DECOMPOSITION OF SPACES OF DISTRIBUTIONS INDUCED BY HERMITE EXPANSIONS

Size: px
Start display at page:

Download "DECOMPOSITION OF SPACES OF DISTRIBUTIONS INDUCED BY HERMITE EXPANSIONS"

Transcription

1 DECOMPOSITION OF SPACES OF DISTRIBUTIONS INDUCED BY HERMITE EXPANSIONS PENCHO PETRUSHEV AND YUAN XU Abstract. Decomoston systems wth radly decayng elements needlets) based on Hermte functons are ntroduced and exlored. It s roved that the Trebel-Lzorkn and Besov saces on R d nduced by Hermte exansons can be characterzed n terms of the needlet coeffcents. It s also shown that the Hermte Trebel-Lzorkn and Besov saces are, n general, dfferent from the resectve classcal saces. 1. Introducton The urose of ths aer s to extend the fundamental results of Frazer and Jawerth [4, 5] on the ϕ-transform to the case of Hermte exansons on R d. In the srt of [4, 5] we wll construct a ar of dual frames n terms of Hermte functons and use them to characterze the Hermte-Trebel-Lzorkn and Hermte- Besov saces. Let {h n } n=0 be the L 2 R d ) normalzed unvarate Hermte functons see 2.1). The d-dmensonal Hermte functons are defned by H α x) := h α1 x 1 ) h αd x d ). Then the kernel of the orthogonal roector of L 2 onto W n := san {H α : α = n} s gven by H n x, y) := α =n H αx)h α y). Our constructon of Hermte frames hnges on the fundamental fact that for comactly suorted C functons â the kernels Λ n x, y) := â n )H x, y) decay radly away from the man dagonal n R d. Ths fact was establshed n [3] for dmenson d = 1 and n [1] n general. We obtan a more recse estmate n Theorem 2.2 below. We utlze kernels of such knd for the constructon of a ar of dual frames {ϕ ξ } ξ X, {ψ ξ } ξ X, where X s a multlevel ndex set. The frame elements have almost exonental localzaton see 3.11)) whch romted us to call them needlets. The needlet systems of ths artcle can be vewed as an analogue of the ϕ-transform of Frazer and Jawerth [4, 5]. Frames of the same nature n the case d = 1 have been revously ntroduced n [3]. Our rmary goal s to utlze needlets to the characterzaton of the Trebel- Lzorkn and Besov saces n the context of Hermte exansons. To be more secfc, assume that â C, su â [1/4, 4], and â > c on [1/3, 3], and defne 1.1) Φ 0 := H 0 and Φ := ν=0 ν ) â 4 1 H ν, Mathematcs Subect Classfcaton. 42B35, 42C15. Key words and hrases. Localzed kernels, frames, Hermte olynomals, Trebel-Lzorkn saces, Besov saces. The second author was artally suorted by the NSF under Grant DMS

2 2 PENCHO PETRUSHEV AND YUAN XU Then for all arorate ndces we defne the Hermte-Trebel-Lzorkn sace F H) as the set of all temered dstrbutons f such that F f F := 2 α Φ f ) ) q) 1/q <, = where Φ fx) := f, Φx, ) see Defnton 4.1). We defne the Hermte-Besov saces B = B H) by the norm ) q ) 1/q. f B := 2 α Φ f One normally uses bnary dlatons n 1.1) see e.g. [17, 10.3] and also [1, 2]). We dlate â by factors of 4 nstead snce then the Hermte F- and B-saces embed ust as the classcal F- and B-saces. Our man results assert that the Hermte-Trebel-Lzorkn and Hermte Besov saces can be characterzed n terms of resectve sequence norms of the needlet coeffcents of the dstrbutons Theorems 4.5, 5.7). Furthermore, we use these results to show that the Hermte-F- and B-saces of essentally ostve smoothness are dfferent from the resectve classcal F- and B-saces on R d. Our develoment here s a art of a bgger roect for needlet characterzaton of Trebel-Lzorkn and Besov saces on nonclasscal domans such as the unt shere [10], the nterval wth Jacob weghts [7], and the unt ball [8]. The rest of the aer s organzed as follows: Secton 2 contans some background materal. The needlets are ntroduced n 3. In 4 the Hermte-Trebel-Lzorln saces are defned and characterzed va needlets. The Hermte-Besov saces are ntroduced and characterzed n 5. Secton 6 contans the roofs of a number of lemmas and theorems from 2-5. Some useful notaton: f := f L R ); for a measurable set E R d, E d denotes the Lebesgue measure of E and 1 E s the characterstc functon of E. Also, for x R d, x s the Eucldean norm of x, x := max 1 d x, and dx, E) := nf y E x y s the l dstance of x from E R d. Postve constants are denoted by c, c 1,... and they may vary at every occurrence; A B means c 1 A B c 2 A. 2. Prelmnares 2.1. Localzed kernels nduced by Hermte functons. We begn wth a revew of some basc roertes of Hermte olynomals and functons. For background nformaton we refer the reader to [16].) The Hermte olynomals are defned by H n t) = 1) n e t2 d ) n e t2), n = 0, 1,.... dt These olynomals are orthogonal wth resect to e t2 on R. We wll denote the L 2 -normalzed Hermte functons by One has R h n t) := 2 n n! π ) 1/2 Hn t)e t2 /2. h n t)h m t)dt = 2 n n! π ) 1 R H n t)h m t)e t2 dt = δ n,m.

3 DECOMPOSITION OF SPACES INDUCED BY HERMITE EXPENSIONS 3 As s well known the Hermte functons form an orthonormal bass for L 2 R). As already mentoned, the d-dmensonal Hermte functons H α are defned by 2.1) H α x) := h α1 x 1 ) h αd x d ), α = α 1,..., α d ). Evdently e x 2 /2 H α x) s a olynomal of degree α := α α d. The Hermte functons form an orthonormal bass for L 2 R d ). Moreover, H α are egenfunctons of the Hermte oerator D := + x 2 and 2.2) DH α = 2 α + d)h α, where s the Lalacan. The oerator D can be wrtten n the form 2.3) D = 1 2 d A A + A A ), where A = + x, A = + x. x x =1 Let e denote the th coordnate vector n R d. Then the oerators A and A satsfy 2.4) A H α = 2α + 2) 1 2 Hα+e and A H α = 2α ) 1 2 Hα e. Combnng these two relatons shows that {H α } satsfy the recurrence relaton 2.5) x H α x) = and also 2.6) x H α x) = ) 1 α +1 2 ) 1 α H α+e x) + α 2 2 H α+e x) + α 2 ) 1 2 H α e x) ) 1 2 H α e x). Let W n := san {H α : α = n} and V n := n W. The kernels of orthogonal roectors on W n and V n are gven by 2.7) H n x, y) := n H α x)h α y) and K n x, y) := H x, y), α =n resectvely. An mortant role wll be layed by oerators whose kernels are obtaned by smoothng out the coeffcents of the kernel K n by samlng a comactly suorted C functon â. For our uroses we wll be consderng smoothng functons â that satsfy: Defnton 2.1. A functon â C [0, ) s sad to be admssble of tye a) f su â [0, 1 + v] v > 0) and ât) = 1 on [0, 1], and of tye b) f su â [u, 1 + v], where 0 < u < 1, v > 0. For an admssble functon â we consder the kernel 2.8) Λ n x, y) := â H x, y). n) It wll be crtcal for our further develoment that the kernels Λ n x, y) and ther dervatves decay radly away from the man dagonal y = x n R d R d :

4 4 PENCHO PETRUSHEV AND YUAN XU Theorem 2.2. Suose â s admssble n the sense of Defnton 2.1 and let α N d 0. Then for any k 1 there exsts a constant c k deendng only on k, α, d, and â such that 2.9) α x α Λ n α 2 [K n+[vn]+ α +k x, x)] 1 2 [K n+[vn]+k y, y)] 1 2 nx, y) c k. 1 + n 1 2 x y ) k Here the deendence of c k on â s of the form c k = ck, α, u, d) max 0 l k â l). We relegate the somewhat lengthy roof of ths theorem to 6.1. The functon ) λ n x) := K n x, x) s termed Chrstoffel functon and t s known see e.g. [9]) to have the followng asymtotc n dmenson d = 1: { 2.11) λ n x) n max 1/2 n 2/3, 1 x }) 1/2 2n unformly for n 1 and x 2n1 + c n 2/3 ), where c > 0 s any fxed constant. Consequently, for d = 1 we have { 2.12) K n x, x) n max 1/2 n 2/3, 1 x }) 1/2, x 2n1 + c n 2/3 ). 2n For d 2 one has see [16,. 70]) 2.13) H n x, x) cn d/2 1, x R d. Ths along wth 2.7) leads to 2.14) K n x, x) cn d/2, x R d, d 1. On the other hand, t s well known that see e.g. [16,. 26]) 2.15) h n x) ce γx2, x 4n + 2) 1/2, γ > 0, and h n cn 1/12, whch readly mly 2.16) K n x, x) ce γ x 2, f x := max 1 d x 4n + 2) 1/2, where γ > 0 deends only on d. Now, combnng 2.9) wth 2.14) and 2.16) settng γ := γ /2) we arrve at Corollary 2.3. Under the hyothess of Theorem 2.2 we have 2.17) α x α Λ n α +d 2 nx, y) c k 1 + n 1 2 x y ), x k Rd, 2.18) and 2.19) α x α Λ e γ x 2 nx, y) c k 1 + n 1 2 x y ), f x k 4n+[vn]+ α +k)+2) 1/2, α x α Λ e γ y 2 nx, y) c k 1 + n 1 2 x y ), f y k 4n + [vn] + k) + 2) 1/2.

5 DECOMPOSITION OF SPACES INDUCED BY HERMITE EXPENSIONS 5 Note that an estmate smlar to 2.17) s roved n [3] when d = 1 and α = 0 and n the general case n [1]. Estmate 2.9) s new. We now turn to a lower bound estmate. Theorem 2.4. Let â be admssble n the sense of Defnton 2.1 and ât) > c > 0 on [1, 1 + τ], τ > 0. Then for any ε > 0, R d Λ n x, y) 2 dy c n d/2 for x 1 ε) 21 + τ)n, where c > 0 deends only on τ, ε, c, and d. Ths theorem rovdes a lower bound for the range where estmate 2.17) wth α = 0) s shar. To ndcate the deendence of K n on d, we wrte K n,d = K n. Theorem 2.4 s an mmedate consequence of 2.12) and the followng lemma. Lemma 2.5. If 0 < λ < 1, 0 < ρ < 1, and d 1, then there exsts a constant c > 0 such that for n 2/λ n 2.20) Hmx, 2 x) cn d 1 2 K[ρn],1 t, t) f t := x 2 2n + 1. m=[1 λ)n] The roof of ths lemma s gven n Norm relaton. For future use we gve here the well known relaton between dfferent norms of functons from V n see e.g. [9]): For 0 <, q 2.21) g cn d 2 1/q 1/ g q for g V n, wth c > 0 deendng only on, q, and d. Ths estmate can be roved by means of the kernels from 2.8) wth â admssble of tye a) Cubature formula. In order to defne our frame elements, we need a cubature formula exact for roducts fg wth f, g V n. Such a formula, however, s readly avalable usng the Gaussan quadrature formula. Prooston 2.6. [15] Denote by t ν,n, ν = 1, 2,..., n, the zeros of the Hermte olynomal H n t). The Gaussan quadrature formula n 2.22) ft)e t2 dt w ν,n ft ν,n ), w ν,n := λ n t ν,n )e t2 ν,n, R ν=1 s exact for all olynomals of degree 2n 1. Here λ n ) s the Chrstoffel functon defned n 2.10). The roduct nature of e x 2 enables us to obtan the desred cubature formula on R d rght away. Prooston 2.7. Let ξ α,n := t α1,n,..., t αd,n) and λ α,n := d ν=1 λ nt αν,n). The cubature formula n n 2.23) fx)gx)dx λ α,n fξ α,n )gξ α,n ) R d α 1=1 α d =1 s exact for all f V l, g V m wth l + m 2n 1.

6 6 PENCHO PETRUSHEV AND YUAN XU We next record some well known roertes of the zeros of Hermte olynomals. Suose {ξ ν } are the zeros of H n t) wth n even) ordered so that 2.24) ξ n 2 < < ξ 1 < 0 < ξ 1 < < ξ n 2, ξ ν = ξ ν. From [9] we have ξ n 2n + 1 n 1/6 and unformly for ν n/2 1 2 { 2.25) ξ ν+1 ξ ν 1 n max 1/2 n 2/3, 1 ξ }) 1/2 ν. 2n Consequently, on account of 2.11) 2.26) λ n ξ ν ) ξ ν 1 ξ ν+1, ν n/2 1 ξ 0 := 0). By [15, )] 2.27) πν 1 2 ) 2n + 1) < ξ 1/2 ν < 4ν + 3, ν = 1,..., n/2. 2n + 1) 1/2 From ths and 2.25) we have, for any ε > 0, 2.28) ξ ν+1 ξ ν 1 n 1/2 f ν 1/2 ε)n, and 2.29) c 1 n 1/2 ξ ν ξ ν 1 c 2 n 1/6 f 1/2 ε)n < ν n/2. Here the constants deend on ε. It also follows by 2.25) that 2.30) ξ ν+1 ξ ν 1 ξ ν ξ ν 2, n/2 + 2 ν n/2 1. For the constructon of our frames n Secton 3 we need the cubature formulae from Prooston 2.7 wth 2.31) n = 2N, where N := [1 + 11δ)4/π) 2 4 ] + 3 and 0 < δ < 1/37 s an arbtrary but fxed) constant. Gven 0, let as above ξ ν, ν = ±1,..., ±N, be the zeros of H 2N t). Let X be the set of all nodes of cubature 2.23) wth n = 2N,.e. X s the set of all onts ξ α := ξ α1,..., ξ αd ), where 0 < α ν N. Also, for ξ = ξ α we denote brefly λ ξ := λ α,n. Note that #X = 2N ) d 4 d. An mmedate consequence of Prooston 2.7 s the followng Corollary 2.8. The cubature formula 2.32) fx)gx)dx λ ξ fξ)gξ), λ ξ := R d ξ X s exact for all f V l, g V m wth l + m 4N 1. d λ 2N ξ αν ), For later use we now ntroduce tles {R ξ } nduced by the onts of X. Set I 1 := [0, ξ 1 + ξ 2 )/2], I 1 := I 1, ν=1 I ν := [ξ ν 1 + ξ ν )/2, ξ ν + ξ ν+1 )/2], ν = ±2,..., ±N 1, and I N := [ξ N 1 + ξ N )/2, ξ N + 2 /6 ], I N := I N. For each ξ = ξ α = ξ α1,..., ξ αd ) n X we set 2.33) R ξ := I α1 I α2 I αd,

7 DECOMPOSITION OF SPACES INDUCED BY HERMITE EXPENSIONS 7 and also 2.34) Q := [ξ N 2 /6, ξ N + 2 /6 ] d = ξ X R ξ. Thus we have assocated to each ξ X 0) a tle R ξ so that dfferent tles do not overla have dsont nterors) and they cover the cube Q [ 2, 2 ] d. Observe that by the constructon of the tles {R ξ } and 2.26) we have 2.35) λ ξ R ξ, ξ X. By 2.28) R ξ 2 d f ξ = ξ α X wth α 1/2 δ/2)2n = 1 δ)n. Assume that ξ α 1 + 4δ)2 +1. By 2.27) ξ α > π α 1/2) 4N + 1) 1/2 and hence π α 1/2) 4N + 1) 1/2 < 1 + 4δ)2+1. Usng the defnton of N n 2.31) t s easy to show that the above nequalty mles α 1 δ)n. Consequently, for ξ X, 2.36) R ξ ξ + [ 2, 2 ] d and R ξ 2 d f ξ 1 + 4δ)2 +1. On the other hand, by 2.28)-2.29) t follows that, n general, 2.37) ξ + [ c 1 2, c 1 2 ] d R ξ ξ + [ c 2 2 /3, c 2 2 /3 ] d, ξ X, and hence 2.38) c 2 d R ξ c 2 d/3. Fnally, note that snce the zeros of H n and H n+1 nterlace, each R η X +l, l 1, may ntersect at most fntely many deendng only on d) tles R ξ, ξ X Maxmal oerator. Let M s be the maxmal oerator, defned by 1/s ) M s fx) := su fy) dy) s, x R d, Q: x Q Q Q where the su s over all cubes Q n R d wth sdes arallel to the coordnate axes whch contan x. We wll need the Fefferman-Sten vector-valued maxmal nequalty see [14]): If 0 < <, 0 < q, and 0 < s < mn{, q}, then for any sequence of functons f 1, f 2,... on R d 2.40) ) 1/q [M s f )] q c f ) q) 1/q, =1 where c = c, q, s, d) Dstrbutons on R d. As s customary, we wll denote by S the Schwartz class of all functons φ C R d ) such that 2.41) P β,γ φ) := su x γ D β φx) < for all γ, β. x The toology on S s defned by the sem-norms P β,γ. Then the sace S of all temerate dstrbutons s defned as the set of all contnuous lnear functonals on S. The arng of f S and φ S wll be denoted by f, φ := fφ) whch s consstent wth the nner roduct f, g := R d fgdx n L 2 R d ). As a convenent notaton we ntroduce the followng convoluton : =1

8 8 PENCHO PETRUSHEV AND YUAN XU Defnton 2.9. For functons Φ : R d R d C and f : R d C, we wrte 2.42) Φ fx) := Φx, y)fy) dy. R d More generally, assumng that f S and Φ : R d R d C s such that Φx, y) belongs to S as a functon of y Φx, ) S), we defne Φ f by 2.43) Φ fx) := f, Φx, ), where on the rght f acts on Φx, y) as a functon of y. We next record some roertes of the above convoluton that are well known and easy to rove. Lemma a) If f S and Φ, ) SR d R d ), then Φ f S. Furthermore H n f V n. b) If f S, Φ, ) SR d R d ), and φ S, then Φ f, φ = f, Φ φ. c) If f S, Φ, ), Ψ, ) SR d R d ), and Φy, x) = Φx, y), Ψy, x) = Ψx, y), then 2.44) Ψ Φ fx) = Ψx, ), Φ, ) f. Evdently the Hermte functons {H α } belong to the sace of test functons S. More mortantly the functons n S can be characterzed by the coeffcents n ther Hermte exansons. Denote 2.45) Pr φ) := n+1) r H n φ 2 = n+1) r φ, H α 2) 1/2, r 0. n=0 Lemma We have n=0 α =n 2.46) φ S φ, H α c k α + 1) k for all α and all k. Moreover, the toology n S can be equvalently defned by the sem-norms Pr from above. Proof. a) Assume frst that the rght-hand sde estmates n 2.46) hold. Alyng reeatedly denttes 2.5)-2.6) one easly derves the estmate 2.47) su x γ D β H α x) c α + 1) γ + β )/2 max x ω α + β + γ H ω for all ndces β and γ, whch mles φ S usng that φ = n=0 n L 2. α =n φ, H α H α b) Suose φ S. Usng 2.2) we have 1 φ, H α = H α x)φx)dx = + x 2 )H α x)φx)dx R 2 α + d d R d 1 = H α x) φ + x 2 φx))dx, 2 α + d R d where for the last equalty we used ntegraton by arts. Reeatng the above rocedure k tmes we obtan a reresentaton for φ, H α of the form ) φ, H α = 2 α + d) k H α x) C β,γ x γ D β φx)dx R d β 2k, γ 2k whch yelds the rght-hand sde estmates n 2.46).

9 DECOMPOSITION OF SPACES INDUCED BY HERMITE EXPENSIONS 9 The equvalence of the toologes n S nduced by the sem-norms from 2.41) and 2.45) follows easly by 2.47) and 2.48). 3. Constructon of buldng blocks Needlets) We utlze the localzed kernels from Theorem 2.2 and the cubature formula from Corollary 2.8 to the constructon of a ar of dual frames consstng of localzed functons on R d. Let â, b satsfy the condtons: 3.1) â, b C R), su â, su b [1/4, 4], 3.2) ât), bt) > c > 0 f t [1/3, 3], 3.3) ât) bt) + â4t) b4t) = 1 f t [1/4, 1]. Consequently, 3.4) â4 ν t) b4 ν t) = 1, t [1, ). ν=0 It s easy to see that see e.g. [5]) f â satsfes 3.1)-3.2), then there exsts b satsfyng 3.1)-3.2) such that 3.3) holds true. Assumng that â, b satsfy 3.1)-3.3), we defne ν ) 3.5) Φ 0 := H 0, Φ := â 4 1 H ν, 1, and 3.6) Ψ 0 := H 0, Ψ := ν=0 ν=0 ν ) b 4 1 H ν, 1. Let X be the set of the nodes of cubature formula 2.32) from Corollary 2.8 and let λ ξ be the coeffcents of that cubature formula. We now defne the th level needlets by 3.7) ϕ ξ x) := λ 1/2 ξ Φ x, ξ) and ψ ξ x) := λ 1/2 ξ Ψ x, ξ), ξ X. Wrte X := X, where equal onts from dfferent levels X are consdered as dstnct elements of X. We use X as an ndex set to defne a ar of dual needlet systems Φ and Ψ by 3.8) Φ := {ϕ ξ } ξ X, Ψ := {ψ ξ } ξ X. Accordng to ther further roles, we wll call {ϕ ξ } analyss needlets and {ψ ξ } synthess needlets. The almost exonental localzaton of the needlets wll be crtcal for our further develoment. Indeed, by 2.17) we have 3.9) Φ ξ, x), Ψ ξ, x) c k 2 d x ξ ) k, x Rd, k, Fx L > 0. Then by 2.19) t follows that for any k > ) Φ ξ, x), Ψ ξ, x) Here c k deends on L and δ as well. c k 2 L x ξ ) k, x Rd, f ξ > 1 + δ)2 +1.

10 10 PENCHO PETRUSHEV AND YUAN XU From above and 2.35)-2.37) we nfer 3.11) ϕ ξ x), ψ ξ x) and 3.12) ϕ ξ x), ψ ξ x) c k 2 d/ x ξ ) k, f ξ 1 + δ)2 +1, c k 2 L x ξ ) k, f ξ > 1 + δ)2 +1. The followng rooston rovdes a dscrete decomoston of S and L R d ) va needlets. Prooston 3.1. a) If f S, then 3.13) f = Ψ Φ f n S and 3.14) f = ξ X f, ϕ ξ ψ ξ n S. b) If f L, 1, then 3.13) 3.14) hold n L. 1 < <, then the convergence n 3.13) 3.14) s uncondtonal. Moreover, f Proof. a) By the defnton of Φ and Ψ n 3.5)-3.6) t follows that Ψ 0 Φ 0 = H 0 and 4 ν ν ) Ψ Φ x, y) = ) b H ν x, y), 1. ν=4 2 â Note that Ψ x, y) and Φ x, y) are symmetrc functons e.g. Ψ y, x) = Ψ x, y)) snce H ν x, y) are symmetrc and hence Ψ Φ x, y) s well defned. Now, 3.4) and Lemma 2.11 yeld 3.13). To establsh 3.14), we note that Ψ x, ) and Φ y, ) belong to V 4 and alyng the cubature formula from Corollary 2.8, we obtan Ψ Φ x, y) = Ψ x, u)φ y, u) dy R d = λ ξ Ψ x, ξ)φ y, ξ) = ψ ξ x)ϕ ξ y). ξ X ξ X Consequently, Ψ Φ f = ξ X f, ϕ ξ ψ ξ. Ths along wth 3.13) mles 3.14). b) Reresentaton 3.13) n L follows easly by the rad decay of the kernels of the nth artal sums. We omt the detals. Then 3.14) n L follows as above. The uncondtonal convergence n L, 1 < <, follows by Prooston 4.3 and Theorem 4.5 below. Remark 3.2. It s well known that there exsts a functon â 0 satsfyng 3.1) 3.2) such that â 2 t)+â 2 4t) = 1, t [1/4, 1]. Suose that n the above constructon

11 DECOMPOSITION OF SPACES INDUCED BY HERMITE EXPENSIONS 11 b = â and â 0. Then ϕξ = ψ ξ. Now 3.14) becomes f = ξ X f, ψ ξ ψ ξ. It s easy to see that ths reresentaton holds n L 2 and f L 2 = f, ψ ξ 2) 1/2, f L 2, ξ X.e. {ψ ξ } ξ X s a tght frame for L 2 R d ). 4. Hermte-Trebel-Lzorkn saces F-saces) In ths secton we ntroduce the analogue of Trebel-Lzorkn saces n the context of Hermte exansons followng the general aroach descrbed n [17, 10.3] and show that they can be characterzed va needlets. In our treatment of Hermte- Trebel-Lzorkn saces we wll utlze the scheme of Frazer and Jawerth from [5] see also [6]) Defnton of Hermte-Trebel-Lzorkn saces. Let the kernels {Φ } be defned by ν ) 4.1) Φ 0 := H 0 and Φ := â 4 1 H ν, 1, ν=0 where {H ν } are from 2.7) and â obeys the condtons: 4.2) 4.3) â C [0, ), su â [1/4, 4], ât) > c > 0, f t [1/3, 3]. Defnton 4.1. The Hermte-Trebel-Lzorkn sace F := F H), where α R, 0 < <, 0 < q, s defned as the set of all f S such that 4.4) f F := 2 α Φ f ) ) q) 1/q <, where the l q -norm s relaced by the su norm when q =. As wll be shown n Theorem 4.5, the above defnton of Trebel-Lzorkn saces s ndeendent of the secfc selecton of â satsfyng 4.2)-4.3) n the defnton of Φ n 4.1). Prooston 4.2. The Hermte-Trebel-Lzorkn sace F whch s contnuously embedded n S F S ). s a quas-banach sace Proof. We wll only establsh that F S. Then the comleteness of F follows by a standard argument usng n addton Fatou s lemma and Prooston 3.1. As n Defnton 4.1, let {Φ } be defned by a functon â obeyng 4.2)-4.3). As already ndcated there exsts a functon b such that 3.1)-3.3) hold. Let {Ψ } be defned as n 3.6) usng ths functon. After ths rearaton, let {ϕ ξ } and {ψ ξ } be needlet systems defned as n 3.7)-3.7) usng these {Φ } and {Ψ }. Let f F. By Prooston 3.1 f = Ψ Φ f n S and hence f, φ = Ψ Φ f, φ = Φ f, Ψ φ, φ S.

12 12 PENCHO PETRUSHEV AND YUAN XU Alyng the Cauchy-Schwarz nequalty and 2.21) we obtan, for 2, 4 Φ f, Ψ φ Φ f 2 Ψ φ 2 c2 d/ Φ f c2 f F P r φ), ν=4 2 H φ 2 whenever r α + d/ + 1. Ths leads to f, φ c f F P r φ), whch yelds the clamed embeddng. Prooston 4.3. We have the followng dentfcaton: 4.5) F 02 L, 1 < <, wth equvalent norms. The roof of ths rooston can be carred out as the roof of Prooston 4.3 n [10] n the case of shercal harmoncs and wll be omtted. It emloys the exstng L multlers for Hermte exansons see e.g. [16]) Needlet decomoston of Hermte-Trebel-Lzorkn saces. In the followng we wll use the multlevel set X := X from 3 and the tles {R ξ } ntroduced n 2.33). Defnton 4.4. Let α R, 0 < <, 0 < q. The Hermte-Trebel- Lzorkn sequence sace f s defned as the set of all sequences of comlex numbers s = {s ξ } ξ X such that 4.6) s f := 2 [ q ) 1/q s ξ R ξ 1/2 1 Rξ )] < ξ X wth the usual modfcaton when q =. Assumng that {ϕ ξ }, {ψ ξ } s a dual ar of analyss and synthess needlets see 3.7)-3.8)), we ntroduce the oerators: S ϕ : f { f, ϕ ξ } ξ X Analyss oerator) and T ψ : {s ξ } ξ X ξ X s ξψ ξ Synthess oerator). We now come to our man result on Hermte-Trebel-Lzorkn saces. Theorem 4.5. If α R and 0 < <, 0 < q, then the oerators S ϕ : F f and T ψ : f F are bounded and T ϕ S ϕ = Id. Consequently, assumng that f S, we have f F f and only f { f, ϕ ξ } f and 4.7) f F Furthermore, the defnton of F satsfyng 4.2) 4.3). { f, ϕ ξ } f. s ndeendent of the secfc selecton of â For the roof of ths theorem we adat some technques from [5]. Defnton 4.6. For any collecton of comlex numbers {a ξ } ξ X, we defne 4.8) a x) := a η η x ) σ η X and 4.9) a ξ := a ξ), ξ X, where σ > d s suffcently large and wll be secfed later on.

13 DECOMPOSITION OF SPACES INDUCED BY HERMITE EXPENSIONS 13 We wll need a coule of lemmas whose roofs are gven n 6. Lemma 4.7. Suose s > 0 and σ > d max{2, 1/s}. Let {b ω } ω X, 0, be a set of comlex numbers. Then ) 4.10) b x) cm s b ω 1 Rω x), x R d. ω X Moreover, for ξ X, 4.11) b ξ1 Rξ x) cm s ω X b ω 1 Rω Here the constants deend only on d, δ, σ, and s. Lemma 4.8. Let g V 4 and denote ) x), x R d. M ξ := su x R ξ gx), ξ X, and m λ := nf x R λ gx), λ X +l. Then there exsts l 1, deendng only d, δ, and σ, such that for any ξ X 4.12) M ξ cm λ for all λ X +l, R λ R ξ, and hence 4.13) M ξ 1 Rξ x) c λ X +l,r λ R ξ where c > 0 deends only on d, δ, and σ. m λ1 Rλ x), x R d, Proof of Theorem 4.5. Suose q < the case q = s easer) and ck s, σ, and k so that 0 < s < mn{, q} and k σ > d max{1, 1/s}. Let {Φ } be from the defnton of Hermte-Trebel-Lzorkn saces see 4.1)- 4.3)). As already ndcated n the begnnng of 3, there exsts a functon b satsfyng 3.1)-3.2) such that 3.3) holds as well. We use ths functon to defne {Ψ } exactly as n 3.6). We further use {Φ } and {Ψ } to defne ust as n 3.7) a ar of dual needlet systems {ϕ η } and {ψ η }. Let { ϕ η }, { ψ η } be a second ar of needlet systems, defned as n 3.5)-3.7) from another ar of kernels { Φ }, { Ψ }. Our frst ste s to establsh the boundedness of the oerator T ψ : f F, defned by T ψs := ξ X s ψ ξ ξ. Prooston 4.2 and the fact that fntely suorted sequences are dense n f mly that t suffces to rove the boundedness of T ψ only for fntely suorted sequence. So, assume s = {s ξ } ξ X s a fntely suorted sequence and let f := T ψs. Evdently Φ ψ ξ = 0 f ξ X ν and ν 2, and hence Φ f = +1 ν= 1 ξ X ν s ξ Φ ψ ξ X 1 := ). Let ξ X ν, 1 ν + 1, and ξ 1 + δ)2 ν+1. Then usng 3.9)-3.11) we get Φ ψ ξ x) c2 3d/2 1 R x y ) k ξ y ) k dy c2 d/ ξ x ) k. d

14 14 PENCHO PETRUSHEV AND YUAN XU Hence, on account of 2.36) 4.14) Φ ψ ξ x) c R ξ 1/ ξ x ) k, x Rd. If ξ X ν, 1 ν + 1, and ξ > 1 + δ)2 ν+1, then by 3.9)-3.12) Φ ψ ξ x) c2 L R d 2 d x y ) k ξ y ) k dy c2 L ξ x ) k for any L > 0. Consequently, n vew of 2.37), estmate 4.14) holds agan. Denote S ξ := s ξ R ξ 1/2. Then by 4.14) we have 4.15) Φ fx) c +1 ν= 1 +1 ν= 1 ξ X ν s ξ Φ ψ ξ x) c S νx) S 1 := 0), +1 ν= 1 ξ X ν s ξ R ξ 1/ ν ξ x ) k where S νx) s defned as n 4.8). We nsert ths n 4.4) and aly Lemma 4.7 and the maxmal nequalty 2.40) to obtan f F 2 α S ) ) q) 1/q [M s 2 )] q ) 1/q α s ξ R ξ 1/2 1 Rξ c {s η } f. ξ X Hence the oerator T ψ : f F s bounded. Assumng that the sace F s defned va {Φ } nstead of {Φ } we next rove the boundedness of the oerator S ϕ : F f. Let f F and set M ξ := su x R ξ Φ fx), ξ X, and m λ := nf x R λ Φ fx), λ X +l, where l s the constant from Lemma 4.8. We have f, ϕ ξ c R ξ 1/2 Φ fξ) c R ξ 1/2 M ξ c R ξ 1/2 M ξ. By Lemma 2.10, Φ f V 4, and alyng Lemma 4.8 see 4.13)), we have Mξ 1 Rξ x) c m λ1 Rλ x), x R d. λ X +l,r λ R ξ

15 DECOMPOSITION OF SPACES INDUCED BY HERMITE EXPENSIONS 15 We use the above, Lemma 4.7, and the maxmal nequalty 2.40) to obtan { f, ϕ ξ } f c 2 ) q ) 1/q M ξ 1 Rξ ξ X c 2 ) q ) 1/q m λ1 Rλ λ X +l ) q ) 1/q c M s 2 α m λ 1 Rλ λ X +l ) q ) 1/q c 2 α m λ 1 Rξ λ X +l c 2 Φ f q) 1/q = c f F. Here for the second nequalty we used that each tle R λ, λ X +l, ntersects no more that fntely many deendng only on d) tles R η, η X. The above confrms the boundedness of the oerator S ϕ : F f. The dentty T ψ S ϕ = Id follows by Theorem 3.1. We fnally show the ndeendence of the defnton of Trebel-Lzorkn saces from the secfc selecton of â satsfyng 4.2)-4.3). Let {Φ }, { Φ } be two sequences of kernels as n the defnton of Trebel-Lzorkn saces defned by two dfferent functons â satsfyng 4.2)-4.3). As n the begnnng of ths roof, there exst two assocated needlet systems {Φ }, {Ψ }, {ϕ ξ }, {ψ ξ } and { Φ }, { Ψ }, { ϕ ξ }, { ψ ξ }. Denote by f F Φ) and f F Φ) the F -norms defned va {Φ } and { Φ }. Then from above t follows that f F Φ) c { f, ϕ ξ } f c f. F Φ) The clamed ndeendence of the defnton of F of the secfc selecton of â n the defnton of the functons {Φ } follows by nterchangng the roles of {Φ } and { Φ } and ther comlex conugates. The Hermte-F-saces embed n one another smlarly as the classcal F-saces. Prooston 4.9. a) If 0 < <, 0 < q, q 1, α R and ε > 0, then 4.16) F α+ε,q F 1. b) Let 0 < < 1 <, 0 < q, q 1, and < α 1 < α <. Then we have the contnuous embeddng 4.17) F F α1q1 1 f α d/ = α 1 d/ 1. The roof of ths embeddng result uses estmate 2.21) and Theorem 4.5 and can be carred out exactly as n the classcal case on R n see e.g. [17],. 47 and. 129). We omt t Comarson of Hermte-F-saces wth classcal F-saces. We next use needlet decomostons to show that the Hemte-Trebel-Lzorkn saces of essentally ostve smoothness are dfferent from the corresondng classcal Trebel- Lzorkn saces on R d.

16 16 PENCHO PETRUSHEV AND YUAN XU Theorem Let 0 < <, 0 < q, and α > d1/ 1) +. Then there exsts a functon f F such that f F H) = and hence f F H). Here F and F H) are the resectve classcal and Hermte Trebel-Lzorkn saces. Proof. For any y R d and a functon f we defne 4.18) f F y := q ) 1/q 2 R ξ 1/2 f, ϕ ξ 1 Rξ )). ξ X, ξ y > y /2 Choose a functon h C R d ) such that h = 1 and su h B0, 1), where B0, 1) := {x R d : x < 1}. Theorem 4.10 wll follow easly by the followng lemma whose roof s gven n 6.2. Lemma Wth the notaton from above, we have 4.19) 4.20) h y) F H) as y, and h y) F y 0 as y. By ths lemma t follows that there exsts a sequence {y } 1 R d such that 0 < y 1 < y 2 <... and y +1 > 3 y, h y ) F α q H) > 2 2, and h y ) F y < 1, = 1, 2,.... We now defne fx) := =1 f x), where f x) := 2 hx y ), x R d. Set τ := mn{, q, 1}. Evdently, h belongs to all classcal Trebel-Lzorkn saces, whch are shft nvarant, and hence f τ Fq α 2 τ h y ) τ Fq α = h τ Fq α 2 τ c h τ Fq α <. =1 Here we use that g τ Fq α On the other hand, for any l 1, f τ F α q g τ F α q H) c 2 ν ν=0 ξ X ν, ξ y l y l /2 c f l τ Fq α H) ) f τ Fy =1 = c2 lτ h y l ) τ F α q > c2 lτ c =1. Thus f F α q. ) q ) 1/q R ξ 1/2 f, ϕ ξ 1 τ Rξ ) H) c 2 τ h y ) τ Fy =1 2 τ c2 lτ c. =1 Here for the second nequalty we used that f ξ y l y l /2, then ξ y > y /2 for all l. Consequently, f τ F α q H) =. 5. Hermte-Besov saces B-saces) Besov tye saces are natural to ntroduce n the context of Hermte exansons see e.g. [17, 10.3]). We wll call them Hermte-Besov saces. To characterze these sace va needlets we use the aroach of Frazer and Jawerth [4] see also

17 DECOMPOSITION OF SPACES INDUCED BY HERMITE EXPENSIONS 17 [6]) to the classcal Besov saces. We refer to [11, 17] as general references for Besov saces Defnton of Hermte-Besov saces. Defnton 5.1. Let the kernels {Φ } be defned by 4.1) wth â satsfyng 4.2) 4.3). The Hermte-Besov sace B := B H), where α R, 0 <, q, s defned as the set of all f S such that ) q ) 1/q 5.1) f B := 2 α Φ f <, where the l q -norm s relaced by the su-norm f q =. Smlarly as for Hermte-Trebel-Lzorkn saces 4) Theorem 5.3 below mles that the above defnton of Hermte-Besov saces s ndeendent of the secfc selecton of â; also B s a quas-banach sace whch s contnuously embedded n S Needlet decomoston of Hermte-Besov saces. As for the Hermte- Trebel-Lzorkn saces we emloy the tles {R ξ } ntroduced n 2.33) n the followng. Also as before X := X. Defnton 5.2. The Hermte-Besov sequence sace b, where α R, 0 <, q, s defned as the set of all sequences of comlex numbers s = {s ξ } ξ X such that 5.2) s b := [2 α ) 1/ ] q ) 1/q R ξ 1 /2 s ξ < ξ X wth obvous modfcatons when = or q =. In the followng, we assume that {Φ }, {Ψ }, {ϕ ξ }, {ψ ξ } s a needlet system defned by 3.5)-3.8). Recall the analyss oerator: S ϕ : f { f, ϕ ξ } ξ X, and the synthess oerator: T ψ : {s ξ } ξ X ξ X s ξψ ξ. Theorem 5.3. If α R and 0 <, q, then the oerators S ϕ : B b and T ψ : b B are bounded and T ϕ S ϕ = Id. Consequently, assumng that f S, we have f B 5.3) f B Furthermore, the defnton of B 4.2) 4.3). f and only f { f, ϕ ξ } b { f, ϕ ξ } b. and s ndeendent of the choce of â satsfyng For the roof of ths theorem we need one addtonal lemma. Lemma 5.4. For any g V 4, 0, and 0 < 5.4) R ξ max gx) ) 1/ c g. x R ξ ξ X The roof of ths lemma s gven n 6. Proof of Theorem 5.3. Let 0 < s < and σ > d max{1, 1/s}. Just as n the roof of Theorem 4.5 we assume that {Φ }, {Ψ }, {ϕ η }, {ψ η } and { Φ }, { Ψ }, { ϕ η }, { ψ η } are two needlet systems, defned as n 3.5)-3.7), whch orgnate from two comletely dfferent functons â satsfyng 4.2)-4.3).

18 18 PENCHO PETRUSHEV AND YUAN XU We frst rove the boundedness of the oerator T ψ : b B, defned by T ψs := ξ X s ψ ξ ξ, assumng that B s defned by {Φ }. As n the Trebel- Lzorkn case due to the embeddng B S t suffces to consder only the case of a fntely suorted sequence s = {s ξ } ξ X. Let f := T ψs. By 4.15) and Lemma 4.7 we get Φ f c whch leads to f B c +1 ν= 1 +1 ν= 1 c {s η } b M s ω X ν R ω 1/2 s ω 1 Rω ) ) 1/ R ω 1 /2 s ω X 1 := ), ω X ν To rove the boundedness of the oerator S ϕ : B s defned n terms of {Φ }. Observng that and hence to the boundedness of T ψ. b f, ϕ ξ = λ 1/2 ξ Φ fξ) R ξ 1/2 Φ fξ), ξ X, we assume that B and Φ f V 4, we get usng Lemma 5.4 ) 1/ ) 1/ R ξ 1 /2 f, ϕ ξ c R ξ Φ fξ) c Φ f. ξ X ξ X Ths yelds { f, ϕ ξ } b c f B and hence the oerator S ϕ s bounded. The dentty T ψ S ϕ = Id s a consequence of Prooston 3.1. The ndeendence of the defnton of B from the artcular selecton of â follows from above exactly as n the case of Trebel-lzorkn saces see the roof of Theorem 4.5). The Hermte-Besov saces embed smlarly as the classcal Besov saces. Prooston 5.5. a) If 0 <, q, q 1, α R and ε > 0, then 5.5) B α+ε,q B 1. b) Let 0 < < 1 <, 0 < q, and < α 1 < α <. Then we have the contnuous embeddng 5.6) B B α1q 1 f α d/ = α 1 d/ 1. c) If 0 < <, 0 < q, α R, then 5.7) B α,mn{,q} F B α,max{,q}. Part b) of ths rooston follows readly by estmate 2.21). The roofs of arts a) and c) are as n the classcal case. We now show that under some restrcton on the ndces the Hemte-Besov saces are essentally dfferent from the classcal Besov saces on R d. Theorem 5.6. Let 0 <, q <, and α > d1/ 1) +. Then there exsts a functon f B such that f B H) = and then f B H). Here B and H) are the resectve classcal and Hermte Besov saces. B

19 DECOMPOSITION OF SPACES INDUCED BY HERMITE EXPENSIONS 19 Proof. We roceed qute smlarly as n the roof of Theorem Gven y R d and a functon f we defne 5.8) f B y := 2 R ξ 1 /2 f, ϕ ξ ) q/) 1/q. ξ X, ξ y > y /2 Pck h C R d ) such that h = 1 and su h B0, 1). The theorem follows easly by the followng: 5.9) 5.10) h y) B H) as y, and h y) B y 0 as y. To rove 5.9) we wll show that there exst ε > 0 and r > 1 such that 5.11) B H) Fr ε2 H). Then the result follows by the argument from the roof of Lemma Let > 1. Pck ε > 0 so that α > 2ε. Then by Proostons 4.9, 5.5 we have the followng embeddngs B B 2ε, F 2ε, F ε2 whch confrms 5.11). Let 1. Then α > d1 1/) and hence as n the roof of Lemma 4.11) there exst ε, δ > 0 such that α d/ = 3ε d/1 + δ). Then Proostons 4.9, 5.5 gve us the followng embeddngs B B 3ε,q 1+δ B2ε,1+δ 1+δ F 2ε,1+δ 1+δ F1+δ ε2, whch leads agan to 5.11). The roof of 5.10) s smlar to the roof of 4.20) and wll be omtted. We fnally want to lnk the Hermte-Besov saces wth the L -aroxmaton from lnear combnatons of Hermte functons. Denote by E n f) the best aroxmaton of f L from V n,.e. 5.12) E n f) := nf g V n f g. Let A be the aroxmaton sace of all functons f L for whch 2 α E 2 f) ) q) 1/q < 5.13) f A := f + wth the usual modfcaton when q =. Prooston 5.7. If α > 0, 1, 0 < q, then B equvalent norms. = A α/2,q Proof. Let f B. We frst observe that under the condtons on α,, and q, B s contnuously mbedded n L,.e. f can be dentfed as a functon n L and f c f B. The roof of ths s easy and standard and wll be omtted. By a well known and easy constructon there exsts a functon â 0 satsfyng 4.2)-4.3) such that ât)+â4t) = 1 for t [1/4, 1] and hence ν=0 â4 ν t) = 1 for t [1, ). Assume that {Φ } are defned by 4.1) usng ths functon â. By Theorem 5.3 the defnton of the Besov saces B s ndeendent of the selecton of â and hence they can be defned va these functons {Φ }. Smlarly as n Prooston 3.1 f = Φ f for f L. Now, usng that Φ f V 4, we have E 4 mf) =m+1 Φ f, m 0. A standard argument emloyng ths leads to the estmate f A c f A α/2,q B. To rove the estmate n the other drecton, let g V 4 2 2). Evdently, Φ f = Φ f g) and the rad decay of Φ yelds Φ f c f g. wth

20 20 PENCHO PETRUSHEV AND YUAN XU Consequently, Φ f ce 4 2f), 2, and Φ f c f. These lead to f B c f A. A α/2,q 6.1. Proofs for Secton Proofs Proof of Theorem 2.2. Ths roof hnges on an mortant lemma from [16]. Let ψ be a unvarate functon. The forward dfferences of ψ are defned by ψt) = ψt + 1) ψt) and k ψ = k 1 ψ), k 2. For a gven functon ψ we defne M ψ x, y) := ψν)h ν x, y), and then M k ψ = k ψν)h ν. ν=0 Lemma 6.1. [16,. 72] Let A x) and A y) x and y varables, resectvely. Then for any k 1, 6.1) 2 k x y ) k M ψ x, y) = k/2 l k ν=0 denote the oerator A aled to the c l,k A y) where c l,k are constants gven by ) k c l,k = 1) k l 4 k l 2k 2l 1)!!. 2l k ) 2l k A x) M ψx, y), l Proof. Ths lemma s roved n [16] excet that the constants c l,k are not determned exlctly there. For k = 1 one has [16, )] ) 6.2) 2x y )M ψ x, y) = A y) A x) M ψ x, y). The general result s obtaned by nducton usng the dentty [16, )] ) r ) r ) r 1 6.3) x y ) A y) A x) A y) A x) x y ) = 2r A y) A x). Assume that 6.1) holds for some k 0. Then usng 6.3) we get 2 k+1 x y ) k+1 M ψ = 2x y ) c l,k A y) A x) = k/2 l k 42l k) = c k,k A y) + c l,k [ A y) k+1)/2 l k where c l,k := 0 f l < k/2. relatons A y) A x) k/2 l k ) 2l k 2x y )M l ψ ) 2l k 1 A x) M ψ] l ) k+1 A x) M k+1 ψ [c l 1,k 42l k)c l,k ] A y) ) 2l k M l ψ ) 2l k 1 A x) M ψ, l Consequently, the coeffcents satsfy the recurrence c k+1,k+1 = c k,k, c l,k+1 = c l 1,k 42l k)c l,k, k + 1)/2 l k.

21 DECOMPOSITION OF SPACES INDUCED BY HERMITE EXPENSIONS 21 It follows from ths and 6.2) that c k,k = 1 for all k. Furthermore, the recurrence relaton above shows that c k,k = 4 k 1 ν=2 1 ν 2 + 2)c ν +1,ν, from whch one uses nducton and the fact that k 1 ν ν= ) = k +1) to derve the stated dentty for c l,k. The case α = 0,..., 0). Assume k 2. By Lemma 6.1, we have 2 k x y ) k Λ n x, y) = c l,k k/2 l k ν=0 ν l â A n) y) ) 2l k A x) Hν x, y), where l â ν n ) s the lth forward dfference aled wth resect to ν. Note frst that alyng the Cauchy-Schwarz nequalty, we have H α+β x)h α+γ y) 2 H α+β x) 2 H α+γ y) 2 6.4) α =n α =n α =n H n+ β x, x)h n+ γ y, y). We next consder the acton of A on H x, y). Usng reeatedly 2.4) we get ) mhβ m 6.5) = [2β + r) + 2] 1/2 H β+me. A x) Ths along wth 6.4) leads to r=0 A y) ) A x) ) m H ν x, y) 2 = α =ν r=0 1 m 1 [2α + r) + 2] 1/2 [2α + r) + 2] 1/2 r=0 H α+e y)h α+me x) 2 [2ν + + m)] +m H α+e y)h α+me x) α =ν [2ν + + m)] +m H ν+ y, y)h ν+m x, x). Hence the bnomal theorem and the Cauchy-Schwarz nequalty gve ) 2l k 2l k A y) A x) ) 2l k A y) Hν x, y) ) A x) ) 2l k H ν x, y) =0 2l k ) 2l k 2ν + 4l 2k) 2l k)/2 [H ν+ y, y)] 1 2 [H ν+2l k x, x)] 1 2 [ 2l k cν 2l k)/2 =0 =0 ] 1 [ 2 2l k H ν+ y, y) H ν+ x, x) A well known roerty of the dfference oerator gves ν ) 6.6) l â = n l â l) ξ) n l â l). n =0 ]

22 22 PENCHO PETRUSHEV AND YUAN XU By Defnton 2.1 t follow that l â ν n ) = 0 f 0 ν un l or ν n + vn, where 0 < u 1 and v > 0. Here u = 1 f â s of tye a).) Usng ths, the above estmates, and the Cauchy-Schwarz nequalty we nfer x y ) k Λ n x, y) c c l,k n l â l) n 2l k)/2 k/2 l k n+[vn] ν=[un] l [ 2l k =0 ] 1 [ 2 2l k H ν+ y, y) H ν+ x, x) =0 c k n k/2 [ K n+[vn]+k y, y) ] 1 2 [ K n+[vn]+k x, x) ] 1 2, where c k > 0 s of the form c k = ck, u, d) max 0 l k â l). Consequently, 6.7) Λ n x, y) c k [ Kn+[vn]+k y, y) ] 1 2 [ K n+[vn]+k x, x) ] 1 2 n x y ) k, x y. We also need estmate Λ n x, y) whenever x and y are close to one another. Alyng the Cauchy-Schwarz nequalty to the sum n 2.7) whch defnes H ν we get Λ n x, y) n+[vn] ν=0 ν ) Hν â x, x) 1/2 H ν y, y) 1/2 n c â [ Kn+[vn] y, y) ] 1 2 [ K n+[vn] x, x) ] 1 2, whch couled wth 6.7) yelds 2.9) n the case under consderaton. The case α > 0. We wll make use of the relaton = x A, where as usual := x. In the followng we agan denote by A x) the oerator A actng on the x varables, and A 0 s understood as the dentty oerator; by defnton A x) ) α := A x) 1 )α 1... A x) d )α d. We also dentfy the oerator of multlcaton by x wth x. In order to use Lemma 6.1, we wll need two commutng relatons. Lemma 6.2. Let k, r, s be nonnegatve ntegers. Then ) k r ) 6.8) x r A x) A y) r k! ) k = A x) A y) x r k )! and 6.9) x y ) k A x) where k!/k )! := 0 f k <. =0 ) s = s =0 ) s k! k )! Proof. To rove 6.8) we start from the dentty: ) k 6.10) x A x) A y) = k A x) A y) For k = 1 ths follows from the obvous denttes 6.11) x A x) In general t follows readly by nducton. A x) ) s x y ) k, ) k 1 + A x) = Id + A x) x and x A y) = A y) x. ) k A y) x. ] 1 2

23 DECOMPOSITION OF SPACES INDUCED BY HERMITE EXPENSIONS 23 We now roceed by nducton on r. Suose 6.8) holds for some r 1 and all k 1. Then ) k r ) x r+1 A x) A y) r k! ) k = x A x) A y) x r k )! =0 r ) [ r k! = A x) A y) k )! =0 r+1 [ ) )] r r k! = + A x) 1 k )! =0 whch comletes the nducton ste as ) r + r To roof 6.9), we start from x y ) k A x) ) k x r+1 + k ) A x) ) k A y) x r+1, 1 A y) ) ] k 1 x r ) = r+1 ). Thus 6.8) s establshed. = kx y ) k 1 + A x) x y ) k. For k = 1 ths dentty follows from 6.11) and, n general, by nducton. Fnally, one roves 6.9) by nducton on s smlarly as above. We omt the detals. The next lemma s nstrumental n the roof of Theorem 2.2 n the case α > 0. Lemma 6.3. If α, β N d 0 and k 1, then A x)) α [ x β n α + β 2 Kn+[vn]+ α + β +k x, x) ] 1 [ 2 K n+[vn]+k y, y) ] 1 2 Λ n x, y) c k 1 +. n x y ) k Proof. We frst show that for 1 d x y ) k A x)) α x β 6.12) Λ n x, y) c k n k+ α + β )/2 [ K n+[vn]+ α + β +k x, x) ] 1 [ 2 K n+[vn]+k y, y) ] 1 2. Clearly A x)) α x β = A x)) ) α α e x β β e A x) α x β and the two oerators searated by a dot commute. Usng 6.9) and Lemma 6.1, we have 6.13) 2 k x y ) k A x)) α x β Λ n x, y) = 2 k A x)) α α e x β βe α ) α k! ) α A x) x β k )! x y ) k Λ n x, y) α ) α 2 k! ν = c l,k â k )! n) k )/2 l k A x)) α e x β A y) ν=0 A x) ) 2l k+ Hν x, y). Furthermore, by 6.8), A x) ) α x β A y) ) 2l k+ β A x) = A x) µ=0 ) α A y) ) β 1) µ 2l k + )! µ 2l k + µ)! ) 2l k+ µ A x) x β µ.

24 24 PENCHO PETRUSHEV AND YUAN XU As A x)) α α e x β β e commutes wth A x) and x, we then conclude that A x)) α e x β A y) ) 2l k+ A x) Hν x, y) = A x)) α e A y) A x) β µ=0 ) β 1) µ 2l k + )! µ 2l k + µ)! ) 2l k+ µ x β µe H ν x, y). Usng relaton 2.5) reeatedly, t follows readly that r x r H λ x) = b m,r λ )H λ+r 2m)e x), m=0 where b m,r λ ) are ostve numbers satsfyng b m,r λ ) λ r/2. Alyng ths dentty to all varables we obtan 6.14) x β µe H ν x, y) = γ 1 ω 1=0... γ d ω d =0 b ω,γ λ) H λ+γ 2ω x)h λ y), λ =ν where γ = β for and γ = β µ, and b ω,γ λ) = b ω1,γ 1 λ 1 )... b ωd,γ d λ d ). Clearly, b ω,γ λ) cν β µ)/2. We now use the bnomal formula and 6.5) to obtan A x)) α e A y) c 2l k+ µ q=0 A y) A x) ) 2l k+ µ Hλ+γ 2ω x)h λ y) ) 2l k+ µ q A x)) α e +qe Hλ+γ 2ω x)h λ y) 2l k+ µ c ν α +2l k µ)/2 H λ+γ 2ω e+qe x)h λ+2l k+ µ q)e y) q=0 and hence A x)) α e x β A y) γ 1 ω 1 =0 γ d ω d =0 A x) 2l k+ µ q=0 λ =ν ) 2l k+ µ Hν x, y) β c µ=0 ν 2l k+ α + β 2µ)/2 H λ+γ 2ω+α e +qe x)h λ+2l k+ µ q)e y).

25 DECOMPOSITION OF SPACES INDUCED BY HERMITE EXPENSIONS 25 As before l â ν n ) = 0 f 0 ν un l or ν n + vn, where 0 < u 1 and v > 0. Also, by 6.6) l â ν n ) n l â ). We use all of the above to conclude that ) α x y ) k A x) x β Λ α nx, y) cn k+ α + β )/2 c l,k n+[vn] β γ 1 ν=[un] l µ=0 ω 1=0 [ 2l k+ µ q=0 γ d ω d =0 [ 2l k+ µ q=0 ] 1 2 H ν+2l k+ µ q y, y) k )/2 l k H ν+ α + γ 2ω +q x, x) cn k+ α + β )/2 [ K n+[vn]+ α + β +k x, x) ] 1 2 [ K n+[vn]+k y, y) ] 1 2, where we agan used the Cuachy-Schwarz nequalty. Ths roves 6.12). On the other hand, usng 6.14) wth µ = 0 and 6.5) we can wrte A x)) α x β Λ n x, y) = β 1 λ =ν ω 1=0... β d ω d =0 A x)) α n+[vn] ν=0 ν â n) b ω,β λ)c α λ) H λ+α+β 2ω x)h λ y), where c α λ) λ α /2. Hence, usng the fact that b ω,β λ) λ β /2 we conclude that A x)) α n+[vn] x β ν Λ n x, y) c â ν n) α + β )/2 ν=0 β 1 λ =ν ω 1 =0... β d ω d =0 H λ+α+β 2ω x)h λ y) cn α + β )/2 [ K n+[vn]+ α + β x, x) ] 1 2 [ K n+[vn] y, y) ] 1 2. Ths along wth 6.12) comletes the roof of Lemma 6.3. The last ste n the roof of Theorem 2.2 s to show that the oerator α can be reresented n the form 6.15) α = c βγ A β x γ, β+γ α where β+γ α means β +γ α for 1 d, and c βγ are constants deendng only on α, β, γ). By 2.3) = x A and hence r = x A ) r. The oerators x multlcaton by x ) and A do not commute, but t s easy to see that x s A = sx s 1 +A x s. Alyng ths reeatedly one fnds the reresentaton r = c νµ A ν x µ. 0 ν+µ r Snce the oerator A ν xµ commutes wth As xl f, ths readly mles reresentaton 6.15). ] 1 2

26 26 PENCHO PETRUSHEV AND YUAN XU Evdently, Lemma 6.3 and 6.15) yeld 2.9) whenever α > 0. The roof of Theorem 2.2 s comlete. Proof of Lemma 2.5. Observe frst that t suffces to rove 2.20) only for n suffcently large snce t holds trvally f 2/λ n c. We next rove 2.20) for d = 1. The Chrstoh-Darboux formula for the Hermte olynomals [15, 5.59)]) shows that K m x, x) = 2 m+1 m!) 1 [ H m+1x)h m x) H mx)h m+1 x) ]. Usng the fact that H m+1x) = 2m + 1)H m x) [15, )]) and H m+1 x) = 2xH m x) 2mH m 1 x) [15, 5.5.8)]), we can rewrte K m x, x) as K m x, x) = 2 m+1 m!) 1 [ 2m + 1)H 2 mx) 4mxH m 1 x)h m x) + 2m 2 H 2 m 1x) ]. Wrtten n terms of the orthonormal Hermte functons, the above dentty becomes m + 1)h 2 mx) + mh 2 m 1x) = K m x, x) + 2m x h m x)h m 1 x). In artcular, t follows that for x 2 2m + 1 m + 1)h 2 mx) + mh 2 m 1x) K m x, x) 2 2m 2m + 1 h m x) h m 1 x). and hence 3m + 2)h 2 mx) + 3mh 2 m 1x) K m x, x) + 2m + 1 hm x) 2m h m 1 x) Consequently, for x 2 2n + 1 n h 2 mx) 1 2 m=[1 λ)n] K m x, x). c n whch roves 2.20) when d = 1. n m=[1 λ)n] n m=[1 λ)n] h 2 m x) + h 2 m+1x) ) K m x, x) c 1 K [ρn] x, x), For d > 1 we need the followng dentty whch follows from the generatng functon of Hermte olynomals see e.g. [16]): H k x, x)r k = π d/2 1 r 2 ) d/2 e 1 r 1+r x 2 := F d r, t), k=0 where t = x. Let us denote H k,d x, x) = H k x, x) for x R d n order to ndcate the deendence on d. Then t follows from above that 6.16) r k [H k,d x, x) H k 2,d x, x)] = 1 r 2 ) r k H k,d x, x) k=0 k=0 = 1 r 2 )F d r, t) = π 1 F d 2 r, t). Notce that H n,d x, x) s a radal functon and hence a functon of t. Thus comarng the coeffcents of r k n both sde shows that H k,d x, x) H k 2,d x, x) = π 1 H k,d 2 x, x), ) 2

27 DECOMPOSITION OF SPACES INDUCED BY HERMITE EXPENSIONS 27 whch mles H k,d x, x) + H k 1,d x, x) = π 1 Now, summng over k we get n H k,d x, x) 1 2 k=[1 λ)n] c n k=[1 λ)n]+1 n k=[1 λ)n]+1 k H,d 2 x, x) = π 1 K k,d 2 x, x). [H k,d x, x) + H k 1,d x, x)] K k,d 2 x, x) c nk [1 ε)n],d 2 x, x) c n [1 ε)n] k=[1 λ)n] H k,d 2 x, x), where ε := 1 ρ)/d. Evdently, by nducton ths estmate yelds 2.20) for d odd. To establsh the result for d even, we only have to rove estmate 2.20) for d = 2. By the defnon of F d r, t), we have F 0 r, t) = e 1 r 1+r t2 = π 1/2 1 r 2 ) 1/2 F 1 r, t) ) 1/2 = π 1/2 1) r 2 h 2 kt)r k = π 1/2 n=0 k=0 where a 2 = 1) ) 1/2 and a2 1 = 0. Hence, usng 6.16) k a k h 2 x) r k, Consequently, k H k,2 x, x) H k 2,2 x, x) = π 1/2 a k h 2 t). H k,2 x, x)+h k 1,2 x, x) = π 1/2 A smle combnatoral formula shows that k a l = l=0 [k )/2] l=0 k l=0 l a l h 2 t) = k k h 2 t) a l. ) 1/2 1) l = Γ [ k 2 ]) l Γ 1 k 2 )Γ1 + [ 2 ]), whch s ostve for all 0 k. Furthermore, by Γk + a)/γk + 1) k a 1 t follows that k l=0 a l ck 1/2 for 0 αk for any α < 1. Therefore, αk H k,2 x, x) + H k 1,2 x, x) ck 1/2 h 2 t) and summng over k we get n k=[1 λ)n] H k,2 x, x) cn 1/2 K [ρn],1 t, t), l=0 whch establshes 2.20) for d = 2.

28 28 PENCHO PETRUSHEV AND YUAN XU 6.2. Proofs for Sectons 4-5. Proof of Lemma 4.7. Let 6.17) b x) := b η dx, R η )) σ, η X where dx, E) stands for the l dstance of x from E R d. Evdently, 6.18) b x) cb x) and b ξ1 Rξ x) cb x), x R d, ξ X. We wll show that 6.19) b x) cm s ω X b ω 1 Rω ) x), x R d. In vew of 6.18) ths mles 4.10)-4.11), and hence Lemma 4.7. By the constructon of the tles {R ξ } n 2.33)-2.34) t follows that there exsts a constant c > 0 deendng only on d such that Q := ξ X R ξ [ c 2, c 2 ] d. Fx x R d. To rove 6.19) we consder two cases for x. Case 1. x > 2c 2. Then dx, R η ) > x /2 for η X and hence b x) = b η dx, R η )) σ c 2 x ) σ b η η X η X 6.20) c4dλ 2 x ) σ b η s η X ) 1/s, where λ := 1 mn{1, 1/s} and for the last estmate we use Hölder s nequalty f s > 1 and the s-trangle nequalty f s < 1. Denote Q x := [ x, x ] d. Notce that Q Q x. From above we nfer b x) c4dλ x d/s 1 ) sdy ) 1/s 2 x ) σ b η 1 Rξ y) Q x Q x η X ) c2 2dλ σ) x d/s σ M s b η 1 Rξ )x) cm s b η 1 Rξ x) η X η X as clamed. Here we used the fact that σ d max{2, 1/s}. Case 2. x 2c 2. To make the argument more transarent we frst subdvde the tles {R η } η X nto boxes of almost equal sdes of length 2. By the constructon of the tles see 2.33)) there exsts a constant c > 0 such that the mnmum sde of each tle R η s c2. Now, evdently each tle R η can be subdvded nto a dsont unon of boxes R θ wth centers θ such that θ + [ c2 1, c2 1 ] R θ θ + [ c2, c2 ]. Denote by X the set of centers of all boxes obtaned by subdvdng the tles from X. Also, set b θ := b η f R θ R η. Evdently, 6.21) b x) := b η dx, R η )) σ η X θ X b θ dx, R θ )) σ

29 DECOMPOSITION OF SPACES INDUCED BY HERMITE EXPENSIONS 29 and 6.22) b η 1 Rη = b θ 1 Rθ. η X η X Denote Y 0 := {θ X : 2 θ x c}, Y m := {θ X : c2 m 1 2 θ x c2 m }, and Q m := {y R d : y x c2 m + 1)2 }, m 1. Clearly, #Y m c2 md, θ Ym R θ Q m, and X = m 0 Y m. Smlarly as n 6.20) b θ dx, R θ )) σ ) 1/s c2 mσ b θ c2 mσ 2 mdλ b θ s θ Y m θ Y m θ Y m c2 mσ dλ d/s) 1 ) sdy ) 1/s b θ 1 Rθ y) Q m Q m θ Y m ) c2 mσ d max{1,1/s}) M s b η 1 Rη x), η X where we used 6.22). Summng u over m 0, takng nto account that σ > d max{2, 1/s}), and also usng 6.21) we arrve at 6.19). Proof of Lemma 4.8. For ths roof we wll need an addtonal lemma. Lemma 6.4. Suose g V 4 and ξ X. Then for any k > 0 and L > 0 we have for x, x 2R ξ 6.23) gx ) gx ) c2 x x gη) ξ η ) k and 6.24) gx ) gx ) ĉ2 L x x η X η X gη) ξ η ) k, f ξ > 1+2δ)2 +1. Here c, ĉ deend on k, d, and δ, and ĉ deends on L as well; 2R ξ R d s the set obtaned by dlatng R ξ by a factor of 2 and wth the same center. Proof. Let Λ 4 be the kernel from 2.8) wth n = 4, where â s admssble of tye a) wth v := δ. Then Λ 4 g = g, snce g V 4, and Λ 4 x, ) V [1+δ)4 ]. Note that [1 + δ)4 ] + 4 2N 1. Therefore, we can use the cubature formula from Corollary 2.8 to obtan gx) = Λ 4 x, y)gy)dy = λ η Λ 4 x, η)gη), R d η X where the weghts λ ξ obey 2.35) see also 2.36)-2.37)). Hence, for x, x 2R ξ gx ) gx ) η X λ η Λ 4 x, η) Λ 4 x, η) gη) 6.25) c x x η X λ η su Λ 4 x, η) gη). x 2R ξ

30 30 PENCHO PETRUSHEV AND YUAN XU Note that 4[1 + δ)4 ] + k + 1) + 2) 1/2 1 + δ)2 +1 for suffcently large deendng on k and δ). Therefore, we have from 2.17)-2.18) 6.26) Λ 4 x, η) and for any L > 0 we need L k) 6.27) Λ 4 x, η) c2 d+1) x η ) k, x Rd, η X, c2 2L x η ) k, f x > 1+δ)2 +1 or η > 1+δ)2 +1. Suose ξ > 1 + 2δ)2 +1, then 2R ξ {x R d : x > 1 + δ)2 +1 } for suffcently large. Combnng 6.25) wth 6.27) and 2.37) we get 6.28) gx ) gx ) c2 d/3 x x su x 2R η X ξ 2 k+l) x η ) k gη), where we used that dam 2R ξ ) c2 /3. However, for any x 2R ξ we have ξ η ξ x + x η ) c2 /3 + x η ) c x η ). We use ths n 6.28) to obtan 6.24) for suffcently large. One roves 6.23) n a smlar fashon. In the case c estmates 6.23)-6.24) follow easly by 6.25). We are now reared to rove Lemma 4.8. Let g V 4. Pck l 1 suffcently large to be determned later on) and denote for ξ X 6.29) X +l ξ) := {η X +l : R η R ξ } and 6.30) d ξ := su{ gx ) gx ) : x, x R η for some η X +l ξ)}. We frst estmate d ξ, ξ X. Case A: ξ 1 + 3δ)2 +1. By 2.36) t follows that for suffcently large l deendng only on d and δ) η X+l ξ)r η 2R ξ. Hence, usng Lemma 6.4 see 6.23)) wth k σ, we get 6.31) d ξ c2 l η X gη) ξ η ) σ, for suffcently large deendng only on d and δ), where c > 0 s a constant ndeendent of l. Case B: ξ > 1+3δ)2 +1. By 2.36) x > 1+2δ)2 +1 for x η X+l ξ)r η f s suffcently large. We aly estmate 6.24) of Lemma 6.4 wth k σ and L = 1 to obtan 6.32) d ξ c2 gη) ξ η ) σ. η X To estmate M ξ, ξ X, we consder two cases for ξ. Case 1: ξ 1 + 4δ)2 +1. By 2.36), we have for suffcently large : 6.33) R ξ ξ + [ 2, 2 ] d, and R η η + [ 2 l, 2 l ] d, η X +l ξ).

APPENDIX A Some Linear Algebra

APPENDIX A Some Linear Algebra APPENDIX A Some Lnear Algebra The collecton of m, n matrces A.1 Matrces a 1,1,..., a 1,n A = a m,1,..., a m,n wth real elements a,j s denoted by R m,n. If n = 1 then A s called a column vector. Smlarly,

More information

Supplementary Material for Spectral Clustering based on the graph p-laplacian

Supplementary Material for Spectral Clustering based on the graph p-laplacian Sulementary Materal for Sectral Clusterng based on the grah -Lalacan Thomas Bühler and Matthas Hen Saarland Unversty, Saarbrücken, Germany {tb,hen}@csun-sbde May 009 Corrected verson, June 00 Abstract

More information

A NOTE ON THE DISCRETE FOURIER RESTRICTION PROBLEM

A NOTE ON THE DISCRETE FOURIER RESTRICTION PROBLEM A NOTE ON THE DISCRETE FOURIER RESTRICTION PROBLEM XUDONG LAI AND YONG DING arxv:171001481v1 [mathap] 4 Oct 017 Abstract In ths aer we establsh a general dscrete Fourer restrcton theorem As an alcaton

More information

More metrics on cartesian products

More metrics on cartesian products More metrcs on cartesan products If (X, d ) are metrc spaces for 1 n, then n Secton II4 of the lecture notes we defned three metrcs on X whose underlyng topologes are the product topology The purpose of

More information

LECTURE 5: FIBRATIONS AND HOMOTOPY FIBERS

LECTURE 5: FIBRATIONS AND HOMOTOPY FIBERS LECTURE 5: FIBRATIONS AND HOMOTOPY FIBERS In ts lecture we wll ntroduce two mortant classes of mas of saces, namely te Hurewcz fbratons and te more general Serre fbratons, wc are bot obtaned by mosng certan

More information

Some congruences related to harmonic numbers and the terms of the second order sequences

Some congruences related to harmonic numbers and the terms of the second order sequences Mathematca Moravca Vol. 0: 06, 3 37 Some congruences related to harmonc numbers the terms of the second order sequences Neşe Ömür Sbel Koaral Abstract. In ths aer, wth hels of some combnatoral denttes,

More information

BOUNDEDNESS OF THE RIESZ TRANSFORM WITH MATRIX A 2 WEIGHTS

BOUNDEDNESS OF THE RIESZ TRANSFORM WITH MATRIX A 2 WEIGHTS BOUNDEDNESS OF THE IESZ TANSFOM WITH MATIX A WEIGHTS Introducton Let L = L ( n, be the functon space wth norm (ˆ f L = f(x C dx d < For a d d matrx valued functon W : wth W (x postve sem-defnte for all

More information

ON A DETERMINATION OF THE INITIAL FUNCTIONS FROM THE OBSERVED VALUES OF THE BOUNDARY FUNCTIONS FOR THE SECOND-ORDER HYPERBOLIC EQUATION

ON A DETERMINATION OF THE INITIAL FUNCTIONS FROM THE OBSERVED VALUES OF THE BOUNDARY FUNCTIONS FOR THE SECOND-ORDER HYPERBOLIC EQUATION Advanced Mathematcal Models & Applcatons Vol.3, No.3, 2018, pp.215-222 ON A DETERMINATION OF THE INITIAL FUNCTIONS FROM THE OBSERVED VALUES OF THE BOUNDARY FUNCTIONS FOR THE SECOND-ORDER HYPERBOLIC EUATION

More information

Affine transformations and convexity

Affine transformations and convexity Affne transformatons and convexty The purpose of ths document s to prove some basc propertes of affne transformatons nvolvng convex sets. Here are a few onlne references for background nformaton: http://math.ucr.edu/

More information

n α j x j = 0 j=1 has a nontrivial solution. Here A is the n k matrix whose jth column is the vector for all t j=0

n α j x j = 0 j=1 has a nontrivial solution. Here A is the n k matrix whose jth column is the vector for all t j=0 MODULE 2 Topcs: Lnear ndependence, bass and dmenson We have seen that f n a set of vectors one vector s a lnear combnaton of the remanng vectors n the set then the span of the set s unchanged f that vector

More information

PARTIAL QUOTIENTS AND DISTRIBUTION OF SEQUENCES. Department of Mathematics University of California Riverside, CA

PARTIAL QUOTIENTS AND DISTRIBUTION OF SEQUENCES. Department of Mathematics University of California Riverside, CA PARTIAL QUOTIETS AD DISTRIBUTIO OF SEQUECES 1 Me-Chu Chang Deartment of Mathematcs Unversty of Calforna Rversde, CA 92521 mcc@math.ucr.edu Abstract. In ths aer we establsh average bounds on the artal quotents

More information

Power-sum problem, Bernoulli Numbers and Bernoulli Polynomials.

Power-sum problem, Bernoulli Numbers and Bernoulli Polynomials. Power-sum roblem, Bernoull Numbers and Bernoull Polynomals. Arady M. Alt Defnton 1 Power um Problem Fnd the sum n : 1... n where, n N or, usng sum notaton, n n n closed form. Recurrence for n Exercse Usng

More information

Solutions for Tutorial 1

Solutions for Tutorial 1 Toc 1: Sem-drect roducts Solutons for Tutoral 1 1. Show that the tetrahedral grou s somorhc to the sem-drect roduct of the Klen four grou and a cyclc grou of order three: T = K 4 (Z/3Z). 2. Show further

More information

Math 217 Fall 2013 Homework 2 Solutions

Math 217 Fall 2013 Homework 2 Solutions Math 17 Fall 013 Homework Solutons Due Thursday Sept. 6, 013 5pm Ths homework conssts of 6 problems of 5 ponts each. The total s 30. You need to fully justfy your answer prove that your functon ndeed has

More information

Exercise Solutions to Real Analysis

Exercise Solutions to Real Analysis xercse Solutons to Real Analyss Note: References refer to H. L. Royden, Real Analyss xersze 1. Gven any set A any ɛ > 0, there s an open set O such that A O m O m A + ɛ. Soluton 1. If m A =, then there

More information

THE WEIGHTED WEAK TYPE INEQUALITY FOR THE STRONG MAXIMAL FUNCTION

THE WEIGHTED WEAK TYPE INEQUALITY FOR THE STRONG MAXIMAL FUNCTION THE WEIGHTED WEAK TYPE INEQUALITY FO THE STONG MAXIMAL FUNCTION THEMIS MITSIS Abstract. We prove the natural Fefferman-Sten weak type nequalty for the strong maxmal functon n the plane, under the assumpton

More information

FINITELY-GENERATED MODULES OVER A PRINCIPAL IDEAL DOMAIN

FINITELY-GENERATED MODULES OVER A PRINCIPAL IDEAL DOMAIN FINITELY-GENERTED MODULES OVER PRINCIPL IDEL DOMIN EMMNUEL KOWLSKI Throughout ths note, s a prncpal deal doman. We recall the classfcaton theorem: Theorem 1. Let M be a fntely-generated -module. (1) There

More information

Lecture 9: Converse of Shannon s Capacity Theorem

Lecture 9: Converse of Shannon s Capacity Theorem Error Correctng Codes: Combnatorcs, Algorthms and Alcatons (Fall 2007) Lecture 9: Converse of Shannon s Caacty Theorem Setember 17, 2007 Lecturer: Atr Rudra Scrbe: Thanh-Nhan Nguyen & Atr Rudra In the

More information

Appendix B. Criterion of Riemann-Stieltjes Integrability

Appendix B. Criterion of Riemann-Stieltjes Integrability Appendx B. Crteron of Remann-Steltes Integrablty Ths note s complementary to [R, Ch. 6] and [T, Sec. 3.5]. The man result of ths note s Theorem B.3, whch provdes the necessary and suffcent condtons for

More information

MATH 241B FUNCTIONAL ANALYSIS - NOTES EXAMPLES OF C ALGEBRAS

MATH 241B FUNCTIONAL ANALYSIS - NOTES EXAMPLES OF C ALGEBRAS MATH 241B FUNCTIONAL ANALYSIS - NOTES EXAMPLES OF C ALGEBRAS These are nformal notes whch cover some of the materal whch s not n the course book. The man purpose s to gve a number of nontrval examples

More information

REAL ANALYSIS I HOMEWORK 1

REAL ANALYSIS I HOMEWORK 1 REAL ANALYSIS I HOMEWORK CİHAN BAHRAN The questons are from Tao s text. Exercse 0.0.. If (x α ) α A s a collecton of numbers x α [0, + ] such that x α

More information

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

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

More information

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

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

More information

8.4 COMPLEX VECTOR SPACES AND INNER PRODUCTS

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

More information

The Order Relation and Trace Inequalities for. Hermitian Operators

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

More information

ON THE EXTENDED HAAGERUP TENSOR PRODUCT IN OPERATOR SPACES. 1. Introduction

ON THE EXTENDED HAAGERUP TENSOR PRODUCT IN OPERATOR SPACES. 1. Introduction ON THE EXTENDED HAAGERUP TENSOR PRODUCT IN OPERATOR SPACES TAKASHI ITOH AND MASARU NAGISA Abstract We descrbe the Haagerup tensor product l h l and the extended Haagerup tensor product l eh l n terms of

More information

6. Hamilton s Equations

6. Hamilton s Equations 6. Hamlton s Equatons Mchael Fowler A Dynamcal System s Path n Confguraton Sace and n State Sace The story so far: For a mechancal system wth n degrees of freedom, the satal confguraton at some nstant

More information

Lecture 3. Ax x i a i. i i

Lecture 3. Ax x i a i. i i 18.409 The Behavor of Algorthms n Practce 2/14/2 Lecturer: Dan Spelman Lecture 3 Scrbe: Arvnd Sankar 1 Largest sngular value In order to bound the condton number, we need an upper bound on the largest

More information

DECOUPLING THEORY HW2

DECOUPLING THEORY HW2 8.8 DECOUPLIG THEORY HW2 DOGHAO WAG DATE:OCT. 3 207 Problem We shall start by reformulatng the problem. Denote by δ S n the delta functon that s evenly dstrbuted at the n ) dmensonal unt sphere. As a temporal

More information

FACTORIZATION IN KRULL MONOIDS WITH INFINITE CLASS GROUP

FACTORIZATION IN KRULL MONOIDS WITH INFINITE CLASS GROUP C O L L O Q U I U M M A T H E M A T I C U M VOL. 80 1999 NO. 1 FACTORIZATION IN KRULL MONOIDS WITH INFINITE CLASS GROUP BY FLORIAN K A I N R A T H (GRAZ) Abstract. Let H be a Krull monod wth nfnte class

More information

Modelli Clamfim Equazione del Calore Lezione ottobre 2014

Modelli Clamfim Equazione del Calore Lezione ottobre 2014 CLAMFIM Bologna Modell 1 @ Clamfm Equazone del Calore Lezone 17 15 ottobre 2014 professor Danele Rtell danele.rtell@unbo.t 1/24? Convoluton The convoluton of two functons g(t) and f(t) s the functon (g

More information

Complete subgraphs in multipartite graphs

Complete subgraphs in multipartite graphs Complete subgraphs n multpartte graphs FLORIAN PFENDER Unverstät Rostock, Insttut für Mathematk D-18057 Rostock, Germany Floran.Pfender@un-rostock.de Abstract Turán s Theorem states that every graph G

More information

Inner Product. Euclidean Space. Orthonormal Basis. Orthogonal

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

More information

NECESSARY AND SUFFICIENT CONDITIONS FOR ALMOST REGULARITY OF UNIFORM BIRKHOFF INTERPOLATION SCHEMES. by Nicolae Crainic

NECESSARY AND SUFFICIENT CONDITIONS FOR ALMOST REGULARITY OF UNIFORM BIRKHOFF INTERPOLATION SCHEMES. by Nicolae Crainic NECESSARY AND SUFFICIENT CONDITIONS FOR ALMOST REGULARITY OF UNIFORM BIRKHOFF INTERPOLATION SCHEMES by Ncolae Cranc Abstract: In ths artcle usng a combnaton of the necessary and suffcent condtons for the

More information

Uniqueness of Weak Solutions to the 3D Ginzburg- Landau Model for Superconductivity

Uniqueness of Weak Solutions to the 3D Ginzburg- Landau Model for Superconductivity Int. Journal of Math. Analyss, Vol. 6, 212, no. 22, 195-114 Unqueness of Weak Solutons to the 3D Gnzburg- Landau Model for Superconductvty Jshan Fan Department of Appled Mathematcs Nanjng Forestry Unversty

More information

2.3 Nilpotent endomorphisms

2.3 Nilpotent endomorphisms s a block dagonal matrx, wth A Mat dm U (C) In fact, we can assume that B = B 1 B k, wth B an ordered bass of U, and that A = [f U ] B, where f U : U U s the restrcton of f to U 40 23 Nlpotent endomorphsms

More information

Lecture 12: Discrete Laplacian

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

More information

General viscosity iterative method for a sequence of quasi-nonexpansive mappings

General viscosity iterative method for a sequence of quasi-nonexpansive mappings Avalable onlne at www.tjnsa.com J. Nonlnear Sc. Appl. 9 (2016), 5672 5682 Research Artcle General vscosty teratve method for a sequence of quas-nonexpansve mappngs Cuje Zhang, Ynan Wang College of Scence,

More information

Weighted anisotropic product Hardy spaces and boundedness of sublinear operators

Weighted anisotropic product Hardy spaces and boundedness of sublinear operators Math. Nachr. 83, No. 3, 39 44 00 / DOI 0.00/mana.0090078 Weghted ansotroc roduct Hardy saces and boundedness of sublnear oerators Marcn Bownk, Baode L, Dachun Yang, and Yuan Zhou Deartment of Mathematcs,

More information

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

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

More information

Managing Capacity Through Reward Programs. on-line companion page. Byung-Do Kim Seoul National University College of Business Administration

Managing Capacity Through Reward Programs. on-line companion page. Byung-Do Kim Seoul National University College of Business Administration Managng Caacty Through eward Programs on-lne comanon age Byung-Do Km Seoul Natonal Unversty College of Busness Admnstraton Mengze Sh Unversty of Toronto otman School of Management Toronto ON M5S E6 Canada

More information

Canonical transformations

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

More information

Dr. Shalabh Department of Mathematics and Statistics Indian Institute of Technology Kanpur

Dr. Shalabh Department of Mathematics and Statistics Indian Institute of Technology Kanpur Analyss of Varance and Desgn of Exerments-I MODULE II LECTURE - GENERAL LINEAR HYPOTHESIS AND ANALYSIS OF VARIANCE Dr. Shalabh Deartment of Mathematcs and Statstcs Indan Insttute of Technology Kanur 3.

More information

KNOTS AND THEIR CURVATURES

KNOTS AND THEIR CURVATURES KNOTS AND THEIR CURVATURES LIVIU I. NICOLAESCU ABSTRACT. I dscuss an old result of John Mlnor statng roughly that f a closed curve n sace s not too curved then t cannot be knotted. CONTENTS. The total

More information

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

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

More information

Mechanics Physics 151

Mechanics Physics 151 Mechancs hyscs 151 Lecture Canoncal Transformatons (Chater 9) What We Dd Last Tme Drect Condtons Q j Q j = = j, Q, j, Q, Necessary and suffcent j j for Canoncal Transf. = = j Q, Q, j Q, Q, Infntesmal CT

More information

Linear, affine, and convex sets and hulls In the sequel, unless otherwise specified, X will denote a real vector space.

Linear, affine, and convex sets and hulls In the sequel, unless otherwise specified, X will denote a real vector space. Lnear, affne, and convex sets and hulls In the sequel, unless otherwse specfed, X wll denote a real vector space. Lnes and segments. Gven two ponts x, y X, we defne xy = {x + t(y x) : t R} = {(1 t)x +

More information

Solutions to exam in SF1811 Optimization, Jan 14, 2015

Solutions to exam in SF1811 Optimization, Jan 14, 2015 Solutons to exam n SF8 Optmzaton, Jan 4, 25 3 3 O------O -4 \ / \ / The network: \/ where all lnks go from left to rght. /\ / \ / \ 6 O------O -5 2 4.(a) Let x = ( x 3, x 4, x 23, x 24 ) T, where the varable

More information

DIFFERENTIAL FORMS BRIAN OSSERMAN

DIFFERENTIAL FORMS BRIAN OSSERMAN DIFFERENTIAL FORMS BRIAN OSSERMAN Dfferentals are an mportant topc n algebrac geometry, allowng the use of some classcal geometrc arguments n the context of varetes over any feld. We wll use them to defne

More information

( ) 2 ( ) ( ) Problem Set 4 Suggested Solutions. Problem 1

( ) 2 ( ) ( ) Problem Set 4 Suggested Solutions. Problem 1 Problem Set 4 Suggested Solutons Problem (A) The market demand functon s the soluton to the followng utlty-maxmzaton roblem (UMP): The Lagrangean: ( x, x, x ) = + max U x, x, x x x x st.. x + x + x y x,

More information

Randić Energy and Randić Estrada Index of a Graph

Randić Energy and Randić Estrada Index of a Graph EUROPEAN JOURNAL OF PURE AND APPLIED MATHEMATICS Vol. 5, No., 202, 88-96 ISSN 307-5543 www.ejpam.com SPECIAL ISSUE FOR THE INTERNATIONAL CONFERENCE ON APPLIED ANALYSIS AND ALGEBRA 29 JUNE -02JULY 20, ISTANBUL

More information

Random Walks on Digraphs

Random Walks on Digraphs Random Walks on Dgraphs J. J. P. Veerman October 23, 27 Introducton Let V = {, n} be a vertex set and S a non-negatve row-stochastc matrx (.e. rows sum to ). V and S defne a dgraph G = G(V, S) and a drected

More information

MAT 578 Functional Analysis

MAT 578 Functional Analysis MAT 578 Functonal Analyss John Qugg Fall 2008 Locally convex spaces revsed September 6, 2008 Ths secton establshes the fundamental propertes of locally convex spaces. Acknowledgment: although I wrote these

More information

GELFAND-TSETLIN BASIS FOR THE REPRESENTATIONS OF gl n

GELFAND-TSETLIN BASIS FOR THE REPRESENTATIONS OF gl n GELFAND-TSETLIN BASIS FOR THE REPRESENTATIONS OF gl n KANG LU FINITE DIMENSIONAL REPRESENTATIONS OF gl n Let e j,, j =,, n denote the standard bass of the general lnear Le algebra gl n over the feld of

More information

INTEGRAL p-adic HODGE THEORY, TALK 14 (COMPARISON WITH THE DE RHAMWITT COMPLEX)

INTEGRAL p-adic HODGE THEORY, TALK 14 (COMPARISON WITH THE DE RHAMWITT COMPLEX) INTEGRAL -ADIC HODGE THEORY, TALK 4 (COMPARISON WITH THE DE RHAMWITT COMPLEX) JOAQUIN RODRIGUES JACINTO (NOTES BY JAMES NEWTON). Recollectons and statement of theorem Let K be a erfectod eld of characterstc

More information

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

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

More information

STEINHAUS PROPERTY IN BANACH LATTICES

STEINHAUS PROPERTY IN BANACH LATTICES DEPARTMENT OF MATHEMATICS TECHNICAL REPORT STEINHAUS PROPERTY IN BANACH LATTICES DAMIAN KUBIAK AND DAVID TIDWELL SPRING 2015 No. 2015-1 TENNESSEE TECHNOLOGICAL UNIVERSITY Cookevlle, TN 38505 STEINHAUS

More information

Affine and Riemannian Connections

Affine and Riemannian Connections Affne and Remannan Connectons Semnar Remannan Geometry Summer Term 2015 Prof Dr Anna Wenhard and Dr Gye-Seon Lee Jakob Ullmann Notaton: X(M) space of smooth vector felds on M D(M) space of smooth functons

More information

On the Connectedness of the Solution Set for the Weak Vector Variational Inequality 1

On the Connectedness of the Solution Set for the Weak Vector Variational Inequality 1 Journal of Mathematcal Analyss and Alcatons 260, 15 2001 do:10.1006jmaa.2000.7389, avalable onlne at htt:.dealbrary.com on On the Connectedness of the Soluton Set for the Weak Vector Varatonal Inequalty

More information

SL n (F ) Equals its Own Derived Group

SL n (F ) Equals its Own Derived Group Internatonal Journal of Algebra, Vol. 2, 2008, no. 12, 585-594 SL n (F ) Equals ts Own Derved Group Jorge Macel BMCC-The Cty Unversty of New York, CUNY 199 Chambers street, New York, NY 10007, USA macel@cms.nyu.edu

More information

Report on Image warping

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

More information

Edge Isoperimetric Inequalities

Edge Isoperimetric Inequalities November 7, 2005 Ross M. Rchardson Edge Isopermetrc Inequaltes 1 Four Questons Recall that n the last lecture we looked at the problem of sopermetrc nequaltes n the hypercube, Q n. Our noton of boundary

More information

3 Basic boundary value problems for analytic function in the upper half plane

3 Basic boundary value problems for analytic function in the upper half plane 3 Basc boundary value problems for analytc functon n the upper half plane 3. Posson representaton formulas for the half plane Let f be an analytc functon of z throughout the half plane Imz > 0, contnuous

More information

Not-for-Publication Appendix to Optimal Asymptotic Least Aquares Estimation in a Singular Set-up

Not-for-Publication Appendix to Optimal Asymptotic Least Aquares Estimation in a Singular Set-up Not-for-Publcaton Aendx to Otmal Asymtotc Least Aquares Estmaton n a Sngular Set-u Antono Dez de los Ros Bank of Canada dezbankofcanada.ca December 214 A Proof of Proostons A.1 Proof of Prooston 1 Ts roof

More information

An efficient algorithm for multivariate Maclaurin Newton transformation

An efficient algorithm for multivariate Maclaurin Newton transformation Annales UMCS Informatca AI VIII, 2 2008) 5 14 DOI: 10.2478/v10065-008-0020-6 An effcent algorthm for multvarate Maclaurn Newton transformaton Joanna Kapusta Insttute of Mathematcs and Computer Scence,

More information

h-analogue of Fibonacci Numbers

h-analogue of Fibonacci Numbers h-analogue of Fbonacc Numbers arxv:090.0038v [math-ph 30 Sep 009 H.B. Benaoum Prnce Mohammad Unversty, Al-Khobar 395, Saud Araba Abstract In ths paper, we ntroduce the h-analogue of Fbonacc numbers for

More information

An Introduction to Morita Theory

An Introduction to Morita Theory An Introducton to Morta Theory Matt Booth October 2015 Nov. 2017: made a few revsons. Thanks to Nng Shan for catchng a typo. My man reference for these notes was Chapter II of Bass s book Algebrac K-Theory

More information

Online Classification: Perceptron and Winnow

Online Classification: Perceptron and Winnow E0 370 Statstcal Learnng Theory Lecture 18 Nov 8, 011 Onlne Classfcaton: Perceptron and Wnnow Lecturer: Shvan Agarwal Scrbe: Shvan Agarwal 1 Introducton In ths lecture we wll start to study the onlne learnng

More information

Some basic inequalities. Definition. Let V be a vector space over the complex numbers. An inner product is given by a function, V V C

Some basic inequalities. Definition. Let V be a vector space over the complex numbers. An inner product is given by a function, V V C Some basc nequaltes Defnton. Let V be a vector space over the complex numbers. An nner product s gven by a functon, V V C (x, y) x, y satsfyng the followng propertes (for all x V, y V and c C) (1) x +

More information

MMA and GCMMA two methods for nonlinear optimization

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

More information

Lecture 10 Support Vector Machines II

Lecture 10 Support Vector Machines II Lecture 10 Support Vector Machnes II 22 February 2016 Taylor B. Arnold Yale Statstcs STAT 365/665 1/28 Notes: Problem 3 s posted and due ths upcomng Frday There was an early bug n the fake-test data; fxed

More information

arxiv: v1 [math.co] 1 Mar 2014

arxiv: v1 [math.co] 1 Mar 2014 Unon-ntersectng set systems Gyula O.H. Katona and Dánel T. Nagy March 4, 014 arxv:1403.0088v1 [math.co] 1 Mar 014 Abstract Three ntersecton theorems are proved. Frst, we determne the sze of the largest

More information

SMARANDACHE-GALOIS FIELDS

SMARANDACHE-GALOIS FIELDS SMARANDACHE-GALOIS FIELDS W. B. Vasantha Kandasamy Deartment of Mathematcs Indan Insttute of Technology, Madras Chenna - 600 036, Inda. E-mal: vasantak@md3.vsnl.net.n Abstract: In ths aer we study the

More information

CONDITIONS FOR INVARIANT SUBMANIFOLD OF A MANIFOLD WITH THE (ϕ, ξ, η, G)-STRUCTURE. Jovanka Nikić

CONDITIONS FOR INVARIANT SUBMANIFOLD OF A MANIFOLD WITH THE (ϕ, ξ, η, G)-STRUCTURE. Jovanka Nikić 147 Kragujevac J. Math. 25 (2003) 147 154. CONDITIONS FOR INVARIANT SUBMANIFOLD OF A MANIFOLD WITH THE (ϕ, ξ, η, G)-STRUCTURE Jovanka Nkć Faculty of Techncal Scences, Unversty of Nov Sad, Trg Dosteja Obradovća

More information

Combinatorial Identities for Incomplete Tribonacci Polynomials

Combinatorial Identities for Incomplete Tribonacci Polynomials Avalable at http://pvamu.edu/aam Appl. Appl. Math. ISSN: 1932-9466 Vol. 10, Issue 1 (June 2015, pp. 40 49 Applcatons and Appled Mathematcs: An Internatonal Journal (AAM Combnatoral Identtes for Incomplete

More information

Professor Terje Haukaas University of British Columbia, Vancouver The Q4 Element

Professor Terje Haukaas University of British Columbia, Vancouver  The Q4 Element Professor Terje Haukaas Unversty of Brtsh Columba, ancouver www.nrsk.ubc.ca The Q Element Ths document consders fnte elements that carry load only n ther plane. These elements are sometmes referred to

More information

Matrix Approximation via Sampling, Subspace Embedding. 1 Solving Linear Systems Using SVD

Matrix Approximation via Sampling, Subspace Embedding. 1 Solving Linear Systems Using SVD Matrx Approxmaton va Samplng, Subspace Embeddng Lecturer: Anup Rao Scrbe: Rashth Sharma, Peng Zhang 0/01/016 1 Solvng Lnear Systems Usng SVD Two applcatons of SVD have been covered so far. Today we loo

More information

12 MATH 101A: ALGEBRA I, PART C: MULTILINEAR ALGEBRA. 4. Tensor product

12 MATH 101A: ALGEBRA I, PART C: MULTILINEAR ALGEBRA. 4. Tensor product 12 MATH 101A: ALGEBRA I, PART C: MULTILINEAR ALGEBRA Here s an outlne of what I dd: (1) categorcal defnton (2) constructon (3) lst of basc propertes (4) dstrbutve property (5) rght exactness (6) localzaton

More information

Genericity of Critical Types

Genericity of Critical Types Genercty of Crtcal Types Y-Chun Chen Alfredo D Tllo Eduardo Fangold Syang Xong September 2008 Abstract Ely and Pesk 2008 offers an nsghtful characterzaton of crtcal types: a type s crtcal f and only f

More information

Digital PI Controller Equations

Digital PI Controller Equations Ver. 4, 9 th March 7 Dgtal PI Controller Equatons Probably the most common tye of controller n ndustral ower electroncs s the PI (Proortonal - Integral) controller. In feld orented motor control, PI controllers

More information

Difference Equations

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

More information

AN ASYMMETRIC GENERALIZED FGM COPULA AND ITS PROPERTIES

AN ASYMMETRIC GENERALIZED FGM COPULA AND ITS PROPERTIES Pa. J. Statst. 015 Vol. 31(1), 95-106 AN ASYMMETRIC GENERALIZED FGM COPULA AND ITS PROPERTIES Berzadeh, H., Parham, G.A. and Zadaram, M.R. Deartment of Statstcs, Shahd Chamran Unversty, Ahvaz, Iran. Corresondng

More information

PHYS 705: Classical Mechanics. Calculus of Variations II

PHYS 705: Classical Mechanics. Calculus of Variations II 1 PHYS 705: Classcal Mechancs Calculus of Varatons II 2 Calculus of Varatons: Generalzaton (no constrant yet) Suppose now that F depends on several dependent varables : We need to fnd such that has a statonary

More information

princeton univ. F 17 cos 521: Advanced Algorithm Design Lecture 7: LP Duality Lecturer: Matt Weinberg

princeton univ. F 17 cos 521: Advanced Algorithm Design Lecture 7: LP Duality Lecturer: Matt Weinberg prnceton unv. F 17 cos 521: Advanced Algorthm Desgn Lecture 7: LP Dualty Lecturer: Matt Wenberg Scrbe: LP Dualty s an extremely useful tool for analyzng structural propertes of lnear programs. Whle there

More information

Fuzzy approach to solve multi-objective capacitated transportation problem

Fuzzy approach to solve multi-objective capacitated transportation problem Internatonal Journal of Bonformatcs Research, ISSN: 0975 087, Volume, Issue, 00, -0-4 Fuzzy aroach to solve mult-objectve caactated transortaton roblem Lohgaonkar M. H. and Bajaj V. H.* * Deartment of

More information

Determinants Containing Powers of Generalized Fibonacci Numbers

Determinants Containing Powers of Generalized Fibonacci Numbers 1 2 3 47 6 23 11 Journal of Integer Sequences, Vol 19 (2016), Artcle 1671 Determnants Contanng Powers of Generalzed Fbonacc Numbers Aram Tangboonduangjt and Thotsaporn Thanatpanonda Mahdol Unversty Internatonal

More information

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

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

More information

The Jacobsthal and Jacobsthal-Lucas Numbers via Square Roots of Matrices

The Jacobsthal and Jacobsthal-Lucas Numbers via Square Roots of Matrices Internatonal Mathematcal Forum, Vol 11, 2016, no 11, 513-520 HIKARI Ltd, wwwm-hkarcom http://dxdoorg/1012988/mf20166442 The Jacobsthal and Jacobsthal-Lucas Numbers va Square Roots of Matrces Saadet Arslan

More information

Mechanics Physics 151

Mechanics Physics 151 Mechancs Physcs 151 Lecture 22 Canoncal Transformatons (Chater 9) What We Dd Last Tme Drect Condtons Q j Q j = = j P, Q, P j, P Q, P Necessary and suffcent P j P j for Canoncal Transf. = = j Q, Q, P j

More information

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

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

More information

Dr. Shalabh Department of Mathematics and Statistics Indian Institute of Technology Kanpur

Dr. Shalabh Department of Mathematics and Statistics Indian Institute of Technology Kanpur Analyss of Varance and Desgn of Exerments-I MODULE III LECTURE - 2 EXPERIMENTAL DESIGN MODELS Dr. Shalabh Deartment of Mathematcs and Statstcs Indan Insttute of Technology Kanur 2 We consder the models

More information

Chapter 4 The Wave Equation

Chapter 4 The Wave Equation Chapter 4 The Wave Equaton Another classcal example of a hyperbolc PDE s a wave equaton. The wave equaton s a second-order lnear hyperbolc PDE that descrbes the propagaton of a varety of waves, such as

More information

Linear Approximation with Regularization and Moving Least Squares

Linear Approximation with Regularization and Moving Least Squares Lnear Approxmaton wth Regularzaton and Movng Least Squares Igor Grešovn May 007 Revson 4.6 (Revson : March 004). 5 4 3 0.5 3 3.5 4 Contents: Lnear Fttng...4. Weghted Least Squares n Functon Approxmaton...

More information

1 Matrix representations of canonical matrices

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

More information

Existence results for a fourth order multipoint boundary value problem at resonance

Existence results for a fourth order multipoint boundary value problem at resonance Avalable onlne at www.scencedrect.com ScenceDrect Journal of the Ngeran Mathematcal Socety xx (xxxx) xxx xxx www.elsever.com/locate/jnnms Exstence results for a fourth order multpont boundary value problem

More information

Errors for Linear Systems

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

More information

The Dirac Equation for a One-electron atom. In this section we will derive the Dirac equation for a one-electron atom.

The Dirac Equation for a One-electron atom. In this section we will derive the Dirac equation for a one-electron atom. The Drac Equaton for a One-electron atom In ths secton we wll derve the Drac equaton for a one-electron atom. Accordng to Ensten the energy of a artcle wth rest mass m movng wth a velocty V s gven by E

More information

Yong Joon Ryang. 1. Introduction Consider the multicommodity transportation problem with convex quadratic cost function. 1 2 (x x0 ) T Q(x x 0 )

Yong Joon Ryang. 1. Introduction Consider the multicommodity transportation problem with convex quadratic cost function. 1 2 (x x0 ) T Q(x x 0 ) Kangweon-Kyungk Math. Jour. 4 1996), No. 1, pp. 7 16 AN ITERATIVE ROW-ACTION METHOD FOR MULTICOMMODITY TRANSPORTATION PROBLEMS Yong Joon Ryang Abstract. The optmzaton problems wth quadratc constrants often

More information

Bernoulli Numbers and Polynomials

Bernoulli Numbers and Polynomials Bernoull Numbers and Polynomals T. Muthukumar tmk@tk.ac.n 17 Jun 2014 The sum of frst n natural numbers 1, 2, 3,..., n s n n(n + 1 S 1 (n := m = = n2 2 2 + n 2. Ths formula can be derved by notng that

More information

9 Characteristic classes

9 Characteristic classes THEODORE VORONOV DIFFERENTIAL GEOMETRY. Sprng 2009 [under constructon] 9 Characterstc classes 9.1 The frst Chern class of a lne bundle Consder a complex vector bundle E B of rank p. We shall construct

More information