establsh quas-optmal convergence of the natural hp-bem on meshes of shape-regular elements. An essental ngredent for the proofs n these papers are the proectonbased nterpolaton operators developed by Demkowcz and co-authors [19, 20, 18]. These operators also facltate a pror error analyss, where one needs hp-approxmaton theory n specfc trace spaces. In [4] we prove an a pror error estmate of the hp-bem on quas-unform meshes of affne elements under the assumpton that the regularty of the soluton to the EFIE s gven n Sobolev spaces on Γ. The latter result states that the method converges wth the same rate r n both h and p 1 (here, r denotes the Sobolev regularty order, and p s assumed to be large enough). However, t s evdent from the numercal results reported n [23] that the p-bem converges faster than the h-bem for the EFIE. Ths s smlar to what was observed and proved for the p-bem and the h-bem appled to ellptc problems n three-dmensons (cf. [27, 21, 3, 5]). Wth ths paper we fll a gap n the theory of the BEM for the EFIE on polyhedral surfaces by provng a precse error estmate for the hp-bem on quas-unform meshes. Smlarly to the ellptc case we make use of explct expressons for sngulartes n electromagnetcs felds (these expressons are avalable from [17, 6]). The establshed error estmate shows that the convergence rates n h and p depend on the strongest sngularty exponent, and that the p-bem converges twce as fast as the h-bem on quas-unform meshes. Ths extends the results of [6], where the analyss was restrcted to the p-bem on a plane open surface. It s now well known that approprate decompostons of vector felds (on both the contnuous and dscrete level) are crtcal for the analyss of electromagnetc problems and ther approxmatons. In the case of the EFIE the man dea s to solate the kernel of the dv Γ - operator such that the complementary feld possesses an enhanced smoothness (cf. [10, 9, 7]). The correspondng decompostons of sngular vector felds greatly facltate our analyss n ths paper as well. In partcular, the complementary vector felds are sngular vector functons belongng to H 1/2 (Γ). Then, n the case of the BDM-based BEM, these vector functons can be approxmated component-wse by contnuous pecewse polynomals, and the desred hp-error estmates are derved by usng the correspondng results n [3] for scalar sngulartes (belongng to H 1/2 (Γ)). However, ths smple approach does not work for the RT-based BEM because the dmenson of the underlyng RT-space on the reference element s smaller than the dmenson of the BDM-counterpart. That s why, the results of [3] cannot be appled drectly, and addtonal techncal arguments are needed (see the proof of Lemma 4.2 n Secton 5.2). 2

3 The rest of the paper s organsed as follows. In the next secton we formulate the EFIE n ts varatonal form and recall the typcal structure of the soluton to ths model problem. In Secton 3 we ntroduce the hp-verson of the BEM for the EFIE and formulate the man result of the paper (Theorem 3.1) statng convergence rates of the method. Ths result follows from the general approxmaton theorem (Theorem 4.1) establshed n Secton 4. The proof of Theorem 4.1 reles, n partcular, on two techncal lemmas whch are proved n Secton 5. Throughout the paper, C denotes a generc postve constant ndependent of h and p. 2 Formulaton of the problem Let Γ be a Lpschtz polyhedral surface n R 3. Throughout the paper we wll use exactly the same notaton as n [4] for all nvolved dfferental and boundary ntegral operators as well as for Sobolev spaces of scalar functons and tangental vector felds on Γ (all essental defntons are gven n [4, Secton 3.1]). In partcular, we use boldface symbols for vector felds, and the spaces (or sets) of vector felds are denoted n boldface as well. For a gven wave number κ > 0, we denote by Ψ κ (resp., Ψ κ ) the scalar (resp., the vectoral) sngle layer boundary ntegral operator on Γ for the Helmholtz operator κ 2 (see [15, Secton 5]). The varatonal formulaton for the EFIE wll be posed n the followng Hlbert space of tangental vector felds on Γ: X = H 1/2 (dv Γ, Γ) := {u H 1/2 (Γ); dv Γ u H 1/2 (Γ)}, whch s the trace space of H(curl, Ω), where Ω R 3 s a Lpschtz polyhedron such that Γ = Ω (we refer to [11, 12, 14] for the defnton and propertes of ths and other trace spaces on Γ). Let X be the dual space of X (wth dualty parng extendng the L 2 (Γ)-nner product for tangental vector felds). Then, for a gven source functonal f X the varatonal formulaton of the EFIE reads as: fnd a complex tangental feld u X such that a(u, v) := Ψ κ dv Γ u, dv Γ v κ 2 Ψ κ u, v = f, v v X. (2.1) Here, the brackets, denote dualtes assocated wth H 1/2 (Γ) and H 1/2 (Γ). To ensure the unqueness of the soluton to (2.1) we always assume that κ 2 s not an electrcal egenvalue of the nteror problem. Let us recall the typcal structure of the soluton u to problem (2.1), provded that the source functonal f s suffcently smooth (we note that ths regularty assumpton s satsfed for the electromagnetc scatterng wth plane ncdent wave). We use the results of [6, Appendx A]. These results were derved from the regularty theory n [17] for the boundary value problems for Maxwell s equatons n 3D by makng use of trace arguments and some techncal calculatons n the sprt of [29] and [28]. Let V = {v} and E = {e} denote the sets of vertces and edges of Γ, respectvely. For v V, let E(v) denote the set of edges wth v as an end pont. Then the soluton u of (2.1) can be wrtten as u = u reg + u sng, (2.2) 3

4 where u reg X k := {u H k (Γ); dv Γ u H k (Γ)} wth k > 0 (2.3) (the spaces H k (Γ) and H k (Γ) are defned n a pecewse fashon by localsaton to each face of Γ, and the space X k s equpped wth ts graph norm X k, see [4]), u sng = u e + u v + u ev, (2.4) e E v V v V e E(v) and u e, u v, and u ev are the edge, vertex, and edge-vertex sngulartes, respectvely. In order to wrte explct expressons for the above sngulartes, let us fx a vertex v V and an edge e E(v). Then, on each face Γ ev Γ such that e Γ ev we wll use local polar and Cartesan coordnate systems (r v, θ v ) and (x e1, x e2 ), both wth the orgn at v, such that e = {(x e1, x e2 ); x e2 = 0, x e1 > 0} and for a suffcently small neghbourhood B τ of v there holds Γ ev B τ {(r v, θ v ); 0 < θ v < ω v }. Here, ω v denotes the nteror angle (on Γ ev ) between the edges meetng at v. For smplcty of notaton we wrte out here only the leadng sngulartes n u e, u v, and u ev on the face Γ ev, thus omttng the correspondng terms of hgher regularty (see [6, Appendx A] for complete expansons). For the edge sngulartes u e one has u e = curl Γ ev ( x γe 1 e2 log x e2 se 1 b e 1 (x e1 ) χ e 1(x e1 ) χ e 2(x e2 ) ) + x γe 2 e2 log x e2 se 2 b e 2 (x e1 ) χ e 1(x e1 ) χ e 2(x e2 ), (2.5) where curl Γ ev = ( / x e2, / x e1 ) s the tangental vector curl operator curl Γ restrcted to the face Γ ev (cf. [11, 12]), γ1 e, γe 2 > 1 2, and se 1, se 2 0 are ntegers. Here, χe 1, χe 2 are C cut-off functons wth χ e 1 = 1 n a certan dstance to the end ponts of e and χe 1 = 0 n a neghbourhood of these vertces. Moreover, χ e 2 = 1 for 0 x e2 δ e and χ e 2 = 0 for x e2 2δ e wth some δ e (0, 1 2 ). The functons be 1 χe 1 Hm 1 (e) and b e 2 χe 1 Hm 2 (e) for m 1 and m 2 as large as requred. The vertex sngulartes u v have the form ( ) u v = curl Γ ev r λv 1 v log r v qv 1 χ v (r v ) χ v 1(θ v ) + r λv 2 v log r v qv 2 χ v (r v ) χ v 2(θ v ), (2.6) where λ v 1, λv 2 > 1 2 are real numbers, qv 1, qv 2 0 are ntegers, χv s a C cut-off functon wth χ v = 1 for 0 r v τ v and χ v = 0 for r v 2τ v wth some τ v (0, 1 2 ). The functons χv 1, χv 2 are such that χ v 1 Ht 1 (0, ω v ), χ v 2 Ht 2 (0, ω v ) for t 1, t 2 as large as requred. For the combned edge-vertex sngularty u ev one has u ev = u ev 1 + u ev 2, where ( ) u ev 1 = curl Γ ev x λv 1 γe 1 e1 x γe 1 e2 log x e1 β 1 log x e2 β 2 χ v (r v )χ ev (θ v ) 4

5 ( + x λv 2 γe 2 e1 x γe 2 e2 log x e1 β 3 log x e2 β 4 χ v (r v )χ ev 0 (θ v ) 1 ) (2.7) and ( ) u ev 2 = curl Γ ev x γe 1 e2 log x e2 se 1 b e 3 (x e1, x e2 ) χ e 2(x e2 ) + x γe 2 e2 log x e2 se 2 b e 4 (x e1, x e2 ) χ e 2(x e2 ). (2.8) Here, λ v, γe, se ( = 1, 2), χv, and χ e 2 are as above, β k 0 (k = 1..., 4) are ntegers, β 1 + β 2 = s e 1 + qv 1, β 3 + β 4 = s e 2 + qv 2 wth qv 1, qv 2 beng as n (2.6), χev s a C cut-off functon wth χ ev = 1 for 0 θ v β v and χ ev = 0 for 3 2 β v θ v ω v for some β v (0, mn{ω v /2, π/8}]. The functons b e 3 and be 4, when extended by zero onto IR2+ := {(x e1, x e2 ); x e2 > 0}, le n H m 1 (IR 2+ ) and H m 2 (IR 2+ ), respectvely, wth m 1, m 2 as large as requred. Fnally, the supports of u ev 1 are subsets of the sector S ev = {(r v, θ v ); 0 r v 2τ v, 0 θ v 3 2 β v}. and uev 2 Remark 2.1 () The exponents γ e ( = 1, 2) of the edge and vertex-edge sngulartes n (2.5), (2.7), (2.8) satsfy γ e > 1 2. However, for our approxmaton analyss below t suffces to requre that γ e > 0 ( = 1, 2). Note that γe > 0 and λv > 1 2 ( = 1, 2) are the mnmum requrements to guarantee u X. () As mentoned above, the terms of hgher regularty (.e., wth greater sngularty exponents) are omtted n (2.5) (2.8). These terms are necessary to obtan the regular part u reg X k of decomposton (2.2) as smooth as requred. Ths can be done by consderng suffcently many (omtted) sngularty terms of each type. Remark 2.2 () By (2.5) (2.8) we conclude that any sngular vector feld u s n (2.4) (s = e, v, or ev) can be wrtten as u s = curl Γ w s + v s = curl Γ w s + (v s 1, v s 2) (2.9) wth correspondng (scalar) sngular functons w s, v1 s, vs 2 beng defned on the whole surface Γ. Note that these scalar functons are H 1/2 -regular on Γ,.e., w s H 1/2 (Γ), v s = (v s 1, v s 2) H 1/2 (Γ), s = e, v, ev. () It s mportant to observe that the functons w s, v1 s, vs 2 (s = e, v, or ev) n (2.9) are typcal scalar sngulartes nherent to solutons of the boundary ntegral equatons wth hypersngular operator for the Laplacan on Γ and wth possbly sngular rght-hand sde. Contnuous pecewse polynomal approxmatons of these scalar sngulartes n fractonal-order Sobolev spaces were analysed n [3, 1] and wll be used to prove the man result of the present paper. 3 The hp-verson of the BEM and the man result For the approxmate soluton of (2.1) we apply the hp-verson of the BEM on quas-unform trangulatons of Γ. Our BEM s based on Galerkn dscretsatons wth an approprate famly of 5

6 H(dv Γ, Γ)-conformng surface elements (here, H(dv Γ, Γ) := {u L 2 (Γ); dv Γ u L 2 (Γ)}). In what follows, h > 0 and p 1 wll always specfy the mesh parameter and a polynomal degree, respectvely. For any Ω R n we wll denote ρ Ω = sup{dam(b); B s a ball n Ω}. Let T = { h } be a famly of meshes h = {Γ ; = 1,..., J}. Each mesh s a partton of Γ nto trangular elements Γ such that Γ = J =1 Γ, and the ntersecton of any two elements Γ, Γ k ( k) s ether a common vertex, an entre sde, or empty. In the followng we always dentfy a face of the polyhedron Γ wth a subdoman of R 2. We denote h = dam(γ ) for any Γ h. Furthermore, any element Γ s the mage of the reference trangle K = {(ξ 1, ξ 2 ); 0 < ξ 1 < 1, 0 < ξ 2 < 1 ξ 1 } under an affne mappng T, more precsely Γ = T ( K), x = T (ξ) = B ξ + b, where B R 2 2, b R 2, x = (x 1, x 2 ) Γ, and ξ = (ξ 1, ξ 2 ) K. Then, the Jacoban matrx of T s B R 2 2, and ts determnant J := det(b ) satsfes the relaton J h 2. We consder a famly T of quas-unform shape-regular meshes h on Γ n the sense that there exst postve constants σ 1, σ 2 ndependent of h = max h such that for any Γ h and arbtrary h T there holds h σ 1 ρ Γ, h σ 2 h. (3.1) Whereas the mappng T ntroduced above s used to assocate scalar functons defned on the real element Γ and on the reference trangle K, the Pola transformaton s used to transform vector-valued functons between K and Γ : v = M (ˆv) = 1 J B ˆv T 1, ˆv = M 1 (v) = J B 1 v T. (3.2) Let us ntroduce the needed polynomal sets. By P p (K) we denote the set of polynomals of total degree p on the reference trangle K. We wll use two famles of H(dv Γ, Γ)-conformng surface elements: the Ravart-Thomas (RT) and Brezz-Douglas-Marn (BDM) elements. The correspondng spaces of degree p 1 on the reference trangle K wll be denoted as follows (see, e.g., [8, 26]): P RT p (K) = (P p 1 (K)) 2 ξp p 1 (K); P BDM p (K) = (P p (K)) 2. We wll use the unfed notaton P p (K) whch refers to ether the RT- or BDM-space on K for p 1. Accordngly, all results n ths paper are formulated n a unfed way and are vald for both the RT- and BDM-based boundary element spaces defned on the trangulaton of Γ. Note, however, that n some cases we wll need to provde arguments separately for each type of these boundary elements. Usng transformatons (3.2), we set X hp := {v H(dv Γ, Γ); M 1 (v Γ ) P p (K), = 1,..., J}. (3.3) 6

7 Note that only one type of surface elements (.e., ether the RT- or BDM-elements) s used n (3.3) for all trangles Γ. We wll denote by N = N(h, p) the dmenson of the dscrete space X hp. One has N h 2 for fxed p and N p 2 for fxed h. The hp-verson of the Galerkn BEM for the EFIE reads as: Fnd u hp X hp such that a(u hp, v) = f, v v X hp. (3.4) Due to the nfnte-dmensonal kernel of the dv Γ -operator, the blnear form a(, ) n (2.1) s not X-coercve, and, hence, the unque solvablty of (3.4) cannot be proved by standard arguments. However, the refned analyss n [7] shows that the unque BEM-soluton u hp X hp does exst, and t converges quas-optmally to the exact soluton u X of the EFIE as N(h, p). Ths result s formulated n the followng proposton. Proposton 3.1 [7, Theorem 1.2] There exsts N 0 1 such that for any f X and for arbtrary mesh-degree combnaton satsfyng N(h, p) N 0 the dscrete problem (3.4) s unquely solvable and the hp-verson of the Galerkn BEM converges quas-optmally,.e., u u hp X C nf{ u v X ; v X hp }. (3.5) Here, u X s the soluton of (2.1), u hp X hp s the soluton of (3.4), X denotes the norm n X, and C > 0 s a constant ndependent of h and p. The next theorem s the man result of ths paper. It states convergence rates of the hp-bem wth quas-unform meshes for the EFIE. These convergence rates (n both the mesh parameter h and the polynomal degree p) are gven explctly n terms of the sngularty exponents of the vector felds n (2.5) (2.8). Theorem 3.1 Let u X and u hp X hp be the solutons of (2.1) and (3.4), respectvely. We assume that the source functonal f n (2.1) s suffcently smooth such that representaton (2.2) (2.8) holds for the soluton u of (2.1). Let v 0 V, e 0 E(v 0 ) be a vertex-edge par such that mn{λ v /2, λv /2, γe 0 1, γe 0 2 } = mn v V,e E(v) mn {λv 1 + 1/2, λ v 2 + 1/2, γ e 1, γ e 2} wth λ v and γ e ( = 1, 2) beng as n (2.5) (2.8). Then for any h > 0 and for every p mn {λ v 0 1, λv 0 2, γe , γe } there holds ( ) v h mn{λ /2,λv /2,γe 0 1,γe 0 }( 2 u u hp X C p log p ) β+ν, (3.6) h where β = max {q v se , qv se } f λv 0 = γ e for = 1, 2, max {q v se , qv se 0 2 } f λv 0 1 = γe , λv 0 2 γe , max {q v se 0 1, qv se } f λv 0 1 γe , λv 0 2 = γe , max {q v se 0 1, qv se 0 2 } otherwse 7 (3.7)

8 wth the numbers s e 0 and q v 0 ( = 1, 2) gven n (2.5) and (2.6), respectvely, and ν = { 1 2 f p = mn {λ v 0 1, λv 0 2, γe 0 1 1/2, γe 0 2 1/2}, 0 otherwse. (3.8) If 1 p < mn {λ v 0 1, λv 0 2, γe , γe }, then for any h > 0 there holds u u hp X C h p+1/2. (3.9) Proof. Consderng enough sngularty terms n representaton (2.4) the functon u reg n (2.3) s as regular as needed. Then, due to the quas-optmal convergence (3.5) of the hp-bem wth quas-unform meshes, the asserton follows mmedately from the general approxmaton result gven n Theorem 4.1 below. Remark 3.1 We have only consdered meshes of trangular elements on Γ. If the meshes contan also shape-regular parallelogram elements (.e., affne mages of the reference square), then a pror error estmates of Theorem 3.1 reman vald only n the case of the RT-based BEM. Ths s because all the auxlary results needed for the proof are vald n ths case. Ths, however, s not true for the BDM-based BEM. In partcular, the arguments n the proofs of Proposton 3.1 above and Proposton 4.1 below rely essentally on the fact that the nvolved polynomal spaces form the exact curl dv sequence (on the reference element), a property the BDM-spaces fal to satsfy on the reference square (see, e.g., [18]). Remark 3.2 We have assumed that Γ s a polyhedral (closed) surface. However, all arguments n our proofs carry over only wth mnor modfcatons to the case of a pecewse plane orentable open surface Γ. Note that n ths case there are no restrctons needed on the wave number κ to ensure the unqueness of the soluton to the EFIE; the strongest edge sngulartes n (2.5) have the exponents γ e = 1 1/2 2 ( = 1, 2); the energy space for the EFIE s H 0 (dv Γ, Γ) and the boundary element space X hp conssts of H(dv Γ, Γ)-conformng polynomal vector felds wth normal components vanshng on Γ (see [6, 4] for detals). 4 General approxmaton result In ths secton we prove the followng general hp-approxmaton result for the vector feld u gven by formulas (2.2) (2.8) wth the sngularty exponents γ e and λ v ( = 1, 2) satsfyng the mnmum requrements to guarantee u X. Theorem 4.1 Let the vector feld u be gven by (2.2) (2.8) on Γ wth γ e 1, γe 2 > 0 and λv 1, λv 2 > 1 2 for each edge e and every vertex v. Also, let v 0 V, e 0 E(v 0 ) be such that mn{λ v /2, λv /2, γe 0 1, γe 0 2 } = mn v V,e E(v) mn {λv 1 + 1/2, λ v 2 + 1/2, γ e 1, γ e 2} 8

9 wth λ v and γ e ( = 1, 2) beng as n (2.5) (2.8). Then for any h > 0 and for every p mn {λ v 0 1, λv 0 2, γe , γe }, there exsts uhp X hp such that { u u hp X C max h mn{k,p}+1/2 p (k+1/2), ( ) v h mn{λ /2,λv /2,γe 0 1,γe 0 } 2 p log p ) β+ν, (4.1) h where β and ν are defned by (3.7) and (3.8), respectvely. If 1 p < mn {λ v 0 1, λv 0 2, γe , γe }, then for any h > 0 there exsts uhp X hp such that u u hp X C h mn{k,p}+1/2. (4.2) In order to prove ths theorem one needs to fnd dscrete vector felds belongng to X hp and approxmatng the smooth and sngular parts of u such that the approxmaton errors satsfy the upper bounds n (4.1) and (4.2). We start wth formulatng the followng hp-approxmaton result for regular vector felds on Γ. Ths result wll be used, n partcular, to approxmate the vector feld u reg X k n (2.2). Proposton 4.1 Let P hp : X X hp be the orthogonal proecton wth respect to the norm n X. If u X k wth k > 0, then u P hp u X C h mn{k,p}+1/2 p (k+1/2) u X k (4.3) wth a postve constant C ndependent of h, p, and u. The proof s gven n [4, Theorem 4.1] for the case of RT-spaces, and t carres over wthout essental modfcatons to the case of BDM-spaces on trangular elements (cf. Remark 3.1). Now, we wll study approxmatons of the sngular part u sng n representaton (2.2). By (2.4) (2.8) and due to the arguments n Remark 2.2 (), we conclude that u sng can be wrtten as u sng = curl Γ w + v = curl Γ w + (v 1, v 2 ), (4.4) where w H 1/2 (Γ) and v H 1/2 (Γ). Let us defne the followng dscrete space (of contnuous pecewse polynomals) over the mesh h : S hp (Γ) := {v C 0 (Γ); v Γ T P p (K), = 1,..., J}. We wll also need the followng functons of h and p: f (h, p) := h α p 2α (1 + log(p/h)) β +ν, (4.5) where = 1, 2, α := mn {λ v 0 + 1/2, γ e 0 }, (4.6) 9

10 β := { q v 0 + s e f λ v 0 = γ e 0 1 2, q v 0 + s e 0 otherwse, { 1 ν := 2 f p = α 1 2, 0 otherwse, and the numbers γ e 0, λv 0, se 0, qv 0 ( = 1, 2) are gven n (2.5) (2.8) for the vertex-edge par (v 0, e 0 ) ntroduced n the formulaton of Theorem 4.1. In the next two lemmas we formulate approxmaton results for the vector felds curl Γ w and v on the rght-hand sde of (4.4). The proofs are gven n Secton 5 below. Lemma 4.1 Let w H 1/2 (Γ) be the scalar sngular functon n representaton (4.4). Then for any h > 0 and p 1, there exsts w hp S hp (Γ) such that curl Γ w hp X hp and there holds { C f1 curl Γ w curl Γ w hp (h, p) f p α X, C h p+1/2 f 1 p < α 1 1 2, (4.9) where f 1 (h, p) and α 1 are defned by (4.5) and (4.6), respectvely. Lemma 4.2 Let v H 1/2 (Γ) be the sngular vector feld n representaton (4.4). Then for any h > 0 and p 1, there exsts v hp X hp satsfyng { C f2 v v hp (h, p) f p α X, C h p+1/2 f 1 p < α 2 1 2, (4.10) where f 2 (h, p) and α 2 are defned by (4.5) and (4.6), respectvely. Now we are able to prove Theorem 4.1. Proof of Theorem 4.1. For the regular vector feld u reg X k n (2.2) we use the orthogonal proecton P hp : X X hp wth respect to the norm n X to defne u hp reg := P hp u reg X hp. Then we have by Proposton 4.1 (4.7) (4.8) u reg u hp reg X C h mn{k,p}+1/2 p (k+1/2). (4.11) Snce the sngular part of decomposton (2.2) can be wrtten as n (4.4), we use approxmatons w hp S hp (Γ) and v hp X hp from the above two lemmas to defne u hp sng := curl Γ w hp + v hp X hp. Then, applyng the trangle nequalty, we obtan by (4.9) and (4.10) { C max u sng u hp {f1 sng (h, p), f 2 (h, p)} f p mn {α 1, α 2 } 1 2 X, C h p+1/2 f 1 p < mn {α 1, α 2 } 1 2. (4.12) Now, we set u hp := u hp reg + u hp sng X hp. Combnng estmates (4.11) and (4.12), applyng the trangle nequalty, and usng expressons (4.5) and (4.6) for the functons f (h, p) and the parameters α, respectvely, we prove the desred estmates n (4.1) and (4.2). 10

11 5 Proofs of techncal lemmas In ths secton we prove Lemmas 4.1 and Proof of Lemma 4.1 Recallng Remark 2.2 (), we use the results of [3, 1] to fnd the desred pecewse polynomal w hp S hp (Γ) such that the norm w w hp H 1/2 (Γ) s bounded as n (4.9) (see Theorems 5.1, 5.2 n [3] and Theorem 4.1 n [1]). Then, recallng the fact that the the operator curl Γ : H 1/2 (Γ) H 1/2 (Γ) s contnuous (see [12]), we derve the estmate n (4.9): curl Γ w curl Γ w hp X = curl Γ (w w hp ) 1/2 H (Γ) C w w hp H 1/2 (Γ) { C f1 (h, p) f p α 1 1 2, C h p+1/2 f 1 p < α It remans to prove that curl Γ w hp X hp. In fact, t s easy to check (see [6, p. 615]) that ( M 1 (curl Γ w hp Γ ) = curl (w hp Γ T ), where curl = ξ 2, ξ 1 ). Hence, recallng that w hp Γ T P p (K), we conclude that M 1 (curl Γ w hp Γ ) (P p 1 (K)) 2 P p (K) (ths s because (P p 1 (K)) 2 s a subset of both P BDM p (K) and P RT p (K)). Moreover, curl Γ w hp H(dv Γ, Γ), because dv Γ (curl Γ w hp ) 0 on Γ. Therefore, curl Γ w hp X hp, and the proof s fnshed. 5.2 Proof of Lemma 4.2 Let v = (v 1, v 2 ) be the second term n decomposton (4.4) of the sngular vector feld u sng. Agan, usng the results n [3, 1], we fnd contnuous pecewse polynomal approxmatons to the scalar components v 1, v 2 of v: for any h > 0 and every p 1, there exst v hp 1, vhp 2 S hp (Γ) such that for = 1, 2 there holds v v hp H 1/2 (Γ) { C f2 (h, p) f p α 2 1 2, C h p+1/2 f 1 p < α (5.1) wth postve constants C > 0 ndependent of h and p. Let v hp = (v hp 1, vhp 2 ). Then vhp H(dv Γ, Γ). Furthermore, n the case of BDM-elements we observe that for any Γ there holds M 1 (v hp Γ ) = J B 1 (v hp Γ ) T (P p (K)) 2 = P BDM (K). p 11

12 Therefore, v hp X hp n ths case. Moreover, snce v H 1/2 (Γ) and v hp S hp (Γ), we use estmate (5.1) and the contnuty of the operator dv Γ : H 1/2 (Γ) H 1/2 (Γ) to obtan v v hp X = v v hp 1/2 H (Γ) + dv Γ(v v hp ) H 1/2 (Γ) C v v hp H 1/2 (Γ) C 2 =1 v v hp H 1/2 (Γ) { C f2 (h, p) f p α 2 1 2, C h p+1/2 f 1 p < α (5.2) Unfortunately, n the case of RT-elements ths component-wse approxmaton of v does not work snce the dmenson of the RT-space on the reference trangle s smaller than the dmenson of the BDM-space, and, n general, M 1 (v hp Γ ) / P RT p (K), so that v hp = (v hp 1, vhp 2 ) / X hp. In ths case we follow the procedure descrbed n [3] (for the scalar case). More precsely, we use approprate h-scaled cut-off functons and represent the vector feld v as the sum of a sngular vector feld ϕ wth small support n the vcnty of the edges and a suffcently smooth vector feld ψ. In the rest of ths subsecton we wll demonstrate how ths procedure works n the vector case. We wll gve a concse step-by-step outlne of the procedure referrng frequently to [3] for partcular error estmates and other techncal detals. Step 1: decomposton of v. Let h 0 = (σ 1 σ 2 ) 1 h wth σ 1, σ 2 from (3.1). Usng the scaled cut-off functons χ e 2 (x e2/h 0 ) and χ v (r v /h 0 ) wth χ e 2 and χv from (2.5) and (2.6), respectvely, one can decompose v as follows (cf. [3, eqs. (5.4), (5.21), (6.4)]) v = ϕ + ψ, (5.3) where supp ϕ v V e E(v) (Āe Āv), ϕ H 1/2 (Γ), ψ X m (see (2.3) for the notaton), and ψ vanshes n small (h-dependent) neghbourhoods of each vertex and each edge of Γ. Here, A e s the unon of elements at one edge e,.e., Ā e := { Γ ; Γ e ø} (note that the endponts of e are not ncluded n e), A v s the unon of elements at a vertex v,.e., Ā v := { Γ ; v Γ }, m s suffcently large and depends on the parameters t 2 and m 2 specfed for the sngulartes u v and u ev 2, respectvely. Step 2: approxmaton of ϕ for p = 1. If p = 1, then one can approxmate ϕ by zero. One has ϕ hp 0 X hp, and, recallng the contnuty of the operator dv Γ, we derve ϕ ϕ hp X = ϕ H 1/2 (Γ) + dv Γ ϕ H 1/2 (Γ) C ϕ H 1/2 (Γ). (5.4) We wll obtan h-estmates for the norms of ϕ n H s (Γ) wth s = 0 and s = 1 2 +ε (for suffcently small ε > 0) by usng the fact that ϕ has a small support. Frst, we apply Lemma 3.1 of [2] and Lemma 3.5 of [3] to localse the norm ϕ H 1/2+ε(Γ) to the faces Γ f Γ and to the elements Γ Γ f, respectvely. Then, we use scalng on each element Γ supp ϕ (cf. [3, Lemma 3.1, eqs. (5.6), (5.11), (5.23), (6.7)]). As a result we have ϕ 2 H s (Γ) C ϕ 2 H s (Γ f ) f: Γ f Γ 12

13 C f: Γ f Γ : Γ Γ f ( ) h 2s ϕ 2 L 2 (Γ ) + ϕ 2 H s (Γ ) Ch 2(α 2+1/2 s) (1 + log(1/h)) 2 β 2 for s {0, 1/2 + ε}. Here, α 2 and β 2 are defned by (4.6) and (4.7), respectvely. between H 0 (Γ) and H 1/2+ε (Γ), we obtan by (5.4) Hence, usng the nterpolaton ϕ ϕ hp X Ch α 2 (1 + log(1/h)) β 2. (5.5) Step 3: approxmaton of ϕ for p 2. We approxmate each component ϕ of ϕ by a pecewse polynomal ϕ hp S h,p 1 (Γ) ( = 1, 2). Here we can use the results of [3, 1] for each type of sngularty: there exst ϕ hp S h,p 1 (Γ) such that for = 1, 2 there holds (cf. [3, eqs. (5.12), (5.13), (5.24), (6.9)]) ϕ ϕ hp H 1/2 (Γ) C hα 2 (p 1) 2α 2 (1 + log p 1 h ) β 2 C h α 2 p 2α 2 (1 + log p h ) β 2, (5.6) where α 2 and β 2 are the same as n (5.5). Settng ϕ hp = (ϕ hp 1, ϕhp 2 ) we observe that ϕhp H(dv Γ, Γ), and for any element Γ one has M 1 (ϕ hp Γ ) = J B 1 (ϕ hp Γ ) T (P p 1 (K)) 2 P RT (K). Hence, ϕ hp X hp. Moreover, snce ϕ H 1/2 (Γ) and ϕ hp S h,p 1 (Γ) for = 1, 2, we estmate by analogy wth (5.4) and usng (5.6): p ϕ ϕ hp X C ϕ ϕ hp H 1/2 (Γ) 2 C ϕ ϕ hp H 1/2 (Γ) C hα 2 p 2α 2 (1 + log p h ) β 2. (5.7) =1 Comparng (5.5) and (5.7) we observe that one can use estmate (5.7) also n the case p = 1. Step 4: approxmaton of ψ. Recallng that ψ X m we apply Proposton 4.1: there exsts ψ hp X hp such that ψ ψ hp X C h mn{k,p}+1/2 p (k+1/2) ψ X k, 0 k m. (5.8) We also recall that ψ vanshes n the h-neghbourhood of all edges and vertces of Γ. Therefore, havng explct expressons of the sngular vector feld v (and thus, of ψ) on each face Γ f Γ, one can derve upper bounds for the norms ψ H k (Γ f ) and dv Γ f ψ H k (Γ f ) n terms of the mesh sze h, the parameter k, and the sngularty exponents. For the norm of ψ ths can be done component-wse usng the same calculatons as n [3] (see, e.g., nequaltes (5.15) and (6.10) theren): ψ H k (Γ f ) C h α2+1/2 k (log(1/h)) β 2 + ν 2, α k m, (5.9) 13

14 where α 2 and β 2 are defned by (4.6) and (4.7), respectvely, ν 2 = 1 2 f k = α and ν 2 = 0 otherwse. To estmate the norm dv Γ f ψ H k (Γ f ) we observe that the operator dv Γ f reduces all sngularty exponents by one (whle preservng the structure of the correspondng sngularty). Then, usng smlar calculatons as ndcated above, we obtan dv Γ f ψ H k (Γ f ) C h α 2 1/2 k (log(1/h)) β 2 + ν 2, α k m, (5.10) where α 2, β 2 are the same as n (5.9), whereas ν 2 = 1 2 f k = α and ν 2 = 0 otherwse. By (5.8)-(5.10) we conclude that ψ ψ hp X C h mn{k,p}+α 2 k p (k+1/2) (log(1/h)) β 2 + ν 2, α k m (5.11) wth the same α 2, β 2, and ν 2 as n (5.10). Let p > 2α Snce m s large enough, we can select an nteger k satsfyng 2α < k mn {m, p}. Then mn {k, p} = k, and p (k+1/2) p 2α 2. If α < p 2α (.e., p s bounded), we choose an nteger k (α 2 1 2, p], and f p = α 2 1 2, then we take k = p = α In both these cases mn {k, p} = k, and p (k+1/2) C p 2α 2. Thus, for any p α 2 1 2, selectng k as ndcated above we fnd by (5.11) ψ ψ hp X C h α 2 p 2α 2 (log(1/h)) β 2 +ν 2 (5.12) wth α 2, β 2, and ν 2 beng defned by (4.6), (4.7), and (4.8), respectvely. Step 5: approxmaton of v = ϕ + ψ. Let us defne v hp := ϕ hp + ψ hp X hp, where ϕ hp and ψ hp are approxmatons constructed above (see Steps 2 4). Then combnng estmates (5.7) and (5.12) and usng the trangle nequalty we prove (4.10) n the case p α It remans to consder the case 1 p < α In ths case one does not need decomposton (5.3). Observe that for each face Γ f one has dv Γ f v H k (Γ f ) wth 1 k < α 1 2. Therefore, v X k wth 1 k < α 1 2, and applyng Proposton 4.1 we fnd vhp X hp satsfyng v v hp X C h mn {k,p}+1/2 v X k. Hence, selectng k [p, α 1 2 ) we arrve at the desred upper bound n (4.10), and the proof s fnshed. References [1] A. Bespalov, A note on the polynomal approxmaton of vertex sngulartes n the boundary element method n three dmensons, J. Integral Equatons Appl., 21 (2009), pp

15 [2] A. Bespalov and N. Heuer, The p-verson of the boundary element method for hypersngular operators on pecewse plane open surfaces, Numer. Math., 100 (2005), pp [3], The hp-verson of the boundary element method wth quas-unform meshes n three dmensons, ESAIM: Mathematcal Modellng and Numercal Analyss, 42 (2008), pp [4], The hp-bem wth quas-unform meshes for the electrc feld ntegral equaton on polyhedral surfaces: a pror error analyss, Appl. Numer. Math., 60 (2010), pp [5], The hp-verson of the boundary element method wth quas-unform meshes for weakly sngular operators on surfaces, IMA J. Numer. Anal., 30 (2010), pp [6], Natural p-bem for the electrc feld ntegral equaton on screens, IMA J. Numer. Anal., 30 (2010), pp [7] A. Bespalov, N. Heuer, and R. Hptmar, Convergence of the natural hp-bem for the electrc feld ntegral equaton on polyhedral surfaces, SIAM J. Numer. Anal., 48 (2010), pp [8] F. Brezz and M. Fortn, Mxed and Hybrd Fnte Element Methods, no. 15 n Sprnger Seres n Computatonal Mathematcs, Sprnger-Verlag, New York, [9] A. Buffa, Remarks on the dscretzaton of some noncoercve operator wth applcatons to heterogeneous Maxwell equatons, SIAM J. Numer. Anal., 43 (2005), pp [10] A. Buffa and S. H. Chrstansen, The electrc feld ntegral equaton on Lpschtz screens: defntons and numercal approxmaton, Numer. Math., 94 (2003), pp [11] A. Buffa and P. Carlet, Jr., On traces for functonal spaces related to Maxwell s equatons, Part I: An ntegraton by parts formula n Lpschtz polyhedra, Math. Methods Appl. Sc., 24 (2001), pp [12], On traces for functonal spaces related to Maxwell s equatons, Part II: Hodge decompostons on the boundary of Lpschtz polyhedra and applcatons, Math. Methods Appl. Sc., 24 (2001), pp [13] A. Buffa, M. Costabel, and C. Schwab, Boundary element methods for Maxwell s equatons on non-smooth domans, Numer. Mat., 92 (2002), pp [14] A. Buffa, M. Costabel, and D. Sheen, On traces for H(curl, Ω) n Lpschtz domans, J. Math. Anal. Appl., 276 (2002), pp [15] A. Buffa and R. Hptmar, Galerkn boundary element methods for electromagnetc scatterng, n Topcs n computatonal wave propagaton, vol. 31 of Lect. Notes Comput. Sc. Eng., Sprnger, Berln, 2003, pp

16 [16] A. Buffa, R. Hptmar, T. von Petersdorff, and C. Schwab, Boundary element methods for Maxwell transmsson problems n Lpschtz domans, Numer. Math., 95 (2003), pp [17] M. Costabel and M. Dauge, Sngulartes of electromagnetc felds n polyhedral domans, Arch. Ratonal Mech. Anal., 151 (2000), pp [18] L. Demkowcz, Polynomal exact sequences and proecton-based nterpolaton wth applcatons to Maxwell equatons, n Mxed Fnte Elements, Compatblty Condtons and Applcatons, D. Boff, F. Brezz, L. Demkowcz, R. Duran, R. Falk, and M. Fortn, eds., vol of Lecture Notes n Mathematcs, Sprnger-Verlag, Berln, 2008, pp [19] L. Demkowcz and I. Babuška, p nterpolaton error estmates for edge fnte elements of varable order n two dmensons, SIAM J. Numer. Anal., 41 (2003), pp [20] L. Demkowcz and A. Buffa, H 1, H(curl) and H(dv)-conformng proecton-based nterpolaton n three dmensons. Quas-optmal p-nterpolaton estmates, Comput. Methods Appl. Mech. Engrg., 194 (2005), pp [21] N. Heuer, M. Maschak, and E. P. Stephan, Exponental convergence of the hp-verson for the boundary element method on open surfaces, Numer. Math., 83 (1999), pp [22] R. Hptmar and C. Schwab, Natural boundary element methods for the electrc feld ntegral equaton on polyhedra, SIAM J. Numer. Anal., 40 (2002), pp [23] F. Leydecker, hp-verson of the boundary element method for electromagnetc problems, PhD thess, Insttut für Angewandte Mathematk, Unverstät Hannover, Hannover, Germany, Electroncally publshed on [24] R. C. MacCamy and E. P. Stephan, A boundary element method for an exteror problem for three-dmensonal Maxwell s equatons, Appl. Anal., 16 (1983), pp [25], Soluton procedures for three-dmensonal eddy current problems, J. Math. Anal. Appl., 101 (1984), pp [26] R. E. Roberts and J.-M. Thomas, Mxed and hybrd methods, n Handbook of Numercal Analyss. Vol. II, P. G. Carlet and J. L. Lons, eds., Amsterdam, 1991, North-Holland, pp [27] C. Schwab and M. Sur, The optmal p-verson approxmaton of sngulartes on polyhedra n the boundary element method, SIAM J. Numer. Anal., 33 (1996), pp [28] T. von Petersdorff, Randwertprobleme der Elastztätstheore für Polyeder Sngulartäten und Approxmaton mt Randelementmethoden, PhD thess, Technsche Hochschule Darmstadt, Germany,

17 [29] T. von Petersdorff and E. P. Stephan, Decompostons n edge and corner sngulartes for the soluton of the Drchlet problem of the Laplacan n a polyhedron, Math. Nachr., 149 (1990), pp

8.5 UNITARY AND HERMITIAN MATRICES. The conjugate transpose of a complex matrix A, denoted by A*, is given by

8.5 UNITARY AND HERMITIAN MATRICES. The conjugate transpose of a complex matrix A, denoted by A*, is given by 6 CHAPTER 8 COMPLEX VECTOR SPACES 5. Fnd the kernel of the lnear transformaton gven n Exercse 5. In Exercses 55 and 56, fnd the mage of v, for the ndcated composton, where and are gven by the followng

More information

1 Example 1: Axis-aligned rectangles

1 Example 1: Axis-aligned rectangles COS 511: Theoretcal Machne Learnng Lecturer: Rob Schapre Lecture # 6 Scrbe: Aaron Schld February 21, 2013 Last class, we dscussed an analogue for Occam s Razor for nfnte hypothess spaces that, n conjuncton

More information

Recurrence. 1 Definitions and main statements

Recurrence. 1 Definitions and main statements Recurrence 1 Defntons and man statements Let X n, n = 0, 1, 2,... be a MC wth the state space S = (1, 2,...), transton probabltes p j = P {X n+1 = j X n = }, and the transton matrx P = (p j ),j S def.

More information

Luby s Alg. for Maximal Independent Sets using Pairwise Independence

Luby s Alg. for Maximal Independent Sets using Pairwise Independence Lecture Notes for Randomzed Algorthms Luby s Alg. for Maxmal Independent Sets usng Parwse Independence Last Updated by Erc Vgoda on February, 006 8. Maxmal Independent Sets For a graph G = (V, E), an ndependent

More information

Ring structure of splines on triangulations

Ring structure of splines on triangulations www.oeaw.ac.at Rng structure of splnes on trangulatons N. Vllamzar RICAM-Report 2014-48 www.rcam.oeaw.ac.at RING STRUCTURE OF SPLINES ON TRIANGULATIONS NELLY VILLAMIZAR Introducton For a trangulated regon

More information

v a 1 b 1 i, a 2 b 2 i,..., a n b n i.

v a 1 b 1 i, a 2 b 2 i,..., a n b n i. SECTION 8.4 COMPLEX VECTOR SPACES AND INNER PRODUCTS 455 8.4 COMPLEX VECTOR SPACES AND INNER PRODUCTS All the vector spaces we have studed thus far n the text are real vector spaces snce the scalars are

More information

BERNSTEIN POLYNOMIALS

BERNSTEIN POLYNOMIALS On-Lne Geometrc Modelng Notes BERNSTEIN POLYNOMIALS Kenneth I. Joy Vsualzaton and Graphcs Research Group Department of Computer Scence Unversty of Calforna, Davs Overvew Polynomals are ncredbly useful

More information

Embedding lattices in the Kleene degrees

Embedding lattices in the Kleene degrees F U N D A M E N T A MATHEMATICAE 62 (999) Embeddng lattces n the Kleene degrees by Hsato M u r a k (Nagoya) Abstract. Under ZFC+CH, we prove that some lattces whose cardnaltes do not exceed ℵ can be embedded

More information

Generalizing the degree sequence problem

Generalizing the degree sequence problem Mddlebury College March 2009 Arzona State Unversty Dscrete Mathematcs Semnar The degree sequence problem Problem: Gven an nteger sequence d = (d 1,...,d n ) determne f there exsts a graph G wth d as ts

More information

Support Vector Machines

Support Vector Machines Support Vector Machnes Max Wellng Department of Computer Scence Unversty of Toronto 10 Kng s College Road Toronto, M5S 3G5 Canada wellng@cs.toronto.edu Abstract Ths s a note to explan support vector machnes.

More information

We are now ready to answer the question: What are the possible cardinalities for finite fields?

We are now ready to answer the question: What are the possible cardinalities for finite fields? Chapter 3 Fnte felds We have seen, n the prevous chapters, some examples of fnte felds. For example, the resdue class rng Z/pZ (when p s a prme) forms a feld wth p elements whch may be dentfed wth the

More information

n + d + q = 24 and.05n +.1d +.25q = 2 { n + d + q = 24 (3) n + 2d + 5q = 40 (2)

n + d + q = 24 and.05n +.1d +.25q = 2 { n + d + q = 24 (3) n + 2d + 5q = 40 (2) MATH 16T Exam 1 : Part I (In-Class) Solutons 1. (0 pts) A pggy bank contans 4 cons, all of whch are nckels (5 ), dmes (10 ) or quarters (5 ). The pggy bank also contans a con of each denomnaton. The total

More information

Calculation of Sampling Weights

Calculation of Sampling Weights Perre Foy Statstcs Canada 4 Calculaton of Samplng Weghts 4.1 OVERVIEW The basc sample desgn used n TIMSS Populatons 1 and 2 was a two-stage stratfed cluster desgn. 1 The frst stage conssted of a sample

More information

Production. 2. Y is closed A set is closed if it contains its boundary. We need this for the solution existence in the profit maximization problem.

Production. 2. Y is closed A set is closed if it contains its boundary. We need this for the solution existence in the profit maximization problem. Producer Theory Producton ASSUMPTION 2.1 Propertes of the Producton Set The producton set Y satsfes the followng propertes 1. Y s non-empty If Y s empty, we have nothng to talk about 2. Y s closed A set

More information

Extending Probabilistic Dynamic Epistemic Logic

Extending Probabilistic Dynamic Epistemic Logic Extendng Probablstc Dynamc Epstemc Logc Joshua Sack May 29, 2008 Probablty Space Defnton A probablty space s a tuple (S, A, µ), where 1 S s a set called the sample space. 2 A P(S) s a σ-algebra: a set

More information

What is Candidate Sampling

What is Candidate Sampling What s Canddate Samplng Say we have a multclass or mult label problem where each tranng example ( x, T ) conssts of a context x a small (mult)set of target classes T out of a large unverse L of possble

More information

Loop Parallelization

Loop Parallelization - - Loop Parallelzaton C-52 Complaton steps: nested loops operatng on arrays, sequentell executon of teraton space DECLARE B[..,..+] FOR I :=.. FOR J :=.. I B[I,J] := B[I-,J]+B[I-,J-] ED FOR ED FOR analyze

More information

where the coordinates are related to those in the old frame as follows.

where the coordinates are related to those in the old frame as follows. Chapter 2 - Cartesan Vectors and Tensors: Ther Algebra Defnton of a vector Examples of vectors Scalar multplcaton Addton of vectors coplanar vectors Unt vectors A bass of non-coplanar vectors Scalar product

More information

benefit is 2, paid if the policyholder dies within the year, and probability of death within the year is ).

benefit is 2, paid if the policyholder dies within the year, and probability of death within the year is ). REVIEW OF RISK MANAGEMENT CONCEPTS LOSS DISTRIBUTIONS AND INSURANCE Loss and nsurance: When someone s subject to the rsk of ncurrng a fnancal loss, the loss s generally modeled usng a random varable or

More information

Imperial College London

Imperial College London F. Fang 1, C.C. Pan 1, I.M. Navon 2, M.D. Pggott 1, G.J. Gorman 1, P.A. Allson 1 and A.J.H. Goddard 1 1 Appled Modellng and Computaton Group Department of Earth Scence and Engneerng Imperal College London,

More information

Product-Form Stationary Distributions for Deficiency Zero Chemical Reaction Networks

Product-Form Stationary Distributions for Deficiency Zero Chemical Reaction Networks Bulletn of Mathematcal Bology (21 DOI 1.17/s11538-1-9517-4 ORIGINAL ARTICLE Product-Form Statonary Dstrbutons for Defcency Zero Chemcal Reacton Networks Davd F. Anderson, Gheorghe Cracun, Thomas G. Kurtz

More information

Module 2 LOSSLESS IMAGE COMPRESSION SYSTEMS. Version 2 ECE IIT, Kharagpur

Module 2 LOSSLESS IMAGE COMPRESSION SYSTEMS. Version 2 ECE IIT, Kharagpur Module LOSSLESS IMAGE COMPRESSION SYSTEMS Lesson 3 Lossless Compresson: Huffman Codng Instructonal Objectves At the end of ths lesson, the students should be able to:. Defne and measure source entropy..

More information

+ + + - - This circuit than can be reduced to a planar circuit

+ + + - - This circuit than can be reduced to a planar circuit MeshCurrent Method The meshcurrent s analog of the nodeoltage method. We sole for a new set of arables, mesh currents, that automatcally satsfy KCLs. As such, meshcurrent method reduces crcut soluton to

More information

PERRON FROBENIUS THEOREM

PERRON FROBENIUS THEOREM PERRON FROBENIUS THEOREM R. CLARK ROBINSON Defnton. A n n matrx M wth real entres m, s called a stochastc matrx provded () all the entres m satsfy 0 m, () each of the columns sum to one, m = for all, ()

More information

REGULAR MULTILINEAR OPERATORS ON C(K) SPACES

REGULAR MULTILINEAR OPERATORS ON C(K) SPACES REGULAR MULTILINEAR OPERATORS ON C(K) SPACES FERNANDO BOMBAL AND IGNACIO VILLANUEVA Abstract. The purpose of ths paper s to characterze the class of regular contnuous multlnear operators on a product of

More information

NMT EE 589 & UNM ME 482/582 ROBOT ENGINEERING. Dr. Stephen Bruder NMT EE 589 & UNM ME 482/582

NMT EE 589 & UNM ME 482/582 ROBOT ENGINEERING. Dr. Stephen Bruder NMT EE 589 & UNM ME 482/582 NMT EE 589 & UNM ME 482/582 ROBOT ENGINEERING Dr. Stephen Bruder NMT EE 589 & UNM ME 482/582 7. Root Dynamcs 7.2 Intro to Root Dynamcs We now look at the forces requred to cause moton of the root.e. dynamcs!!

More information

Institute of Informatics, Faculty of Business and Management, Brno University of Technology,Czech Republic

Institute of Informatics, Faculty of Business and Management, Brno University of Technology,Czech Republic Lagrange Multplers as Quanttatve Indcators n Economcs Ivan Mezník Insttute of Informatcs, Faculty of Busness and Management, Brno Unversty of TechnologCzech Republc Abstract The quanttatve role of Lagrange

More information

The Greedy Method. Introduction. 0/1 Knapsack Problem

The Greedy Method. Introduction. 0/1 Knapsack Problem The Greedy Method Introducton We have completed data structures. We now are gong to look at algorthm desgn methods. Often we are lookng at optmzaton problems whose performance s exponental. For an optmzaton

More information

How Sets of Coherent Probabilities May Serve as Models for Degrees of Incoherence

How Sets of Coherent Probabilities May Serve as Models for Degrees of Incoherence 1 st Internatonal Symposum on Imprecse Probabltes and Ther Applcatons, Ghent, Belgum, 29 June 2 July 1999 How Sets of Coherent Probabltes May Serve as Models for Degrees of Incoherence Mar J. Schervsh

More information

Series Solutions of ODEs 2 the Frobenius method. The basic idea of the Frobenius method is to look for solutions of the form 3

Series Solutions of ODEs 2 the Frobenius method. The basic idea of the Frobenius method is to look for solutions of the form 3 Royal Holloway Unversty of London Department of Physs Seres Solutons of ODEs the Frobenus method Introduton to the Methodology The smple seres expanson method works for dfferental equatons whose solutons

More information

How To Understand The Results Of The German Meris Cloud And Water Vapour Product

How To Understand The Results Of The German Meris Cloud And Water Vapour Product Ttel: Project: Doc. No.: MERIS level 3 cloud and water vapour products MAPP MAPP-ATBD-ClWVL3 Issue: 1 Revson: 0 Date: 9.12.1998 Functon Name Organsaton Sgnature Date Author: Bennartz FUB Preusker FUB Schüller

More information

Energies of Network Nastsemble

Energies of Network Nastsemble Supplementary materal: Assessng the relevance of node features for network structure Gnestra Bancon, 1 Paolo Pn,, 3 and Matteo Marsl 1 1 The Abdus Salam Internatonal Center for Theoretcal Physcs, Strada

More information

The descriptive complexity of the family of Banach spaces with the π-property

The descriptive complexity of the family of Banach spaces with the π-property Arab. J. Math. (2015) 4:35 39 DOI 10.1007/s40065-014-0116-3 Araban Journal of Mathematcs Ghadeer Ghawadrah The descrptve complexty of the famly of Banach spaces wth the π-property Receved: 25 March 2014

More information

THE DISTRIBUTION OF LOAN PORTFOLIO VALUE * Oldrich Alfons Vasicek

THE DISTRIBUTION OF LOAN PORTFOLIO VALUE * Oldrich Alfons Vasicek HE DISRIBUION OF LOAN PORFOLIO VALUE * Oldrch Alfons Vascek he amount of captal necessary to support a portfolo of debt securtes depends on the probablty dstrbuton of the portfolo loss. Consder a portfolo

More information

PSYCHOLOGICAL RESEARCH (PYC 304-C) Lecture 12

PSYCHOLOGICAL RESEARCH (PYC 304-C) Lecture 12 14 The Ch-squared dstrbuton PSYCHOLOGICAL RESEARCH (PYC 304-C) Lecture 1 If a normal varable X, havng mean µ and varance σ, s standardsed, the new varable Z has a mean 0 and varance 1. When ths standardsed

More information

Project Networks With Mixed-Time Constraints

Project Networks With Mixed-Time Constraints Project Networs Wth Mxed-Tme Constrants L Caccetta and B Wattananon Western Australan Centre of Excellence n Industral Optmsaton (WACEIO) Curtn Unversty of Technology GPO Box U1987 Perth Western Australa

More information

NON-CONSTANT SUM RED-AND-BLACK GAMES WITH BET-DEPENDENT WIN PROBABILITY FUNCTION LAURA PONTIGGIA, University of the Sciences in Philadelphia

NON-CONSTANT SUM RED-AND-BLACK GAMES WITH BET-DEPENDENT WIN PROBABILITY FUNCTION LAURA PONTIGGIA, University of the Sciences in Philadelphia To appear n Journal o Appled Probablty June 2007 O-COSTAT SUM RED-AD-BLACK GAMES WITH BET-DEPEDET WI PROBABILITY FUCTIO LAURA POTIGGIA, Unversty o the Scences n Phladelpha Abstract In ths paper we nvestgate

More information

How To Calculate The Accountng Perod Of Nequalty

How To Calculate The Accountng Perod Of Nequalty Inequalty and The Accountng Perod Quentn Wodon and Shlomo Ytzha World Ban and Hebrew Unversty September Abstract Income nequalty typcally declnes wth the length of tme taen nto account for measurement.

More information

Boundary Element Domain Decomposition Methods Challenges and Applications

Boundary Element Domain Decomposition Methods Challenges and Applications Boundary Element Doman Decomposton Methods Challenges and Applcatons Olaf Stenbach Insttut für Numersche Mathematk Technsche Unverstät Graz SFB 404 Mehrfeldprobleme n der Kontnuumsmechank, Stuttgart n

More information

The OC Curve of Attribute Acceptance Plans

The OC Curve of Attribute Acceptance Plans The OC Curve of Attrbute Acceptance Plans The Operatng Characterstc (OC) curve descrbes the probablty of acceptng a lot as a functon of the lot s qualty. Fgure 1 shows a typcal OC Curve. 10 8 6 4 1 3 4

More information

1. Fundamentals of probability theory 2. Emergence of communication traffic 3. Stochastic & Markovian Processes (SP & MP)

1. Fundamentals of probability theory 2. Emergence of communication traffic 3. Stochastic & Markovian Processes (SP & MP) 6.3 / -- Communcaton Networks II (Görg) SS20 -- www.comnets.un-bremen.de Communcaton Networks II Contents. Fundamentals of probablty theory 2. Emergence of communcaton traffc 3. Stochastc & Markovan Processes

More information

COLLOQUIUM MATHEMATICUM

COLLOQUIUM MATHEMATICUM COLLOQUIUM MATHEMATICUM VOL. 74 997 NO. TRANSFERENCE THEORY ON HARDY AND SOBOLEV SPACES BY MARIA J. CARRO AND JAVIER SORIA (BARCELONA) We show that the transference method of Cofman and Wess can be extended

More information

FINITE HILBERT STABILITY OF (BI)CANONICAL CURVES

FINITE HILBERT STABILITY OF (BI)CANONICAL CURVES FINITE HILBERT STABILITY OF (BICANONICAL CURVES JAROD ALPER, MAKSYM FEDORCHUK, AND DAVID ISHII SMYTH* To Joe Harrs on hs sxteth brthday Abstract. We prove that a generc canoncally or bcanoncally embedded

More information

Linear Circuits Analysis. Superposition, Thevenin /Norton Equivalent circuits

Linear Circuits Analysis. Superposition, Thevenin /Norton Equivalent circuits Lnear Crcuts Analyss. Superposton, Theenn /Norton Equalent crcuts So far we hae explored tmendependent (resste) elements that are also lnear. A tmendependent elements s one for whch we can plot an / cure.

More information

DEFINING %COMPLETE IN MICROSOFT PROJECT

DEFINING %COMPLETE IN MICROSOFT PROJECT CelersSystems DEFINING %COMPLETE IN MICROSOFT PROJECT PREPARED BY James E Aksel, PMP, PMI-SP, MVP For Addtonal Informaton about Earned Value Management Systems and reportng, please contact: CelersSystems,

More information

The Noether Theorems: from Noether to Ševera

The Noether Theorems: from Noether to Ševera 14th Internatonal Summer School n Global Analyss and Mathematcal Physcs Satellte Meetng of the XVI Internatonal Congress on Mathematcal Physcs *** Lectures of Yvette Kosmann-Schwarzbach Centre de Mathématques

More information

Optimal resource capacity management for stochastic networks

Optimal resource capacity management for stochastic networks Submtted for publcaton. Optmal resource capacty management for stochastc networks A.B. Deker H. Mlton Stewart School of ISyE, Georga Insttute of Technology, Atlanta, GA 30332, ton.deker@sye.gatech.edu

More information

A Novel Methodology of Working Capital Management for Large. Public Constructions by Using Fuzzy S-curve Regression

A Novel Methodology of Working Capital Management for Large. Public Constructions by Using Fuzzy S-curve Regression Novel Methodology of Workng Captal Management for Large Publc Constructons by Usng Fuzzy S-curve Regresson Cheng-Wu Chen, Morrs H. L. Wang and Tng-Ya Hseh Department of Cvl Engneerng, Natonal Central Unversty,

More information

Joint Scheduling of Processing and Shuffle Phases in MapReduce Systems

Joint Scheduling of Processing and Shuffle Phases in MapReduce Systems Jont Schedulng of Processng and Shuffle Phases n MapReduce Systems Fangfe Chen, Mural Kodalam, T. V. Lakshman Department of Computer Scence and Engneerng, The Penn State Unversty Bell Laboratores, Alcatel-Lucent

More information

Finite difference method

Finite difference method grd ponts x = mesh sze = X NÜÆ Fnte dfference method Prncple: dervatves n the partal dfferental eqaton are approxmated by lnear combnatons of fncton vales at the grd ponts 1D: Ω = (0, X), (x ), = 0,1,...,

More information

Application of Quasi Monte Carlo methods and Global Sensitivity Analysis in finance

Application of Quasi Monte Carlo methods and Global Sensitivity Analysis in finance Applcaton of Quas Monte Carlo methods and Global Senstvty Analyss n fnance Serge Kucherenko, Nlay Shah Imperal College London, UK skucherenko@mperalacuk Daro Czraky Barclays Captal DaroCzraky@barclayscaptalcom

More information

POLYSA: A Polynomial Algorithm for Non-binary Constraint Satisfaction Problems with and

POLYSA: A Polynomial Algorithm for Non-binary Constraint Satisfaction Problems with and POLYSA: A Polynomal Algorthm for Non-bnary Constrant Satsfacton Problems wth and Mguel A. Saldo, Federco Barber Dpto. Sstemas Informátcos y Computacón Unversdad Poltécnca de Valenca, Camno de Vera s/n

More information

NPAR TESTS. One-Sample Chi-Square Test. Cell Specification. Observed Frequencies 1O i 6. Expected Frequencies 1EXP i 6

NPAR TESTS. One-Sample Chi-Square Test. Cell Specification. Observed Frequencies 1O i 6. Expected Frequencies 1EXP i 6 PAR TESTS If a WEIGHT varable s specfed, t s used to replcate a case as many tmes as ndcated by the weght value rounded to the nearest nteger. If the workspace requrements are exceeded and samplng has

More information

Can Auto Liability Insurance Purchases Signal Risk Attitude?

Can Auto Liability Insurance Purchases Signal Risk Attitude? Internatonal Journal of Busness and Economcs, 2011, Vol. 10, No. 2, 159-164 Can Auto Lablty Insurance Purchases Sgnal Rsk Atttude? Chu-Shu L Department of Internatonal Busness, Asa Unversty, Tawan Sheng-Chang

More information

THE METHOD OF LEAST SQUARES THE METHOD OF LEAST SQUARES

THE METHOD OF LEAST SQUARES THE METHOD OF LEAST SQUARES The goal: to measure (determne) an unknown quantty x (the value of a RV X) Realsaton: n results: y 1, y 2,..., y j,..., y n, (the measured values of Y 1, Y 2,..., Y j,..., Y n ) every result s encumbered

More information

An Alternative Way to Measure Private Equity Performance

An Alternative Way to Measure Private Equity Performance An Alternatve Way to Measure Prvate Equty Performance Peter Todd Parlux Investment Technology LLC Summary Internal Rate of Return (IRR) s probably the most common way to measure the performance of prvate

More information

Power-of-Two Policies for Single- Warehouse Multi-Retailer Inventory Systems with Order Frequency Discounts

Power-of-Two Policies for Single- Warehouse Multi-Retailer Inventory Systems with Order Frequency Discounts Power-of-wo Polces for Sngle- Warehouse Mult-Retaler Inventory Systems wth Order Frequency Dscounts José A. Ventura Pennsylvana State Unversty (USA) Yale. Herer echnon Israel Insttute of echnology (Israel)

More information

21 Vectors: The Cross Product & Torque

21 Vectors: The Cross Product & Torque 21 Vectors: The Cross Product & Torque Do not use our left hand when applng ether the rght-hand rule for the cross product of two vectors dscussed n ths chapter or the rght-hand rule for somethng curl

More information

A Probabilistic Theory of Coherence

A Probabilistic Theory of Coherence A Probablstc Theory of Coherence BRANDEN FITELSON. The Coherence Measure C Let E be a set of n propostons E,..., E n. We seek a probablstc measure C(E) of the degree of coherence of E. Intutvely, we want

More information

8 Algorithm for Binary Searching in Trees

8 Algorithm for Binary Searching in Trees 8 Algorthm for Bnary Searchng n Trees In ths secton we present our algorthm for bnary searchng n trees. A crucal observaton employed by the algorthm s that ths problem can be effcently solved when the

More information

An alternate point-wise scheme for Electric Field Integral Equations

An alternate point-wise scheme for Electric Field Integral Equations An alternate pont-wse scheme for Electrc Feld Integral Equatons Gabrele Rosat, Juan R. Mosg Laboratory of Electromagnetcs and Acoustcs (LEMA) Ecole Polytechnque Fédérale de Lausanne (EPFL), CH-05 Lausanne,

More information

Do Hidden Variables. Improve Quantum Mechanics?

Do Hidden Variables. Improve Quantum Mechanics? Radboud Unverstet Njmegen Do Hdden Varables Improve Quantum Mechancs? Bachelor Thess Author: Denns Hendrkx Begeleder: Prof. dr. Klaas Landsman Abstract Snce the dawn of quantum mechancs physcst have contemplated

More information

Conversion between the vector and raster data structures using Fuzzy Geographical Entities

Conversion between the vector and raster data structures using Fuzzy Geographical Entities Converson between the vector and raster data structures usng Fuzzy Geographcal Enttes Cdála Fonte Department of Mathematcs Faculty of Scences and Technology Unversty of Combra, Apartado 38, 3 454 Combra,

More information

L10: Linear discriminants analysis

L10: Linear discriminants analysis L0: Lnear dscrmnants analyss Lnear dscrmnant analyss, two classes Lnear dscrmnant analyss, C classes LDA vs. PCA Lmtatons of LDA Varants of LDA Other dmensonalty reducton methods CSCE 666 Pattern Analyss

More information

Stability, observer design and control of networks using Lyapunov methods

Stability, observer design and control of networks using Lyapunov methods Stablty, observer desgn and control of networks usng Lyapunov methods von Lars Naujok Dssertaton zur Erlangung des Grades enes Doktors der Naturwssenschaften - Dr. rer. nat. - Vorgelegt m Fachberech 3

More information

A hybrid global optimization algorithm based on parallel chaos optimization and outlook algorithm

A hybrid global optimization algorithm based on parallel chaos optimization and outlook algorithm Avalable onlne www.ocpr.com Journal of Chemcal and Pharmaceutcal Research, 2014, 6(7):1884-1889 Research Artcle ISSN : 0975-7384 CODEN(USA) : JCPRC5 A hybrd global optmzaton algorthm based on parallel

More information

Feature selection for intrusion detection. Slobodan Petrović NISlab, Gjøvik University College

Feature selection for intrusion detection. Slobodan Petrović NISlab, Gjøvik University College Feature selecton for ntruson detecton Slobodan Petrovć NISlab, Gjøvk Unversty College Contents The feature selecton problem Intruson detecton Traffc features relevant for IDS The CFS measure The mrmr measure

More information

Brigid Mullany, Ph.D University of North Carolina, Charlotte

Brigid Mullany, Ph.D University of North Carolina, Charlotte Evaluaton And Comparson Of The Dfferent Standards Used To Defne The Postonal Accuracy And Repeatablty Of Numercally Controlled Machnng Center Axes Brgd Mullany, Ph.D Unversty of North Carolna, Charlotte

More information

On Leonid Gurvits s proof for permanents

On Leonid Gurvits s proof for permanents On Leond Gurvts s proof for permanents Monque Laurent and Alexander Schrver Abstract We gve a concse exposton of the elegant proof gven recently by Leond Gurvts for several lower bounds on permanents.

More information

Logistic Regression. Lecture 4: More classifiers and classes. Logistic regression. Adaboost. Optimization. Multiple class classification

Logistic Regression. Lecture 4: More classifiers and classes. Logistic regression. Adaboost. Optimization. Multiple class classification Lecture 4: More classfers and classes C4B Machne Learnng Hlary 20 A. Zsserman Logstc regresson Loss functons revsted Adaboost Loss functons revsted Optmzaton Multple class classfcaton Logstc Regresson

More information

Fisher Markets and Convex Programs

Fisher Markets and Convex Programs Fsher Markets and Convex Programs Nkhl R. Devanur 1 Introducton Convex programmng dualty s usually stated n ts most general form, wth convex objectve functons and convex constrants. (The book by Boyd and

More information

On Lockett pairs and Lockett conjecture for π-soluble Fitting classes

On Lockett pairs and Lockett conjecture for π-soluble Fitting classes On Lockett pars and Lockett conjecture for π-soluble Fttng classes Lujn Zhu Department of Mathematcs, Yangzhou Unversty, Yangzhou 225002, P.R. Chna E-mal: ljzhu@yzu.edu.cn Nanyng Yang School of Mathematcs

More information

Chapter 4 ECONOMIC DISPATCH AND UNIT COMMITMENT

Chapter 4 ECONOMIC DISPATCH AND UNIT COMMITMENT Chapter 4 ECOOMIC DISATCH AD UIT COMMITMET ITRODUCTIO A power system has several power plants. Each power plant has several generatng unts. At any pont of tme, the total load n the system s met by the

More information

TENSOR GAUGE FIELDS OF DEGREE THREE

TENSOR GAUGE FIELDS OF DEGREE THREE TENSOR GAUGE FIELDS OF DEGREE THREE E.M. CIOROIANU Department of Physcs, Unversty of Craova, A. I. Cuza 13, 2585, Craova, Romana, EU E-mal: manache@central.ucv.ro Receved February 2, 213 Startng from a

More information

"Research Note" APPLICATION OF CHARGE SIMULATION METHOD TO ELECTRIC FIELD CALCULATION IN THE POWER CABLES *

Research Note APPLICATION OF CHARGE SIMULATION METHOD TO ELECTRIC FIELD CALCULATION IN THE POWER CABLES * Iranan Journal of Scence & Technology, Transacton B, Engneerng, ol. 30, No. B6, 789-794 rnted n The Islamc Republc of Iran, 006 Shraz Unversty "Research Note" ALICATION OF CHARGE SIMULATION METHOD TO ELECTRIC

More information

Global stability of Cohen-Grossberg neural network with both time-varying and continuous distributed delays

Global stability of Cohen-Grossberg neural network with both time-varying and continuous distributed delays Global stablty of Cohen-Grossberg neural network wth both tme-varyng and contnuous dstrbuted delays José J. Olvera Departamento de Matemátca e Aplcações and CMAT, Escola de Cêncas, Unversdade do Mnho,

More information

Subexponential time relations in the class group of large degree number fields

Subexponential time relations in the class group of large degree number fields Subexponental tme relatons n the class group of large degree number felds Jean-Franços Basse Unversty of Calgary 2500 Unversty Drve NW Calgary, Alberta, Canada T2N 1N4 Abstract Hafner and McCurley descrbed

More information

Mean Molecular Weight

Mean Molecular Weight Mean Molecular Weght The thermodynamc relatons between P, ρ, and T, as well as the calculaton of stellar opacty requres knowledge of the system s mean molecular weght defned as the mass per unt mole of

More information

Solving Factored MDPs with Continuous and Discrete Variables

Solving Factored MDPs with Continuous and Discrete Variables Solvng Factored MPs wth Contnuous and screte Varables Carlos Guestrn Berkeley Research Center Intel Corporaton Mlos Hauskrecht epartment of Computer Scence Unversty of Pttsburgh Branslav Kveton Intellgent

More information

Time Domain simulation of PD Propagation in XLPE Cables Considering Frequency Dependent Parameters

Time Domain simulation of PD Propagation in XLPE Cables Considering Frequency Dependent Parameters Internatonal Journal of Smart Grd and Clean Energy Tme Doman smulaton of PD Propagaton n XLPE Cables Consderng Frequency Dependent Parameters We Zhang a, Jan He b, Ln Tan b, Xuejun Lv b, Hong-Je L a *

More information

The Development of Web Log Mining Based on Improve-K-Means Clustering Analysis

The Development of Web Log Mining Based on Improve-K-Means Clustering Analysis The Development of Web Log Mnng Based on Improve-K-Means Clusterng Analyss TngZhong Wang * College of Informaton Technology, Luoyang Normal Unversty, Luoyang, 471022, Chna wangtngzhong2@sna.cn Abstract.

More information

HÜCKEL MOLECULAR ORBITAL THEORY

HÜCKEL MOLECULAR ORBITAL THEORY 1 HÜCKEL MOLECULAR ORBITAL THEORY In general, the vast maorty polyatomc molecules can be thought of as consstng of a collecton of two electron bonds between pars of atoms. So the qualtatve pcture of σ

More information

Risk-based Fatigue Estimate of Deep Water Risers -- Course Project for EM388F: Fracture Mechanics, Spring 2008

Risk-based Fatigue Estimate of Deep Water Risers -- Course Project for EM388F: Fracture Mechanics, Spring 2008 Rsk-based Fatgue Estmate of Deep Water Rsers -- Course Project for EM388F: Fracture Mechancs, Sprng 2008 Chen Sh Department of Cvl, Archtectural, and Envronmental Engneerng The Unversty of Texas at Austn

More information

On the Optimal Control of a Cascade of Hydro-Electric Power Stations

On the Optimal Control of a Cascade of Hydro-Electric Power Stations On the Optmal Control of a Cascade of Hydro-Electrc Power Statons M.C.M. Guedes a, A.F. Rbero a, G.V. Smrnov b and S. Vlela c a Department of Mathematcs, School of Scences, Unversty of Porto, Portugal;

More information

Availability-Based Path Selection and Network Vulnerability Assessment

Availability-Based Path Selection and Network Vulnerability Assessment Avalablty-Based Path Selecton and Network Vulnerablty Assessment Song Yang, Stojan Trajanovsk and Fernando A. Kupers Delft Unversty of Technology, The Netherlands {S.Yang, S.Trajanovsk, F.A.Kupers}@tudelft.nl

More information

Quantization Effects in Digital Filters

Quantization Effects in Digital Filters Quantzaton Effects n Dgtal Flters Dstrbuton of Truncaton Errors In two's complement representaton an exact number would have nfntely many bts (n general). When we lmt the number of bts to some fnte value

More information

CHOLESTEROL REFERENCE METHOD LABORATORY NETWORK. Sample Stability Protocol

CHOLESTEROL REFERENCE METHOD LABORATORY NETWORK. Sample Stability Protocol CHOLESTEROL REFERENCE METHOD LABORATORY NETWORK Sample Stablty Protocol Background The Cholesterol Reference Method Laboratory Network (CRMLN) developed certfcaton protocols for total cholesterol, HDL

More information

SIMPLE LINEAR CORRELATION

SIMPLE LINEAR CORRELATION SIMPLE LINEAR CORRELATION Smple lnear correlaton s a measure of the degree to whch two varables vary together, or a measure of the ntensty of the assocaton between two varables. Correlaton often s abused.

More information

Section 2 Introduction to Statistical Mechanics

Section 2 Introduction to Statistical Mechanics Secton 2 Introducton to Statstcal Mechancs 2.1 Introducng entropy 2.1.1 Boltzmann s formula A very mportant thermodynamc concept s that of entropy S. Entropy s a functon of state, lke the nternal energy.

More information

Nonbinary Quantum Error-Correcting Codes from Algebraic Curves

Nonbinary Quantum Error-Correcting Codes from Algebraic Curves Nonbnary Quantum Error-Correctng Codes from Algebrac Curves Jon-Lark Km and Judy Walker Department of Mathematcs Unversty of Nebraska-Lncoln, Lncoln, NE 68588-0130 USA e-mal: {jlkm, jwalker}@math.unl.edu

More information

Compiling for Parallelism & Locality. Dependence Testing in General. Algorithms for Solving the Dependence Problem. Dependence Testing

Compiling for Parallelism & Locality. Dependence Testing in General. Algorithms for Solving the Dependence Problem. Dependence Testing Complng for Parallelsm & Localty Dependence Testng n General Assgnments Deadlne for proect 4 extended to Dec 1 Last tme Data dependences and loops Today Fnsh data dependence analyss for loops General code

More information

GRAVITY DATA VALIDATION AND OUTLIER DETECTION USING L 1 -NORM

GRAVITY DATA VALIDATION AND OUTLIER DETECTION USING L 1 -NORM GRAVITY DATA VALIDATION AND OUTLIER DETECTION USING L 1 -NORM BARRIOT Jean-Perre, SARRAILH Mchel BGI/CNES 18.av.E.Beln 31401 TOULOUSE Cedex 4 (France) Emal: jean-perre.barrot@cnes.fr 1/Introducton The

More information

Nordea G10 Alpha Carry Index

Nordea G10 Alpha Carry Index Nordea G10 Alpha Carry Index Index Rules v1.1 Verson as of 10/10/2013 1 (6) Page 1 Index Descrpton The G10 Alpha Carry Index, the Index, follows the development of a rule based strategy whch nvests and

More information

Implementation of Deutsch's Algorithm Using Mathcad

Implementation of Deutsch's Algorithm Using Mathcad Implementaton of Deutsch's Algorthm Usng Mathcad Frank Roux The followng s a Mathcad mplementaton of Davd Deutsch's quantum computer prototype as presented on pages - n "Machnes, Logc and Quantum Physcs"

More information

Rotation Kinematics, Moment of Inertia, and Torque

Rotation Kinematics, Moment of Inertia, and Torque Rotaton Knematcs, Moment of Inerta, and Torque Mathematcally, rotaton of a rgd body about a fxed axs s analogous to a lnear moton n one dmenson. Although the physcal quanttes nvolved n rotaton are qute

More information

Sngle Snk Buy at Bulk Problem and the Access Network

Sngle Snk Buy at Bulk Problem and the Access Network A Constant Factor Approxmaton for the Sngle Snk Edge Installaton Problem Sudpto Guha Adam Meyerson Kamesh Munagala Abstract We present the frst constant approxmaton to the sngle snk buy-at-bulk network

More information

1.1 The University may award Higher Doctorate degrees as specified from time-to-time in UPR AS11 1.

1.1 The University may award Higher Doctorate degrees as specified from time-to-time in UPR AS11 1. HIGHER DOCTORATE DEGREES SUMMARY OF PRINCIPAL CHANGES General changes None Secton 3.2 Refer to text (Amendments to verson 03.0, UPR AS02 are shown n talcs.) 1 INTRODUCTION 1.1 The Unversty may award Hgher

More information

Downlink Power Allocation for Multi-class. Wireless Systems

Downlink Power Allocation for Multi-class. Wireless Systems Downlnk Power Allocaton for Mult-class 1 Wreless Systems Jang-Won Lee, Rav R. Mazumdar, and Ness B. Shroff School of Electrcal and Computer Engneerng Purdue Unversty West Lafayette, IN 47907, USA {lee46,

More information

Formulating & Solving Integer Problems Chapter 11 289

Formulating & Solving Integer Problems Chapter 11 289 Formulatng & Solvng Integer Problems Chapter 11 289 The Optonal Stop TSP If we drop the requrement that every stop must be vsted, we then get the optonal stop TSP. Ths mght correspond to a ob sequencng

More information

LECTURES on COMPUTATIONAL NUMERICAL ANALYSIS of PARTIAL DIFFERENTIAL EQUATIONS

LECTURES on COMPUTATIONAL NUMERICAL ANALYSIS of PARTIAL DIFFERENTIAL EQUATIONS LECTURES on COMPUTATIONAL NUMERICAL ANALYSIS of PARTIAL DIFFERENTIAL EQUATIONS J. M. McDonough Departments of Mechancal Engneerng and Mathematcs Unversty of Kentucky c 1985, 00, 008 Contents 1 Introducton

More information