IZABELLA LABA AND HONG WANG

Size: px
Start display at page:

Download "IZABELLA LABA AND HONG WANG"

Transcription

1 DECOUPLING AND NEAR-OPTIMAL RESTRICTION ESTIMATES FOR CANTOR SETS IZABELLA LABA AND HONG WANG Abstract. For any α (0, d), we construct Cantor sets n R d of Hausdorff dmenson α such that the assocated natural measure µ obeys the restrcton estmate fdµ p C p f L 2 (µ) for all p > 2d/α. Ths range s optmal except for the endpont. Ths extends the earler wor of Chen-Seeger and Shmern-Suomala, where a smlar result was obtaned by dfferent methods for α = d/ wth N. Our proof s based on the decouplng technques of Bourgan-Demeter and a theorem of Bourgan on the exstence of Λ(p) sets. 1. ntroducton We defne the Fourer transform f(ξ) = e 2πx ξ f(x)dx ξ R d. If µ s a measure on R d, we wll also wrte fdµ(ξ) = e 2πx ξ f(x)dµ(x) ξ R d. (1) We are nterested n estmates of the form ĝdµ p C g L q (µ) g L q (µ), where the constant may depend on the measure µ and on the exponents p, q, but not on f. If µ s a probablty measure, we trvally have ĝdµ g L 1 (dµ) g L q (dµ), so that (1) holds wth p = and all q [1, ]. In general, t s not possble to say more than that. However, the problem becomes more nterestng f we restrct attenton to specfc well-behaved classes of measures. There s a vast lterature on restrcton estmates for smooth manfolds (see e.g. [19], [22], [23] for an overvew and a selecton of references). It s well nown that (1) cannot hold wth p < (and any q) when µ s supported on a flat manfold such as a hyperplane. On the other hand, such estmates are possble f µ s the surface measure on a curved manfold M, wth the range of exponents p, q dependng on the geometry of M, n partcular on ts dmenson, smoothness and curvature. Date: July 27,

2 2 IZABELLA LABA AND HONG WANG In the model case when µ s the Lebesgue measure on the sphere S d 1 R d, the classc Tomas-Sten theorem states that (1) holds wth q = 2 and p 2d+2. It was d 1 furthermore conjectured by Sten that for q =, the range of p could be mproved to p > 2d ; ths has been proved for d = 2, but remans open n hgher dmenson, d 1 wth the current best results due to Guth [9], [10]. The conjectured range p > 2d for the sphere, f true, would be the best possble. d 1 Ths follows by lettng f 1 and usng the well nown statonary phase asymptotcs for dµ. The range of p n the Tomas-Sten theorem s also nown to be optmal. Here, the sharpness example s provded by the Knapp constructon where f s the characterstc functon of a small sphercal cap of dameter δ 0. We are nterested n the case when µ s a fractal measure on R d, sngular wth respect to Lebesgue. Here, agan, addtonal assumptons are necessary to mae nontrval estmates of the form (1) possble. For example, f µ s the natural selfsmlar measure on the Cantor ternary set, an easy calculaton shows that (1) cannot hold for any p <. However, f we assume that µ obeys an addtonal Fourer decay condton, then the followng result s nown. Here and below, we use B(x, r) to denote the closed ball of radus r centered at x. Theorem 1. Let µ be a Borel probablty measure on R d. Assume that there are α, β (0, d) and C 1, C 2 0 such that (2) µ(b(x, r)) C 1 r α x R d, r > 0, (3) µ(ξ) C 2 (1 + ξ ) β/2 ξ R d Then for all p (4d 4α + 2β)/β, the estmate (1) holds wth q = 2. Theorem 1 s due to Mocenhaupt [16] and Mtss [15] n the non-endpont range; the endpont was settled later by Ba and Seeger [1]. In the case α = β = d 1, ths recovers the Tomas-Sten theorem for the sphere. The range of exponents p n Theorem 1 s nown to be the best possble n dmenson 1, n the sense that for any 0 < α β < 1, there exsts a probablty measure µ on R, supported on a set of Hausdorff dmenson α and obeyng (2) and ((3), such that (1) fals for all p < (4 4α + 2β)/β, see [12], [3]. The examples are based on a constructon due to Hambroo and Laba [12]: the dea s to modfy a random Cantor-type constructon so as to embed a lower-dmensonal Cantor subset that has much more arthmetc structure than the rest of the set. Ths can be vewed as an analogue of the Knapp example for fractal sets. A hgher-dmensonal varant of the constructon wth d 1 < α β < d s gven n [13]. On the other hand, there exst specfc measures on R d for whch the range of exponents n (1) can be mproved further. If µ s supported on a set of Hausdorff dmenson α < d, t s easy to see usng energy ntegrals that (1) cannot hold outsde

3 DECOUPLING AND NEAR-OPTIMAL RESTRICTION ESTIMATES FOR CANTOR SETS 3 of the range p 2d/α, even f q =. (See e.g. [12, Secton 1]; the counterexample s provded by the functon f 1.) It turns out that there are measures for whch ths range s n fact realzed, wth examples provded by Chen [2], Shmern and Suomala [17], and Chen and Seeger [4]. In partcular, Chen and Seeger [4] proved that for d 1 and α = d/, where N, there are measures supported on a set of Hausdorff dmenson α, obeyng (2) and (3) wth β = α, for whch (1) holds for all p 2d/α. The proofs are based on regularty of convolutons: assumng that α = d/, the ey ntermedate step s to prove that the -fold self-convoluton µ µ s absolutely contnuous. Ths method, however, does not yeld optmal exponents when α d/ wth nteger. Our man result s as follows. Theorem 2. Let d N and 0 < α < d. Then there exsts a probablty measure supported on a subset of [0, 1] d of Hausdorff dmenson α such that: for every 0 < γ < α, there s a constant C 1 (γ) such that (4) µ(b(x, r)) C 1 (γ)r γ x R d, r > 0 for every β < mn(α/2, 1), there s a constant C 2 (β) > 0 such that (5) µ(ξ) C 2 (β)(1 + ξ ) β ξ R d, (6) for every p > 2d/α, we have the estmate ĝdµ p C 3 (p) g L 2 (µ) g L 2 (µ). Ths complements the results of [2], [4], [17], and provdes a matchng (except for the endpont) result for all dmensons 0 < α < d that are not of the form α = d/. The frst man ngredent of our constructon s Bourgan s theorem on Λ(p) sets [5] (see also Talagrand [20]). In ts full generalty, Bourgan s theorem apples to general bounded orthogonal systems of functons. We state t here n the specfc case of exponental functons on the unt cube n R d. Ths provdes an optmal restrcton estmate on each sngle scale n the Cantor constructon. Theorem 3. (Bourgan [5]) Let p > 2. For every N N suffcently large, there s a set S = S N {0, 1,..., N 1} d of sze t c 0 N 2d/p such that (7) a S c a e 2πa x L p [0,1] d C(p)( a S c a 2 ) 1/2 wth the constants c 0 and C(p) ndependent of N. (The set S s called a Λ(p)-set.) To pass from here to restrcton estmates for multscale Cantor sets, we use the decouplng technques of Bourgan and Demeter [6], [7]. Ths produces localzed restrcton estmates of the form (8) f L 2 (µ) C ɛ R ɛ f L p ([ R,R] d ),

4 4 IZABELLA LABA AND HONG WANG or equvalently by dualty, (9) ĝdµ L p ([ R,R] d ) C ɛ R ɛ g L 2 (µ) for all ɛ > 0, wth constants ndependent of R. The R ɛ factors account for the fact that we lose a constant factor at each step of the teraton. We wll try to mnmze these losses by applyng Bourgan s theorem to an ncreasng sequence of values of N, but we wll not be able to avod them completely. Fnally, we use a varant of Tao s epslon removal lemma [21] to deduce the global restrcton estmate (6) from (8). Ths removes the R ɛ factors, but at the cost of losng the endpont exponent p = 2d/α. It s not clear whether the endpont estmate can be obtaned wth our current methods. Our proof of the localzed restrcton estmate (8) s fully determnstc. However, the epslon removal lemma requres a pontwse Fourer decay estmate for µ. Randomzng our constructon enables us to prove the estmate (5) va an argument borrowed from [14], [17]. Ths proves the Fourer decay part of Theorem 2, and s also suffcent to complete the epslon removal argument. If d 2, or f d 2 and α d 2, the Cantor set supportng µ n Theorem 2 s a Salem set (.e. ts Fourer dmenson s equal to ts Hausdorff dmenson). The condton α d 2 s necessary for ths type of constructons to produce a Salem set, for the same reasons as n [17]. We note, however, that our proof of (6) wth p > 2d/α does not requre optmal Fourer decay and that the estmate (5) for any β > 0 would suffce. 2. The decouplng machnery We wll use the decouplng machnery developed by Bourgan and Demeter [6], [7]. In ths paper, we wll follow the conventons of [7], wth the surface measure on a parabolod replaced by the natural measure on a Cantor set. We use X Y to say that X CY for some constant C > 0, and X Y to say that X Y and X Y. The constants such as C, C, etc. and the mplct constants n may change from lne to lne, and may depend on d and p, but are ndependent of varables or parameters such as x, N, R, j, l. For quanttes that depend on parameters such as ɛ, we wll wrte X(ɛ) ɛ Y (ɛ) as shortcut for for every ɛ > 0 there s a constant C ɛ > 0 such that X(ɛ) C ɛ Y (ɛ). We wrte [N] = {0, 1,..., N 1} and B(x, r) = {y R d : x y r}. We use to use the Eucldean (l 2 ) norm of a vector n R d, the cardnalty of a fnte set, or the d-dmensonal Lebesgue measure of a subset of R d, dependng on the context. Occasonally, we wll also use the l norm on R d : f x = (x 1,..., x d ) R d, we wrte x = max( x 1,..., x d ). We wll also sometmes use F for the Fourer transform, so that Ff = f.

5 DECOUPLING AND NEAR-OPTIMAL RESTRICTION ESTIMATES FOR CANTOR SETS 5 Followng [7], we wll use cube-adjusted weghts. An R-cube wll be a d-dmensonal cube of sde length R, wth all sdes parallel to coordnate hyperplanes. Unless stated otherwse, we wll assume R-cubes to be closed. If I s an R-cube centered at c, we defne w I (x) = ( 1 + ) 100 x c R and ( ) 1/p 1 F L p (w I) = F p w I. I If η : R [0, ) s a functon (usually Schwartz), and I s as above, we wll wrte ( ) x c η I (x) = η. R If g : R C s a functon, I s an nterval, and σ s a measure (whch wll usually be clear from context), we wll wrte E I g = F 1 (1 I gdσ). We wll use the followng tools from Bourgan-Demeter, whch we restate here n a verson adapted to our settng. Lemma 1. (Reverse Hölder nequalty, [7, Corollary 4.2]). Let 1 p q. If I s a 1/R-cube and J s an R-cube, then (10) E I g L q (w J ) E I g L p (w J ) wth the mplct constant ndependent of R, I, J, g. Lemma 2. (L 2 decouplng, [7, Proposton 6.1]). Let I be a /R-cube for some N, and let I = I 1 I be a tlng of I by 1/R-cubes dsjont except for ther boundares. Then for any R-cube J we have (11) E I g 2 L 2 (w J ) j E Ij g 2 L 2 (w J ). Lemma 3. (Band-lmted functons are locally constant, cf. [8, 2.2]) There s a non-negatve functon η L 1 (R d ) such that the followng holds. For every R > 0, and every ntegrable functon h : R d C supported on a 1/R-cube I, there s a functon H : R [0, ) such that: H s constant on each sem-closed R-cube J ν := Rν + [0, R) d, ν Z d, ĥ(x) H(x) ( ĥ η R)(x) for all x R, where η R (y) = 1 η( y ). In R d R partcular, (12) H L 1 (R d ) η L 1 (R d ) ĥ L 1 (R d ).

6 6 IZABELLA LABA AND HONG WANG Proof. Replacng h by h( c) and ĥ(x) by e 2πc x ĥ(x) f necessary, we may assume that I = [0, 1 R ]d. Let χ be a non-negatve Schwartz functon such that χ 1 on [0, 1] d and that χ(x) s non-negatve, radally symmetrc and decreasng n x. Then χ(r ) 1 on I, and χ(r ) = 1 χ( ). Defne R d R and η(x) := sup y x 1 χ(y) H(x) := sup{ ĥ(y) : x, y belong to the same J ν}. Clearly, η s ntegrable and H s constant on each J ν. We have ĥ(x) H(x) by defnton. To prove the second nequalty, we note that h = hχ(r ), so that ĥ = ĥ χ(r ). Suppose that x J ν for some ν Z d, then for each y J ν we have ĥ(y) ĥ(z) 1 R χ(y z d R )dz Snce x y R, we have y z x z R R = y x R 1, so that by the defnton of η we have η( x z y z ) χ( ). Hence R R ĥ(y) ĥ(z) 1 R η(x z d R )dz = ( ĥ η R)(x), and the desred nequalty follows upon tang the supremum over y J ν. Fnally, by Fubn s theorem and rescalng we have H L 1 (R d ) ĥ L 1 (R d ) η R L 1 (R d ) = ĥ L 1 (R d ) η L 1 (R d ). Corollary 1. For every R > 0, M N, every ntegrable functon h : R d supported on an (MR) 1 -cube I, and every R-cube J, we have C (13) ĥ L 1 (w J ) 1 M d ĥ L 1 (R d ). Proof. Let L ν = MRν + [0, MR) d for ν Z d. Let H be the functon provded by Lemma 3 wth R replaced by MR, so that on each L ν we have H(x) H ν for some

7 DECOUPLING AND NEAR-OPTIMAL RESTRICTION ESTIMATES FOR CANTOR SETS 7 constant H ν 0. Then ĥ L 1 (w J ) H(x)w J (x)dx = H ν w J (x)dx ν L ν H ν w J (x)dx ν R = H ν R d w [0,1] d(x)dx. ν R Let C 1 := R d w [0,1] d(x)dx, then ĥ L 1 (w J ) C 1 H ν R d = C 1 H M d ν (MR) d ν ν = C 1 H(x)dx M d 1 M d ĥ L 1 (R d ), where at the last step we used (12). R 3. Sngle-scale decouplng We begn wth a sngle-scale decouplng nequalty for Cantor sets wth Λ(p) alphabets. We wll need the followng contnuous verson of Theorem 3. Lemma 4. Let p > 2, and let S [N] d be as n Theorem 3. Then for all h supported on E := S + [0, 1] d we have the nequalty (14) ĥ L p ([0,1] d ) C(p) h L 2 (R d ) Proof. We have a S By (7), t follows that c a e 2πa x L = sup f, c a e 2πa x p ([0,1] d ) f L p ([0,1] d =1 ) a S = sup f, c a δ a f L p ([0,1] d =1 ) a S f L p ([0,1] d =1 ) = sup sup sup c a l 2 (S) =1 f L p ([0,1] d =1 ) a S a S c a f(a) c a f(a) C(p),

8 8 IZABELLA LABA AND HONG WANG so that f(a) l 2 (S) C(p) f L p ([0,1] d ) Smlarly, for any translate S + z of S we have Integratng n z [0, 1] d, we get f(a) l 2 (S+z) C(p) f L p ([0,1] d ) (15) f 2 L 2 (E) = [0,1] d f(a) l 2 (S+z)dz C(p) 2 f 2 L p ([0,1] d ). Argung agan by dualty, we have f L 2 (E) = sup g L 2 (E) =1 = sup g L 2 (E) =1 g1 E fdx F(g1 E ) fdx Usng (15), and tang the supremum over f wth f L p ([0,1] d ) 1, we get whch s (14) wth h = g1 E. F(g1 E ) L p ([0,1] d ) C(p) g L 2 (E), We note that the concluson of Lemma 4 remans true f we assume that h s supported on S + [ 1/2, 3/2] d nstead of E. Ths s proved by wrtng h as a sum of 2 d functons supported on translates of E and applyng Lemma 4 to each of them. We can now prove our frst decouplng nequalty. Lemma 5. Let S [N] d be a Λ(p)-set as n Theorem 3, and let E = S + [0, 1] d. Let f : R C be a functon such that g := f = s supported on E. For each a S, let g a = g1 a+[0,1] d, and defne f a va f a = g a. Then (16) f 2 L p (w I ) C(p) 2 a S for any 1-cube I. f a 2 L p (w I ) Proof. We frst rewrte the rght-hand sde of (16) usng Lemmas 2 and 1 wth σ equal to the Lebesgue measure. We have so that E a+[0,1] dg = ĝ a = f a, E [0,N] dg = ĝ = f, f 2 L 2 (w I ) a S f a 2 L 2 (w I ) a S f a 2 L p (w I ).

9 DECOUPLING AND NEAR-OPTIMAL RESTRICTION ESTIMATES FOR CANTOR SETS 9 Therefore to prove (16), t suffces to prove that (17) f 2 L p (w I ) C(p) 2 f 2 L 2 (w I ) Let η be a nonnegatve Schwartz functon such that η(x) = η( x), η 1 on [ 1, 1] d and supp η [ 1 2, 1 2 ]d. We wll prove that (18) f 2 L p (I) C(p) 2 f 2 L 2 (η I ) for every 1-cube I. By a coverng argument [7, Lemma 4.1], ths mples (17). We may assume that I = [0, 1] d. (If I = z +[0, 1] d ], we may replace f by f z = f( + z) and observe that f z (ξ) = e 2πz ξ f(ξ) s agan supported n E.) Let h = g ( η)ˇ, so that ĥ = f η and h s supported on S + [ 1/2, 3/2] d. Snce η 1 on [0, 1] d, we have f L p ([0,1] d ) f η L p ([0,1] d ). By Lemma 4 and the remar after ts proof appled to h, as clamed. f η L p ([0,1] d ) C(p) h L 2 (R) = f η L 2 (R) = f L 2 (η) 4. The Cantor set constructon Our proof of Theorem 2 s based on the constructon of a multscale Λ(p) Cantor set of dmenson α. Let α (0, d), p = 2d/α, and let {n j } j N be a sequence of postve ntegers. For the constructon of the measure µ n Theorem 2, we wll assume the followng condtons on n j : (19) n 1 n 2..., n, (20) ɛ > 0 C ɛ > 0 N n +1 C ɛ (n 1... n ) ɛ. However, large parts of our proof wor under weaer assumptons. In partcular, our localzed restrcton estmate n Lemma 8 contnues to hold f n j = n for all j. We also note here that n order for (20) to hold, t s enough to assume that (19) holds and that n j grow slowly enough, for example (21) wll suffce. For each j N, let n j+1 j + 1 n j j Σ j = Σ j (n j, t j, c 0, C(p)) = {S [n j ] d : S = t j and (7) holds wth N = n j } By Theorem 3, there are c 0, C(p) > 0 (ndependent of j) and t j wth t j c 0 n 2d/p j such that Σ j s non-empty for all j. Henceforth, we fx these values of c 0, C(p) and

10 10 IZABELLA LABA AND HONG WANG t j. By the well nown upper bounds on the sze of Λ(p) sets (see [5]), we must n fact have (22) c 0 n 2d/p j t j c 1 n 2d/p j for some constant c 1 ndependent of j. Let N = n 1... n and T = t 1... t. We construct a Cantor set E of Hausdorff dmenson α as follows. Defne A 1 = N 1 1 S 1, E 1 = A 1 + [0, N 1 1 ] d, for some S 1 Σ 1. For every a A 1, choose a Λ(p) set S 2,a Σ 2 wth S 2,a = t 2, and let A 2,a = a + N2 1 S 2,a, A 2 = A 2,a, E 2 = A 2 + [0, N2 1 ] d. a A 1 We contnue by nducton. Let 2, and suppose that we have defned the sets A j and E j, j = 1, 2,...,. For every a A, choose S +1,a Σ +1 wth S +1,a = t +1, and let A +1,a = a + N 1 +1 S +1,a, A +1 = a A A +1,a, E +1 = A +1 + [0, N 1 +1 ]d. Ths produces a sequence of sets [0, 1] d E 1 E 2 E 3..., where each E j conssts of T j cubes of sde length N 1 j. For each j, let µ j = 1 E j 1 E j. We wll dentfy the functons µ j wth the absolutely contnuous measures µ j dx. It s easy to see that µ j converge wealy as j to a probablty measure µ supported on the Cantor set E := j=1 E j. We note that for each N 1 j -cube τ of E j, and for all l > j, we have µ j (τ) = µ l (τ) = µ(τ) = T 1 j. For the tme beng, the specfc choce of the sets A,a does not matter, as long as they are Λ(p)-sets of the prescrbed cardnalty. Our multscale decouplng nequalty n Proposton 1 and the localzed restrcton estmate n Corollary 2 do not requre any addtonal condtons. However, addtonal randomzaton of these choces wll become mportant later n provng our global restrcton estmate. Lemma 6. Assume that (19) and (20) hold. Then the set E has Hausdorff dmenson α. Moreover, for every 0 γ < α there s a constant C 1 (γ) such that (23) µ(b(x, r)) C 1 (γ)r γ x R d, r > 0. Proof. We frst note that (19), (20) and (22) mply that (24) N α 2ɛ j+1 ɛ N α ɛ j ɛ T j ɛ N α+ɛ j ɛ N α+2ɛ j 1.

11 DECOUPLING AND NEAR-OPTIMAL RESTRICTION ESTIMATES FOR CANTOR SETS 11 Indeed, from (22) we have c j 0Nj α T j c j 1Nj α, whch mples N α ɛ j ɛ T j ɛ N α+ɛ j by (19). The remanng two nequaltes n (24) follow from (20). We frst prove (23). If r > N1 1, then (23) holds trvally wth C 1 = N γ 1. Suppose now that N 1 j+1 < r N 1 j for some j 1. Then any ball B(x, r) ntersects at most a bounded number of the N 1 j -cubes of E j, so that µ(b(x, R)) T 1 j γ N γ j+1 rγ for all γ < α (the second nequalty n the sequence follows from (24)). To prove the dmenson statement, we only need to show that E has Hausdorff dmenson at most α, snce the lower bound s provded by (23). To ths end, t suffces to chec that for every ɛ > 0, and for all r > 0, the set E can be covered by C ɛ r α ɛ balls of radus r. Agan, t suffces to consder r > N1 1. Suppose that N 1 j+1 < r N 1 j, then E E j+1 can be covered by T j+1 balls of radus N 1 j+1, hence also of radus r. Snce T j+1 ɛ N α+ɛ j r α ɛ, the desred bound follows. 5. Multscale decouplng Our goal n ths secton s to derve the followng multscale decouplng nequalty for fnte teratons of Cantor sets. For a A, let τ,a = a + [0, N 1 ]d. If f : R d C s a functon, we defne f,a va f,a = 1 τ,a f. Proposton 1. There s a constant C 0 (p) (ndependent of ) such that for any N -cube J, and for any functon f wth supp f E, we have ( ) 1/p ( ) 1/2 (25) f p L p (w I ) C 0 (p) f,a 2 L p (w J ), I I a A where J = I I I s a tlng of J by 1-cubes. Proof. The dea s to terate Lemma 5. Applyng t to the set N 1 E 1 and a rescalng of f by N 1, we see that there s a constant C 1 (p) such that for any functon f wth supp f E 1, and for any N 1 -cube J, we have (26) f 2 L p (w J ) C 1 (p) 2 a A 1 f 1,a 2 L p (w J ). Smlarly, applyng Lemma 5 to a rescalng of f j,a by N j+1 for each a A j, we see that for any N j+1 -cube J and for any functon f wth supp f E j+1 we have (27) f j,a 2 L p (w J ) C 1 (p) 2 b A j+1,a f j+1,b 2 L p (w J ) wth the same constant C 1 (p). To connect the steps of the teraton, we wll need a smple lemma on mxed norms.

12 12 IZABELLA LABA AND HONG WANG Lemma 7. Let {c j } be a double-ndexed sequence (fnte or nfnte) wth c j 0. Then for p > 2, ( ) p/2 ( ( 2/p ) p/2. (28) cj) p c 2 j j j Proof. Let F j () = c 2 j, and G() = j F j() = j c2 j, so that ( ( p/2 ) 2/p. G p/2 = cj) 2 j On the other hand, by Mnows s nequalty G p/2 j F j () p/2 = j ( c p j) 2/p, and the lemma follows. We wll prove (25) by nducton n. For an m-cube J wth m N, let J = I I(J) I be a tlng of J by 1-cubes. Let C 2 be a constant such that (29) w I C 2 w J. I I(J) It s easy to see that such a constant exsts and can be chosen ndependently of J. We wll prove that (25) holds wth C 0 (p) = C 1 (p)c 1/p 2. To start the nducton, let J be an N 1 -cube, then by (29) and (26), f p L p (w I ) C 2 f p L p (w J ) I I(J) C 1 (p) p C 2 ( a A 1 f 1,a 2 L p (w J ) Ths s (25) for = 1. Suppose now that we have proved (25) for = j. Let J be an N j+1 -cube, and let J = L L I be a tlng of J by N j-cubes. Then f p L p (w I ) I I(J) f p L p (w I ) = L L I I(L) C 0 (p) jp L L ( a A j f j,a 2 L p (w L ) ) p/2 ) p/2

13 DECOUPLING AND NEAR-OPTIMAL RESTRICTION ESTIMATES FOR CANTOR SETS 13 by our nductve assumpton. Usng Lemma 7, a rescalng of (29), and (27), we see that f p L p (w I ) C 0(p) jp ( ) 2/p p/2 f j,a p L p (w L ) I I(J) a Aj L L C 0 (p) jp C 2 f j,a 2 L p (w J ) a Aj p/2 C 0 (p) jp C 2 C 1 (p) f j+1,a 2 L p (w J ) a A j+1 Ths ends the nductve step and the proof of the proposton. p/2. 6. From decouplng to localzed restrcton Lemma 8. Let E and µ be as n Secton 5. Let J be an N -cube. Then for all g L 2 (dµ), we have ĝdµ L p (J) C 0 (p) N d/p wth the mplct constants ndependent of. T 1/2 g L 2 (dµ) Proof. It suffces to prove that for all l >, and for all g L 2 (R) supported on E l, we have gdµ l L p (J) C 0 (p) N d/p T 1/2 g L 2 (dµ l ) wth the mplct constants ndependent of and l. The clam then follows by tang the lmt l. We contnue to use the Cantor set notaton from Sectons 4 and 5. For a A j, let g j,a = 1 a+[0,+n 1 j ] g. By Proposton 1 and Lemma 1, we have d gdµ ( ) 1/2 l L p (J) C 0 (p) g,a dµ l 2 L p (w J ) a A ( C 0 (p) N d p d ) 1/2 2 g,a dµ l 2 L 2 (w J ) a A For each a A, let B l,a be the set of l-th level descendants of a (more precsely, B l,a = {b A l : b + [0, N 1 l ] d a + [0, N 1 ]d }. Note that B l,a = T l /T. By

14 14 IZABELLA LABA AND HONG WANG Cauchy-Schwartz, so that g,a dµ l L 2 (w J ) ĝ l,b dµ l L 2 (w J ) b B l,a gdµ l L p (J) C 0 (p) N d p d 2 C 0 (p) N d p d 2 ( Tl T ( Tl T ( Tl T ) 1/2 ( ) 1/2 ( ĝ l,b dµ l 2 L 2 (w J ) b A l ) 1/2 ( ) d/2 N ( N l b B l,a ĝ l,b dµ l 2 L 2 (w J ) ) 1/2 b A l ĝ l,b dµ l 2 L 2 (R) ) 1/2. At the last step, we appled Corollary 1 to the functons (g l,b dµ l )( ) (g l,b dµ l )( ) supported on 2N 1 l -cubes. Snce ĝ l,b dµ l 2 L 2 (R) = g l,b dµ l 2 L 2 (R) b A l b A l we fnally have = gdµ l 2 L 2 (R) = N d l T 1 l g 2 L 2 (µ l ) ) 1/2 gdµ l L p (J) C 0 (p) N d p 1 2 ( Tl T ) 1/2 ( N N l = C 0 (p) N d/p T 1/2 g L 2 (dµ l ) ) d/2 ( ) N d 1/2 l g L2(µl) T l as clamed. Corollary 2. (Localzed restrcton estmate) Assume that (19) holds. Then for any ɛ > 0 we have the estmate (30) ĝdµ L p (J) C ɛ R ɛ g L 2 (dµ). for all R n 1 and for all R-cubes J. The constant C ɛ depends on ɛ, but not on g, R or J. Equvalently, for any f supported n J, we have (31) f L 2 (dµ) C ɛ R ɛ f L p (J).

15 DECOUPLING AND NEAR-OPTIMAL RESTRICTION ESTIMATES FOR CANTOR SETS 15 Proof. Suppose that N < R N +1, and let J be an N +1 -cube contanng J. By Lemma 8 and (19), we have ĝdµ L p (J) ĝdµ L p (J ) C 0 (p) +1 N d/p +1 T 1/2 +1 g L 2 (dµ) C 0 (p) +1 N d/p +1 (c+1 0 N 2d/p +1 ) 1/2 g L 2 (dµ) (C 0 (p)c 1/2 0 ) +1 g L 2 (dµ) ɛ R ɛ g L 2 (dµ) as clamed. The second part (31) follows by dualty. 7. Global restrcton estmate Proposton 2. Assume that n obey (19) and (20). Suppose furthermore that µ obeys a pontwse decay estmate (32) µ(x) (1 + x ) β for some β > 0. Then for any q > p we have the estmate (33) ĝdµ L q (R) g L 2 (dµ). The mplct constant depends on the measure µ and on q, but not on g. Equvalently, (34) f L 2 (dµ) f L q (R). To prove ths, we adapt Tao s epslon-removal argument, see [21, Theorem 1.2]. It suffces to prove Lemma 9 below; once ths s done, the proof of the proposton s completed exactly as n [21], wth Lemma 9 replacng Tao s Lemma 3.2. Lemma 9. Assume that n, t, µ are as n Theorem 2, and let R > 0 be large enough. Suppose that {I 1,..., I M } s a sparse collecton of R-cubes, n the sense that ther centers x 1,..., x M are R B M B -separated for some large enough constant B (dependng on β). Then for any f supported on M j=1 I j, we have f L 2 (dµ) ɛ R Cɛ f L p (R). Here and below, the constant C n the exponent may depend on B, and may change from lne to lne, but s ndependent of R, M, ɛ, or f. Proof. We follow the outlne of Tao s argument, wth modfcatons necessary to adapt t to our settng. We frst note the followng estmate: f f s supported n an R-cube J wth R N l, then (35) f L 2 (dµ l ) ɛ R ɛ f L p (J).

16 16 IZABELLA LABA AND HONG WANG The mplct constant depends on ɛ, but not on f, R or J. Ths s proved as n Lemma 8 and Corollary 2, except that we do not tae the lmt l n the proof of Lemma 8. Let N be such that N R < N +1. We have E = T N d ; by (24), ths mples that (36) R 2d p d ɛ ɛ E ɛ R 2d p d+ɛ. Let f = f φ, where supp f I and φ = φ I for a fxed Schwartz functon φ such that φ 0, φ 1 on [ 1, 1] d, and supp φ [ 1, 1] d. Note that φ (x) = R d φ(rx), and n partcular φ s supported n [ R 1, R 1 ] d. Then f = φ f = φ f. By the support propertes of φ, for x E we actually have f(x) = φ ( f 1 E )(x), where we abuse the notaton slghtly and use E to denote both the -th stage set from the Cantor teraton and a CN 1 -neghbourhood of E. Ths s harmless snce ether set can be covered by a bounded number of translates of the other. We clam that the followng holds: for all r [1, 2], and for any collecton of functons F 1,..., F M L 2 (R d ), we have (37) F φ r ɛ R Cɛ E r/2 F r L 2 L (µ) 2 (R d ). Assumng the clam (37), we complete the proof of the lemma as follows. F = f 1 E, and observe that E r/2 F r L 2 (R d ) = f r L 2 (µ ). Applyng (37) to F wth r = p, and then usng (35), we get p f L 2 (dµ) = F φ p L 2 (µ) ɛ R Cɛ E p /2 ɛ R Cɛ f p L 2 (µ ) F p L 2 (R d ) Let ɛ R Cɛ f p L p (R d ) as requred. R Cɛ f p L p (R d )

17 DECOUPLING AND NEAR-OPTIMAL RESTRICTION ESTIMATES FOR CANTOR SETS 17 It remans to prove (37). We wll do so by nterpolatng between r = 1 and r = 2. For r = 1, t suffces to prove that for each, (38) F φ L 2 (µ) ɛ R Cɛ E 1/2 F L 2 (R d ), snce ths mples (37) by trangle nequalty. To prove (38), we nterpolate between L 1 and L estmates. Frst, we have by Fubn s theorem F φ L 1 (µ) F (x y) φ (y) dy dµ(x) ( ) = F (u) φ (v u) dµ(v) du R d sup µ(x + [ R 1, R 1 ] d ) F L 1 (R d ) x ɛ R d R 2d p +ɛ F L 1 (R d ) ɛ R Cɛ E 1 F L 1 (R d ) where at the last step we used (36). Interpolatng ths wth the pontwse bound sup F φ (x) F L (R d ) φ L 1 (R d ) F L (R d ) x we get (38). To complete the argument, we need to prove (37) wth r = 2. Defne the functons g va ĝ = F, so that g L 2 (R) = F L 2 (R) and F φ = ĝφ. We thus need to prove that (39) 2 ĝ φ ɛ R Cɛ E 1 g 2 L 2 L (µ) 2 (R d ). By translatonal nvarance and the rapd decay of φ, t suffces to prove (39) wth φ replaced by 1 B. Let R be the operator R(h) = ĥ E. By Corollary 2, R s a bounded operator from L p (J) to L 2 (µ) for any bounded cube J (wth norm dependng on J). We have to prove that R( ( ) 1/2 g 1 I ) ɛ R Cɛ E 1/2 g 2 L 2 L (µ) 2 (R d ) By the T T argument, t suffces to prove that ( 1 ) 2 1/2 ( I R R g j 1 Ij ɛ R Cɛ E 1 g 2 L 2 (R d ) j L 2 (R d ) ) 1/2

18 18 IZABELLA LABA AND HONG WANG By Schur s test, ths follows from (40) sup 1 I R R1 Ij h L j 2 (R d ) ɛ R Cɛ E 1 h L 2 (R ). d We clam that (41) 1 I R R1 I h L 2 (R d ) ɛ R Cɛ E 1 h L 2 (R d ) and (42) 1I R R1 Ij h L 2 (R d ) M 2 h L 2 (R d ), j, wth constants ndependent of, j. Together, these two mply (40). We frst prove (41), By Lemma 2, Hölder s nequalty, and by (36), we have R1 I h L 2 (dµ) ɛ R ɛ 1 I h L p (R d ) ɛ R Cɛ R d 2 d p h L 2 (R d ) ɛ R Cɛ E 1/2 h L 2 (R d ). Ths mples (41) by the T T argument wth a fxed. For j, we note that R Rh = h µ, so that 1 I R R1 Ij s an ntegral operator wth the ernel K j (x, y) = 1 I (x)1 Ij (y) µ(x y). By (32), K(x, y) dy I j M Bβ R Bβ M 2 f B was chosen large enough dependng on β. The clamed estmate (42) now follows from Schur s test. 8. Fourer decay To complete the proof of Theorem 2, t now suffces to prove that the Cantor set n Secton 4 can be constructed so that (19), (20), and (5) all hold. Snce (5) mples (32), the restrcton estmate (6) wll follow from Proposton 2. In all our ntermedate results so far, t dd not matter how the Λ(p) alphabet sets S,a were chosen, as long as they had the prescrbed cardnaltes. Here, however, t s crucal to randomze the choce of S,a. Theorem 4. Let {n } N and {t } N be two determnstc sets of ntegers such that (19), (20), (22) all hold, and that Σ s non-empty for each (as provded by Bourgan s theorem). Let {µ } N {0} be a sequence of random measures on [0, 1] d such that: µ 0 = 1 [0,1] d, µ 1, µ 2,... are constructed nductvely va the teratve process descrbed n Secton 4,

19 DECOUPLING AND NEAR-OPTIMAL RESTRICTION ESTIMATES FOR CANTOR SETS 19 for each N, the sets S,a are chosen randomly and ndependently from Σ, wth probablty dstrbuton such that (43) E(µ (x) E n ) = µ 1 (x) x [0, 1] d. Then the lmtng Cantor measure µ almost surely obeys all conclusons of Theorem 2. An example of a random constructon of µ that meets the condton (43) s as follows. Choose n and t as ndcated n the theorem (recall that for (20) to hold, t suffces to assume (21)). For each N, choose a Λ(p) set B [n ] d such that B = t c 0 n 2d/p and (7) holds wth n = n, for some c 0, C(p) ndependent of. Let B = {B,v : v [n ] d }, B,v [n ] d, B,v = v + B mod (n Z) d Then B Σ (n, t, c 0, 2 d C(p)), snce any functon supported on B,v s a sum of at most 2 d functons supported on translates of B. Set A 0 = {0}. Let now 1, and assume that A 1 has been constructed. For each a A 1, choose a random v(, a) [n ] d so that P(v(, a) = v) = n d for each v [n ] d and the choces are ndependent for dfferent a A 1. Let S,a = B,v(a), a random translate of B, and contnue the constructon as n Secton 4. Then (43) holds by translatonal averagng, and all other assumptons of the theorem hold wth C(p) replaced by 2 d C(p). Instead of usng random translates of a sngle set B for each, we could choose a set B,a Σ for each a A 1, then let S,a = B(, a) + v(, a) mod (n Z) d, where v(, a) s a random translaton vector n [n ] d as above, chosen ndependently of B,a and ndependently of the choces made for all other a. Bourgan s theorem [5] shows that a generc subset of [N] d of sze about N 2d/p s a Λ(p) set, so that Σ (wth an approprate choce of c 0 and C(p)) should be large for most values of t j n the ndcated range, provdng many sets avalable for the constructon. Other varants are possble. We now turn to the proof of the theorem. Proof. By Lemma 6, E = supp µ has Hausdorff dmenson α, and µ obeys (4) for all 0 < γ < α. The Fourer decay estmate (5) s proved by a calculaton almost dentcal to that n [14, Secton 6] for a specal case n dmenson 1, and n [17, Theorem 14.1] (see also [18, Theorem 4.2]) for more general measures n hgher dmensons. The proof n [17, Theorem 14.1] can be followed here almost word for word, except for the trval changes n parameters to allow a varable sequence {n j } nstead of a constant one (see e.g. [3]). It s easy to chec that the proof goes through as long as log N +1 ɛ N ɛ, N. Snce log N +1 = log N +log n +1 ɛ N ɛ +nɛ +1, ths s a weaer condton than (20).

20 20 IZABELLA LABA AND HONG WANG Fnally, the restrcton estmate (6) holds by Proposton 2 and by (5). 9. Acnowledgements Ths wor was started whle the frst author was vstng the Insttute for Computatonal and Expermental Research n Mathematcs (ICERM). The frst author was supported by the NSERC Dscovery Grant 22R We would le to than Laura Clade, Semyon Dyatlov, Larry Guth, Mar Lewo, Pablo Shmern and Josh Zahl for helpful conversatons. References [1] J.-G. Ba, A. Seeger, Extensons of the Sten-Tomas theorem, Math. Res. Lett. 18 (2011), no. 4, [2] X. Chen, A Fourer restrcton theorem based on convoluton powers, Proc. Amer. Math. Soc. 142 (2014), [3] X. Chen, Sets of Salem type and sharpness of the L 2 -Fourer restrcton theorem, Trans. Amer. Math. Soc. 368 (2016), [4] X. Chen, A. Seeger, Convoluton powers of Salem measures wth applcatons, preprnt, 2015, [5] J. Bourgan, Bounded orthogonal systems and the Λ(p)-set problem, Acta Math. 162 (1989), [6] J. Bourgan, C. Demeter, The proof of the l 2 decouplng conjecture, Ann. Math. 182 (2015), [7] J. Bourgan, C. Demeter, A study gude for the l 2 decouplng theorem, preprnt, 2016 [8] J. Bourgan, L. Guth, Bounds on oscllatory ntegral operators based on multlnear estmates, Geom. Funct. Anal. 21 (2011), [9] L. Guth, Restrcton estmates usng polynomal parttonng, J. Amer. Math. Soc. 29 (2016), [10] L. Guth, Restrcton estmates usng polynomal parttonng II, preprnt, 2016, [11] K.Hambroo, Restrcton theorems and Salem sets, Ph.D. thess, Unversty of Brtsh Columba (2015). [12] K. Hambroo, I. Laba, On the sharpness of Mocenhaupt s restrcton theorem, Geom. Funct. Anal. 23 (2013), no. 4, [13] K. Hambroo, I. Laba, Sharpness of the Mocenhaupt-Mtss-Ba-Seeger Restrcton Theorem n Hgher Dmensons, to appear n Bull. London Math. Soc. [14] I. Laba, M. Praman, Arthmetc progressons n sets of fractonal dmenson, Geom. Funct. Anal. 19 (2009), no. 2, [15] T. Mtss, A Sten-Tomas restrcton theorem for general measures, Publ. Math. Debrecen 60 (2002),

21 DECOUPLING AND NEAR-OPTIMAL RESTRICTION ESTIMATES FOR CANTOR SETS 21 [16] G. Mocenhaupt, Salem sets and restrcton propertes of Fourer transforms, Geom. Funct. Anal. 10 (2000), [17] P. Shmern, V. Suomala, Spatally ndependent martngales, ntersectons, and applcatons, Memors of the Amer. Math. Soc. to appear, [18] P. Shmern, V. Suomala, A class of random Cantor measures, wth applcatons, preprnt, 2016, [19] E. M. Sten, Harmonc Analyss, Prnceton Unv. Press, [20] M. Talagrand, Sectons of smooth convex bodes va majorng measures, Acta Math. 175 (1995), [21] T. Tao, The Bochner-Resz conjecture mples the restrcton conjecture, Due Math. J. 96 (1999), [22] T. Tao, Some Recent Progress on the Restrcton Conjecture, Fourer Analyss and Convexty, , Appl. Numer. Harmon. Anal., Brhuser Boston, Boston, MA, [23] T. Tao, Recent progress on the restrcton conjecture, preprnt, 2003, [24] P. A. Tomas, A restrcton theorem for the Fourer transform, Bull. Amer. Math. Soc. 81 (1975), [25] P. A. Tomas, Restrcton theorems for the Fourer transform, n Harmonc Analyss n Eucldean Spaces (Proc. Sympos. Pure Math. 35, Amer. Math. Soc., 1979, vol I), Department of Mathematcs, UBC, Vancouver, B.C. V6T 1Z2, Canada laba@math.ubc.ca Department of Mathematcs, MIT, Cambrdge, MA 02139, USA hongwang@mt.edu

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

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

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

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

Math 702 Midterm Exam Solutions

Math 702 Midterm Exam Solutions Math 702 Mdterm xam Solutons The terms measurable, measure, ntegrable, and almost everywhere (a.e.) n a ucldean space always refer to Lebesgue measure m. Problem. [6 pts] In each case, prove the statement

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

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

SUCCESSIVE MINIMA AND LATTICE POINTS (AFTER HENK, GILLET AND SOULÉ) M(B) := # ( B Z N)

SUCCESSIVE MINIMA AND LATTICE POINTS (AFTER HENK, GILLET AND SOULÉ) M(B) := # ( B Z N) SUCCESSIVE MINIMA AND LATTICE POINTS (AFTER HENK, GILLET AND SOULÉ) S.BOUCKSOM Abstract. The goal of ths note s to present a remarably smple proof, due to Hen, of a result prevously obtaned by Gllet-Soulé,

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

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

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

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

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

Dirichlet s Theorem In Arithmetic Progressions

Dirichlet s Theorem In Arithmetic Progressions Drchlet s Theorem In Arthmetc Progressons Parsa Kavkan Hang Wang The Unversty of Adelade February 26, 205 Abstract The am of ths paper s to ntroduce and prove Drchlet s theorem n arthmetc progressons,

More information

Another converse of Jensen s inequality

Another converse of Jensen s inequality Another converse of Jensen s nequalty Slavko Smc Abstract. We gve the best possble global bounds for a form of dscrete Jensen s nequalty. By some examples ts frutfulness s shown. 1. Introducton Throughout

More information

arxiv: v1 [math.ca] 9 Apr 2019

arxiv: v1 [math.ca] 9 Apr 2019 HAUSDORFF DIMENSION OF THE LARGE VALUES OF WEYL SUMS arxv:1904.04457v1 [math.ca] 9 Apr 2019 CHANGHAO CHEN AND IGOR E. SHPARLINSKI Abstract. The authors have recently obtaned a lower bound of the Hausdorffdmenson

More information

CSCE 790S Background Results

CSCE 790S Background Results CSCE 790S Background Results Stephen A. Fenner September 8, 011 Abstract These results are background to the course CSCE 790S/CSCE 790B, Quantum Computaton and Informaton (Sprng 007 and Fall 011). Each

More information

a b a In case b 0, a being divisible by b is the same as to say that

a b a In case b 0, a being divisible by b is the same as to say that Secton 6.2 Dvsblty among the ntegers An nteger a ε s dvsble by b ε f there s an nteger c ε such that a = bc. Note that s dvsble by any nteger b, snce = b. On the other hand, a s dvsble by only f a = :

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

Foundations of Arithmetic

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

More information

20. Mon, Oct. 13 What we have done so far corresponds roughly to Chapters 2 & 3 of Lee. Now we turn to Chapter 4. The first idea is connectedness.

20. Mon, Oct. 13 What we have done so far corresponds roughly to Chapters 2 & 3 of Lee. Now we turn to Chapter 4. The first idea is connectedness. 20. Mon, Oct. 13 What we have done so far corresponds roughly to Chapters 2 & 3 of Lee. Now we turn to Chapter 4. The frst dea s connectedness. Essentally, we want to say that a space cannot be decomposed

More information

Spectral Graph Theory and its Applications September 16, Lecture 5

Spectral Graph Theory and its Applications September 16, Lecture 5 Spectral Graph Theory and ts Applcatons September 16, 2004 Lecturer: Danel A. Spelman Lecture 5 5.1 Introducton In ths lecture, we wll prove the followng theorem: Theorem 5.1.1. Let G be a planar graph

More information

Feature Selection: Part 1

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

More information

Maximizing the number of nonnegative subsets

Maximizing the number of nonnegative subsets Maxmzng the number of nonnegatve subsets Noga Alon Hao Huang December 1, 213 Abstract Gven a set of n real numbers, f the sum of elements of every subset of sze larger than k s negatve, what s the maxmum

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

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

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

10-801: Advanced Optimization and Randomized Methods Lecture 2: Convex functions (Jan 15, 2014)

10-801: Advanced Optimization and Randomized Methods Lecture 2: Convex functions (Jan 15, 2014) 0-80: Advanced Optmzaton and Randomzed Methods Lecture : Convex functons (Jan 5, 04) Lecturer: Suvrt Sra Addr: Carnege Mellon Unversty, Sprng 04 Scrbes: Avnava Dubey, Ahmed Hefny Dsclamer: These notes

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

Anti-van der Waerden numbers of 3-term arithmetic progressions.

Anti-van der Waerden numbers of 3-term arithmetic progressions. Ant-van der Waerden numbers of 3-term arthmetc progressons. Zhanar Berkkyzy, Alex Schulte, and Mchael Young Aprl 24, 2016 Abstract The ant-van der Waerden number, denoted by aw([n], k), s the smallest

More information

PROCEEDINGS OF THE AMERICAN MATHEMATICAL SOCIETY Volume 125, Number 7, July 1997, Pages 2119{2125 S (97) THE STRONG OPEN SET CONDITION

PROCEEDINGS OF THE AMERICAN MATHEMATICAL SOCIETY Volume 125, Number 7, July 1997, Pages 2119{2125 S (97) THE STRONG OPEN SET CONDITION PROCDINGS OF TH AMRICAN MATHMATICAL SOCITY Volume 125, Number 7, July 1997, Pages 2119{2125 S 0002-9939(97)03816-1 TH STRONG OPN ST CONDITION IN TH RANDOM CAS NORBRT PATZSCHK (Communcated by Palle. T.

More information

A note on almost sure behavior of randomly weighted sums of φ-mixing random variables with φ-mixing weights

A note on almost sure behavior of randomly weighted sums of φ-mixing random variables with φ-mixing weights ACTA ET COMMENTATIONES UNIVERSITATIS TARTUENSIS DE MATHEMATICA Volume 7, Number 2, December 203 Avalable onlne at http://acutm.math.ut.ee A note on almost sure behavor of randomly weghted sums of φ-mxng

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

Min Cut, Fast Cut, Polynomial Identities

Min Cut, Fast Cut, Polynomial Identities Randomzed Algorthms, Summer 016 Mn Cut, Fast Cut, Polynomal Identtes Instructor: Thomas Kesselhem and Kurt Mehlhorn 1 Mn Cuts n Graphs Lecture (5 pages) Throughout ths secton, G = (V, E) s a mult-graph.

More information

The Second Eigenvalue of Planar Graphs

The Second Eigenvalue of Planar Graphs Spectral Graph Theory Lecture 20 The Second Egenvalue of Planar Graphs Danel A. Spelman November 11, 2015 Dsclamer These notes are not necessarly an accurate representaton of what happened n class. The

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

Stanford University CS359G: Graph Partitioning and Expanders Handout 4 Luca Trevisan January 13, 2011

Stanford University CS359G: Graph Partitioning and Expanders Handout 4 Luca Trevisan January 13, 2011 Stanford Unversty CS359G: Graph Parttonng and Expanders Handout 4 Luca Trevsan January 3, 0 Lecture 4 In whch we prove the dffcult drecton of Cheeger s nequalty. As n the past lectures, consder an undrected

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

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

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

More information

P exp(tx) = 1 + t 2k M 2k. k N

P exp(tx) = 1 + t 2k M 2k. k N 1. Subgaussan tals Defnton. Say that a random varable X has a subgaussan dstrbuton wth scale factor σ< f P exp(tx) exp(σ 2 t 2 /2) for all real t. For example, f X s dstrbuted N(,σ 2 ) then t s subgaussan.

More information

Case A. P k = Ni ( 2L i k 1 ) + (# big cells) 10d 2 P k.

Case A. P k = Ni ( 2L i k 1 ) + (# big cells) 10d 2 P k. THE CELLULAR METHOD In ths lecture, we ntroduce the cellular method as an approach to ncdence geometry theorems lke the Szemeréd-Trotter theorem. The method was ntroduced n the paper Combnatoral complexty

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

Supplement to Clustering with Statistical Error Control

Supplement to Clustering with Statistical Error Control Supplement to Clusterng wth Statstcal Error Control Mchael Vogt Unversty of Bonn Matthas Schmd Unversty of Bonn In ths supplement, we provde the proofs that are omtted n the paper. In partcular, we derve

More information

On the Operation A in Analysis Situs. by Kazimierz Kuratowski

On the Operation A in Analysis Situs. by Kazimierz Kuratowski v1.3 10/17 On the Operaton A n Analyss Stus by Kazmerz Kuratowsk Author s note. Ths paper s the frst part slghtly modfed of my thess presented May 12, 1920 at the Unversty of Warsaw for the degree of Doctor

More information

European Journal of Combinatorics

European Journal of Combinatorics European Journal of Combnatorcs 0 (009) 480 489 Contents lsts avalable at ScenceDrect European Journal of Combnatorcs journal homepage: www.elsever.com/locate/ejc Tlngs n Lee metrc P. Horak 1 Unversty

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

Vapnik-Chervonenkis theory

Vapnik-Chervonenkis theory Vapnk-Chervonenks theory Rs Kondor June 13, 2008 For the purposes of ths lecture, we restrct ourselves to the bnary supervsed batch learnng settng. We assume that we have an nput space X, and an unknown

More information

Polynomials. 1 More properties of polynomials

Polynomials. 1 More properties of polynomials Polynomals 1 More propertes of polynomals Recall that, for R a commutatve rng wth unty (as wth all rngs n ths course unless otherwse noted), we defne R[x] to be the set of expressons n =0 a x, where a

More information

Finding Dense Subgraphs in G(n, 1/2)

Finding Dense Subgraphs in G(n, 1/2) Fndng Dense Subgraphs n Gn, 1/ Atsh Das Sarma 1, Amt Deshpande, and Rav Kannan 1 Georga Insttute of Technology,atsh@cc.gatech.edu Mcrosoft Research-Bangalore,amtdesh,annan@mcrosoft.com Abstract. Fndng

More information

TAIL BOUNDS FOR SUMS OF GEOMETRIC AND EXPONENTIAL VARIABLES

TAIL BOUNDS FOR SUMS OF GEOMETRIC AND EXPONENTIAL VARIABLES TAIL BOUNDS FOR SUMS OF GEOMETRIC AND EXPONENTIAL VARIABLES SVANTE JANSON Abstract. We gve explct bounds for the tal probabltes for sums of ndependent geometrc or exponental varables, possbly wth dfferent

More information

Eigenvalues of Random Graphs

Eigenvalues of Random Graphs Spectral Graph Theory Lecture 2 Egenvalues of Random Graphs Danel A. Spelman November 4, 202 2. Introducton In ths lecture, we consder a random graph on n vertces n whch each edge s chosen to be n the

More information

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

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

More information

Rapid growth in finite simple groups

Rapid growth in finite simple groups Rapd growth n fnte smple groups Martn W. Lebeck, Gl Schul, Aner Shalev March 1, 016 Abstract We show that small normal subsets A of fnte smple groups grow very rapdly namely, A A ɛ, where ɛ > 0 s arbtrarly

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

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

j) = 1 (note sigma notation) ii. Continuous random variable (e.g. Normal distribution) 1. density function: f ( x) 0 and f ( x) dx = 1

j) = 1 (note sigma notation) ii. Continuous random variable (e.g. Normal distribution) 1. density function: f ( x) 0 and f ( x) dx = 1 Random varables Measure of central tendences and varablty (means and varances) Jont densty functons and ndependence Measures of assocaton (covarance and correlaton) Interestng result Condtonal dstrbutons

More information

Math 426: Probability MWF 1pm, Gasson 310 Homework 4 Selected Solutions

Math 426: Probability MWF 1pm, Gasson 310 Homework 4 Selected Solutions Exercses from Ross, 3, : Math 26: Probablty MWF pm, Gasson 30 Homework Selected Solutons 3, p. 05 Problems 76, 86 3, p. 06 Theoretcal exercses 3, 6, p. 63 Problems 5, 0, 20, p. 69 Theoretcal exercses 2,

More information

Notes on Frequency Estimation in Data Streams

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

More information

Bezier curves. Michael S. Floater. August 25, These notes provide an introduction to Bezier curves. i=0

Bezier curves. Michael S. Floater. August 25, These notes provide an introduction to Bezier curves. i=0 Bezer curves Mchael S. Floater August 25, 211 These notes provde an ntroducton to Bezer curves. 1 Bernsten polynomals Recall that a real polynomal of a real varable x R, wth degree n, s a functon of the

More information

MASSACHUSETTS INSTITUTE OF TECHNOLOGY 6.265/15.070J Fall 2013 Lecture 12 10/21/2013. Martingale Concentration Inequalities and Applications

MASSACHUSETTS INSTITUTE OF TECHNOLOGY 6.265/15.070J Fall 2013 Lecture 12 10/21/2013. Martingale Concentration Inequalities and Applications MASSACHUSETTS INSTITUTE OF TECHNOLOGY 6.65/15.070J Fall 013 Lecture 1 10/1/013 Martngale Concentraton Inequaltes and Applcatons Content. 1. Exponental concentraton for martngales wth bounded ncrements.

More information

Every planar graph is 4-colourable a proof without computer

Every planar graph is 4-colourable a proof without computer Peter Dörre Department of Informatcs and Natural Scences Fachhochschule Südwestfalen (Unversty of Appled Scences) Frauenstuhlweg 31, D-58644 Iserlohn, Germany Emal: doerre(at)fh-swf.de Mathematcs Subject

More information

Supplement: Proofs and Technical Details for The Solution Path of the Generalized Lasso

Supplement: Proofs and Technical Details for The Solution Path of the Generalized Lasso Supplement: Proofs and Techncal Detals for The Soluton Path of the Generalzed Lasso Ryan J. Tbshran Jonathan Taylor In ths document we gve supplementary detals to the paper The Soluton Path of the Generalzed

More information

Problem Set 9 Solutions

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

More information

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

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

DOUBLE POINTS AND THE PROPER TRANSFORM IN SYMPLECTIC GEOMETRY

DOUBLE POINTS AND THE PROPER TRANSFORM IN SYMPLECTIC GEOMETRY DOUBLE POINTS AND THE PROPER TRANSFORM IN SYMPLECTIC GEOMETRY JOHN D. MCCARTHY AND JON G. WOLFSON 0. Introducton In hs book, Partal Dfferental Relatons, Gromov ntroduced the symplectc analogue of the complex

More information

Introductory Cardinality Theory Alan Kaylor Cline

Introductory Cardinality Theory Alan Kaylor Cline Introductory Cardnalty Theory lan Kaylor Clne lthough by name the theory of set cardnalty may seem to be an offshoot of combnatorcs, the central nterest s actually nfnte sets. Combnatorcs deals wth fnte

More information

DONALD M. DAVIS. 1. Main result

DONALD M. DAVIS. 1. Main result v 1 -PERIODIC 2-EXPONENTS OF SU(2 e ) AND SU(2 e + 1) DONALD M. DAVIS Abstract. We determne precsely the largest v 1 -perodc homotopy groups of SU(2 e ) and SU(2 e +1). Ths gves new results about the largest

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

Bézier curves. Michael S. Floater. September 10, These notes provide an introduction to Bézier curves. i=0

Bézier curves. Michael S. Floater. September 10, These notes provide an introduction to Bézier curves. i=0 Bézer curves Mchael S. Floater September 1, 215 These notes provde an ntroducton to Bézer curves. 1 Bernsten polynomals Recall that a real polynomal of a real varable x R, wth degree n, s a functon of

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

arxiv: v1 [math.ca] 31 Jul 2018

arxiv: v1 [math.ca] 31 Jul 2018 LOWE ASSOUAD TYPE DIMENSIONS OF UNIFOMLY PEFECT SETS IN DOUBLING METIC SPACE HAIPENG CHEN, MIN WU, AND YUANYANG CHANG arxv:80769v [mathca] 3 Jul 08 Abstract In ths paper, we are concerned wth the relatonshps

More information

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

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

More information

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

First day August 1, Problems and Solutions

First day August 1, Problems and Solutions FOURTH INTERNATIONAL COMPETITION FOR UNIVERSITY STUDENTS IN MATHEMATICS July 30 August 4, 997, Plovdv, BULGARIA Frst day August, 997 Problems and Solutons Problem. Let {ε n } n= be a sequence of postve

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

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

REGULAR POSITIVE TERNARY QUADRATIC FORMS. 1. Introduction

REGULAR POSITIVE TERNARY QUADRATIC FORMS. 1. Introduction REGULAR POSITIVE TERNARY QUADRATIC FORMS BYEONG-KWEON OH Abstract. A postve defnte quadratc form f s sad to be regular f t globally represents all ntegers that are represented by the genus of f. In 997

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

MATH 5707 HOMEWORK 4 SOLUTIONS 2. 2 i 2p i E(X i ) + E(Xi 2 ) ä i=1. i=1

MATH 5707 HOMEWORK 4 SOLUTIONS 2. 2 i 2p i E(X i ) + E(Xi 2 ) ä i=1. i=1 MATH 5707 HOMEWORK 4 SOLUTIONS CİHAN BAHRAN 1. Let v 1,..., v n R m, all lengths v are not larger than 1. Let p 1,..., p n [0, 1] be arbtrary and set w = p 1 v 1 + + p n v n. Then there exst ε 1,..., ε

More information

Harmonic Analysis and Additive Combinatorics on Fractals

Harmonic Analysis and Additive Combinatorics on Fractals Harmonic Analysis and Additive Combinatorics on Fractals Etta Z. Falconer Lecture Mathfest, 2016 Fractals: random vs. structured Example: Ternary Cantor set Dimension (similarity and Hausdorff) log 2 log

More information

Solutions to the 71st William Lowell Putnam Mathematical Competition Saturday, December 4, 2010

Solutions to the 71st William Lowell Putnam Mathematical Competition Saturday, December 4, 2010 Solutons to the 7st Wllam Lowell Putnam Mathematcal Competton Saturday, December 4, 2 Kran Kedlaya and Lenny Ng A The largest such k s n+ 2 n 2. For n even, ths value s acheved by the partton {,n},{2,n

More information

Curvature and isoperimetric inequality

Curvature and isoperimetric inequality urvature and sopermetrc nequalty Julà ufí, Agustí Reventós, arlos J Rodríguez Abstract We prove an nequalty nvolvng the length of a plane curve and the ntegral of ts radus of curvature, that has as a consequence

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

A CHARACTERIZATION OF ADDITIVE DERIVATIONS ON VON NEUMANN ALGEBRAS

A CHARACTERIZATION OF ADDITIVE DERIVATIONS ON VON NEUMANN ALGEBRAS Journal of Mathematcal Scences: Advances and Applcatons Volume 25, 2014, Pages 1-12 A CHARACTERIZATION OF ADDITIVE DERIVATIONS ON VON NEUMANN ALGEBRAS JIA JI, WEN ZHANG and XIAOFEI QI Department of Mathematcs

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

= z 20 z n. (k 20) + 4 z k = 4

= z 20 z n. (k 20) + 4 z k = 4 Problem Set #7 solutons 7.2.. (a Fnd the coeffcent of z k n (z + z 5 + z 6 + z 7 + 5, k 20. We use the known seres expanson ( n+l ( z l l z n below: (z + z 5 + z 6 + z 7 + 5 (z 5 ( + z + z 2 + z + 5 5

More information

On C 0 multi-contractions having a regular dilation

On C 0 multi-contractions having a regular dilation SUDIA MAHEMAICA 170 (3) (2005) On C 0 mult-contractons havng a regular dlaton by Dan Popovc (mşoara) Abstract. Commutng mult-contractons of class C 0 and havng a regular sometrc dlaton are studed. We prove

More information

Caps and Colouring Steiner Triple Systems

Caps and Colouring Steiner Triple Systems Desgns, Codes and Cryptography, 13, 51 55 (1998) c 1998 Kluwer Academc Publshers, Boston. Manufactured n The Netherlands. Caps and Colourng Stener Trple Systems AIDEN BRUEN* Department of Mathematcs, Unversty

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

Lecture 17 : Stochastic Processes II

Lecture 17 : Stochastic Processes II : Stochastc Processes II 1 Contnuous-tme stochastc process So far we have studed dscrete-tme stochastc processes. We studed the concept of Makov chans and martngales, tme seres analyss, and regresson analyss

More information

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

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

More information

Supplementary material: Margin based PU Learning. Matrix Concentration Inequalities

Supplementary material: Margin based PU Learning. Matrix Concentration Inequalities Supplementary materal: Margn based PU Learnng We gve the complete proofs of Theorem and n Secton We frst ntroduce the well-known concentraton nequalty, so the covarance estmator can be bounded Then we

More information

The proximal average for saddle functions and its symmetry properties with respect to partial and saddle conjugacy

The proximal average for saddle functions and its symmetry properties with respect to partial and saddle conjugacy The proxmal average for saddle functons and ts symmetry propertes wth respect to partal and saddle conjugacy Rafal Goebel December 3, 2009 Abstract The concept of the proxmal average for convex functons

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

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

Lecture Space-Bounded Derandomization

Lecture Space-Bounded Derandomization Notes on Complexty Theory Last updated: October, 2008 Jonathan Katz Lecture Space-Bounded Derandomzaton 1 Space-Bounded Derandomzaton We now dscuss derandomzaton of space-bounded algorthms. Here non-trval

More information

COS 521: Advanced Algorithms Game Theory and Linear Programming

COS 521: Advanced Algorithms Game Theory and Linear Programming COS 521: Advanced Algorthms Game Theory and Lnear Programmng Moses Charkar February 27, 2013 In these notes, we ntroduce some basc concepts n game theory and lnear programmng (LP). We show a connecton

More information

On a theorem of V. Bernik in the metrical theory of Diophantine approximation

On a theorem of V. Bernik in the metrical theory of Diophantine approximation Acta Arthmetca 117, no.1 (2005) 71-80 On a theorem of V. Bern n the metrcal theory of Dophantne approxmaton by V. Beresnevch (Mns) 1 1. Introducton. We begn by ntroducng some notaton: #S wll denote the

More information

The Expectation-Maximization Algorithm

The Expectation-Maximization Algorithm The Expectaton-Maxmaton Algorthm Charles Elan elan@cs.ucsd.edu November 16, 2007 Ths chapter explans the EM algorthm at multple levels of generalty. Secton 1 gves the standard hgh-level verson of the algorthm.

More information