Dynamic Policy Programming with Function Approximation: Supplementary Material
|
|
- Gary Jordan
- 5 years ago
- Views:
Transcription
1 Dyamic Policy Programmig with Fuctio pproximatio: Supplemetary Material Mohammad Gheshlaghi zar Radboud Uiversity Nijmege Geert Grooteplei Noord EZ Nijmege Netherlads Viceç Gómez Radboud Uiversity Nijmege Geert Grooteplei Noord EZ Nijmege Netherlads Hilbert J. Kappe Radboud Uiversity Nijmege Geert Grooteplei Noord EZ Nijmege Netherlads ppedices Defiitios.1 Markov Decisio Processes ad Bellma Equatio statioary MDP is a 5-tuple S,,R,T,γ, where S,,R are, respectively, the set of all system states, the set of actios that ca be take ad the set of rewards that may be issued, such that rss a deotes the reward of the ext state s give that the curret state is s ad the actio is a. T is a set of matrices of dimesio S S, oe for each a such that Tss a deotes the probability of the ext state s give that the curret state is s ad the actio is a. γ 0,1 deotes the discout factor. ssumptio 1. We assume that for every 3-tuple s,a,s S S, the magitude of the immediate reward, r a ss is bouded from above by R. statioary policy is a mappig π that assigs to each state s a probability distributio over the actio space, oe for each s,a S such that π s a deotes the probability of the actio a give the curret state is s. Give the policy π, its correspodig value fuctio V π deotes the expected value of the log-term discouted sum of rewards i each state s, whe the actio is chose by policy π. The goal is to fid a policy π that attais the optimal value fuctio V s, such that V s satisfies a Bellma equatio: V s = s π s a π a Tss a ra ss +γv s, s S. 1 s S
2 Mohammad Gheshlaghi zar, Viceç Gómez.2 Matrix otatio We fid it coveiet to use matrix ad vector otatio for aalysis of theorem 1, 2 ad 3. We begi by re-defiig the variables of the MDP. T ad r are, respectively, a S S trasitio matrix with etries Tsa,s = Tss a ad a S S reward matrix with etries rsa,s = rss a. r is a S 1 colum vector of expected rewards with etries rsa = s S Ta ss ra ss. The policy ca also be re-expressed as a S 1 vector π with etries πsa = π s a. I additio, it will be coveiet to itroduce a S S matrix Π, give by: π s1 a 1 π s1 a π s2 a 1 π s2 a Π =..., 2 πs S a1 πs S a where Π cosists of S row blocks, each of legth, which are arraged diagoally. The policy matrix Π is related to the policy vector π by π = Π T 1 with 1 deotes a S 1 vector of all 1s. Further, oe ca easily verify that the matrix-product ΠT gives the state to state trasitio matrix for the policy π. Oe ca also preset the value fuctio i vector space. v π is a S 1 vector with etries v π s = V π s. The Bellma equatio 1 ca ow be re-expressed i matrix otatio as: v = M r+γtv, 3 wherem istheoperatorothe S 1vector r+γtv,suchthatm r+γtv s = a rsa+ γtv sa. Ofte it is coveiet to associate value fuctios ot with states but with state-actio pairs. Therefore, we itroduce a S 1 vector of the actio-values q π whose etries q π sa deotes the expected value of the sum of future rewards for all state-actio s,a S, provided the future actios are chose by the policy π. q π satisfies the followig Bellma equatio: q π = T π q π = r+γtπq π, 4 where we itroduce the Bellma operator T π ad q π deotes the fixed poit of T π. The optimal q, q, also satisfies a Bellma equatio: q = T q = r+γtm q, 5 where we itroduce the Bellma operator T. q deotes the fixed poit of the operator T. Both T ad T π are cotractio mappigs with the factor γ?, chap. 1. I other words, for ay two vectors q ad q, we have: T q T q q q, T π q T π q q q, 6 where deotes a L -orm o R S. The operator O defied ca be writte i matrix otatio as: Op = p+ r+γtm η p ΞM η p. 7 Here, p deotes a S 1 vector of the actio prefereces. M η is the soft- operator o the vector p with M η ps = M η Ps. Ξ is a S S matrix give by: 1 1 Ξ = 1 1 T,
3 Mohammad Gheshlaghi zar, Viceç Gómez where Ξ cosists of S colum blocks of 1s, each of legth, which are arraged diagoally. Ξ trasforms the S 1 vector M η p to the S 1 vector ΞM η p with ΞM η psa = M η ps. Further, oe ca easily verify that for ay policy matrix defied by 2: πs 1 a a πs 2 a ΠΞ = a. = I, 9.. πs S a where I is a S S idetity matrix. We kow that lim η M η p = M p. Further, It is ot difficult to show that the followig iequality holds for M p M η p : Lemma 1. Let p be a S 1 vector of actio prefereces ad η be a positive costat, the a upper boud for M η p M p ca be obtaied as follows: a M η p M p 1 η log. Proof. sketch First, we ote that M η p M p = M η p M p 0. The we apply equatio 12 of the mai article to each etries of p M p. Now, oe ca easily show that L η p M ps 0 for all s S. This combied with the fact that for the probability distributio π: H π s log derive 0 M p M η p 1 η log. The result the follows by takig the sup-orm. Defiig p as the actio preferece resulted by the th iteratio of 7, we have: p = p 1 + r+γtm η p 1 ΞM η p 1 = p 1 + r+γtπ 1 p 1 ΞΠ 1 p 1, = 1,2,3,, 10 where p 0 = p ad Π is a policy distributio matrix associated with the policy distributio vector π give by: π sa = expηp sa, s,a S, 11 Zs ad Zs = a expηp sa is the ormalizatio factor. B Proof of Lemma 1 of The Mai rticle The optimal value value fuctio is defied as follows: V π s = s π s Π s a π a Tss a ra ss +γv π s 1 η KLπ s π s s S 12 The imizatio i 12 ca be performed i closed form usig Lagrage multipliers: Ls,λ s = π s a Tss a ra ss +γv πs 1 η KLπ s π s λ s s a 1. a s S a π The ecessary coditio for the extremum with respect to π s is: 0 = Ls,λ s π s a = s S Tss a ra ss +γv πs 1 η 1 η log πs a λ s. π s a
4 Mohammad Gheshlaghi zar, Viceç Gómez So, the solutio is: π s a = π s aexp ηλ s 1exp η ss s ST a ra ss +γv πs, s S. 13 The Lagrage multipliers ca be solved from the costraits: 1 = π s a = exp ηλ s 1 s aexp η a a π ss s ST a ra ss +γv πs, λ s = 1 η log a π s aexp η ss s ST a ra ss +γv π s 1 η. 14 By pluggig 14 i to 13 we have: π s aexp η Tss a ra ss +γv πs π s a = a s S Substitutig policy 15 i 12, we obtai: π s aexp η, s,a S. 15 Tss a ra ss +γv πs s S V πs = 1 η log a π s aexp η s ST a ss ra ss +γv πs. C Proof of Theorem 1 This sectio provides a formal aalysis of the covergece behavior of DPP. Our objective is to establish a rate of covergece for the value fuctio of the policy iduced by DPP. Our mai result is that at iteratio of DPP we have: q π q < δ, where δ = O Here, q π is a vector of the actio-values uder the policy π ad π is the policy iduced by DPP at iteratio. Equatio 16 implies that, asymptotically, the policy iduced by DPP, π = lim π, coverges to the optimal policy π. To derive 16 oe eeds to relate q π to the optimal q. Ufortuately, fidig a direct relatio betwee q π ad q is ot a easy task. Istead, we ca relate q π to q via a auxiliary q, which we defie later i this sectio. I the remaider of this sectio, we first express π i terms of q. The, we obtai a upper boud o the ormed error q q i lemma 2. Fially, we use these two results to derive a boud o the ormed error q π q. I ordertoexpressπ i termsofq, weexpadthe correspodigactiopreferecep byrecursivesubstitutio of 10: p = p 1 + r+γtπ 1 p 1 ΞΠ 1 p 1 = p 2 +2 r+γtπ 2 p 2 ΞΠ 2 p 2 +γtπ 1 p 2 +γtπ 1 r +γ 2 TΠ 1 TΠ 2 p 2 γtπ 1 ΞΠ 2 p 2 ΞΠ 1 p 2 ΞΠ 1 r 17 γξπ 1 TΠ 2 p 2 +ΞΠ 1 ΞΠ 2 p 2 = 2 r+γtπ 1 r+p 2 +γtπ 1 p 2 +γ 2 TΠ 1 TΠ 2 p 2 ΞΠ 1 r ΞΠ 1 p 2 γξπ 1 TΠ 2 p 2 by 9.
5 Mohammad Gheshlaghi zar, Viceç Gómez Equatio 17 expresses p i terms of p 2, where i the last step we have caceled some terms because of 9. To express p i terms of p 0, we proceed with the expasio of 10: p = r+ 1 kγk k TΠ j r+p 0 + γk k TΠ j p 0 j=1 j=1 ΞΠ 1 1 r+ 2 k 1γk k TΠ j 1 r j=1 ΞΠ 1 p γk k TΠ j 1 p 0 j=1 = r+ 1 γk k TΠ j r γ 1 kγk 1 k TΠ j r+p 0 j=1 j=1 + γk k TΠ j p 0 1ΞΠ 1 r+ 2 γk k TΠ j 1 r j=1 j=1 +ΞΠ 1 γ 2 kγk 1 k TΠ j 1 r p 0 1 γk k TΠ j 1 p 0. j=1 j=1 18 We defie the auxiliary actio-values q γk k j=1 TΠ j p0 ad write 18 as: = r + 1 γk k j=1 TΠ j r ad q p = p 0 + p = q γ q γ +qp ΞΠ 1 1q 1 γ q 1 +q P 1 = q γ +c Ξd. 19 Here, c = q P γ q / γ is a S 1 vector bouded by: c c = γ R p 0, = 1,2,3,, 20 ad d = Π 1 1q 1 γ q 1/ γ+q P 1 is a S 1 vector. From its defiitio, it is easy to see that q satisfies the followig Bellma equatio: q = r+γtπ 1q Note the differece betwee 21 ad 4: whereas q π evolves accordig to a fixed policy π, q evolves accordig to a policy that evolves accordig to DPP. Fially, we express π i terms of q. By pluggig 19 i 11 ad takig i to accout that Ξd sa = d s, we obtai: π sa = exp η q sa+c sa d s = exp η q sa+c sa Zs Z, s,a S, 22 s where Z s = Zsexpηd s is the ormalizatio factor. Equatio 22 establishes the relatio betwee π ad q. To relate q ad q we state the followig lemma, that establishes a boud o q q : Lemma 2 boud o q q. Let assumptio 1 hold, deotes the cardiality of ad be a positive iteger, also, for keepig the represetatio succict, assume that both r ad p 0 are bouded from above by some costat L > 0, the the followig iequality holds: q q δ 1, where δ 1 is give by: δ 1 = 2γ 2 log /η +2L 4 +2γ 1 L. 23
6 Mohammad Gheshlaghi zar, Viceç Gómez Proof. q q = 1 T k 1 q k+1 T k q k +T 1 q 1 q 1 T k 1 q k+1 T k q k + T 1 q 1 q 1 k 1 q k+1 T q k +γ 1 q 1 q by For 1 k 1 we obtai: q k+1 T q k TΠ k q k TM q k by 21 ad 5 Π k q k M q k by Hölder s iequality. 25 By mootoicity property for ay actio-value vector q ad distributio matrix Π we have: Πqs = a πsaqsa M qs, s S. 26 By comparig 26 with 25, we obtai: q k+1 T q k M q ks Π k q ks s S M q k + c k s+ c s S k k Π kq ks M q k + c k Π k q 1 k + γc k M q k + c k Π k k q k k + c M η k q k + c k k + 2γc k M k q k + c + 2γc k 27 log /η +2c k by lemma 1. By substitutio of 27 i 24, we obtai: It is ot difficult to show that 1 q q 1 γ k log /η +2c k + 2γ 1 L. 28 γ k k 2γ/ 2. Combiig this with 28 yields: q q 2γ log /η +2c 2 +2γ 1 L 2γ 2 log /η +2L 4 +2γ 1 L by 37. Equatio 22 expresses the policy π i terms of the auxiliary q. Lemma 2 provides a upper boud o the ormed-error q q. We use these two results to derive a boud o the ormed-error q π q :
7 Mohammad Gheshlaghi zar, Viceç Gómez We start by otig that q π is obtaied by ifiite applicatio of the operator T π to a arbitrary iitial q-vector, for which we take q, the: q π q m = lim T k m π q T π k 1 T q lim m lim m m T k π q T π k 1 T q m γ k 1 T π q T q by 6 1 T π q T q TΠ q TM q Π q M q by Hölder s iequality. 29 log similar lies with the proof of lemma 2, we obtai: q π q Π q M q M q s Π q s s S M q s Π q s +δ 1 s S M q + c s Π q s s S + γ δ 1 + c M q + c Π q + c + γ 2δ 1 + 2c M q + c M η q + c + γ 2δ 1 + 2c log +2δ 1 η + 2c 2 log /η +2L 2 +2δ 1 2 log /η +2L 2 +4γ 2 log /η +2L log /η +2L 4 +4γ 1 L +4γ 1 L by lemma 2 by 37 by lemma 1 by 37 This completes the proof. D Proof of Theorem 2 First, we show that there exists a limit for p i ifiity. The, we compute this limit. Before we proceed with the proof we review the followig assumptio ad corollary from the mai article
8 Mohammad Gheshlaghi zar, Viceç Gómez ssumptio 2. We assume that MDP has a uique determiistic optimal policy π give by: π sa = { 1 a = a s 0 otherwise, s S, where a s = arg a q sa. Corollary 1. The followig relatio holds i limit: lim = q, s,a S. + qπ The covergece of p to a limit ca be established by provig the covergece of all the terms i RHS of 19. We have already prove the covergece of q to q i lemma 2. The, the covergece of q / γ to q / γ is immediate. Corollary 1 together with assumptio 2 yields the covergece of π to π. The covergece of q p to some limit q p the follows. ccordigly, there exists a limit for p. Now, we compute the limit of p. Combiig corollary 1 with 19 ad takig i to accout that v = Π q yields: lim p sa = lim q sa γ q sa +q p sa 1v s+γ v s q p sa γ γ E Proof of Theorem 3 = lim q sa v s v s +γ q sa +q p sa q p sa +v s γ γ { v = s a = a s otherwise. This sectio provides a formal theoretical aalysis of the performace of dyamic policy programmig i the presece of approximatio error. Each iteratio of approximate dyamic programmig ca be characterized by 30. Our objective is to establish a L -orm performace loss boud of the policy iduced by approximate DPP. p = p 1 + r+γtm η p 1 ΞM η p 1 +ǫ 1 = p 1 + r+γtπ 1 p 1 ΞΠ 1 p 1 +ǫ 1, = 1,2,3,, 30 Our mai result that at iteratio of approximate dyamic policy programmig, we have: where: q π q δ, 31 δ = 4γ 2 log /η +4L 5 +4γ L 2 + 2γ γ k ε k. with ε k = 1 / k j=0:k 1 ǫ j. Here, q π is a vector of the actio-values uder the policy π ad π is the policy iduced after iteratio of approximate DPP. To relate q with q π, we ca relate q π to q via a auxiliary q, which we defie later i this sectio. I the remaider of this sectio, we first express π i terms of q i lemma 3. The, we obtai a upper boud o the ormed error q q i lemma 4. Fially, we use these two results to derive 31. We begi our aalysis with the followig lemma:
9 Mohammad Gheshlaghi zar, Viceç Gómez Lemma 3. Let be a positive iteger ad p 0 deotes the iitial actio prefereces. lso, lets defie the auxiliary actio-value fuctios q ad q p as: 1 q = r+ ǫ + γ k kj=1 TΠ j r+ ǫ k q p = p 0 + k=0 γ k kj=1 TΠ j p0 32 with ǫ = 1 l=0 ǫ l/, the we have: p = q γ q γ +qp ΞΠ 1 1q 1 γ q 1 +q P 1 γ 33 Proof. I order to express π i terms of q, we expad the correspodig actio preferece p by recursive substitutio of 30: p = p 1 + r+γtπ 1 p 1 ΞΠ 1 p 1 +ǫ 1 = p 2 +ǫ 2 +2 r+γtπ 2 p 2 +γtπ 2 ǫ 2 ΞΠ 2 p 2 +γtπ 1 p 2 +γ 2 TΠ 1 TΠ 2 p 2 +γtπ 1 r γtπ 1 ΞΠ 2 p 2 ΞΠ 1 p 2 ΞΠ 1 ǫ 2 ΞΠ 1 r γξπ 1 TΠ 2 p 2 +ΞΠ 1 ΞΠ 2 p 2 +ǫ 1 = 2 r+γtπ 1 r+p 2 ++γtπ 1 p 2 +γ 2 TΠ 1 TΠ 2 p 2 ΞΠ 1 r 34 ΞΠ 1 p 2 γξπ 1 TΠ 2 p 2 +ǫ 1 +ǫ 2 +γtπ 2 ǫ 2 ΞΠ 1 ǫ 2 Equatio 34 expresses p i terms of p 2, where i the last step we have caceled some terms because of 9. To express p i terms of p 0, we proceed with the expasio of 30: p = 1 k=0 kγk k j=1 TΠ j r+ k 1 k=0 γk k j=1 TΠ j + 1 γk k TΠ 2 j ǫ l ΞΠ 1 k 1γk k TΠ j 1 k=0 j=1 l=0 k=0 j=1 1 ΞΠ 1 γk k TΠ j 1 p γk k TΠ k 2 j 1 ǫ l k=0 j=1 k=0 j=1 l=0 = 1 γk k TΠ j r γ 1 kγk 1 k TΠ j r k=0 j=1 j=1 + 1 γk k TΠ j ǫ k γ 1 kγk 1 k TΠ j ǫ k k=0 j=1 j=1 + γk k TΠ 2 j p 0 1ΞΠ 1 γk k TΠ j 1 r k=0 j=1 k=0 j=1 2 1ΞΠ 1 ǫ k 1 p 0 γk k TΠ j 1 k=0 j=1 2 +γξπ 1 kγk 1 k TΠ j 1 ǫ k 1 j=1 +ΞΠ 1 γ 2 kγk 1 k TΠ j 1 r 1 γk k TΠ j 1 p 0 j=1 k=0 j=1 r 35 Equatio 33 the follows by comparig 35 with 32. It is easy to see that q, defied i lemma 3, satisfies the followig Bellma equatio: q = r+γtπ 1q 1 + ǫ 36 Note the differece betwee 36 ad 4: whereas q π evolves accordig to a fixed policy π, q to a policy that evolves accordig to approximate DPP. evolves accordig
10 Mohammad Gheshlaghi zar, Viceç Gómez For brevity we itroduce the ew variables c = q P γ q / γ 1 ad d = Π 1 1q 1 γ q 1/ γ + q P 1 ad re-express 33 as: p = q +c Ξd 38 Fially, we express π i terms of q. By pluggig 33 i 11 ad takig i to accout that Ξd sa = d s, we obtai: π sa = exp ηq sa+c sa d s Zs = exp η q sa+c sa s,a S, 39 Z, s where Z s = Zsexpd s is the ormalizatio factor. Equatio 39 establishes the relatio betwee π ad q. To relate q ad q, we state the followig lemma, that establishes a boud o q q : Lemma 4 L boud o q q. Let assumptio 1 hold ad q defied accordig to 32. Let deotes the cardiality of ad be a positive iteger, also, for keepig the represetatio succict, assume that both r ad p 0 are bouded from above by some costat L > 0, the the followig iequality holds: q q δ 2, where δ 2 is give by: with ε k = ǫ k. δ 2 = 2γ 2 log /η +4L 4 +2γ 1 L + γ k ε k. 40 Proof. q q = 1 T k 1 q k+1 T k q k +T 1 q 1 q 1 T k 1 q k+1 T k q k + T 1 q 1 q 1 k 1 q k+1 T q k +γ 1 q 1 q by For all l N : 2 l 1 we obtai: q l+1 T q l = γ TΠl q l TM q l + ǫl+1 by 36 ad 5 + ε l+1 TΠl q l TM q l Πl q l M q l + ε l+1 by Hölder s iequality Note that: c c = γ 2 R 1 +ǫ + p0, = 1,2,3,, 37 with ǫ =,2,3,... ǫ k
11 Mohammad Gheshlaghi zar, Viceç Gómez Takig i to accout that M q l Π l q l s we obtai: q l+1 T q l M q l s Π l q l s + ε l+1 s S M q l + c l s+ c Π l q l s + ε l+1 s S l l M q l + c l Π l q l + γc + ε l+1 l l M q l + c l Π l q l + c l + 2γc + ε l+1 l l l M ηl q l + c l M q l + c l + 2γc + ε l+1 l l l log / η +2c l + ε l+1 by lemma By substitutio of 43 i 41, we obtai: q q log / 1 η +2c γ k k + 2γ 1 L + γ k 1 ε k. 44 It is ot difficult to show that 1 γ k k 2γ/ 2. Combiig this with 44 yields: q q 2γ log /η +2c 2 +2γ 1 L + 2γ 2 log /η +4L 4 γ k ε k +2γ 1 L + γ k ε k. Equatio 39 expresses the policy π i terms of the auxiliary q. Lemma 4 provides a upper-boud o the ormed-error q q. We use these two results to derive a boud o the ormed-error q π q. We start by otig that q π is obtaied by ifiite applicatio of the operator T π to a arbitrary iitial q-vector, for which we take q, the: q π q = lim m lim m T k π q T π k 1 T q m lim m m T k π q T π k 1 T q m γ k 1 T π q T q by 6 1 T π q T q TΠ q TM q Π q M q by Hölder s iequality. 45
12 Mohammad Gheshlaghi zar, Viceç Gómez log similar lies with the proof of lemma 4, we obtai: q π q Π q M q M q s Π q s s S s S M s S M q s Π q s q + c + γ δ 2 + c M q + c Π + γ 2δ 2 + 2c q π q M q + c + γ Π +δ 2 s Π q s q + c q + c 2δ 2 + 2c M q + c M η q + c + γ 2δ 2 + 2c log +2δ 2 + 2c η 2 log /η +4L 2 +2δ log /η +4L 4 +4γ 1 L +2 γ k ε k by lemma 4 by 37 by lemma 1 by 37 This completes the proof of theorem 3. Refereces Bertsekas, D. P Dyamic Programmig ad Optimal Cotrol, volume II. thea Scietific, Belmout, Massachusetts, third editio.
62. Power series Definition 16. (Power series) Given a sequence {c n }, the series. c n x n = c 0 + c 1 x + c 2 x 2 + c 3 x 3 +
62. Power series Defiitio 16. (Power series) Give a sequece {c }, the series c x = c 0 + c 1 x + c 2 x 2 + c 3 x 3 + is called a power series i the variable x. The umbers c are called the coefficiets of
More informationMath 104: Homework 2 solutions
Math 04: Homework solutios. A (0, ): Sice this is a ope iterval, the miimum is udefied, ad sice the set is ot bouded above, the maximum is also udefied. if A 0 ad sup A. B { m + : m, N}: This set does
More informationMarkov Decision Processes
Markov Decisio Processes Defiitios; Statioary policies; Value improvemet algorithm, Policy improvemet algorithm, ad liear programmig for discouted cost ad average cost criteria. Markov Decisio Processes
More informationNEW FAST CONVERGENT SEQUENCES OF EULER-MASCHERONI TYPE
UPB Sci Bull, Series A, Vol 79, Iss, 207 ISSN 22-7027 NEW FAST CONVERGENT SEQUENCES OF EULER-MASCHERONI TYPE Gabriel Bercu We itroduce two ew sequeces of Euler-Mascheroi type which have fast covergece
More informationOptimally Sparse SVMs
A. Proof of Lemma 3. We here prove a lower boud o the umber of support vectors to achieve geeralizatio bouds of the form which we cosider. Importatly, this result holds ot oly for liear classifiers, but
More informationThe log-behavior of n p(n) and n p(n)/n
Ramauja J. 44 017, 81-99 The log-behavior of p ad p/ William Y.C. Che 1 ad Ke Y. Zheg 1 Ceter for Applied Mathematics Tiaji Uiversity Tiaji 0007, P. R. Chia Ceter for Combiatorics, LPMC Nakai Uivercity
More informationBeurling Integers: Part 2
Beurlig Itegers: Part 2 Isomorphisms Devi Platt July 11, 2015 1 Prime Factorizatio Sequeces I the last article we itroduced the Beurlig geeralized itegers, which ca be represeted as a sequece of real umbers
More informationSupplementary Material for Fast Stochastic AUC Maximization with O(1/n)-Convergence Rate
Supplemetary Material for Fast Stochastic AUC Maximizatio with O/-Covergece Rate Migrui Liu Xiaoxua Zhag Zaiyi Che Xiaoyu Wag 3 iabao Yag echical Lemmas ized versio of Hoeffdig s iequality, ote that We
More informationCHAPTER 10 INFINITE SEQUENCES AND SERIES
CHAPTER 10 INFINITE SEQUENCES AND SERIES 10.1 Sequeces 10.2 Ifiite Series 10.3 The Itegral Tests 10.4 Compariso Tests 10.5 The Ratio ad Root Tests 10.6 Alteratig Series: Absolute ad Coditioal Covergece
More informationSequences, Mathematical Induction, and Recursion. CSE 2353 Discrete Computational Structures Spring 2018
CSE 353 Discrete Computatioal Structures Sprig 08 Sequeces, Mathematical Iductio, ad Recursio (Chapter 5, Epp) Note: some course slides adopted from publisher-provided material Overview May mathematical
More informationSOME TRIBONACCI IDENTITIES
Mathematics Today Vol.7(Dec-011) 1-9 ISSN 0976-38 Abstract: SOME TRIBONACCI IDENTITIES Shah Devbhadra V. Sir P.T.Sarvajaik College of Sciece, Athwalies, Surat 395001. e-mail : drdvshah@yahoo.com The sequece
More informationDefinition 4.2. (a) A sequence {x n } in a Banach space X is a basis for X if. unique scalars a n (x) such that x = n. a n (x) x n. (4.
4. BASES I BAACH SPACES 39 4. BASES I BAACH SPACES Sice a Baach space X is a vector space, it must possess a Hamel, or vector space, basis, i.e., a subset {x γ } γ Γ whose fiite liear spa is all of X ad
More informationInfinite Sequences and Series
Chapter 6 Ifiite Sequeces ad Series 6.1 Ifiite Sequeces 6.1.1 Elemetary Cocepts Simply speakig, a sequece is a ordered list of umbers writte: {a 1, a 2, a 3,...a, a +1,...} where the elemets a i represet
More information6.3 Testing Series With Positive Terms
6.3. TESTING SERIES WITH POSITIVE TERMS 307 6.3 Testig Series With Positive Terms 6.3. Review of what is kow up to ow I theory, testig a series a i for covergece amouts to fidig the i= sequece of partial
More informationComplex Analysis Spring 2001 Homework I Solution
Complex Aalysis Sprig 2001 Homework I Solutio 1. Coway, Chapter 1, sectio 3, problem 3. Describe the set of poits satisfyig the equatio z a z + a = 2c, where c > 0 ad a R. To begi, we see from the triagle
More informationMinimax PAC Bounds on the Sample Complexity of Reinforcement Learning with a Generative Model
Noame mauscript No. will be iserted by the editor Miimax PAC Bouds o the Sample Complexity of Reiforcemet Learig with a Geerative Model Mohammad Gheshlaghi Azar Rémi Muos Hilbert J. Kappe Received: date
More information1 Duality revisited. AM 221: Advanced Optimization Spring 2016
AM 22: Advaced Optimizatio Sprig 206 Prof. Yaro Siger Sectio 7 Wedesday, Mar. 9th Duality revisited I this sectio, we will give a slightly differet perspective o duality. optimizatio program: f(x) x R
More informationPhysics 324, Fall Dirac Notation. These notes were produced by David Kaplan for Phys. 324 in Autumn 2001.
Physics 324, Fall 2002 Dirac Notatio These otes were produced by David Kapla for Phys. 324 i Autum 2001. 1 Vectors 1.1 Ier product Recall from liear algebra: we ca represet a vector V as a colum vector;
More informationConvergence of random variables. (telegram style notes) P.J.C. Spreij
Covergece of radom variables (telegram style otes).j.c. Spreij this versio: September 6, 2005 Itroductio As we kow, radom variables are by defiitio measurable fuctios o some uderlyig measurable space
More informationEnumerative & Asymptotic Combinatorics
C50 Eumerative & Asymptotic Combiatorics Stirlig ad Lagrage Sprig 2003 This sectio of the otes cotais proofs of Stirlig s formula ad the Lagrage Iversio Formula. Stirlig s formula Theorem 1 (Stirlig s
More informationMa 530 Introduction to Power Series
Ma 530 Itroductio to Power Series Please ote that there is material o power series at Visual Calculus. Some of this material was used as part of the presetatio of the topics that follow. What is a Power
More informationINFINITE SEQUENCES AND SERIES
11 INFINITE SEQUENCES AND SERIES INFINITE SEQUENCES AND SERIES 11.4 The Compariso Tests I this sectio, we will lear: How to fid the value of a series by comparig it with a kow series. COMPARISON TESTS
More information( a) ( ) 1 ( ) 2 ( ) ( ) 3 3 ( ) =!
.8,.9: Taylor ad Maclauri Series.8. Although we were able to fid power series represetatios for a limited group of fuctios i the previous sectio, it is ot immediately obvious whether ay give fuctio has
More informationChapter 3. Strong convergence. 3.1 Definition of almost sure convergence
Chapter 3 Strog covergece As poited out i the Chapter 2, there are multiple ways to defie the otio of covergece of a sequece of radom variables. That chapter defied covergece i probability, covergece i
More informationarxiv: v1 [math.fa] 3 Apr 2016
Aticommutator Norm Formula for Proectio Operators arxiv:164.699v1 math.fa] 3 Apr 16 Sam Walters Uiversity of Norther British Columbia ABSTRACT. We prove that for ay two proectio operators f, g o Hilbert
More informationProblem Set 2 Solutions
CS271 Radomess & Computatio, Sprig 2018 Problem Set 2 Solutios Poit totals are i the margi; the maximum total umber of poits was 52. 1. Probabilistic method for domiatig sets 6pts Pick a radom subset S
More informationMath 155 (Lecture 3)
Math 55 (Lecture 3) September 8, I this lecture, we ll cosider the aswer to oe of the most basic coutig problems i combiatorics Questio How may ways are there to choose a -elemet subset of the set {,,,
More informationREGRESSION WITH QUADRATIC LOSS
REGRESSION WITH QUADRATIC LOSS MAXIM RAGINSKY Regressio with quadratic loss is aother basic problem studied i statistical learig theory. We have a radom couple Z = X, Y ), where, as before, X is a R d
More informationMASSACHUSETTS INSTITUTE OF TECHNOLOGY 6.436J/15.085J Fall 2008 Lecture 19 11/17/2008 LAWS OF LARGE NUMBERS II THE STRONG LAW OF LARGE NUMBERS
MASSACHUSTTS INSTITUT OF TCHNOLOGY 6.436J/5.085J Fall 2008 Lecture 9 /7/2008 LAWS OF LARG NUMBRS II Cotets. The strog law of large umbers 2. The Cheroff boud TH STRONG LAW OF LARG NUMBRS While the weak
More informationOPTIMAL STOPPING AND EXIT TIMES FOR SOME CLASSES OF RANDOM PROCESSES. Vladyslav Tomashyk
NATIONAL TARAS SHEVCHENKO UNIVERSITY OF KYIV UKRAINE OPTIMAL STOPPING AND EXIT TIMES FOR SOME CLASSES OF RANDOM PROCESSES Vladyslav Tomashyk Mechaics ad Mathematics Faculty Departmet of Probability Theory,
More informationReview Problems 1. ICME and MS&E Refresher Course September 19, 2011 B = C = AB = A = A 2 = A 3... C 2 = C 3 = =
Review Problems ICME ad MS&E Refresher Course September 9, 0 Warm-up problems. For the followig matrices A = 0 B = C = AB = 0 fid all powers A,A 3,(which is A times A),... ad B,B 3,... ad C,C 3,... Solutio:
More informationSupplementary Material for Fast Stochastic AUC Maximization with O(1/n)-Convergence Rate
Supplemetary Material for Fast Stochastic AUC Maximizatio with O/-Covergece Rate Migrui Liu Xiaoxua Zhag Zaiyi Che Xiaoyu Wag 3 iabao Yag echical Lemmas ized versio of Hoeffdig s iequality, ote that We
More informationRegression with quadratic loss
Regressio with quadratic loss Maxim Ragisky October 13, 2015 Regressio with quadratic loss is aother basic problem studied i statistical learig theory. We have a radom couple Z = X,Y, where, as before,
More informationIt is often useful to approximate complicated functions using simpler ones. We consider the task of approximating a function by a polynomial.
Taylor Polyomials ad Taylor Series It is ofte useful to approximate complicated fuctios usig simpler oes We cosider the task of approximatig a fuctio by a polyomial If f is at least -times differetiable
More informationSieve Estimators: Consistency and Rates of Convergence
EECS 598: Statistical Learig Theory, Witer 2014 Topic 6 Sieve Estimators: Cosistecy ad Rates of Covergece Lecturer: Clayto Scott Scribe: Julia Katz-Samuels, Brado Oselio, Pi-Yu Che Disclaimer: These otes
More informationSupport vector machine revisited
6.867 Machie learig, lecture 8 (Jaakkola) 1 Lecture topics: Support vector machie ad kerels Kerel optimizatio, selectio Support vector machie revisited Our task here is to first tur the support vector
More informationApplication to Random Graphs
A Applicatio to Radom Graphs Brachig processes have a umber of iterestig ad importat applicatios. We shall cosider oe of the most famous of them, the Erdős-Réyi radom graph theory. 1 Defiitio A.1. Let
More informationEntropy Rates and Asymptotic Equipartition
Chapter 29 Etropy Rates ad Asymptotic Equipartitio Sectio 29. itroduces the etropy rate the asymptotic etropy per time-step of a stochastic process ad shows that it is well-defied; ad similarly for iformatio,
More informationMulti parameter proximal point algorithms
Multi parameter proximal poit algorithms Ogaeditse A. Boikayo a,b,, Gheorghe Moroşau a a Departmet of Mathematics ad its Applicatios Cetral Europea Uiversity Nador u. 9, H-1051 Budapest, Hugary b Departmet
More informationMath 25 Solutions to practice problems
Math 5: Advaced Calculus UC Davis, Sprig 0 Math 5 Solutios to practice problems Questio For = 0,,, 3,... ad 0 k defie umbers C k C k =! k!( k)! (for k = 0 ad k = we defie C 0 = C = ). by = ( )... ( k +
More informationIntroduction to Optimization Techniques. How to Solve Equations
Itroductio to Optimizatio Techiques How to Solve Equatios Iterative Methods of Optimizatio Iterative methods of optimizatio Solutio of the oliear equatios resultig form a optimizatio problem is usually
More informationb i u x i U a i j u x i u x j
M ath 5 2 7 Fall 2 0 0 9 L ecture 1 9 N ov. 1 6, 2 0 0 9 ) S ecod- Order Elliptic Equatios: Weak S olutios 1. Defiitios. I this ad the followig two lectures we will study the boudary value problem Here
More informationAssignment 2 Solutions SOLUTION. ϕ 1 Â = 3 ϕ 1 4i ϕ 2. The other case can be dealt with in a similar way. { ϕ 2 Â} χ = { 4i ϕ 1 3 ϕ 2 } χ.
PHYSICS 34 QUANTUM PHYSICS II (25) Assigmet 2 Solutios 1. With respect to a pair of orthoormal vectors ϕ 1 ad ϕ 2 that spa the Hilbert space H of a certai system, the operator  is defied by its actio
More informationChapter 6 Infinite Series
Chapter 6 Ifiite Series I the previous chapter we cosidered itegrals which were improper i the sese that the iterval of itegratio was ubouded. I this chapter we are goig to discuss a topic which is somewhat
More informationRademacher Complexity
EECS 598: Statistical Learig Theory, Witer 204 Topic 0 Rademacher Complexity Lecturer: Clayto Scott Scribe: Ya Deg, Kevi Moo Disclaimer: These otes have ot bee subjected to the usual scrutiy reserved for
More informationAn Introduction to Randomized Algorithms
A Itroductio to Radomized Algorithms The focus of this lecture is to study a radomized algorithm for quick sort, aalyze it usig probabilistic recurrece relatios, ad also provide more geeral tools for aalysis
More informationMAT1026 Calculus II Basic Convergence Tests for Series
MAT026 Calculus II Basic Covergece Tests for Series Egi MERMUT 202.03.08 Dokuz Eylül Uiversity Faculty of Sciece Departmet of Mathematics İzmir/TURKEY Cotets Mootoe Covergece Theorem 2 2 Series of Real
More informationECE 901 Lecture 12: Complexity Regularization and the Squared Loss
ECE 90 Lecture : Complexity Regularizatio ad the Squared Loss R. Nowak 5/7/009 I the previous lectures we made use of the Cheroff/Hoeffdig bouds for our aalysis of classifier errors. Hoeffdig s iequality
More information2 Markov Chain Monte Carlo Sampling
22 Part I. Markov Chais ad Stochastic Samplig Figure 10: Hard-core colourig of a lattice. 2 Markov Chai Mote Carlo Samplig We ow itroduce Markov chai Mote Carlo (MCMC) samplig, which is a extremely importat
More informationlim za n n = z lim a n n.
Lecture 6 Sequeces ad Series Defiitio 1 By a sequece i a set A, we mea a mappig f : N A. It is customary to deote a sequece f by {s } where, s := f(). A sequece {z } of (complex) umbers is said to be coverget
More informationA Lattice Green Function Introduction. Abstract
August 5, 25 A Lattice Gree Fuctio Itroductio Stefa Hollos Exstrom Laboratories LLC, 662 Nelso Park Dr, Logmot, Colorado 853, USA Abstract We preset a itroductio to lattice Gree fuctios. Electroic address:
More informationA Hadamard-type lower bound for symmetric diagonally dominant positive matrices
A Hadamard-type lower boud for symmetric diagoally domiat positive matrices Christopher J. Hillar, Adre Wibisoo Uiversity of Califoria, Berkeley Jauary 7, 205 Abstract We prove a ew lower-boud form of
More informationProduct measures, Tonelli s and Fubini s theorems For use in MAT3400/4400, autumn 2014 Nadia S. Larsen. Version of 13 October 2014.
Product measures, Toelli s ad Fubii s theorems For use i MAT3400/4400, autum 2014 Nadia S. Larse Versio of 13 October 2014. 1. Costructio of the product measure The purpose of these otes is to preset the
More informationSequences and Series of Functions
Chapter 6 Sequeces ad Series of Fuctios 6.1. Covergece of a Sequece of Fuctios Poitwise Covergece. Defiitio 6.1. Let, for each N, fuctio f : A R be defied. If, for each x A, the sequece (f (x)) coverges
More informationSolutions to HW Assignment 1
Solutios to HW: 1 Course: Theory of Probability II Page: 1 of 6 Uiversity of Texas at Austi Solutios to HW Assigmet 1 Problem 1.1. Let Ω, F, {F } 0, P) be a filtered probability space ad T a stoppig time.
More informationRandom Walks on Discrete and Continuous Circles. by Jeffrey S. Rosenthal School of Mathematics, University of Minnesota, Minneapolis, MN, U.S.A.
Radom Walks o Discrete ad Cotiuous Circles by Jeffrey S. Rosethal School of Mathematics, Uiversity of Miesota, Mieapolis, MN, U.S.A. 55455 (Appeared i Joural of Applied Probability 30 (1993), 780 789.)
More informationEconometrica Supplementary Material
Ecoometrica Supplemetary Material SUPPLEMENT TO NONPARAMETRIC INSTRUMENTAL VARIABLES ESTIMATION OF A QUANTILE REGRESSION MODEL : MATHEMATICAL APPENDIX (Ecoometrica, Vol. 75, No. 4, July 27, 1191 128) BY
More informationACO Comprehensive Exam 9 October 2007 Student code A. 1. Graph Theory
1. Graph Theory Prove that there exist o simple plaar triagulatio T ad two distict adjacet vertices x, y V (T ) such that x ad y are the oly vertices of T of odd degree. Do ot use the Four-Color Theorem.
More informationMa 530 Infinite Series I
Ma 50 Ifiite Series I Please ote that i additio to the material below this lecture icorporated material from the Visual Calculus web site. The material o sequeces is at Visual Sequeces. (To use this li
More informationMath 312 Lecture Notes One Dimensional Maps
Math 312 Lecture Notes Oe Dimesioal Maps Warre Weckesser Departmet of Mathematics Colgate Uiversity 21-23 February 25 A Example We begi with the simplest model of populatio growth. Suppose, for example,
More informationOn the distribution of coefficients of powers of positive polynomials
AUSTRALASIAN JOURNAL OF COMBINATORICS Volume 49 (2011), Pages 239 243 O the distributio of coefficiets of powers of positive polyomials László Major Istitute of Mathematics Tampere Uiversity of Techology
More informationCOURSE INTRODUCTION: WHAT HAPPENS TO A QUANTUM PARTICLE ON A PENDULUM π 2 SECONDS AFTER IT IS TOSSED IN?
COURSE INTRODUCTION: WHAT HAPPENS TO A QUANTUM PARTICLE ON A PENDULUM π SECONDS AFTER IT IS TOSSED IN? DROR BAR-NATAN Follows a lecture give by the author i the trivial otios semiar i Harvard o April 9,
More informationLet us give one more example of MLE. Example 3. The uniform distribution U[0, θ] on the interval [0, θ] has p.d.f.
Lecture 5 Let us give oe more example of MLE. Example 3. The uiform distributio U[0, ] o the iterval [0, ] has p.d.f. { 1 f(x =, 0 x, 0, otherwise The likelihood fuctio ϕ( = f(x i = 1 I(X 1,..., X [0,
More informationRoger Apéry's proof that zeta(3) is irrational
Cliff Bott cliffbott@hotmail.com 11 October 2011 Roger Apéry's proof that zeta(3) is irratioal Roger Apéry developed a method for searchig for cotiued fractio represetatios of umbers that have a form such
More informationSeunghee Ye Ma 8: Week 5 Oct 28
Week 5 Summary I Sectio, we go over the Mea Value Theorem ad its applicatios. I Sectio 2, we will recap what we have covered so far this term. Topics Page Mea Value Theorem. Applicatios of the Mea Value
More informationMath 113, Calculus II Winter 2007 Final Exam Solutions
Math, Calculus II Witer 7 Fial Exam Solutios (5 poits) Use the limit defiitio of the defiite itegral ad the sum formulas to compute x x + dx The check your aswer usig the Evaluatio Theorem Solutio: I this
More informationCHAPTER 1 SEQUENCES AND INFINITE SERIES
CHAPTER SEQUENCES AND INFINITE SERIES SEQUENCES AND INFINITE SERIES (0 meetigs) Sequeces ad limit of a sequece Mootoic ad bouded sequece Ifiite series of costat terms Ifiite series of positive terms Alteratig
More informationStrong Convergence Theorems According. to a New Iterative Scheme with Errors for. Mapping Nonself I-Asymptotically. Quasi-Nonexpansive Types
It. Joural of Math. Aalysis, Vol. 4, 00, o. 5, 37-45 Strog Covergece Theorems Accordig to a New Iterative Scheme with Errors for Mappig Noself I-Asymptotically Quasi-Noexpasive Types Narogrit Puturog Mathematics
More informationINFINITE SEQUENCES AND SERIES
INFINITE SEQUENCES AND SERIES INFINITE SEQUENCES AND SERIES I geeral, it is difficult to fid the exact sum of a series. We were able to accomplish this for geometric series ad the series /[(+)]. This is
More informationThe natural exponential function
The atural expoetial fuctio Attila Máté Brookly College of the City Uiversity of New York December, 205 Cotets The atural expoetial fuctio for real x. Beroulli s iequality.....................................2
More informationMASSACHUSETTS INSTITUTE OF TECHNOLOGY 6.265/15.070J Fall 2013 Lecture 21 11/27/2013
MASSACHUSETTS INSTITUTE OF TECHNOLOGY 6.265/15.070J Fall 2013 Lecture 21 11/27/2013 Fuctioal Law of Large Numbers. Costructio of the Wieer Measure Cotet. 1. Additioal techical results o weak covergece
More informationChapter 8. Euler s Gamma function
Chapter 8 Euler s Gamma fuctio The Gamma fuctio plays a importat role i the fuctioal equatio for ζ(s that we will derive i the ext chapter. I the preset chapter we have collected some properties of the
More informationAlgebra of Least Squares
October 19, 2018 Algebra of Least Squares Geometry of Least Squares Recall that out data is like a table [Y X] where Y collects observatios o the depedet variable Y ad X collects observatios o the k-dimesioal
More informationA 2nTH ORDER LINEAR DIFFERENCE EQUATION
A 2TH ORDER LINEAR DIFFERENCE EQUATION Doug Aderso Departmet of Mathematics ad Computer Sciece, Cocordia College Moorhead, MN 56562, USA ABSTRACT: We give a formulatio of geeralized zeros ad (, )-discojugacy
More informationON POINTWISE BINOMIAL APPROXIMATION
Iteratioal Joural of Pure ad Applied Mathematics Volume 71 No. 1 2011, 57-66 ON POINTWISE BINOMIAL APPROXIMATION BY w-functions K. Teerapabolar 1, P. Wogkasem 2 Departmet of Mathematics Faculty of Sciece
More informationRandom Matrices with Blocks of Intermediate Scale Strongly Correlated Band Matrices
Radom Matrices with Blocks of Itermediate Scale Strogly Correlated Bad Matrices Jiayi Tog Advisor: Dr. Todd Kemp May 30, 07 Departmet of Mathematics Uiversity of Califoria, Sa Diego Cotets Itroductio Notatio
More informationNotes for Lecture 11
U.C. Berkeley CS78: Computatioal Complexity Hadout N Professor Luca Trevisa 3/4/008 Notes for Lecture Eigevalues, Expasio, ad Radom Walks As usual by ow, let G = (V, E) be a udirected d-regular graph with
More informationChimica Inorganica 3
himica Iorgaica Irreducible Represetatios ad haracter Tables Rather tha usig geometrical operatios, it is ofte much more coveiet to employ a ew set of group elemets which are matrices ad to make the rule
More informationMA131 - Analysis 1. Workbook 2 Sequences I
MA3 - Aalysis Workbook 2 Sequeces I Autum 203 Cotets 2 Sequeces I 2. Itroductio.............................. 2.2 Icreasig ad Decreasig Sequeces................ 2 2.3 Bouded Sequeces..........................
More informationPoincaré Problem for Nonlinear Elliptic Equations of Second Order in Unbounded Domains
Advaces i Pure Mathematics 23 3 72-77 http://dxdoiorg/4236/apm233a24 Published Olie Jauary 23 (http://wwwscirporg/oural/apm) Poicaré Problem for Noliear Elliptic Equatios of Secod Order i Ubouded Domais
More informationLinear Elliptic PDE s Elliptic partial differential equations frequently arise out of conservation statements of the form
Liear Elliptic PDE s Elliptic partial differetial equatios frequetly arise out of coservatio statemets of the form B F d B Sdx B cotaied i bouded ope set U R. Here F, S deote respectively, the flux desity
More informationRandom Models. Tusheng Zhang. February 14, 2013
Radom Models Tusheg Zhag February 14, 013 1 Radom Walks Let me describe the model. Radom walks are used to describe the motio of a movig particle (object). Suppose that a particle (object) moves alog the
More informationMeasure and Measurable Functions
3 Measure ad Measurable Fuctios 3.1 Measure o a Arbitrary σ-algebra Recall from Chapter 2 that the set M of all Lebesgue measurable sets has the followig properties: R M, E M implies E c M, E M for N implies
More informationNotes 19 : Martingale CLT
Notes 9 : Martigale CLT Math 733-734: Theory of Probability Lecturer: Sebastie Roch Refereces: [Bil95, Chapter 35], [Roc, Chapter 3]. Sice we have ot ecoutered weak covergece i some time, we first recall
More informationMATH 31B: MIDTERM 2 REVIEW
MATH 3B: MIDTERM REVIEW JOE HUGHES. Evaluate x (x ) (x 3).. Partial Fractios Solutio: The umerator has degree less tha the deomiator, so we ca use partial fractios. Write x (x ) (x 3) = A x + A (x ) +
More informationAverage Reward Optimization Objective In Partially Observable Domains - Supplementary Material
Average Reward Optimizatio Objective I Partially Observable Domais - Supplemetary Material Aother example of a cotrolled system (see Sec. 4.2) Figure 3. The first two plots describe the behavior of the
More informationEntropy and Ergodic Theory Lecture 5: Joint typicality and conditional AEP
Etropy ad Ergodic Theory Lecture 5: Joit typicality ad coditioal AEP 1 Notatio: from RVs back to distributios Let (Ω, F, P) be a probability space, ad let X ad Y be A- ad B-valued discrete RVs, respectively.
More informationDifferentiable Convex Functions
Differetiable Covex Fuctios The followig picture motivates Theorem 11. f ( x) f ( x) f '( x)( x x) ˆx x 1 Theorem 11 : Let f : R R be differetiable. The, f is covex o the covex set C R if, ad oly if for
More informationsubcaptionfont+=small,labelformat=parens,labelsep=space,skip=6pt,list=0,hypcap=0 subcaption ALGEBRAIC COMBINATORICS LECTURE 8 TUESDAY, 2/16/2016
subcaptiofot+=small,labelformat=pares,labelsep=space,skip=6pt,list=0,hypcap=0 subcaptio ALGEBRAIC COMBINATORICS LECTURE 8 TUESDAY, /6/06. Self-cojugate Partitios Recall that, give a partitio λ, we may
More information1 Convergence in Probability and the Weak Law of Large Numbers
36-752 Advaced Probability Overview Sprig 2018 8. Covergece Cocepts: i Probability, i L p ad Almost Surely Istructor: Alessadro Rialdo Associated readig: Sec 2.4, 2.5, ad 4.11 of Ash ad Doléas-Dade; Sec
More informationGeneralized Semi- Markov Processes (GSMP)
Geeralized Semi- Markov Processes (GSMP) Summary Some Defiitios Markov ad Semi-Markov Processes The Poisso Process Properties of the Poisso Process Iterarrival times Memoryless property ad the residual
More informationMath Solutions to homework 6
Math 175 - Solutios to homework 6 Cédric De Groote November 16, 2017 Problem 1 (8.11 i the book): Let K be a compact Hermitia operator o a Hilbert space H ad let the kerel of K be {0}. Show that there
More informationOn forward improvement iteration for stopping problems
O forward improvemet iteratio for stoppig problems Mathematical Istitute, Uiversity of Kiel, Ludewig-Mey-Str. 4, D-24098 Kiel, Germay irle@math.ui-iel.de Albrecht Irle Abstract. We cosider the optimal
More informationSection A assesses the Units Numerical Analysis 1 and 2 Section B assesses the Unit Mathematics for Applied Mathematics
X0/70 NATIONAL QUALIFICATIONS 005 MONDAY, MAY.00 PM 4.00 PM APPLIED MATHEMATICS ADVANCED HIGHER Numerical Aalysis Read carefully. Calculators may be used i this paper.. Cadidates should aswer all questios.
More informationNumerical Method for Blasius Equation on an infinite Interval
Numerical Method for Blasius Equatio o a ifiite Iterval Alexader I. Zadori Omsk departmet of Sobolev Mathematics Istitute of Siberia Brach of Russia Academy of Scieces, Russia zadori@iitam.omsk.et.ru 1
More informationChapter 4. Fourier Series
Chapter 4. Fourier Series At this poit we are ready to ow cosider the caoical equatios. Cosider, for eample the heat equatio u t = u, < (4.) subject to u(, ) = si, u(, t) = u(, t) =. (4.) Here,
More informationSequences and Limits
Chapter Sequeces ad Limits Let { a } be a sequece of real or complex umbers A ecessary ad sufficiet coditio for the sequece to coverge is that for ay ɛ > 0 there exists a iteger N > 0 such that a p a q
More informationStochastic Matrices in a Finite Field
Stochastic Matrices i a Fiite Field Abstract: I this project we will explore the properties of stochastic matrices i both the real ad the fiite fields. We first explore what properties 2 2 stochastic matrices
More informationZeros of Polynomials
Math 160 www.timetodare.com 4.5 4.6 Zeros of Polyomials I these sectios we will study polyomials algebraically. Most of our work will be cocered with fidig the solutios of polyomial equatios of ay degree
More informationThe picture in figure 1.1 helps us to see that the area represents the distance traveled. Figure 1: Area represents distance travelled
1 Lecture : Area Area ad distace traveled Approximatig area by rectagles Summatio The area uder a parabola 1.1 Area ad distace Suppose we have the followig iformatio about the velocity of a particle, how
More information