Large Contests - Online Appendix
|
|
- Juliana Hawkins
- 7 years ago
- Views:
Transcription
1 Large Cotests - Oe Appedx Wojcech Oszewsk ad Ro Sege August 2015 We choose a equbrum for each cotest, ad refer to the sequece whch the -th eemet s the equbrum of the -th cotest as the sequece of equbra. For both our theorems, we w show that every subsequece of ths sequece cotas a further subsequece that satsfes the statemet of the theorem. Ths suffces, because the foowg observato cabeappedwthz beg the set of equbra of cotest. (Subsequece Property) Gve a sequece of sets {Z : =1, 2,...}, supposethatfor every subsequece {Z k : k =1, 2,...}, every sequece {z k : k =1, 2,...} wth z k Z k cotas a subsequece {z k : =1, 2,...} such that every eemet z k has some property. The there exsts a N such that for every N every eemet Z hasthsproperty. 1 1 Proof of Theorem 1 We beg wth a oute of the proof. Gve a subsequece of equbra, each equbrum the subsequece duces for each payer a mappg from bds to expected percete rakgs. We cosder the average of those mappgs, ad G 1 composed wth ths average gves a mappg T from bds to przes. As creases, ths mappg approxmates the equbrum mappgs from bds to przes of a payers the -th cotest. We the use Hey s (1912) seecto theorem to fd a subsequece of T that coverges to some mt mappg T from bds to przes. We show that T s cotuous ad the subsequece of T coverges uformy to T. The, for each aget type we defe the set of optma bds whe T s treated as a Departmet of Ecoomcs, Northwester Uversty, Evasto, IL (e-ma: wo@orthwester.edu) ad The Pesyvaa State Uversty, Uversty Park, IL (e-ma: rus41@psu.edu). 1 Otherwse, there woud be a sequece {z k : k =1, 2,...} wth z k Z k of eemets wthout the property.
2 verse tarff, addefe the correspodece from aget types to sets of optma bds. We cosder a sma eghborhood of the graph of ths correspodece, ad show that for arge every payer s best resposes the -thcotestarethe x -sce of ths eghborhood. Such a sce coud prcpe be arge eve f the set of optma bds of the correspodg aget type s sma (ths woud happe f the set of optma bds of a earby aget type s arge). We show, however, that uder strct sge crossg each aget type has a sge optma bd, whch s cotuous ad weaky creases the aget type. Ths mpes that every payer s best respose set, ad therefore the support of her equbrum strategy, s bouded wth a arbtrary sma terva as creases. We the cocude that the uque mechasm duced by T mpemets the assortatve aocato. Ths demostrates part (b) the statemet of the theorem; part (a) the foows easy. For the proof, we take the subsequece of equbra to be the sequece of equbra (ths smpfes otato ad has o effect o the proofs). We deote the equbrum of the -th cotest by σ =(σ 1,...,σ ), whereσ s payer s equbrum strategy; a strategy of payer s a radom varabe takg vaues X B whose marga dstrbuto o X cocdes wth the dstrbuto of payer s types F. By referrg to payer bddg wth some probabty a subset S of B, we mea the probabty of the set X S,.e., the probabty of S measured by the marga dstrbuto of payer s strategy o B. We deote by R (t) the radom varabe that s the percete ocato of payer the orda rakg of the payers the -th cotest f she bds sghty above t ad the other payers empoy ther equbrum strateges. 2 That s, Ã! R (t) = 1 1+ X k6= 1 {σ k X [0,t]} where 1 {σ X [0,t]} s 1 f σ X [0,t] ad 0 otherwse. Let Ã,! A (t) = 1 1+ X Pr (σ k X [0,t]) k6= be the expected percete rakg of payer. The, by Hoeffdg s equaty, for a t B we have Pr ( R (t) A (t) >δ) < 2exp 2δ 2 ( 1) ª. (1) Fay, et A (t) = 1 X A (t) 2 Ths s the fmumofherrakgfshebdsabovet, whch s equvaet to bddg t ad wg ay tes there. If tes happe wth probabty 0, the ths s equvaet to bddg t. 2 =1
3 be the average of the expected percetes rakgs of the payers the -th cotest f they bd t ad the other payers empoy ther equbrum strateges. Let T be the mappg from bds to przes duced by A.Thats,T (t) =(G ) 1 (A (t)), where (G ) 1 (z) =f{y : G (y) z} for z>0, ad(g ) 1 (0) = f {y : G (y) > 0}. (I words, (G ) 1 (z) s the prze of a aget wth percete rakg z whe przes are dstrbuted accordg to G.) Sce every T s (weaky) creasg, by Hey s (1912) seecto theorem for mootoe fuctos the sequece T cotas a subsequece that coverges potwse to a fucto T : B Y. For the rest of the proof, deote ths subsequece by T. We frst descrbe some propertes of verse tarff T : (1) T s (weaky) creasg, because every T s (weaky) creasg. (2) T (0) = 0, otherwse payers bddg 0 woud have proftabe devatos. 3 (3) T (b max )=1, because A (b max )=1ad therefore T (b max )=1. I addto, we w use the foowg property of dscrete cotest equbra: (No-Gap Property) I ay equbrum, there s o terva (a, b) B of postve egth whch a payers bd wth probabty 0 ad some payer bds [b, b max ] wth postve probabty. Proof: Suppose the cotrary, ad cosder such a maxma terva (a, b). A payer woud oy bd b or sghty hgher tha b f some other payer bds b wth postve probabty. But the payer who bds b wth postve probabty woud be better off ether by sghty creasg her bd (f aother payer bds b ad wg the te eads to a hgher prze) or by decreasg her bd ( the compemetary case). Our frst emma shows that T s cotuous. (I order ot to obscure the structure of the proof, we reegate to the ed of the secto the proofs of a emmas.) Lemma 1 For ay t B ad ay sequeces q m t ad r m t B, wehavem T (q m )= m T (r m )=T(t). The dea of the proof s that f T were dscotuous at some t, theforarge t woud be better to bd sghty above t tha sghty beow t. But f o payer bds sghty beow t, the by the No-Gap Property o payer bds t or above. 3 Ideed, suppose to the cotrary that T (0) > 0. Ths meas that for some δ>0ad arge eough, A (0) >G (0) + δ. Thus, a fracto of at east G (0) + δ payers bd 0 the -th cotest wth postve probabty. Ay oe of them woud be better off bddg sghty above 0, ad wg agast a other payers who bd 0, tha bddg 0 ad wth postve probabty osg to a other payers who bd 0. 3
4 Cotuty ad mootocty of T mpy the foowg resut. Lemma 2 T coverges to T uformy o B. Weowreatetheversetarff T to payers behavor the equbra that correspod to the sequece T. Deote by BR x type x s set of optma bds gve T,.e., the bds t that maxmze U(x, T (t),t). Deote by BR(ε) the ε-eghborhood of the graph of the correspodece that assgs to every type x the set BR x. 4 Deote by BR x (ε) the set of bds t such that (x, t) BR (ε). Note that BR(ε) s a 2-dmesoa ope set, whe each BR x (ε) s a 1-dmesoa sce of BR (ε). Usg sets BR x (ε), we ca characterze payers equbrum behavor. Lemma 3 For every ε>0, there s a N such that for every N, theequbrumof the -th cotest every best respose of every type x of every payer beogs to BR x (ε). Strct sge crossg mpes severa propertes of BR x. Lemma 4 For every x the set BR x s a sgeto. I addto, the fucto br that assgs to x thesgeeemetofbr x s cotuous ad weaky creasg. Lemma 4 mpes that for every ε>0 there s a δ>0such that BR x (δ) [br (x) ε, br (x)+ε] for every type x. We therefore have the foowg coroary of Lemmas 3 ad 4. Coroary 1 For every ε>0, there s a N such that for every N, theequbrum of the -th cotest every best respose of every type x of every payer beogs to (br (x ) ε, br (x )+ε). To prove part (b) of the theorem we eed to show that T br s the assortatve aocato. Ths s doe by the foowg emma. Lemma 5 G 1 (F (x)) = T (br (x)) for a types x. Thus, the mechasm that prescrbes for type x prze T (br (x)) ad bd br (x) s a tarff mechasm that mpemets the assortatve aocato. Moreover, every type ca get at east 0 by bddg 0, adtype0 gets o more tha 0 (because T (br (0)) = 0). To compete the proof, t remas to show part (a) of the theorem. I short, ths part foows from Coroary 1 ad Hoeffdg s equaty (see Appedx 1.6). 4 That s, BR (ε) s the uo over a types x ad bds t BR x of the ope bas of radus ε cetered at (x, t). 4
5 1.1 Proof of Lemma 1 Suppose frst the emma s fase for some t (0,b max ] ad q m t. Let y 0 = m T (q m ) ad y 00 = T (t) (the mt exsts by the mootocty of T ), ad et γ =(y 00 y 0 ) /2. Suppose frst that U (0,y,t) strcty creases y. (Reca that we assumed U (x, y, t) strcty creases y oy for x > 0.) The, by uform cotuty of U, thereexstδ, > 0 such that every type x gas at east from obtag a prze hgher by γ at a bd hgher by δ. More precsey, U (x, y + γ,t + δ) U (x, y, t) (2) for a x, y, adt such that y + γ ad t + δ beog to the doma of U. Ths mpes, as U s bouded, that every type x strcty prefers bddg t + δ ad obtag wth suffcety hgh probabty a prze suffcety cose to y + γ to bddg t ad obtag wth suffcety hgh probabty a prze suffcety cose to y, depedety of the przes obtaed wth the remag probabty. Choose t 0 = q m such that t t 0 <δ. Next, choose arge eough so that T (t) T (t) < γ/2 ad T (t 0 ) T (t 0 ) <γ/2. ThsmpesthatT (t) T (s) >γfor ay bd s t 0. By choosg arge eough, we guaratee (see (1), whch appes uformy to a bds) that R (s), the percete rakg of payer who bds s the -th cotest, s cose to A (s) wth hgh probabty, ad R (t) s cose to A (t) wth hgh probabty. Thus, every type x obtas a prze suffcety cose to T (t) wth a suffcety hgh probabty by bddg (sghty above) t, ad obtas a prze that s wth a suffcety hgh probabty at most sghty hgher tha T (t 0 ) by bddg (sghty above) ay s t 0. 5 Therefore, because t t 0 <δ, o payer bds ay s (t δ, t 0 ] wth postve probabty, so by the No-Gap Property T (t 0 )=1.ButT (t 0 ) T (t 0 ) y 0 <y 00 1, a cotradcto. Whe U (0,y,t) oy weaky creases y, the argumet above shows that for ay ε>0 there exst δ, > 0 for whch (2) hods for every type x [ε, 1]. There aso exsts t 0 = q m such that t t 0 < δ ad T (t) T (t 0 ) > 3γ/2 for arge eough. Lettg t 00 =f{s : T (s) T (t) γ} (t 0,t] we see that oy payers wth types ower tha ε ca bd [t 0,t 00 ). Thus, for sma eough ε (by cotuty of G 1 ad covergece of (G ) 1 to G 1 ), order to crease T (t 0 ) to T (t 0 )+γ/2 mutpe payers wth types ε or hgher must bd t 00 wth postve probabty ad therefore te there. But the ay oe of these payers coud proftaby devate to bddg sghty above t For t = b max, bddg sghty above b max s mpossbe. But by bddg b max a payer ws wth probabty 1, because b max s strcty domated by 0 for a payers. 6 By dog so such a payer woud obta wth hgh probabty a prze of at east T (t 0 )+γ/2 stead of osg the te wth postve probabty ad the obtag wth hgh probabty a prze of at most T (t 0 )+ε. 5
6 Theargumetsaaogousfwesupposethattheemmasfaseforsomet (0,b max ) ad r m t. Ift =0, the the above proof shows that for arge o payer bds t =0wth postve probabty. Ths meas, tur, that suffcety sma bds gve ower payoffs tha t 0 = r m such that t 0 t<δ. Thus, o payer bds cose to t =0wth postve probabty, whch cotradcts the No-Gap Property. 1.2 Proof of Lemma 2 Suppose the cotrary. The, there s some δ>0ad a sequece of tegers 1, 2,... such that for every k there s some bd t k wth T k (tk ) T (t k ) >δ. Passg to a subsequece f ecessary, we assume that t k t. Cosder umbers q 0 ad q 00 such that q 0 <t<q 00 ad T (q 00 ) T (q 0 ) <δ/2; suchumbers exst because T s cotuous. 7 For arge eough vaues of k,wehavethat T k (q 0 ) T (q 0 ) < δ/2 ad T k (q 00 ) T (q 00 ) <δ/2. For ay t 0 [q 0,q 00 ],ether(a)t k (t 0 ) T (t 0 ),or(b)t k (t 0 ) T (t 0 ). By mootocty of T ad T k,wehave T k (t 0 ) T (t 0 ) T k (q 00 ) T (q 0 ) T k (q 00 ) T (q 00 ) + T (q 00 ) T (q 0 ) <δ case (a), ad T (t 0 ) T k (t 0 ) T (q 00 ) T k (q 0 ) T (q 00 ) T (q 0 ) + T (q 0 ) T k (q 0 ) <δ case (b). Sce t k [q 0,q 00 ] for arge eough k, we obta a cotradcto to the assumpto that T k (tk ) T (t k ) >δfor a such k. 1.3 Proof of Lemma 3 Suppose to the cotrary that for arbtrary arge, the equbrum of the -th cotest some type x of some payer has a best respose that beogs to the compemet of BR x (ε). Passg to a coverget subsequece f ecessary, we assume that x x. Note that for every x theresaδ x > 0 such that (uder the verse tarff) aybdfrom the compemet of BR x (ε) gves type x apayoff ower by at east δ x tha ay eemet of BR x does. Let δ = δ x. We have that: 7 If t =0set q 0 =0,adft = b max set q 00 = b max. 6
7 1. The maxma payoff of type x, attaed at ay bd from BR x,scotuousx. Ths foows from Berge s Theorem. 2. For every ρ>0, forsuffcety arge the hghest payoff that type x caobtaby bddg the compemet of BR x (ε) caot exceed by ρ the hghest payoff that type x ca obta by bddg the compemet of BR x (ε). Ideed, suppose that for a sequece k dvergg to type x k obtas by bddg some t k the compemet of BR x k (ε) apayoff at east ρ hgher tha the hghest payoff that type x ca obta by bddg the compemet of BR x (ε). Passg to a coverget subsequece f ecessary, we assume that t k t. Sce every (x k,t k ) beogs to the compemet of BR (ε), so does (x,t); thus,(x,t) beogs to the compemet of BR x (ε). However,bycotuty of the payoff fuctos, bddg t gves type x apayoff by at east ρ hgher tha the hghest payoff that type x ca obta by bddg the compemet of BR x (ε), a cotradcto. By 1 ad 2, for suffcety arge, aybdthecompemetofbr x (ε) gves type x apayoff ower by at east δ/2 tha ay bd BR x. Ideed, by 2 apped to ρ = δ/4, ay bd the compemet of BR x (ε) gves type x apayoff at most δ/4 hgher tha the hghest payoff that type x ca obta by bddg the compemet of BR x (ε). Thsast payoff s tur ower tha the payoff that type x obtas by bddg BR x by at east δ. Adby1,thepayoff that type x obtas by bddg BR x caot be ower by more tha δ/4 tha the payoff that type x obtas by bddg BR x. By uform covergece of T to T, the aaogous statemet, wth δ/2 repaced wth some smaer postve umber ad T repaced wth T, s aso true. Ths meas, however, that for suffcety arge,payer woud be strcty better off bddg sghty above ay bd BR x whe her type s x tha bddg the compemet of BR x (ε). Thssbecause (1)mpesthatforsuffcety arge, by bddg sghty above t the payer obtas a prze arbtrary cose to T (t) wth probabty arbtrary cose to Proof of Lemma 4 Observe that for ay x 0 <x 00,strctsgecrossgmpesthatft 0 BR x 0 ad t 00 BR x 00, the t 0 t 00. Suppose that BR x 0 cotaed two bds, t 1 <t 2,forsometypex. The frst observato ad Lemma 3 mpy that for ay 0 <ε<(t 2 t 1 )/4, forsuffcety arge oy payers wth types I =[max{x 0 ε, 0}, m {x 0 + ε, 1}] may bd the terva [t 1 +(t 2 t 1 )/4,t 2 (t 2 t 1 )/4]. Cosder obtag a prze that s hgher the mt prze dstrbuto 8 by creasg 8 More precsey, gve a ta prze y 0, the prze that s hgher the mt prze dstrbuto s the prze y 00 such that = G(y 00 ) G(y 0 ). The prze that s hgher the prze dstrbuto G s defed 7
8 the bd from t 1 +(t 2 t 1 )/4 to t 2 (t 2 t 1 )/2. If s suffcety sma, the by cotuty of G 1 the crease the prze s sma as we, so the assocated cremet utty s egatve for a types, ad uformy bouded away from 0. Therefore, takg /2 =F (m {x 0 + ε, 1}) F (max {x 0 ε, 0}), fε>0 s suffcety sma, the for suffcety arge every type of every payer s better off bddg t 1 +(t 2 t 1 )/4 tha bddg t 2 (t 2 t 1 )/2. Ths s because wth hgh probabty the hgher bd eads to a prze that s approxmatey oy /2 hgher the prze dstrbuto G. 9 By covergece of (G ) 1 to (G) 1,forsuffcety arge ths prze s ot much more tha /2 hgher the mt prze dstrbuto. Moreover, every type of every payer s better off bddg t 1 +(t 2 t 1 )/4 tha bddg ay bdterva(t 2 (t 2 t 1 )/2,t 2 (t 2 t 1 )/4), because such bds are eve more costy tha t 2 (t 2 t 1 )/2, ad eabe a payer to obta a prze that s wth hgh probabty ot much more tha /2 hgher the mt prze dstrbuto tha the prze the payer obtas by bddg t 2 (t 2 t 1 )/2. Therefore, o payer bds the terva ((t 2 (t 2 t 1 )/2,t 2 (t 2 t 1 )/4)) wth postve probabty, so by the No-Gap Property T (t 2 (t 2 t 1 )/4) = 1 for suffcety arge. Thus, T (t 2 (t 2 t 1 )/4) = 1, sot 2 caot be BR x, because bddg sghty above t 2 (t 2 t 1 )/4 gves type x ahgherpayoff. Cosequety, BR x s a sgeto for ay x, ad by strct sge crossg, br s weaky creasg. A argumet aaogous to the argumet used to show that BR s a sgeto aso shows that br s cotuous Proof of Lemma 5 Cosder a arbtrary type x. Let x m =m{z : br (z) =br (x)} ad x max =max{z : br (z) =br (x)} (x m ad x max are we defed because br s cotuous). Frst, observe that G 1 (F x m )=G 1 (F (x max )). Ideed, by Coroary 1, for suffcety arge a types the terva [x m,x max ] bd the -th cotest cose to br (x). Suppose that G 1 (F x m ) <G 1 (F (x max )), ad cosder the payers whose type beogs smary. 9 Ths foows from the covergece of F to F ad Hoeffdg s equaty apped to radom varabes ½ 1 f m {x Z 0 + ε, 1} x = max {x0 ε, 0}, 0 otherwse, for =1,...,. 10 More precsey, suppose that br s dscotuous at some x, ad appy the argumet to t 1 = br(x 1 ) ad t 2 = br(x 2 ) where x 1 ad x 2 are sghty ower ad hgher, respectvey, tha x. 8
9 to [x m,x max ] wth postve probabty. 11 Amog these payers, the oe whose expected prze s the owest cotget o havg a type ths terva ca proftaby devate to bddg sghty above br (x), thereby outbddg the other payers wth a type ths terva ad obtag a dscretey hgher prze. Suppose that x m > 0. By Coroary 1 for ay δ>0, theresan such that f N, the the equbrum bds of every payer wth type ower tha x m δ are ower tha br(x m ), ad the equbrum bds of every payer wth type hgher tha x m are hgher tha br x m δ. Therefore, a payer who bds br x m outbds a payers wth types ower tha x m δ, sot (br x m ) (G ) 1 (F x m δ ), ad a payer who bds br x m δ s outbd by a payers wth types hgher tha x m,sot br x m δ (G ) 1 (F x m ). Sce T coverges to T, T ad br are cotuous, (G ) 1 coverges to (G 1 ), F coverges to F,adF ad G 1 are cotuous, we obta T br x m = G 1 F x m. Smary, f x max < 1, weobtathatt (br (x max )) = G 1 (F (x max )). Thus, sce br(x) =br(x m )=br(x max ) ad G 1 (F (x)) = G 1 (F (x m )) = G 1 (F (x max )), we have that T (br (x)) = G 1 (F (x)) whe x m > 0 or x max < 1. Fay,tcaotbethat x m =0ad x max =1, because 0=G 1 (F (0)) <G 1 (F (1)) = Proof of Part (a) of Theorem 1 Cosder a type x X. Lett = br (x), adett 0 ad t 00 be such that T (t 0 )=T(t) ε/3 ad T (t 00 )=T(t)+ε/3. Fay, et x 0 ad x 00 be such that t 0 = br (x 0 ) ad t 00 = br (x 00 ).(Take x 0 =0ad t 0 =0f T (t) ε/3 < 0, adx 00 =1ad t 00 = br(1) f T (t)+ε/3 > 1.) By Lemma 3, for suffcety arge, every payer wth type o hgher tha x 0 bds ess tha the payer wth type x, ad every payer wth type o ower tha x 00 bds more tha the payer wth type x. ByHoeffdg s equaty, the payer of type x outbds wth hgh probabty at east a fracto of payers cose to F (x 0 ). Sce F coverges to F, she outbds wth hgh probabty at east a fracto of payers cose to F (x 0 ).So,sce(G ) 1 coverges to G 1, she obtas (wth hgh probabty) a prze o ower tha G 1 (F (x 0 )) ε/3 =T (t) 2ε/3. Smary, for suffcety arge apayerof type x outbds wth hgh probabty at most a fracto of payers cose to F (x 00 ), ad so she obtas (wth hgh probabty) a prze o hgher tha G 1 (F (x 00 )) = T (t)+2ε/3. (These bouds are mmedate f y ε/2 < 0 or f y + ε/2 > 1.) Thus, type x obtas (wth hgh probabty) a prze whch dffers from G 1 (F (x)) by at most 2ε/3. 11 For arge, at east a fracto of payers cose to F (x max ) F x m have types that beog to [x m,x max ] wth postve probabty. 9
10 Ths proves part (a) for a sge x, but we must show that there s a N such that for ay N part (a) hods for a x smutaeousy. Such a N ca be obtaed by takg a fte grd of types x, ad the correspodg grd of bds br (x) such that T (t 1 ) T (t 2 ) <ε/3 for ay par of eghborg eemets t 1, t 2 of the grd, ad takg the argest N amog the N s correspodg to x s from the grd. 2 Proof of Theorem 2 Reca that G 1 (z) =f{y : G (y) z} for z>0ad G 1 (0) = f {y : G (y) > 0}, ad ote that G 1 may be dscotuous (but s eft-cotuous). Dscotutes requre modfyg amost a the argumets used the proof of Theorem 1. As Appedx 1, we reegate to the ed of the secto the proofs of a termedate resuts. Let I 0 =(y0,y 0 u ) be a ogest terva [0,G 1 (1)] to whch G assgs measure 0; et I 1 =(y1,y 1 u ) be a ogest such terva dsjot from I 1, ad so o. The, every ope terva of przes that has measure zero s cotaed oe of the tervas I 0,I 1,... Ad for ay ε>0, theresak such that the egths of I K+1, I K+2,... sum up to ess tha ε. The deftos of R, A, A,adT are as Appedx 1. The defto of T, however, must be chaged. Frst, by Hey s seecto theorem, we take a covergg subsequece of the sequece A ; deote ts mt by A : B [0, 1]. Ths fucto s weaky creasg (because each A s). For the rest of the proof, deote ths covergg subsequece by A (wth the correspodg sequece T =(G ) 1 A ). Let T = G 1 A. SceG may ot have fu support, we ow have that T (0) = f{z : G(z) > 0} ad T (b max )=G 1 (1); addto, T s st (weaky) creasg (compare to propertes (1)-(3) from Appedx 1). I addto, F coverges potwse to F,but(G ) 1 may ot coverge potwse to G 1. It s, however, easy to check that m (G ) 1 (r) =G 1 (r) uess r s the vaue of G o a terva I k =(yk,yu k );moreover,m (G ) 1 (r) G 1 (r) for every r that s the vaue of G o a terva I k =(yk,yu k ), but t ca happe that m (G ) 1 (r) =yk u ad G 1 (r) =yk. The dscotutes G 1 mpy that T mayotbecotuous,solemma1doesot hod. Pots of dscotuty, however, correspod to ope tervas of przes that have measure zero. More precsey, we have the foowg resut. Lemma 6 For ay t>0 B, oe of the foowg codtos hods: 1. T s cotuous at t, that s, for ay sequeces q m t ad r m t, wehavem T (q m )= m T (r m )=T(t). 10
11 2. There s some k =1, 2,... such that for ay sequeces q m t ad r m t, wehave m T (q m )=yk ad m T (rm )=yk u.moreover,m A (qm )=G(yk ) ad m A (rm )= G(yk u). Usg Lemma 6, we defe aother fucto T o B by settg T (t) = m T (r) for some sequece r t (ad T (b max )=G 1 (1)). The mootocty of T guaratees that T (t) s we-defed. I addto, t s easy to check that T s (weaky) creasg, rght-cotuous, ad cotuous at every bd t such that codto 1 from Lemma 6 hods. Note that T may otbeaextesooft, because whe m T (r) 6= T (t), wehavethatt (t) = m T (r) 6= T (t). Cosder ow a bd t > 0 such that codto 2 from Lemma 6 hods. Deote ths bd t by t k,wherek s descrbed codto 2. The, there s a bd t 0 <t k such that A (t 0 )=A(t) =G yk,soa s costat o a terva beow tk. Ideed, f A(t 0 ) <G yk for a t 0 <t k, the, as the proof of Lemma 6, for arge o payer woud bd ay t 0 sghty beow t k. Ths woud be so, because bddg sghty above t k woud amost certay gve aprzeoowerthayk u, whereas bddg t0 woud amost certay gve a prze o hgher tha yk.let t k =f t 0 : A (t 0 )=G yk ª <tk. It s aso true that every maxma terva o whch T s costat wth a vaue ower tha G 1 (1) s t k,t k for some k. Ideed, cosder a maxma otrva terva wth ower boud t ad upper boud t u o whch the vaue of T s y<g 1 (1). It suffces to show that T (t u ) >y, because the codto 2 from Lemma 6 appes to t u, whch mpes that t u = t k for some k; ad the maxmaty of t,t u yeds t = t k. Suppose that T (t u )=y. The, for arge eough bddg t u amost certay gves a prze at most sghty hgher tha y, whereas bddg sghty above t amost certay gves a prze ot much ower tha y. But the, for arge eough, o payer bds some eghborhood of t u, because bddg sghty above t eads to a hgher payoff. Ths cotradcts the No-Gap Property, because y<g 1 (1). Because G 1 may be dscotuous, T eed ot coverge uformy to T or T,eve o the set of pots at whch they are cotuous. I partcuar, for a t t k,t k t may be that T (t) =(G ) 1 (A (t)) yk u for arbtrary arge, whereast (t) =T (t) =yk. Nevertheess, T coverges uformy except o some eghborhoods of a fte umber of tervas [t k,t k]. More precsey, we say that T coverges uformy to T up to β o a set C f there exsts a N such that for every N ad t C we have that T (t) T (t) <β. We the have the foowg modfcato of Lemma 2. 11
12 Lemma 7 For every β>0, there exsts a umber K such that for every γ>0, T coverges uformy to T up to β o the compemet of K[ O γ = (t k γ,t k + γ). k=1 We ow reate payers equbrum behavor arge cotests to the verse tarff T. Defe BR x, BR(ε), adbr x (ε) as Appedx 1 wth T stead of T (the maxma payoff s acheved because T s creasg ad rght-cotuous, so s upper sem-cotuous). Defe the mass expeded ( the -th cotest) a terva of bds I by payers wth type x S as ( P =1 Pr (σ S I)) /. We the have the foowg resut, whch we use provg the remag resuts. Lemma 8 For a k ad ay ε>0 ad L>0, thereexstsγ>0 such that for suffcety arge we have that: () The mass expeded (t k γ,t k + γ) by payers wth types x for whch t k / BR x (ε) s ess tha ε/3l; () The mass expeded (t k γ,t k ] by payers wth types x for whch t k / BR x (ε) s ess tha ε/3l. I addto, for ay α>0, forsuffcety arge we have that: () The mass expeded t k + α, t k α byapayerssessthaε/3l. Lemma 3 must aso to be modfed. Lemma 9 For every ε>0, thereexstk such that for every γ>0, there s a N such that for every N the equbrum of the -th cotest every best respose of every type x of every payer beogs to K[ BR x (ε) (t k γ,t k ). k=1 Strct sge crossg o oger mpes that BR x s a sgeto. Istead, we have the foowg resut. Lemma 10 If strct sge crossg hods, the for a but a coutabe umber of types the set BR x s a sgeto. For those types for whch t s ot a sgeto, BR x cotas precsey two eemets: t k ad t k for some k. The correspodece that assgs to type x the set BR x s weaky creasg (.e., for ay x 0 <x 00,ft 0 BR x 0 ad t 00 BR x 00, thet 0 t 00 )ad upper hem-cotuous. 12
13 Let br (x) =mbr x,adotethatbr s creasg ad eft cotuous, ad s ot rght cotuous precsey at types x for whch BR x s ot a sgeto. We the have the foowg coroary of Lemmas 8, 9, ad 10, whch s a modfcato of Coroary 1. Coroary 2 For every ε>0, there s a N such that for N at east a fracto 1 ε of of payers bd (br (x ) ε, br (x )+ε) wth probabty at east 1 ε. To prove part (b) of the theorem t remas to show that T br s the assortatve aocato. Ths s doe by the foowg emma, whch s a modfcatooflemma 5that accommodates the dscotutes T ad br. Lemma 11 G 1 (F (x)) = T (br (x)) for ay type x>0. To compete the proof, t remas to show (a) the statemet of the theorem. To do so, we use the foowg resut. Lemma 12 For every ε, δ > 0, theresan such that for N, eachtypex from a set whose F -measure s at east 1 ε bds at east wth probabty 1 ε a t, adobtasateast wth probabty 1 ε a y such that for some r BR x, t r <δad y T (r) <δ. To see that Lemma 12 mpes (a) the statemet of the theorem, choose some ε>0. Lemma 12 shows that for every δ>0, theresan such that for N ad for a fracto 1 ε of payers, thef -measure of ther types x that satsfy the codto of Lemma 12 s at east (1 ε). Ths meas that each such payer obtas wth probabty at east 1 ε aprzey that dffers by at most δ from the prze T (r) for some optma bd r of the payer s type. For types x > 0 such that br (x ) s a uque optma bd, ths yeds (a) by Lemma 11. However, by Lemma 9 ad strct sge crossg, there s oy a coutabe umber of other types x.adthef-measure of such types s 0 sce F has o atoms, so the F -measure of such types s arbtrary sma for suffcety arge. 2.1 Proof of Lemma 6 Let m T (q m ) = y 0 ad m T (r m ) = y 00. Both mts y 0 ad y 00 exst ad y 0 y 00 by mootocty. Suppose that y 0 <y 00. If G assgs a postve measure to (y 0,y 00 ),thet assgs a postve measure to ay terva wth edpots suffcety cose to y 0 ad y 00.I 13
14 such a case, we obta a cotradcto by argumets smar to those used the proof of Lemma 1. Ideed, for suffcety arge o bdder woud bd sghty beow t, because bddg sghty above t woud amost certay gve a better prze. Thus, G assgs measure zero to (y 0,y 00 ). Thsmpesthat(y 0,y 00 ) (yk,yu k ) for some k. By defto, T takes vaues [0,yk ] [yu k, 1], soy0 = yk ad y00 = yk u. Moreover, the mootocty of T mpes that k s the same for ay sequeces q m t ad r m t. Itremas to show that m A (q m )=G(yk ) ad m A (rm )=G(yk u). For ths, ote that f m A (q m ) >G(yk ),them T (qm ) >yk. Smary, f m A (rm ) > G(yk u),them T (rm ) >yk u. The equates m A (qm ) <G(yk ) ad m A (rm ) <G(yk u) ca be rued out smary f G does ot have atoms at yk or yu k. Suppose that G has a atom at yk u ad m A (rm ) <G(yk u). Sce m T (rm )=yk u, A (rm ) >G(yk ) for suffcety arge m. Taketwoumbersr m such that G(yk ) <A(rm ) <G(yk u);deotethembyt0 <t 00. The, for suffcety arge ay payer obtas a prze cose to yk u wth arbtrary hgh probabty by bddg ay t [t 0,t 00 ].Thus,forsuffcety arge, opayerwoudbdtheterva [(t 0 + t 00 )/2,t 00 ] wth postve probabty. Ths cotradcts the No-Gap Property. Suppose that G has a atom at yk ad m A (qm ) <G(yk ). The, for suffcety arge bddg q m amost certay gves a prze at most sghty better tha yk. I cotrast, bddg r m amost certay gves a prze at east as good as yk u. Ths foows drecty from (1) f G has a atom at yk u.ifgdoes ot, the ths aga foows from (1) for arge eough, because A (r m ) >G(yk u ) for ay m. For arge eough a cotradcto wth the No-Gap Property s obtaed smary to the ast part of the proof of Lemma 1 that deas wth U (0,y,t) strcty creasg y. 2.2 Proof of Lemma 7 The proof s aaogous to the proof of Lemma 2. Take a K such that the egths of I K+1, I K+2,... sum up to ess tha β/2. Takeayγ>0, ad suppose to the cotrary that there s a creasg sequece of tegers 1, 2,..., m,...such that for every m theressomebd t m / O γ wth T m (t m ) T (t m ) β. Passg to a subsequece f ecessary, we assume that the sequece t m t. Takeq 0 ad q 00 such that q 0 <t<q 00 ad T (q 00 ) T (q 0 ) <β/2, 12 ad K[ [q 0,q 00 ] B [t k,t k ]. Ths s possbe, sce the egths of I K+1, I K+2,... sum up to ess tha β/2. I addto, for arge eough k we have that T k (q 0 ) T (q 0 ) <β/2 ad T k (q 00 ) T (q 00 ) <β/2, sce 12 If t =0,takeq 0 =0. k=1 14
15 the egth of each I K+1, I K+2,... s ess tha β/2. The rest of the proof cocdes wth the proofoflemma Proof of Lemma 8 Frst, observe that the maxma payoff of type x, attaedataybdbr x, s st cotuous x. Ideed, upper sem-cotuty of T s a that s eeded for the cotuty of the maxma payoff. Ths observato mpes that there exsts a δ>0 such that for ay type x ay bd the compemet of BR x (ε) gves type x apayoff ower by at east δ tha ay bd BR x. For (), suppose the cotrary that for ay γ>0there are arbtrary arge such that themassexpeded(t k γ,t k + γ) by payers wth types x for whch t k / BR x (ε) s at east ε/3l. Take γ sma eough so that the payoff that such payers obta by bddg sghty more tha ay bd BR x s hgher by δ/2 tha the payoff that they woud obta by bddg t k γ ad gettg y k.13 Suppose frst that t k > 0. By mootocty of A ad the defto of t k,wehavethat A t k γ <G yk. Take a postve α<ε/6lsuch that A t k γ <G yk α. For ay t t k γ ad suffcety arge, fa (t) <G yk α/2, theopayeroftypex such that t k / BR x (ε) bds t, because by bddg t such a payer woud obta wth hgh probabty a prze o hgher tha yk, ad therefore woud obta a hgher payoff by bddg sghty more tha ay bd BR x. Let γ be defed by t k γ =f t : A (t) G ª yk α.scea t k γ <G yk α ad A (t) s rght-cotuous, we have that γ <γ(for suffcety arge ). Ad sce for every t<t k γ we have A (t) <G yk α/2 (by defto of γ ), payers wth types x for whch t k / BR x (ε) must exped the mass of at east ε/3l [t k γ,t k + γ). If more tha haf of ths mass s expeded t k γ,t k + γ, the we have that A t k + γ >A t k γ + ε/6l G y k α + ε/6l >G y k. Ths caot happe 13 To see why bddg sghty above ay t BR x gves at east a payoff cose to U (x, T (t),t), cosder the foowg two cases: (a) G 1 (A(r)) >T (t) for a r>t; ths case, sce m (G ) 1 (A(r)) G 1 (A(r)), for ay r>t,f s suffcety arge, the (G ) 1 (A(r)) >T (t). Ths mpes that a payer obtas a prze hgher tha T (t) wth arbtrary hgh probabty by bddg r. (b) G 1 (A(r)) = T (t) for r>tcose eough to t; thscase,t (r) =T (t) for a such r. Ths mpes that t = t k for some 0 k0. The cam ow foows from eft-cotuty of G 1 ad the fact that m (G ) 1 (q) G 1 (q) for ay q. 15
16 for suffcety arge, because for t t k,t k s A(t) =G y k. Thus, the payers wth types x for whch t k / BR x (ε) bd precsey t k γ wth probabty at east ε/6l. Sce these payers te wth each other at t k γ, by bddg t k γ they must obta a prze of aspecfc typey wth probabty 1, eve f they ose a tes at t k γ. (Otherwse, each of them coud obta a hgher payoff by bddg sghty above t k γ + γ ad wg the tes at t k γ.) But a payer who oses a tes at t k γ has rak order o hgher tha G yk α, bydefto of γ,soy yk. Therefore, such a payer woud obta a strcty hgher payoff by bddg sghty more tha ay bd BR x. Now suppose that t k =0. The A(t k ) G yk. The case A(t k ) <G yk s haded as the case t k > 0 above. Suppose that A(t k )=G yk. The, for ay γ>0such that t k + γ<t k,forsuffcety arge themassexpeded t k,t k + γ by a payers s smaer tha ε/6l, becausea (t) =G yk for ay t t k + γ,t k. Thus, f () does ot hod, for suffcety arge the mass expeded precsey at t k by the payers wth types x for whch t k / BR x (ε) s at east ε/6l, ad so the rakg of a payer who tes at t k ad oses s at most G yk ε/12l. But ths case each payer of type x for whch t k / BR x (ε) woud strcty prefer bddg sghty more tha ay bd BR x to bddg t k, a cotradcto. Toshow(),otethatft k = t k for some 0 k0, the () foows from (). Thus, suppose that t k 6= t k for ay 0 k0. Suppose the cotrary that for ay γ>0there s a arbtrary arge such that the mass expeded (t k γ,t k ] by payers wth types x for whch t k / BR x (ε) s at east ε/3l. Takeγ sma eough so that the payoff that such payers obta by bddg sghty more tha ay bd BR x s hgher by δ/2 tha the payoff that they woud obta by bddg t k γ ad gettg yk u. Observe that for suffcety arge, by bddg t k ay payer amost certay obtas a prze at most sghty better tha T (t k )=yk u.thssso, because t k 6= t k ad so A(t k) 6= G(y 0 k 0) for ay k0. Therefore, for arge eough apayer wth type x for whch t k / BR x (ε) woud be better off bddg sghty above ay t BR x tha bddg (t k γ,t k ]. Part () foows mmedatey from the fact that the vaue of A o t k,t k s G y k,by the defto of t k. 2.4 Proof of Lemma 9 Take a δ>0 such that for ay type x ay bd the compemet of BR x (ε) gves type x a payoff ower by at east δ tha ay bd BR x.takeβ>0such that for ay type x, bdt, ad przes y 0 ad y 00 wth y 0 y 00 β we have U (x, y 0,t) U (x, y 00,t) δ 3. 16
17 Next, take a K guarateed by Lemma 7 for ths β. I addto, take K arge eough so the egths of I K+1, I K+2,... sum up to ess tha β/2. Fay, for ay λ>0 take a N λ that satsfes the defto of uform covergece up to β o the compemet of O λ. (Note that K s the same for a λ.) Suppose to the cotrary of the statemet of the emma that there s a γ > 0 ad a subsequece of cotests such that a type x of payer the -th cotest has a best respose t to the strateges of the other payers that does ot beog to BR x (ε) [ K k=1 (t k γ,t k). As usua, we assume that the subsequece s the etre sequece; moreover, we assume that x x ad t t. Cosder the foowg two cases: A. (t 6= t k for ay k =1,..., K) I ths case, for some λ>0 there s a eghborhood of t that s dsjot from O λ.byuformcovergeceoft to T up to β o the compemet of O λ, U (x,t (t ),t ) U (x,t (t ),t ) δ 3 for N λ. Ad because t / BR x (ε), for ay t BR x we have U (x,t (t),t) U (x,t (t ),t ) δ. Thus, we obta U (x,t (t),t) U (x,t (t ),t ) 2δ 3. Observe that ay bd t 0 hgher tha t guaratees, for suffcety arge, aprzeotmuch worse tha T (t) wth arbtrary hgh probabty. 14 We w ow show that by bddg t,forsuffcety hgh type x obtas wth arbtrary hgh probabty a prze o better tha T (t )+β. Ideed, sce t does ot beog to t k,t k for ay k K, wehavethata (t 0 ) s bouded away from G y1,...,g y K for t 0 suffcety cose to t. Therefore, A (t ) s aso bouded away from G y1,...,g y K for suffcety arge. Ad for suffcety arge, bddg t gves wth arbtrary hgh probabty a rak order arbtrary cose to A (t ). Sce the egths of I K+1, I K+2,... sum up to ess tha β/2, adforayr other tha G y1,...,g y K ad suffcety arge the dfferece betwee (G ) 1 (r) ad G 1 (r) s o arger tha the egth of I K+1,bybddgt a payer obtas wth arbtrary hgh probabty a prze o better tha (G ) 1 (A (t ))+β. Therefore, by defto of β, wehavethatbybddgt type x obtas a payoff that s hgher tha U (x,t (t ),t ) by at most sghty more tha δ/3. Cosequety, for suffcety arge payer woudobtabybddgsomet 0 >tapayoff strcty hgher tha by bddg t, a cotradcto. 14 Toseewhy,seetheprevousfootote. 17
18 B. (t = t k for some k =1,..., K) The, cosder a t sghty hgher tha t,suchthat t does ot beog to [t k,t k] for k =1,..., K, ad such that: () for suffcety arge the payoff (of ay payer) the -th cotest of bddg t s ot much ower tha the payoff of bddg t ; () for suffcety arge, wehavethatthedfferece betwee U(x,T (t),t) for ay t BR x ad U(x,T (t ),t ) s ot much ower tha δ. Ths atter codto s possbe because, by defto, (x,t ) / BR(ε), ad by rght cotuty of T at t.now, usg (), appy a argumet aaogous to that from case A wth t payg the roe of t, wth a cotradcto obtaed by referrg to (). 2.5 Proof of Lemma 10 Mootocty of the correspodece foows from strct sge crossg, ad upper hemcotuty foows from stadard argumets. 15 Suppose that BR x cotas a par of bds t 1 <t 2. Beow we w show that for ay ε>0 ad ay terva [a, b] such that t 1 <aad b<t 2,forsuffcety arge the mass expeded [a, b] by a payers s at most ε. Ths mpes that the fucto A, adtherefore T, s costat o every such terva [a, b], adthereforeo(t 1,t 2 ). But T (t 2 ) >T (t 1 ) because t 1 <t 2 are BR x,sobydefto of the dscotuty pots t k of T we must have (t 1,t 2 ) t k,t k for some k. Ad because BRx B\ [ k=1 (t k,t k), wehavethatt 1 = t k ad t 2 = t k. It remas to show that for ay ε>0,forsuffcety arge the mass expeded [a, b] by a payers s at most ε. We w show ths for ε/2 ad payers of types ower tha x (a smar argumet appes to types hgher tha x). Choose x 0 <xsuch that F (x) F (x 0 ) <ε/3. For suffcety sma λ>0, sup z x 0BR z (λ) <a. (Ths s because x 0 <xad t 1 BR x, so every bd BR z s at most t 1 <a.) Therefore, by Lemma 9, there s some K such that for every γ>0ad suffcety arge ay bd [a, b] made by a payer of type z x 0 the -th cotest s [ K k=1 (t k γ,t k). Cosder oe of these K tervas for whch (t k γ,t k) [a, b] 6=. Scesup z x 0BR z (λ) < a t k, t k s ot BR z (λ) for ay z x 0.Ift k > sup z x 0BR z (λ), theby()oflemma8 there exsts a γ such that for suffcety arge the mass expeded t k γ,t k by payers of type z x 0 s ess tha ε/6k. Ift k sup z x 0BR z (λ), the by () ad () of Lemma 8, for suffcety arge themassexpeded[a, t k ) by payers of type z x 0 s ess tha ε/6k. 15 More precsey, ths foows from the fact that BR x s the set of a t such that (t, T (t)) maxmzes type x s utty over the cosure of the graph of T, whch s a compact set. 18
19 Therefore, for arge eough the mass expeded [a, b] by payers of type z x 0 s smaer tha ε/6, adbecausef (x) F (x 0 ) <ε/3, the mass expeded [a, b] by payers of type z x s smaer tha ε/ Proof of Coroary 2 Choose ε > 0. Lemma 10 mpes that there s a fte umber of tervas of types wth tota F -mass ε/2, such that for every type x ot oe of these tervas, BR x (br (x) ε, br (x)+ε). 16 Cosder the F -mass 1 ε/2 of types x wth the ast property, ad et K betheoethestatemetoflemma9. The,byLemma9adLemma8 for L = K, forsuffcety arge, atmostaf -mass ε/2 of those types bd outsde of (br (x) ε, br (x)+ε). 2.7 Proof of Lemma 11 The proof s aaogous to that of Lemma 5. Cosder a arbtrary type x. Defe x m = m {z : br (x) BR z } ad x max =max{z : br (x) BR z }. By strct sge crossg, BR z has oy oe eemet br(z) =br(x) for a z (x m,x max );tmayhavetwoeemetsfor z = x m or x max, whch cases br(x) s the hgher oe ad the ower oe of the two, respectvey. The cam that G 1 (F x m )=G 1 (F (x max )) s obtaed by the same argumet as the proof of Lemma 5. The rest of the proof requres the foowg mor chages whe BR x m has two eemets (ad aaogous chages whe BR x max has two eemets): 1. Istead of x m,wecosderx m = m z : br x m ª BR z,adcomparethe equbrum bds of every payer wth type ower tha x m δ to br(x m ), ad the equbrum bds of every payer wth type hgher tha x m to br x m δ. Ths chage does ot affect the argumets, sce G 1 (F x m )=G 1 (F x m ). 2. It may ot be true that the equbrum bds of every payer wth type ower tha x m δ are ower tha br(x m ), or the equbrum bds of every payer wth type hgher tha x m are hgher tha br x m δ, because payers may bd [ K k=1 (t k γ,t k) BR(ε) (see Lemma 9). However, ths happes oy wth vashg probabty as grows arge, so the argumets are aga ot affected. 16 There s a K>0 such that P k>k tk t k <ε. For each k K such that BRxk = t k,t ª k for some type x k, cosder the terva of types [x k λ, x k + λ] [0, 1], whereλ s such that (cotuous) F creases by o more tha ε/2k o ay terva o arger tha 2λ. The sum of the F -massofthesetervasso arger tha ε/2, ad the sum of the jumps of br o the compemet of these tervas s smaer tha ε. 19
20 2.8 Proof of Lemma 12 Take ay λ>0. By Lemma 9, there s a arge K such that for ay γ>0, f s suffcety arge, the equbrum bd of every payer the -th cotest beogs wth probabty 1 to BR x K[ (λ) (t k γ,t k ). k=1 Assume that K s, addto, arge eough so that the egths of I K+1, I K+2,... sum up to ess tha δ/2. We frst cam that for ay t/ (t k γ,t k) for a k =1,..., K, there exsts a N t such that for every N t,apayerwhobdst the -th cotest obtas (wth hgh probabty) a prze y such that y T (t) <δ/2. We w aso show that there exsts a N = N t that s commo for a such bds t. Suppose frst that t 6= t k for ay k =1,..., K. Sce A (t) dffers from G(yk ) ad G(yu k ) for ay k =1,..., K, ay rak order cose to A (t) aso dffers from G(yk ) ad G(yu k ).By(1), for suffcety arge, apayerwhobdst has (wth hgh probabty) a rak order cose to A (t); partcuar, ths rak order dffers from G(yk ) ad G(yu k ). By the assumpto that the egths of I K+1, I K+2,... sum up to ess tha δ/2, ths mpes that the dfferece betwee T (t) ad the prze obtaed by a payer who bds t s ower tha δ/2 (wth hgh probabty). Suppose that t = t k for some k =1,..., K. By a argumet aaogous to the oe used the prevous case, the prze obtaed by a payer who bds t caot, as creases, exceed T (t) by δ/2 wth a probabty that s bouded away from 0. AdT (t) caot exceed ths prze by δ/2 wth a probabty that s bouded away from 0 as creases, because the payer woud proftaby devate by bddg sghty above t, whch woud guaratee a prze o worse tha T (t) wth arbtrary hgh probabty. Now, ote that the umber N t that was chose for ay bd t has the requred property aso for a bds cose eough to t; the case of t = t k for some k =1,..., K, wemeabds cose eough ad hgher tha t. That s, for every t there s a eghborhood W t of that t wth N t that s commo for a bds from ths eghborhood. The famy of sets W t s a ope coverg of the compact set of bds t that satsfy t/ (t k γ,t k) for k =1,..., K. Thus, t cotas a fte subcoverg, ad ay umber N that exceeds umbers N t for a eemets of ths fte subcoverg has the requred property. Ths yeds the emma for bds t / (t k γ,t k) for a k =1,...,K. Ideed, ote that BR x s a sgeto, ad br(x) s ts oy eemet, for a except a coutabe umber of types x. Sce F has o atoms, the set of such types has F -measure 1. Ad for such types x, equbrum bds t/ (t k γ,t k) for a k =1,..., K beog to (br (x) λ, br (x)+λ). If λ 20
21 s suffcety sma, ad x s bouded away from 0, the t r <δfor r = br (x), ad T (t) T (r) δ/2. 17 Ad f λ s suffcety sma, the aso T (r) T (t) δ/2, because t/ (t k γ,t k) for a k =1,...,K ad the egths of I K+1, I K+2,... sum up to ess tha δ/2. Fay, by our frst cam, the prze y obtaed by bddg t must satsfy y T (t) <δ/2, so y T (r) <δ. Now cosder bds t such that t s (t k γ,t k) for some k =1,..., K. By () of Lemma 8, we ca dsregard bds t [t k + γ,t k γ]. Suppose that t s (t k γ,t k ) ad t k 6= t k for 0 a other k 0 =1,..., K. By () of Lemma 8, oe ca assume that t k BR x. 18 We w show that for suffcety sma γ ad for suffcety arge, payer obtas by bddg t (wth arbtrary hgh probabty) a prze y (T (t k ) δ, T (t k )+δ). Frst, ote that payer caot obta by bddg t aprzeowerthat (t k ) δ (wth probabty bouded away from 0), because for sma eough γ t woud be proftabe to devate to bddg sghty above t k, ad obta a prze ot much ower tha T (t k ) wth hgh probabty. Payer caot obta by bddg t a prze hgher tha T (t k )+δ (wth probabty bouded away from 0), because by (1), for ay r>t k ad suffcety arge the rak order of payer s wth arbtrary hgh probabty bouded above by A(r). Thus, the upper boud o the prze foows from the assumpto that t k 6= t k for a other 0 k0 =1,..., K, adtheegthsof I K+1,I K+2,... sum up to ess tha δ/2. Fay, suppose that t s t k γ,t k + γ for some k =1,..., K. By()ofLemma8,oe ca assume that t k BR x. We w show that for suffcety sma γ ad for suffcety arge, equbrum bddg t k γ,t k + γ eads (wth arbtrary hgh probabty) to a prze y (T (t k ) δ, T (t k )+δ), except a sma probabty evet. Ideed, by a argumet smar to that from the prevous case, such a bd caot ead to a prze ower tha T (t k ) δ (wth probabty bouded away from 0). To obta a prze hgher tha T (t k )+δ wth a ovashg probabty, a payer s expected rak order whe bddg t caot be ower tha G yk by a ovashg costat. But, f a ovashg fracto of payers w a prze hgher tha T (t k )+δ wth a ovashg probabty by bddg t k γ,t k + γ, the the crease expected rak order o the terva t k γ,t k + γ s bouded away 17 Ideed, for types bouded away from 0, adforsuffcety sma λ, wehavethatu(x, y, t) >U(x, y 0,t 0 ) wheever y y 0 >δ/2 ad t t 0 <λ. (The assumpto that types are bouded away fro 0 s esseta, becauseweddotassumethatu(0,y,t) strcty creases y.) However, sce r = br (x), we caot have U(x, T (t),t) >U(x, T (r),r). 18 The emma says oy that the mass expeded (t k γ,t k ] by types x for whch t k / BR x (λ) for some sma λ>0 s sma. However, f λ>0 s suffcety sma, the the mass of types x such that t k / BR x but t k BR x (λ) s sma. 21
22 from 0 for a, whch cotradcts the fact that A (t k + γ) approaches G(y k ) as creases. 22
Lecture 7. Norms and Condition Numbers
Lecture 7 Norms ad Codto Numbers To dscuss the errors umerca probems vovg vectors, t s usefu to empo orms. Vector Norm O a vector space V, a orm s a fucto from V to the set of o-egatve reas that obes three
More informationAPPENDIX III THE ENVELOPE PROPERTY
Apped III APPENDIX III THE ENVELOPE PROPERTY Optmzato mposes a very strog structure o the problem cosdered Ths s the reaso why eoclasscal ecoomcs whch assumes optmzg behavour has bee the most successful
More informationOnline Appendix: Measured Aggregate Gains from International Trade
Ole Appedx: Measured Aggregate Gas from Iteratoal Trade Arel Burste UCLA ad NBER Javer Cravo Uversty of Mchga March 3, 2014 I ths ole appedx we derve addtoal results dscussed the paper. I the frst secto,
More informationANOVA Notes Page 1. Analysis of Variance for a One-Way Classification of Data
ANOVA Notes Page Aalss of Varace for a Oe-Wa Classfcato of Data Cosder a sgle factor or treatmet doe at levels (e, there are,, 3, dfferet varatos o the prescrbed treatmet) Wth a gve treatmet level there
More informationSTATISTICAL PROPERTIES OF LEAST SQUARES ESTIMATORS. x, where. = y - ˆ " 1
STATISTICAL PROPERTIES OF LEAST SQUARES ESTIMATORS Recall Assumpto E(Y x) η 0 + η x (lear codtoal mea fucto) Data (x, y ), (x 2, y 2 ),, (x, y ) Least squares estmator ˆ E (Y x) ˆ " 0 + ˆ " x, where ˆ
More informationPreprocess a planar map S. Given a query point p, report the face of S containing p. Goal: O(n)-size data structure that enables O(log n) query time.
Computatoal Geometry Chapter 6 Pot Locato 1 Problem Defto Preprocess a plaar map S. Gve a query pot p, report the face of S cotag p. S Goal: O()-sze data structure that eables O(log ) query tme. C p E
More informationAn Effectiveness of Integrated Portfolio in Bancassurance
A Effectveess of Itegrated Portfolo Bacassurace Taea Karya Research Ceter for Facal Egeerg Isttute of Ecoomc Research Kyoto versty Sayouu Kyoto 606-850 Japa arya@eryoto-uacp Itroducto As s well ow the
More informationSTOCHASTIC approximation algorithms have several
IEEE TRANSACTIONS ON INFORMATION THEORY, VOL 60, NO 10, OCTOBER 2014 6609 Trackg a Markov-Modulated Statoary Degree Dstrbuto of a Dyamc Radom Graph Mazyar Hamd, Vkram Krshamurthy, Fellow, IEEE, ad George
More informationSTOCK INVESTMENT MANAGEMENT UNDER UNCERTAINTY. Madalina Ecaterina ANDREICA 1 Marin ANDREICA 2
"AROACHES IN ORGANISATIONA MANAGEMENT" 15-16 Noveber 01, BCHAREST, ROMANIA STOCK INVESTMENT MANAGEMENT NDER NCERTAINTY Madaa Ecatera ANDREICA 1 Mar ANDREICA ABSTRACT Ths paper presets a stock vestet aageet
More informationInnovation and Production in the Global Economy Online Appendix
Iovato ad Producto te Goba Ecoomy Oe Appedx Costas Aroas Prceto, Yae ad NBER Nataa Ramodo Arzoa State Adrés Rodríguez-Care UC Bereey ad NBER Stepe Yeape Pe State ad NBER December 204 Abstract I ts oe Appedx
More informationT = 1/freq, T = 2/freq, T = i/freq, T = n (number of cash flows = freq n) are :
Bullets bods Let s descrbe frst a fxed rate bod wthout amortzg a more geeral way : Let s ote : C the aual fxed rate t s a percetage N the otoal freq ( 2 4 ) the umber of coupo per year R the redempto of
More informationNumerical Methods with MS Excel
TMME, vol4, o.1, p.84 Numercal Methods wth MS Excel M. El-Gebely & B. Yushau 1 Departmet of Mathematcal Sceces Kg Fahd Uversty of Petroleum & Merals. Dhahra, Saud Araba. Abstract: I ths ote we show how
More informationThe simple linear Regression Model
The smple lear Regresso Model Correlato coeffcet s o-parametrc ad just dcates that two varables are assocated wth oe aother, but t does ot gve a deas of the kd of relatoshp. Regresso models help vestgatg
More informationIn nite Sequences. Dr. Philippe B. Laval Kennesaw State University. October 9, 2008
I ite Sequeces Dr. Philippe B. Laval Keesaw State Uiversity October 9, 2008 Abstract This had out is a itroductio to i ite sequeces. mai de itios ad presets some elemetary results. It gives the I ite Sequeces
More informationOn Savings Accounts in Semimartingale Term Structure Models
O Savgs Accouts Semmartgale Term Structure Models Frak Döberle Mart Schwezer moeyshelf.com Techsche Uverstät Berl Bockehemer Ladstraße 55 Fachberech Mathematk, MA 7 4 D 6325 Frakfurt am Ma Straße des 17.
More informationIDENTIFICATION OF THE DYNAMICS OF THE GOOGLE S RANKING ALGORITHM. A. Khaki Sedigh, Mehdi Roudaki
IDENIFICAION OF HE DYNAMICS OF HE GOOGLE S RANKING ALGORIHM A. Khak Sedgh, Mehd Roudak Cotrol Dvso, Departmet of Electrcal Egeerg, K.N.oos Uversty of echology P. O. Box: 16315-1355, ehra, Ira sedgh@eetd.ktu.ac.r,
More informationThe Gompertz-Makeham distribution. Fredrik Norström. Supervisor: Yuri Belyaev
The Gompertz-Makeham dstrbuto by Fredrk Norström Master s thess Mathematcal Statstcs, Umeå Uversty, 997 Supervsor: Yur Belyaev Abstract Ths work s about the Gompertz-Makeham dstrbuto. The dstrbuto has
More informationThe Digital Signature Scheme MQQ-SIG
The Dgtal Sgature Scheme MQQ-SIG Itellectual Property Statemet ad Techcal Descrpto Frst publshed: 10 October 2010, Last update: 20 December 2010 Dalo Glgorosk 1 ad Rue Stesmo Ødegård 2 ad Rue Erled Jese
More informationRelaxation Methods for Iterative Solution to Linear Systems of Equations
Relaxato Methods for Iteratve Soluto to Lear Systems of Equatos Gerald Recktewald Portlad State Uversty Mechacal Egeerg Departmet gerry@me.pdx.edu Prmary Topcs Basc Cocepts Statoary Methods a.k.a. Relaxato
More informationSAMPLE QUESTIONS FOR FINAL EXAM. (1) (2) (3) (4) Find the following using the definition of the Riemann integral: (2x + 1)dx
SAMPLE QUESTIONS FOR FINAL EXAM REAL ANALYSIS I FALL 006 3 4 Fid the followig usig the defiitio of the Riema itegral: a 0 x + dx 3 Cosider the partitio P x 0 3, x 3 +, x 3 +,......, x 3 3 + 3 of the iterval
More informationADAPTATION OF SHAPIRO-WILK TEST TO THE CASE OF KNOWN MEAN
Colloquum Bometrcum 4 ADAPTATION OF SHAPIRO-WILK TEST TO THE CASE OF KNOWN MEAN Zofa Hausz, Joaa Tarasńska Departmet of Appled Mathematcs ad Computer Scece Uversty of Lfe Sceces Lubl Akademcka 3, -95 Lubl
More information6.7 Network analysis. 6.7.1 Introduction. References - Network analysis. Topological analysis
6.7 Network aalyss Le data that explctly store topologcal formato are called etwork data. Besdes spatal operatos, several methods of spatal aalyss are applcable to etwork data. Fgure: Network data Refereces
More informationSecurity Analysis of RAPP: An RFID Authentication Protocol based on Permutation
Securty Aalyss of RAPP: A RFID Authetcato Protocol based o Permutato Wag Shao-hu,,, Ha Zhje,, Lu Sujua,, Che Da-we, {College of Computer, Najg Uversty of Posts ad Telecommucatos, Najg 004, Cha Jagsu Hgh
More informationBayesian Network Representation
Readgs: K&F 3., 3.2, 3.3, 3.4. Bayesa Network Represetato Lecture 2 Mar 30, 20 CSE 55, Statstcal Methods, Sprg 20 Istructor: Su-I Lee Uversty of Washgto, Seattle Last tme & today Last tme Probablty theory
More information1st International Symposium on Imprecise Probabilities and Their Applications, Ghent, Belgium, 29 June - 2 July 1999
1st Iteratoal Symposum o Imprecse robabltes ad Ther Applcatos, Ghet, Belgum, 29 Jue - 2 July 1999 Sharg Belefs: Betwee Agreeg ad Dsagreeg Atoe Bllot æ, Ala Chateaueuf y, Itzhak Glboa z, Jea-Marc Tallo
More informationChapter Eight. f : R R
Chapter Eght f : R R 8. Itroducto We shall ow tur our atteto to the very mportat specal case of fuctos that are real, or scalar, valued. These are sometmes called scalar felds. I the very, but mportat,
More informationA New Bayesian Network Method for Computing Bottom Event's Structural Importance Degree using Jointree
, pp.277-288 http://dx.do.org/10.14257/juesst.2015.8.1.25 A New Bayesa Network Method for Computg Bottom Evet's Structural Importace Degree usg Jotree Wag Yao ad Su Q School of Aeroautcs, Northwester Polytechcal
More information1. The Time Value of Money
Corporate Face [00-0345]. The Tme Value of Moey. Compoudg ad Dscoutg Captalzato (compoudg, fdg future values) s a process of movg a value forward tme. It yelds the future value gve the relevat compoudg
More informationModels for Selecting an ERP System with Intuitionistic Trapezoidal Fuzzy Information
JOURNAL OF SOFWARE, VOL 5, NO 3, MARCH 00 75 Models for Selectg a ERP System wth Itutostc rapezodal Fuzzy Iformato Guwu We, Ru L Departmet of Ecoomcs ad Maagemet, Chogqg Uversty of Arts ad Sceces, Yogchua,
More informationCommon p-belief: The General Case
GAMES AND ECONOMIC BEHAVIOR 8, 738 997 ARTICLE NO. GA97053 Commo p-belef: The Geeral Case Atsush Kaj* ad Stephe Morrs Departmet of Ecoomcs, Uersty of Pesylaa Receved February, 995 We develop belef operators
More informationOptimal multi-degree reduction of Bézier curves with constraints of endpoints continuity
Computer Aded Geometrc Desg 19 (2002 365 377 wwwelsevercom/locate/comad Optmal mult-degree reducto of Bézer curves wth costrats of edpots cotuty Guo-Dog Che, Guo-J Wag State Key Laboratory of CAD&CG, Isttute
More informationSequences and Series
Secto 9. Sequeces d Seres You c thk of sequece s fucto whose dom s the set of postve tegers. f ( ), f (), f (),... f ( ),... Defto of Sequece A fte sequece s fucto whose dom s the set of postve tegers.
More informationSHAPIRO-WILK TEST FOR NORMALITY WITH KNOWN MEAN
SHAPIRO-WILK TEST FOR NORMALITY WITH KNOWN MEAN Wojcech Zelńsk Departmet of Ecoometrcs ad Statstcs Warsaw Uversty of Lfe Sceces Nowoursyowska 66, -787 Warszawa e-mal: wojtekzelsk@statystykafo Zofa Hausz,
More informationRandomized Load Balancing by Joining and Splitting Bins
Radomzed Load Baacg by Jog ad Spttg Bs James Aspes Ytog Y 1 Itoducto Cosde the foowg oad baacg sceao: a ceta amout of wo oad s dstbuted amog a set of maches that may chage ove tme as maches o ad eave the
More informationECONOMIC CHOICE OF OPTIMUM FEEDER CABLE CONSIDERING RISK ANALYSIS. University of Brasilia (UnB) and The Brazilian Regulatory Agency (ANEEL), Brazil
ECONOMIC CHOICE OF OPTIMUM FEEDER CABE CONSIDERING RISK ANAYSIS I Camargo, F Fgueredo, M De Olvera Uversty of Brasla (UB) ad The Brazla Regulatory Agecy (ANEE), Brazl The choce of the approprate cable
More informationStatistical Pattern Recognition (CE-725) Department of Computer Engineering Sharif University of Technology
I The Name of God, The Compassoate, The ercful Name: Problems' eys Studet ID#:. Statstcal Patter Recogto (CE-725) Departmet of Computer Egeerg Sharf Uversty of Techology Fal Exam Soluto - Sprg 202 (50
More informationAbraham Zaks. Technion I.I.T. Haifa ISRAEL. and. University of Haifa, Haifa ISRAEL. Abstract
Preset Value of Autes Uder Radom Rates of Iterest By Abraham Zas Techo I.I.T. Hafa ISRAEL ad Uversty of Hafa, Hafa ISRAEL Abstract Some attempts were made to evaluate the future value (FV) of the expected
More informationLecture 4: Cauchy sequences, Bolzano-Weierstrass, and the Squeeze theorem
Lecture 4: Cauchy sequeces, Bolzao-Weierstrass, ad the Squeeze theorem The purpose of this lecture is more modest tha the previous oes. It is to state certai coditios uder which we are guarateed that limits
More informationFinite Difference Method
Fte Dfferece Method MEL 87 Computatoa Heat rasfer --4) Dr. Praba audar Assstat Professor Departmet of Mechaca Egeerg II Deh Dscretzato Methods Requred to covert the geera trasport equato to set of agebrac
More information1. C. The formula for the confidence interval for a population mean is: x t, which was
s 1. C. The formula for the cofidece iterval for a populatio mea is: x t, which was based o the sample Mea. So, x is guarateed to be i the iterval you form.. D. Use the rule : p-value
More informationThe Sample Complexity of Exploration in the Multi-Armed Bandit Problem
Joura of Machie Learig Research 5 004) 63-648 Submitted 1/04; Pubished 6/04 The Sampe Compexity of Exporatio i the Muti-Armed Badit Probem Shie Maor Joh N. Tsitsikis Laboratory for Iformatio ad Decisio
More informationOn Application-level Load Balancing in FastReplica
O Appcato-eve Load Baacg FastRepca Jagwo Lee, Gustavo de Vecaa Abstract I the paper, we cosder the probem o dstrbutg arge-sze cotet to a xed set o odes. I cotrast wth the most exstg ed-system soutos to
More informationClassic Problems at a Glance using the TVM Solver
C H A P T E R 2 Classc Problems at a Glace usg the TVM Solver The table below llustrates the most commo types of classc face problems. The formulas are gve for each calculato. A bref troducto to usg the
More informationn. We know that the sum of squares of p independent standard normal variables has a chi square distribution with p degrees of freedom.
UMEÅ UNIVERSITET Matematsk-statstska sttutoe Multvarat dataaalys för tekologer MSTB0 PA TENTAMEN 004-0-9 LÖSNINGSFÖRSLAG TILL TENTAMEN I MATEMATISK STATISTIK Multvarat dataaalys för tekologer B, 5 poäg.
More informationProperties of MLE: consistency, asymptotic normality. Fisher information.
Lecture 3 Properties of MLE: cosistecy, asymptotic ormality. Fisher iformatio. I this sectio we will try to uderstad why MLEs are good. Let us recall two facts from probability that we be used ofte throughout
More informationCurve Fitting and Solution of Equation
UNIT V Curve Fttg ad Soluto of Equato 5. CURVE FITTING I ma braches of appled mathematcs ad egeerg sceces we come across epermets ad problems, whch volve two varables. For eample, t s kow that the speed
More informationA Study of Unrelated Parallel-Machine Scheduling with Deteriorating Maintenance Activities to Minimize the Total Completion Time
Joural of Na Ka, Vol. 0, No., pp.5-9 (20) 5 A Study of Urelated Parallel-Mache Schedulg wth Deteroratg Mateace Actvtes to Mze the Total Copleto Te Suh-Jeq Yag, Ja-Yuar Guo, Hs-Tao Lee Departet of Idustral
More informationFinito: A Faster, Permutable Incremental Gradient Method for Big Data Problems
Fto: A Faster, Permutable Icremetal Gradet Method for Bg Data Problems Aaro J Defazo Tbéro S Caetao Just Domke NICTA ad Australa Natoal Uversty AARONDEFAZIO@ANUEDUAU TIBERIOCAETANO@NICTACOMAU JUSTINDOMKE@NICTACOMAU
More informationHow To Value An Annuity
Future Value of a Auty After payg all your blls, you have $200 left each payday (at the ed of each moth) that you wll put to savgs order to save up a dow paymet for a house. If you vest ths moey at 5%
More informationAn SVR-Based Data Farming Technique for Web Application
A SVR-Based Data Farmg Techque for Web Appcato Ja L 1 ad Mjg Peg 2 1 Schoo of Ecoomcs ad Maagemet, Behag Uversty 100083 Bejg, P.R. Cha Ja@wyu.c 2 Isttute of Systems Scece ad Techoogy, Wuy Uversty, Jagme
More informationChapter 3. AMORTIZATION OF LOAN. SINKING FUNDS R =
Chapter 3. AMORTIZATION OF LOAN. SINKING FUNDS Objectves of the Topc: Beg able to formalse ad solve practcal ad mathematcal problems, whch the subjects of loa amortsato ad maagemet of cumulatve fuds are
More informationA Covariance Analysis Model for DDoS Attack Detection*
A Covarace Aayss Mode or DDoS Attac Detecto* Shuyua J Deartmet o Comutg HogKog Poytechc Uversty HogKog Cha cssy@com.oyu.edu.h Dae S. Yeug Deartmet o Comutg HogKog Poytechc Uversty HogKog Cha csdae@et.oyu.edu.h
More informationBertrand Edgeworth Competition in Differentiated Markets: Hotelling revisited
Bertrad Edgeworth Competto Dfferetated Markets: Hotellg revsted Ncolas Boccard * & Xaver Wauthy ** March 000 A BSTRACT Ths paper deals wth stuatos where frms commt to capactes ad compete prces the market
More informationCompressive Sensing over Strongly Connected Digraph and Its Application in Traffic Monitoring
Compressve Sesg over Strogly Coected Dgraph ad Its Applcato Traffc Motorg Xao Q, Yogca Wag, Yuexua Wag, Lwe Xu Isttute for Iterdscplary Iformato Sceces, Tsghua Uversty, Bejg, Cha {qxao3, kyo.c}@gmal.com,
More informationThe Analysis of Development of Insurance Contract Premiums of General Liability Insurance in the Business Insurance Risk
The Aalyss of Developmet of Isurace Cotract Premums of Geeral Lablty Isurace the Busess Isurace Rsk the Frame of the Czech Isurace Market 1998 011 Scetfc Coferece Jue, 10. - 14. 013 Pavla Kubová Departmet
More information4.3. The Integral and Comparison Tests
4.3. THE INTEGRAL AND COMPARISON TESTS 9 4.3. The Itegral ad Compariso Tests 4.3.. The Itegral Test. Suppose f is a cotiuous, positive, decreasig fuctio o [, ), ad let a = f(). The the covergece or divergece
More informationRQM: A new rate-based active queue management algorithm
: A ew rate-based actve queue maagemet algorthm Jeff Edmods, Suprakash Datta, Patrck Dymod, Kashf Al Computer Scece ad Egeerg Departmet, York Uversty, Toroto, Caada Abstract I ths paper, we propose a ew
More information10.5 Future Value and Present Value of a General Annuity Due
Chapter 10 Autes 371 5. Thomas leases a car worth $4,000 at.99% compouded mothly. He agrees to make 36 lease paymets of $330 each at the begg of every moth. What s the buyout prce (resdual value of the
More informationOn formula to compute primes and the n th prime
Joural's Ttle, Vol., 00, o., - O formula to compute prmes ad the th prme Issam Kaddoura Lebaese Iteratoal Uversty Faculty of Arts ad ceces, Lebao Emal: ssam.addoura@lu.edu.lb amh Abdul-Nab Lebaese Iteratoal
More informationAsymptotic Growth of Functions
CMPS Itroductio to Aalysis of Algorithms Fall 3 Asymptotic Growth of Fuctios We itroduce several types of asymptotic otatio which are used to compare the performace ad efficiecy of algorithms As we ll
More informationConvexity, Inequalities, and Norms
Covexity, Iequalities, ad Norms Covex Fuctios You are probably familiar with the otio of cocavity of fuctios. Give a twicedifferetiable fuctio ϕ: R R, We say that ϕ is covex (or cocave up) if ϕ (x) 0 for
More informationWe investigate a simple adaptive approach to optimizing seat protection levels in airline
Reveue Maagemet Wthout Forecastg or Optmzato: A Adaptve Algorthm for Determg Arle Seat Protecto Levels Garrett va Ryz Jeff McGll Graduate School of Busess, Columba Uversty, New York, New York 10027 School
More informationAverage Price Ratios
Average Prce Ratos Morgstar Methodology Paper August 3, 2005 2005 Morgstar, Ic. All rghts reserved. The formato ths documet s the property of Morgstar, Ic. Reproducto or trascrpto by ay meas, whole or
More informationInfinite Sequences and Series
CHAPTER 4 Ifiite Sequeces ad Series 4.1. Sequeces A sequece is a ifiite ordered list of umbers, for example the sequece of odd positive itegers: 1, 3, 5, 7, 9, 11, 13, 15, 17, 19, 21, 23, 25, 27, 29...
More informationChapter 6: Variance, the law of large numbers and the Monte-Carlo method
Chapter 6: Variace, the law of large umbers ad the Mote-Carlo method Expected value, variace, ad Chebyshev iequality. If X is a radom variable recall that the expected value of X, E[X] is the average value
More informationSection 11.3: The Integral Test
Sectio.3: The Itegral Test Most of the series we have looked at have either diverged or have coverged ad we have bee able to fid what they coverge to. I geeral however, the problem is much more difficult
More informationConstrained Cubic Spline Interpolation for Chemical Engineering Applications
Costraed Cubc Sple Iterpolato or Chemcal Egeerg Applcatos b CJC Kruger Summar Cubc sple terpolato s a useul techque to terpolate betwee kow data pots due to ts stable ad smooth characterstcs. Uortuatel
More informationApplication of GA with SVM for Stock Price Prediction in Financial Market
Iteratoa Joura of Scece ad Research (IJSR) ISSN (Oe): 39-7064 Impact Factor (0): 3.358 Appcato of GA wth SVM for Stock Prce Predcto Faca Market Om Prakash Jea, Dr. Sudarsa Padhy Departmet of Computer Scece
More informationApproximation Algorithms for Scheduling with Rejection on Two Unrelated Parallel Machines
(ICS) Iteratoal oural of dvaced Comuter Scece ad lcatos Vol 6 No 05 romato lgorthms for Schedulg wth eecto o wo Urelated Parallel aches Feg Xahao Zhag Zega Ca College of Scece y Uversty y Shadog Cha 76005
More informationRUSSIAN ROULETTE AND PARTICLE SPLITTING
RUSSAN ROULETTE AND PARTCLE SPLTTNG M. Ragheb 3/7/203 NTRODUCTON To stuatos are ecoutered partcle trasport smulatos:. a multplyg medum, a partcle such as a eutro a cosmc ray partcle or a photo may geerate
More informationFractal-Structured Karatsuba`s Algorithm for Binary Field Multiplication: FK
Fractal-Structured Karatsuba`s Algorthm for Bary Feld Multplcato: FK *The authors are worg at the Isttute of Mathematcs The Academy of Sceces of DPR Korea. **Address : U Jog dstrct Kwahadog Number Pyogyag
More informationSettlement Prediction by Spatial-temporal Random Process
Safety, Relablty ad Rs of Structures, Ifrastructures ad Egeerg Systems Furuta, Fragopol & Shozua (eds Taylor & Fracs Group, Lodo, ISBN 978---77- Settlemet Predcto by Spatal-temporal Radom Process P. Rugbaapha
More informationMDM 4U PRACTICE EXAMINATION
MDM 4U RCTICE EXMINTION Ths s a ractce eam. It does ot cover all the materal ths course ad should ot be the oly revew that you do rearato for your fal eam. Your eam may cota questos that do ot aear o ths
More informationBanking (Early Repayment of Housing Loans) Order, 5762 2002 1
akg (Early Repaymet of Housg Loas) Order, 5762 2002 y vrtue of the power vested me uder Secto 3 of the akg Ordace 94 (hereafter, the Ordace ), followg cosultato wth the Commttee, ad wth the approval of
More informationDepartment of Computer Science, University of Otago
Departmet of Computer Sciece, Uiversity of Otago Techical Report OUCS-2006-09 Permutatios Cotaiig May Patters Authors: M.H. Albert Departmet of Computer Sciece, Uiversity of Otago Micah Colema, Rya Fly
More informationOn Error Detection with Block Codes
BULGARIAN ACADEMY OF SCIENCES CYBERNETICS AND INFORMATION TECHNOLOGIES Volume 9, No 3 Sofa 2009 O Error Detecto wth Block Codes Rostza Doduekova Chalmers Uversty of Techology ad the Uversty of Gotheburg,
More informationThree Dimensional Interpolation of Video Signals
Three Dmesoal Iterpolato of Vdeo Sgals Elham Shahfard March 0 th 006 Outle A Bref reve of prevous tals Dgtal Iterpolato Bascs Upsamplg D Flter Desg Issues Ifte Impulse Respose Fte Impulse Respose Desged
More informationAutomated Event Registration System in Corporation
teratoal Joural of Advaces Computer Scece ad Techology JACST), Vol., No., Pages : 0-0 0) Specal ssue of CACST 0 - Held durg 09-0 May, 0 Malaysa Automated Evet Regstrato System Corporato Zafer Al-Makhadmee
More informationA multi-layer market for vehicle-to-grid energy trading in the smart grid
A mult-layer market for vehcle-to-grd eergy tradg the smart grd Albert Y.S. Lam, Logbo Huag, Aloso Slva, Wald Saad To cte ths verso: Albert Y.S. Lam, Logbo Huag, Aloso Slva, Wald Saad. A mult-layer market
More informationSimple Linear Regression
Smple Lear Regresso Regresso equato a equato that descrbes the average relatoshp betwee a respose (depedet) ad a eplaator (depedet) varable. 6 8 Slope-tercept equato for a le m b (,6) slope. (,) 6 6 8
More informationSpeeding up k-means Clustering by Bootstrap Averaging
Speedg up -meas Clusterg by Bootstrap Averagg Ia Davdso ad Ashw Satyaarayaa Computer Scece Dept, SUNY Albay, NY, USA,. {davdso, ashw}@cs.albay.edu Abstract K-meas clusterg s oe of the most popular clusterg
More informationOne way to organize workers that lies between traditional assembly lines, where workers are specialists,
MANUFACTURING & SERVICE OPERATIONS MANAGEMENT Vol. 7, No. 2, Sprg 2005, pp. 121 129 ss 1523-4614 ess 1526-5498 05 0702 0121 forms do 10.1287/msom.1040.0059 2005 INFORMS Usg Bucket Brgades to Mgrate from
More informationCapacitated Production Planning and Inventory Control when Demand is Unpredictable for Most Items: The No B/C Strategy
SCHOOL OF OPERATIONS RESEARCH AND INDUSTRIAL ENGINEERING COLLEGE OF ENGINEERING CORNELL UNIVERSITY ITHACA, NY 4853-380 TECHNICAL REPORT Jue 200 Capactated Producto Plag ad Ivetory Cotrol whe Demad s Upredctable
More informationBased on PSO cloud computing server points location searching
Iteratoa Worshop o Coud Coput ad Iforato Securty (CCIS 13) Based o PSO coud coput server pots ocato search Che Hua-sha Iforato Techooy Ceter Shaha Iteratoa Studes Uversty Shaha, cha, 1391641563 E-a: huasha@shsu.edu.c
More informationM. Salahi, F. Mehrdoust, F. Piri. CVaR Robust Mean-CVaR Portfolio Optimization
M. Salah, F. Mehrdoust, F. Pr Uversty of Gula, Rasht, Ira CVaR Robust Mea-CVaR Portfolo Optmzato Abstract: Oe of the most mportat problems faced by every vestor s asset allocato. A vestor durg makg vestmet
More informationof the relationship between time and the value of money.
TIME AND THE VALUE OF MONEY Most agrbusess maagers are famlar wth the terms compoudg, dscoutg, auty, ad captalzato. That s, most agrbusess maagers have a tutve uderstadg that each term mples some relatoshp
More informationChapter 3 0.06 = 3000 ( 1.015 ( 1 ) Present Value of an Annuity. Section 4 Present Value of an Annuity; Amortization
Chapter 3 Mathematcs of Face Secto 4 Preset Value of a Auty; Amortzato Preset Value of a Auty I ths secto, we wll address the problem of determg the amout that should be deposted to a accout ow at a gve
More informationOverview of some probability distributions.
Lecture Overview of some probability distributios. I this lecture we will review several commo distributios that will be used ofte throughtout the class. Each distributio is usually described by its probability
More informationThe analysis of annuities relies on the formula for geometric sums: r k = rn+1 1 r 1. (2.1) k=0
Chapter 2 Autes ad loas A auty s a sequece of paymets wth fxed frequecy. The term auty orgally referred to aual paymets (hece the ame), but t s ow also used for paymets wth ay frequecy. Autes appear may
More informationAn Approach to Evaluating the Computer Network Security with Hesitant Fuzzy Information
A Approach to Evaluatg the Computer Network Securty wth Hestat Fuzzy Iformato Jafeg Dog A Approach to Evaluatg the Computer Network Securty wth Hestat Fuzzy Iformato Jafeg Dog, Frst ad Correspodg Author
More informationEfficient Compensation for Regulatory Takings. and Oregon s Measure 37
Effcet Compesato for Regulatory Takgs ad Orego s Measure 37 Jack Scheffer Ph.D. Studet Dept. of Agrcultural, Evrometal ad Developmet Ecoomcs The Oho State Uversty 2120 Fyffe Road Columbus, OH 43210-1067
More informationWe present a new approach to pricing American-style derivatives that is applicable to any Markovian setting
MANAGEMENT SCIENCE Vol. 52, No., Jauary 26, pp. 95 ss 25-99 ess 526-55 6 52 95 forms do.287/msc.5.447 26 INFORMS Prcg Amerca-Style Dervatves wth Europea Call Optos Scott B. Laprse BAE Systems, Advaced
More informationMeasuring the Quality of Credit Scoring Models
Measur the Qualty of Credt cor Models Mart Řezáč Dept. of Matheatcs ad tatstcs, Faculty of cece, Masaryk Uversty CCC XI, Edurh Auust 009 Cotet. Itroducto 3. Good/ad clet defto 4 3. Measur the qualty 6
More informationCH. V ME256 STATICS Center of Gravity, Centroid, and Moment of Inertia CENTER OF GRAVITY AND CENTROID
CH. ME56 STTICS Ceter of Gravt, Cetrod, ad Momet of Ierta CENTE OF GITY ND CENTOID 5. CENTE OF GITY ND CENTE OF MSS FO SYSTEM OF PTICES Ceter of Gravt. The ceter of gravt G s a pot whch locates the resultat
More informationÉditeur Inria, Domaine de Voluceau, Rocquencourt, BP 105 LE CHESNAY Cedex (France) ISSN 0249-6399
Uté de recherche INRIA Lorrae, techopôle de Nacy-Brabos, 615 rue du jard botaque, BP 101, 54600 VILLERS-LÈS-NANCY Uté de recherche INRIA Rees, IRISA, Campus uverstare de Beauleu, 35042 RENNES Cedex Uté
More informationLoad Balancing Control for Parallel Systems
Proc IEEE Med Symposum o New drectos Cotrol ad Automato, Chaa (Grèce),994, pp66-73 Load Balacg Cotrol for Parallel Systems Jea-Claude Heet LAAS-CNRS, 7 aveue du Coloel Roche, 3077 Toulouse, Frace E-mal
More informationGeneralized solutions for the joint replenishment problem with correction factor
Geerzed soutos for the ot repeshet proe wth correcto fctor Astrct Erc Porrs, Roert Deer Ecooetrc Isttute, erge Isttute, Ersus Uversty Rotterd, P.O. Box 73, 3 DR Rotterd, he etherds Ecooetrc Isttute Report
More informationUsing Phase Swapping to Solve Load Phase Balancing by ADSCHNN in LV Distribution Network
Iteratoal Joural of Cotrol ad Automato Vol.7, No.7 (204), pp.-4 http://dx.do.org/0.4257/jca.204.7.7.0 Usg Phase Swappg to Solve Load Phase Balacg by ADSCHNN LV Dstrbuto Network Chu-guo Fe ad Ru Wag College
More informationIntegrating Production Scheduling and Maintenance: Practical Implications
Proceedgs of the 2012 Iteratoal Coferece o Idustral Egeerg ad Operatos Maagemet Istabul, Turkey, uly 3 6, 2012 Itegratg Producto Schedulg ad Mateace: Practcal Implcatos Lath A. Hadd ad Umar M. Al-Turk
More informationCredibility Premium Calculation in Motor Third-Party Liability Insurance
Advaces Mathematcal ad Computatoal Methods Credblty remum Calculato Motor Thrd-arty Lablty Isurace BOHA LIA, JAA KUBAOVÁ epartmet of Mathematcs ad Quattatve Methods Uversty of ardubce Studetská 95, 53
More information