Size: px
Start display at page:

Download ""

Transcription

1 Whch one should I mtate? Karl H. Schlag Projektberech B Dscusson Paper No. B-365 March, 996 I wsh to thank Avner Shaked for helpful comments. Fnancal support from the Deutsche Forschungsgemenschaft, Sonderforschungsberech 303 at the Unversty of Bonn s gratefully acknowledged. Abt. Wrtschaftstheore III, Department of Economcs, Unversty of Bonn, Adenauerallee 4-6, 533 Bonn, Germany.

2 Abstract We consder the model of socal learnng by Schlag [5]. Indvduals must repeatedly choose an acton n a mult-armed bandt. We assume that each ndvdual observes the outcomes of two other ndvduals' choces before her own next choce must be made { the orgnal model only allows for one observaton. Selecton of optmal behavor yelds a varant of the proportonal mtaton rule { the optmal rule based on one observaton. When each ndvdual uses ths rule then the adaptaton of actons n an nnte populaton follows an aggregate monotone dynamc. JEL classcaton numbers: C7, C79. Keywords: socal learnng, mult-armed bandt, mtaton, payo ncreasng, proportonal mtaton rule, aggregate monotone dynamc.

3 Introducton In ths paper we consder a varant of a model by Schlag [5]. Schlag consders a model of socal learnng n whch ndvduals repeatedly face a mult-armed bandt. Between ther choces each ndvdual may observe the performance of one other ndvdual, a stuaton referred to n the followng as sngle samplng. Indvduals forget about observatons n the past. Two alternatve approaches to selectng an optmal ndvdual behavor, a bounded ratonal approach and a populaton-orented approach are suggested. Ether approach leads to the same unque prescrpton, the so-called proportonal mtaton rule, of how to choose future actons: ) follow an mtatve behavor,.e., only change actons through mtatng others, ) never mtate an ndvdual that performed worse than oneself, and ) mtate an ndvdual that performed better wth a probablty that s proportonal to how much better ths ndvdual performed. In ths paper we analyze how the above result changes when an ndvdual s allowed to observe the performance of two other ndvduals between her choces. Ths stuaton wll be referred to as double samplng. In contrast to the sngle samplng settng t turns out that there s no behavor that s better than all other behavoral rules (accordng to ether of the selecton approaches for ndvdual behavor). However, there s a best way of performng better than under sngle samplng. Ths can be acheved by modfyng the proportonal mtaton rule, the resultng rule we call the adjusted proportonal mtaton rule. Ths varant of the proportonal mtaton rule speces addtonally to ) to ) above, v) to be more lkely to mtate the ndvdual n the sample who realzed the hgher payo, and v) to be more lkely to mtate one of the two sampled ndvduals the lower the payo of the other one s, especally not to gnore a sampled ndvdual that realzed a lower payo even though he wll never be mtated. Its smple functonal form and ts performance lead us to selectng the adjusted proportonal mtaton rule as the optmal rule under double sam-

4 plng. Where aggregate behavor of an nnte populaton of ndvduals usng the optmal rule under sngle samplng followed the replcator dynamc (Taylor [6]), under double samplng t follows an aggregate monotone dynamc (as dened by Samuelson and Zhang [4]). The rest of the paper s organzed as follows. In Secton two the basc payo realzaton and samplng scenaro s ntroduced. The feasble behavoral rules for ths settng are presented. In secton three we select among the behavoral rules. Secton four contans the mplcatons optmal behavor has for the populaton dynamcs. In Secton ve we consder an alternatve two populaton matchng scenaro. The Appendx contans the proof of the man theorem whch s stated n Secton three. The Settng Consder the followng dynamc process of choosng actons, samplng and updatng. Let W be a nte populaton (or set) of N ndvduals, N 3. In a sequence of rounds, each ndvdual must choose an acton from a nte set of actons A = f; ; ::; ng where n : Choosng the acton yelds an uncertan payo drawn from a probablty dstrbuton P wth nte support n [;!] where and!; <!, are exogenous parameters. Payos are realzed ndependently of all other events. Let denote the expected payo generated by choosng acton ;.e., = P x[;!] xp (x) ; A. Then the E tuple DA; (P ) A consttutes a mult-armed bandt or a game aganst nature. The set of all mult-armed bandts wth acton set A yeldng payos n [;!] wll be denoted by G (A; [;!]) : A state s A W of the populaton n a gven round t s the descrpton of the acton that each ndvdual s choosng n round t. Let (A) be the set of probablty dstrbutons on A. For a gven state s let p = p (s) (A) denote the probablty dstrbuton that s assocated wth randomly selectng an ndvdual and observng the acton she has chosen for ths round,.e., p (s) = jfc W : s (c) =gj ( A). The set of all such probablty N

5 dstrbutons wll be denoted by N (A),.e., p N (A) and A mples N p N. Gven ths notaton, the average expected payo of the populaton n state s; (s) ; s gven by (s) = P p (s). Before each round of payo realzaton, each ndvdual meets (or samples) two other ndvduals from the populaton and observes the payo each of them receved, together wth the assocated acton, n the prevous round. Gven three derent ndvduals c; d; e W; the probablty that ndvdual `c' samples ndvduals `d' and `e' s denoted by P (c ; fd; eg). In the followng we wll assume that samplng s symmetrc,.e., that P (c ; fd; eg) = P (d ; fc; eg) = P (e ; fc; dg) for all c; d; e W: The stuaton n whch samplngs occurs by choosng two ndvduals randomly from the populaton wll be called random samplng, n ths case P (c ; fd; eg) = all c; d; e W. for (N )(N ) The descrpton of how an ndvdual chooses her next acton n a multarmed bandt n G (A; [;!]) based on her prevous observatons s summarzed by a behavoral rule. We allow for the ndvdual to use a randomzng devce that generates ndependent events when makng ths choce. We restrct attenton to behavoral rules where observatons pror to her last payo realzaton do not nuence her next choce,.e., essentally an ndvdual forgets these observatons. Hence, a behavoral rule s a functon F : A[;!]fA[;!] A [;!]g!(a) where F (; x; fj; y; k; zg) r s the probablty of playng acton r after obtanng payo x wth acton and samplng ndvduals usng acton j and acton k that obtaned payo y and payo z respectvely. For ; j; k; r A, let F r jk := x;y;z[;!] be the so-called swtchng probabltes. F (; x; fj; y; k; zg) r P (x) P j (y) P k (z) ; ; j; k; r A; A class of behavoral rules of specal mportance n our analyss wll be the class of mtatng rules. A behavoral rule F s called mtatng f F (; x; fj; y; k; zg) r = 0 when r = f; j; kg. For an mtatng behavoral rule F let F (; x; fj; y; k; zg) jk denote the probablty of swtchng actons (to ether acton j or k),.e., F (; x; fj; y; k; zg) jk = let F jk jk F(; x; fj; y; k; zg) : Smlarly, be the assocated swtchng probabltes,.e., F jk jk = 3 F jk.

6 3 Examples In the followng we present some examples of behavoral rules. Behavoral rules under sngle samplng can be embedded n the class of behavoral rules under double samplng by randomly selectng one of the two sampled ndvduals and applyng the sngle samplng rule. More speccally, the behavoral rule f under sngle samplng,.e., f : A [;!] A[;!]! (A); s assocated to the behavoral rule F f under double samplng de- ned by F f (; x; fj; y; k; zg) r = f (; x; j; y) + f (; x; k; z) r r ; ; j; k; r A; x; y; z [;!] : Behavoral rules constructed n ths way wll be called sngle samplng rules. An mportant behavoral rule under sngle samplng rule s the mtatng rule f p that satses f p (; x; j; y) j = [y x] where [x] = x when x>0! + + and [x] + = 0 when x 0: Schlag [5] argues that ths so-called proportonal mtaton rule wth rate s the unque optmal rule under sngle samplng.! The assocated sngle samplng rule wll be denoted by F p. The behavoral rule of mportance n the present model of double samplng s the rule we refer to as the adjusted proportonal mtaton rule. Let : [;!]! R + be the lnearly decreasng functon such that () =! and (!) =!,.e., (x) =! +! x for x [;!] : (! ) Consder the behavoral rule ^F such that ^F (; x; fj; y; k; zg) j = (z)[y x] + ; ^F (; x; fj; y; k; zg) k = (y)[z x] + and ^F (; x; fj; y; j; zg) j = (z)[y x] + + (y)[z x] + ; jf; j; kgj = 3. In order for ^F to be n fact a behavoral rule we must show that ^F (; x; fj; y; k; zg) jk when x<yz;ths s true snce ^F (; ; fj; y; k; zg) jk = y! +! z! + z! +! y :! We wll call ^F the adjusted proportonal mtaton rule (based on [;!]). Notce that an ndvdual followng ths rule wll be more lkely to mtate the ndvdual n the sample that realzed the hgher payo. He wll never mtate 4

7 an ndvdual that realzed a lower payo, wll never-the-less use the payo of such an ndvdual to determne how lkely to swtch to the other ndvdual. Some extreme stuatons for how the payo of the one ndvdual nuences the probablty of swtchng to the other are as follows: ^F (; x; fj; y; k; g)j = f p (; x; j; y) j and ^F (; x; fj; y; k;!g) j = f p (; x; j; y) j : A popular rule under sngle samplng s the mtatng rule `mtate f better', where the ndvdual adapts the acton of the observed ndvdual f and only f t acheved a hgher payo. In the lterature ths rule s extended to the framework of multple samplng n the followng two derent ways. `Imtate the best' (Axelrod []) s the mtatng behavoral rule F that satses: F (; x; fj; y; k; zg) j =f y > max fx; zg and F (; x; fj; y; k; zg) = f x max fy; zg ; ; j; k A; x; y; z [;!] : `Imtate the best average' (Bruch []; Ellson and Fudenberg [3]) s the mtatng behavoral rule F that satses F (; x; fj; y; j; zg) j = f (y + z) > x and 0 otherwse, F (; x; f; y; j; zg) j =fz> (x+y) and 0 otherwse, F (; x; fj; y; k; zg) = j f y > max fx; zg ; F (; x; fj; y; k; yg) j = F (; x; fj; y; k; yg) k = f y > x and F (; x; fj; y; k; zg) =fxmax fy; zg ; ;j;k A wth jf; j; kgj =3; x; y; z [;!] : 4 Selecton Among the Rules The so-called expected mprovement EIP F (s) n state s s gven by the followng expresson: EIP F (s):= N j c;d;ew P (c ; fd; eg) F j s(c)s(d)s(e) h j s(c) : Indvduals are assumed to prefer so-called mprovng behavoral rules, these are rules that always generate non negatve expected mprovement. Formally, a behavoral rule F s called mprovng f EIP F (s) 0 for all s N (A) and all mult-armed bandts n G (A; [;!]) : Schlag [5] gves two alternatve scenaros that cause an ndvdual to choose an mprovng rule. ) Indvduals are boundedly ratonal. They enter the populaton by replacng a random ndvdual n the populaton. They adapt the acton last 5

8 chosen by ths ndvdual. In each round an ndvdual evaluates the performance of her behavor as f she just entered. Indvduals prefer a rule that always ncreases expected payos n any mult-armed bandt n G (A; [;!]). ) Indvduals evaluate the performance of ther behavor n a populaton of replcas. An ndvdual consders a populaton n whch each ndvdual s usng her behavoral rule. She prefers a rule that s expected to ncrease average payos n each state and each mult-armed bandt n G (A; [;!]). Schlag [5] characterzes the set of mprovng rules under sngle samplng. Especally t turns out that the proportonal mtaton rule wth rate s! mprovng and that the rule `mtate f better' s not mprovng. Clearly, an mprovng behavoral rule f n the sngle samplng settng s assocated to a sngle samplng rule F f (see Secton 3) that s mprovng n the present double samplng settng. The followng theorem characterzes the entre set of mprovng behavoral rules under double samplng. Theorem The behavoral rule F s mprovng f and only f F s mtatng and for all subsets f; j; kg A wth jf; j; kgj > there exsts a functon f;j;kg :[;!]! R + 0 such that F (; x; fj; y; k; zg) jk F (j; y; f; x; k; zg) F (k; z; f; x; j; yg) = f;j;kg (z)(y x)+ f;j;kg (y)(z x); () f = fj; kg. Proof. (n the Appendx) Theorem and ts proof gve lttle nsght as to whch functons f;j;kg () are assocated to an mprovng rule. Of course, the rght hand sde of () must be bounded above by ; especally f;j;kg (y) max all y [;!] : n ; y (! ) However, Theorem enables us to verfy whether a behavoral rule s mprovng or not. Consder for example the rules `mtate the best average' and `mtate the best'. () mples that nether of these rules s mprovng. In the followng we show ths statement usng a counterexample n order to explctly llustrate how these two rules fal to be mprovng. 6 o for

9 Fx x ; 3 + 3!. Consder a mult-armed bandt n whch P (x) =, P () =and P (!) = for some 0 <<. Then > f and only f <! x! :Notce that the rule `mtate the best' and the rule `mtate the best average' nduce the same swtchng probabltes F = F = and F = F = : Especally, > mples F 3 F > 0 and F F < 0. Followng (8), ths leads to negatve expected mprovement f only acton and acton are played n the populaton, wth postve probablty some ndvdual usng acton observes some ndvdual usng acton and f < <! x. Hence we see that nether `mtate the best' 3! nor `mtate the best average' s mprovng. Under the sngle samplng rules, Schlag [5] shows that the proportonal mtaton rule F p never acheves a lower expected mprovement than any other mprovng sngle samplng rule. Hence, we say that F p domnates the sngle samplng rules. More generally, let F be a set of behavoral rules. We say that a behavoral rule F domnates the set of behavoral rules F f EIP F (s) EIP F 0 (s) for all F 0 F, for any state s and for any mult-armed bandt n G (A; [;!]). Consequently, f F contans an mprovng rule and F domnates F then F s mprovng. In the followng we wll show that mprovng rules under double samplng wth constant swtchng rates f;j;kg () are of no advantage compared to the sngle samplng scenaro. As mentoned above, the hghest expected mprovement s realzed by the proportonal mtaton rule F p : Followng (8), EIP F p (s)= N(! ) c;d;ew P (c ; fd; eg) s(d) s(c) : () Snce f;j;kg (!), followng () and (8), an mprovng rule under double samplng wth constant swtchng rates f;j;kg never acheves a! hgher expected mprovement than F p. The advantage of double samplng les n the fact that swtchng rates of mprovng rules must no longer be constant. The followng theorem states that, unlke under sngle samplng, under double samplng there s no behavoral rule that domnates all other mprovng rules. However, we show that followng the adjusted proportonal mtaton rule ^F s the best way of 7

10 performng better than the proportonal mtaton rule F p. Theorem Let F be the set of sngle samplng rules that are mprovng. Let F be the set of rules that domnate F. Then the adjusted proportonal mtaton rule ^F domnates F. There s no behavoral rule that domnates the set of mprovng rules. In the followng, let F 3 be the set of rules that domnate F : Proof. Consder an mprovng rule F 0 F, let 0 f;j;kg be the assocated swtchng rates. We wll rst show that 0 (!) = f;j;kg : As mentoned! above, 0 (!) : Consder the mult-armed bandt n whch P f;j;kg! () = and P j (!) =P k (!)=. Consder a populaton wth one ndvdual usng, one usng j and the rest usng k: Usng the fact that a j jk = ak jk = 0 f;j;kg (!) t follows that EIP F 0 = N 0 (!)(! f;j;kg ) (! N! ) = EIP F p. Snce F 0 domnates F p and F p F weobtan that 0 (!) =. f;j;kg! Notce that 0 (!)(y )+ f;j;kg 0 (y)(! ) f;j;kg F0 (; ; fj; y; k;!g) jk mples 0 (y) f;j;kg h (y ) 0f;j;kg! (!) =! +! y (! ) = (y) : (3) Hence, (3) and (8) mply EIP ^F EIP F 0 for any state s and any mult-armed bandt n G (A; [;!]) whch means that ^F F 3. Especally, t follows that f;j;kg (y) = (y) for any rule F F 3 : We wll now construct a rule that s not domnated by any rule n F 3 : Ths wll show that there s no rule that domnates all other mprovng rules. Let ~ F be the behavoral rule that s constructed lke ^F usng the functon ~ where ~ (y) = +! when y and ~ (y) = 0 for y > +! : It follows! that F ~ (; ; fj; y; k; zg) jk and hence that F ~ s n fact a behavoral rule. Moreover, by constructon, ~ F s mprovng and ~ (y) > (y) for all <y +! :Hence, ~ F s not domnated by any rule n F3. One can argue that an ndvdual wll choose an mprovng rule that domnates the mprovng rules under sngle samplng,.e., a rule n F. She 8

11 mght as well choose a rule that s best at dong ths,.e., a rule n F 3. We presented such a rule, the adjusted proportonal mtaton rule, that addtonally never mtates lower payos and has a smple form. Ths leads us to selectng ths rule as the optmal rule under double samplng. 5 Populaton Dynamcs In ths secton we consder the aggregate behavor of a populaton n whch each ndvdual uses the optmal rule. We wll restrct attenton to random samplng. Moreover, we wll consder adjustment n nnte populatons as an approxmaton of the short run adjustment of a large populaton. Schlag [5] speces the exact meanng of ths approxmaton for the sngle samplng settng. In an nnte populaton, random samplng means that the probablty that an ndvdual observes acton s equal to the proporton of ndvduals usng ths acton. In ths sense, a descrpton of the proportons p usng acton for each A s sucent to determne the populaton adjustment. Hence we wll dentfy the state of a populaton wth p = (p ) A (A): Straghtforward calculatons show that the adjustment process (p t ) tn of a monomorphc populaton (each ndvdual s followng the same behavor) n whch the underlyng rule s mprovng, gven an ntal state p (A), s gven by p t+ = p t + pt for A and t N. j;k pt j pt k proportonal mtaton rule ^F,weobtan " p t+ = p t + where (p) = P Ap : h a j ( jk k )+a k jk ( j ) If, n addton, the underlyng rule s the adjusted! +! (pt ) (! ) #, p t p t, (4) 9

12 6 A Two Populaton Matchng Scenaro What about a settng n whch the mult-armed bandt s not statonary over tme? We wll consder a popular example for such a stuaton; ndvduals wll be randomly matched to play a game. dsjont sets (populatons) of ndvduals W Consder two nte, and W, each of sze N, also referred to as populaton one and two. Let A be the nte set of actons avalable to an ndvdual n populaton ; = ; : Payos are realzed by matchng ndvduals from derent populatons. When an ndvdual n populaton one usng acton A s matched wth an ndvdual n populaton two usng acton j A ; the ndvdual n populaton k acheves an uncertan payo drawn from a gven, ndependent probablty dstrbuton P k; j k =;. Assocatng player to beng an ndvdual n populaton, the tuple * A ;A ; P j A ; ja P j + A ja denes an asymmetrc two player normal form game. We wll restrct attenton to the class of asymmetrc two player normal form games, denoted by G (A ;A ;[ ;! ];[ ;! ]), n whch player k has acton set A k ;k =;;where P j has nte support n [ ;! ] and P j has nte support n [ ;! ] for all A and j A ; <! and <! are gven. For a gven asymmetrc game, let () and () be the blnear functons on (A ) (A ) where k (; j) s the expected payo to player k when player one s usng acton and player two s usng acton j;.e., k (; j) = P x[ k ;! k ] xp k j (x) ;k=;: Indvduals of opposte populatons are matched at random n pars, for an ndvdual n populaton one ths means the followng. Let s (A ) W be the current state n populaton one and let p N (A ) be the assocated populaton shares. Smlarly let s (A ) W and q N (A ) be dened for populaton two. Then an ndvdual n populaton one s matched wth an ndvdual n populaton two usng acton j A wth probablty q j. Snce we consder random matchng, (; q) speces the expected payo of an ndvdual n populaton one usng acton A and (p; q) speces the average payo n populaton one n ths state. Especally, each ndvdual n populaton one s facng a mult-armed bandt D A ; (P 0 ) AE G(A ;[ ;! ]) that 0

13 depends on the populaton shares n populaton two; P 0 (x) =P jaq j P j (x) for x [ ;! ] and A. Samplng occurs wthn the same populaton and s performed as n the mult-armed bandt settng. A behavoral rule F for an ndvdual n populaton k s a functon F : (A k )[ k ;! k ](A k )[ k ;! k ]!(A k ), k =;: Schlag [5] gves two scenaros n whch an ndvdual prefers to use the same rule n ths populaton matchng settng as n the former mult-armed bandt settng: ) It mght be that ndvduals do not realze that the mult-armed bandt s non statonary or that they smply gnore ths fact. ) An ndvdual mght choose her rule accordng to ts performance n a populaton of replcas and prefers a rule that s expected to ncrease average payos whenever all ndvduals n the opposte populaton do not change ther acton. Hence, we consder the adjusted proportonal mtaton rule based on [ ;! ]tobetheoptmal rule for an ndvdual n populaton n ths populaton matchng settng. In the followng we consder the aggregate behavor of the two populatons under random samplng when each ndvdual uses her optmal rule. As n Secton 5 we consder the lmt of ths adjustment asthe populaton sze N tends to nnty and apply a law of large numbers type of argument. Analogue to (4), the resultng adjustment process (p t ;q t ) tn gven by " = p t + +! (p t ;q t )! " j = q t j + +! (p t ;q t )! p t+ q t+ # h (! ) ; q t p t ;q t p t ; (5) # h (! ) p t ;j p t ;q t q t j, for A ;j A and t N. Accordng to Samuelson and Zhang [4], (5) s called an aggregate monotone dynamc. Under sngle samplng the adjustment generated when each ndvdual s usng her optmal rule (.e., the proportonal mtaton rule wth rate! for populaton ) s approxmated s

14 by the followng dscrete verson of the replcator dynamc (Taylor [6]): p t+ = p t + h ; q t p t ;q t p t! ; (6) q t+ j = q t j +! h p t ;j p t ;q t q t j. Comparng ths to (5) we see that the advantage of double samplng for ndvduals usng ther optmal rule ^F s greatest when average payos n ther own populaton are low. References [] R. M. Axelrod, The Evoluton of Cooperaton, Basc Books, New York, 984. [] E. Bruch, \Evoluton von Kooperaton n Netzwerken", Dplomarbet, Unversty of Bonn, 993. [3] G. Ellson and D. Fudenberg, Word-Of-Mouth Communcaton and Socal Learnng, Quart. J. Econ. 440 (995), [4] L. Samuelson and J. Zhang, Evolutonary Stablty n Asymmetrc Games, J. Econ. Theory 57 (99), [5] K. H. Schlag, \Why Imtate, and f so, How? A Bounded Ratonal Approach to Mult-Armed Bandts," Unversty of Bonn, Dsc. Paper B-36, Bonn, 996. [6] P. Taylor, Evolutonarly Stable Strateges Wth Two Types of Players, J. Appled Prob. 6 (979), A The Proof of Theorem Proof. For ; j; k A and s A W p jk (s) = let c;d;ew fs(c);s(d);s(e)g=f;j;kg P (c ; fd; eg).

15 We wll rst show the `f' statement. EIP F (s) = N + N c;d;ew s(d)=s(e) c;d;ew s(d)6=s(e) hf s(d) s(c)s(d)s(e) = 3N + 3N ;ja P (c ; fd; eg) P (c ; fd; eg) hf s(d) s(c)s(d)s(e) s(d) s(c) s(d) s(c) + F s(e) s(c)s(d)s(e) p jj (s) ;j;ka jf;j;kgj=3 F j jj p jk (s) 6 4 F jj + + F j jk s(e) s(c) ( j ) (7) F k jk F k jk F jk Consder actons ; j; k A such that = fj; kg : Then F kj F j kj ( j ) ( k ) ( k j ) F jk jk F jk F kj = x;y;z P (x) P j (y) P k (z) hf (; x; fj; y; k; zg) jk F (j; y; f; x; k; zg) F (k; z; f; x; j; yg) = P (x) P j (y) P k (z) x;y;z " # = P k (z) f;j;kg (z) z f;j;kg (z)(y x)+ f;j;kg (y)(z x) ( j )+ " y P j (y) f;j;kg (y) # ( k ) and, gven a l jk = P y P l (y) f;j;kg (y), l f; j; kg, weobtan h F j jk F jk ( j )+ F k jk F kj ( k )+ = = = F jk jk F jk F kj + F k jk F j jk F j kj j + F j kj F k jk F k jk F j kj ( k ) a j + f;j;kg ( j ) a k f;j;kg + ( j ) a k + f;j;kg ( k j ) a f;j;kg j + ( k ) a j + f;j;kg ( j k ) a f;j;kg k h ( j ) a k f;j;kg +( k ) a j +( f;j;kg j k ) af;j;kg 0: 3 ( k j ) k F k jk

16 Hence, (7) smples to EIP F (s)= N c;d;ew and t follows that EIP F 0: P (c ; s(e) fd; eg) s(d) s(c) a (8) fs(c);s(d);s(e)g Wenow come to the proof of the `only f' statement. In order to smplfy the presentaton of the proof we wll assume that P (c ; fd; eg) > 0 for all c; d; e W (jfc; d; egj = 3). The proof can be easly adjusted to the more general case. The fact that F s mtatng follows just lke under sngle samplng (Schlag [5]). Assume that F (; x; fj; y; k; zg) r > 0 for some r = f; j; kg and x; y; z [;!]. Consder amult-armed bandt n whch P (x) =P (y)=p (!)= 3 ; P j P k P and P l () = for all l = f; j; kg. In a state s whch only actons n f; j; kg are beng played t follows that EIP F (s) < 0: Next we wll show () for 6= j = k. Consder a populaton state n whch one ndvdual s playng acton and the rest are playng acton j. Then Let EIP F (s)= N F j jj F jj ( j ) : (9) g (x; y; z) =F(; x; fj; y; j; zg) j F (j; y; f; x; j; zg) F (j; z; f; x; j; yg) ; x; y; z [;!] : Let y = z. We now follow the same arguments as n the proof of Theorem n Schlag [5] to shows that there exsts jj :[;!]! R + 0 that such g (x; y; y) = jj (x)(y x) for all x; y [;!] : (0) For gven x; y [;!] ; consder the mult-armed bandt where P (x) = P j (y) =. Then F j jj F jj = g (x; y; y) and hence followng (9), g (x; y) 0 and g (y; x) 0 whenever y>x: () Moreover, usng arguments nvolvng symmetry t follows that g (x; x) = 0 for all x [;!]. Next we wll show that g (x; y; y) y x = g (x; z; z) z x 8y <x<z. () 4

17 Gven y < x < z; consder a mult-armed bandt where P (x) = ; P j (y) = and P j (z) = ; 0 : Then j > f and only f < z x =: z y. Agan followng (9), we obtan F j jj F jj = g (x; y)+( )g(x; z) 0 f < and Therefore, g (x; y)+( that () s true. g (x; y)+( )g(x; z) 0 f > )g(x; z) =0;whch, after smplcaton, shows Followng () there exsts jj : (;!)! R + 0 such that g (x; y; y) = jj (x) (y x) for all x; y (;!). Lookng back at the above proof we see that the explct values of and! dd not enter the argument. Hence, (0) holds for all x; y [;!]. Consder now a mult-armed bandt wth P (x) = ; P j (y) = and P j (z) = for y<x<zand 0. Followng (9) and gven I () = F j jj F jj (3) = jj (y)(y x)+( )g(x; y; z)+( ) jj (z)(z x) we obtan that I 0 f and only f j : Hence I = 0 f and only f j = f and only f = z x =: z y. Snce I ( ) = 0, (z x) jj (y) + g(x; y; z)+(x y) jj (z) =0and hence g (x; y; z) = jj (z)(y x)+ jj (y)(z x): (4) We wll now derve g (x; y; z) for y z < x. Consder a mult-armed bandt wth P j (x) = ; P (y) = ; P (z) = and P (z 0 ) = for z 0 >x:then I = jj (y)(y x)+ jj (z)(z x)+g (x; y; z) (5) + ( )[ jj (z 0 )(y x)+ jj (y)(z 0 x)] + ( )[ jj (z 0 )(z x)+ jj (z)(z 0 x)] +( ) jj (z 0 )(z 0 x). As before, I =0f and only f = j f and only f (x y)+(x z)= ( )(z 0 x). Settng (x y)+(x z)=( )(z 0 x), together wth (5) mples that (4) also holds for y z<x. 5

18 Repeatng such calculatons for the remanng values of (x; y; z) not yet consdered nally yelds that (4) holds for all x; y; z [;!] : Ths completes the proof of the `only f' statement for j = k: We now proceed wth the case where j 6= k: Consder a populaton n whch one ndvdual s playng, one s playng k and the rest are playng j. Consder a mult-armed bandt n whch j = k : Then F j jj EIP F (s) = 3N p jj (s) + 3N p jk (s) F jj ( j ) h F jk jk F jk F kj ( j ) Snce F j jj F = 0 f jj = j t follows that F jk jk F jk F = 0 f kj and only f = j must hold. Followng the same arguments as n the proof where j = k we obtan that there exsts jk kj :[;!]! R + such that F (; x; fj; y; k; zg) jk F (j; y; f; x; k; zg) F (k; z; f; x; j; yg) = jk (z)(y x)+ jk (y)(z x) holds for all x; y; z [;!] : The only thng remanng to show s that jk s ndependent of a permutaton of, j and k. Consder a mult-armed bandt wth P (x) = P j (y) = P k (z) = and a populaton n whch one ndvdual s playng, one s playng k and the rest are playng j. Followng the calculatons when provng the `f' statement, we obtan EIP F (s) = 3N p jj (s) jj (y)(y x) + 3N p kjj (s) kjj (y)(y z) (6) 3N p jk (s) 6 4 Settng y = z ths smples to h EIP F (s) = 3N p jj (s) jj (y)(y x) x jk (y) + 3N p jk (s) x jk (z)(y h x)+ jk (y)(z x) +y jk (z)(x h y)+ jk (x)(z y) +z kj (x)(y z)+ kj (y)(x z) 6 y jk (y) y kj (y) 3 7 5

19 whch mples, when settng y = x 6= 0, that jk (x) jk (x) kj (x) = 0 for any x 6= 0: Smlarly, settng x = y n (6) leads to kj (x) jk (x) jk (x) = 0 for any x 6= 0: Together ths means that kj (x) = jk (x) = jk (x) for all x 6= 0: The specal case of x = 0 s easly shown usng more general mult-armed bandts and hence the proof s complete. 7

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

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

More information

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

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

More information

Recurrence. 1 Definitions and main statements

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

More information

Luby s Alg. for Maximal Independent Sets using Pairwise Independence

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

More information

PSYCHOLOGICAL RESEARCH (PYC 304-C) Lecture 12

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

More information

Extending Probabilistic Dynamic Epistemic Logic

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

More information

The OC Curve of Attribute Acceptance Plans

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

More information

An Alternative Way to Measure Private Equity Performance

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

More information

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

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

More information

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

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

More information

8 Algorithm for Binary Searching in Trees

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

More information

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

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

More information

What is Candidate Sampling

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

More information

Generalizing the degree sequence problem

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

More information

Calculation of Sampling Weights

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

More information

Answer: A). There is a flatter IS curve in the high MPC economy. Original LM LM after increase in M. IS curve for low MPC economy

Answer: A). There is a flatter IS curve in the high MPC economy. Original LM LM after increase in M. IS curve for low MPC economy 4.02 Quz Solutons Fall 2004 Multple-Choce Questons (30/00 ponts) Please, crcle the correct answer for each of the followng 0 multple-choce questons. For each queston, only one of the answers s correct.

More information

1 Example 1: Axis-aligned rectangles

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

More information

Stochastic epidemic models revisited: Analysis of some continuous performance measures

Stochastic epidemic models revisited: Analysis of some continuous performance measures Stochastc epdemc models revsted: Analyss of some contnuous performance measures J.R. Artalejo Faculty of Mathematcs, Complutense Unversty of Madrd, 28040 Madrd, Span A. Economou Department of Mathematcs,

More information

Addendum to: Importing Skill-Biased Technology

Addendum to: Importing Skill-Biased Technology Addendum to: Importng Skll-Based Technology Arel Bursten UCLA and NBER Javer Cravno UCLA August 202 Jonathan Vogel Columba and NBER Abstract Ths Addendum derves the results dscussed n secton 3.3 of our

More information

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

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

More information

Joe Pimbley, unpublished, 2005. Yield Curve Calculations

Joe Pimbley, unpublished, 2005. Yield Curve Calculations Joe Pmbley, unpublshed, 005. Yeld Curve Calculatons Background: Everythng s dscount factors Yeld curve calculatons nclude valuaton of forward rate agreements (FRAs), swaps, nterest rate optons, and forward

More information

The Cox-Ross-Rubinstein Option Pricing Model

The Cox-Ross-Rubinstein Option Pricing Model Fnance 400 A. Penat - G. Pennacc Te Cox-Ross-Rubnsten Opton Prcng Model Te prevous notes sowed tat te absence o arbtrage restrcts te prce o an opton n terms o ts underlyng asset. However, te no-arbtrage

More information

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

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

More information

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

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

More information

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

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

More information

The Greedy Method. Introduction. 0/1 Knapsack Problem

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

More information

Bandwdth Packng E. G. Coman, Jr. and A. L. Stolyar Bell Labs, Lucent Technologes Murray Hll, NJ 07974 fegc,stolyarg@research.bell-labs.com Abstract We model a server that allocates varyng amounts of bandwdth

More information

Dscrete-Tme Approxmatons of the Holmstrom-Mlgrom Brownan-Moton Model of Intertemporal Incentve Provson 1 Martn Hellwg Unversty of Mannhem Klaus M. Schmdt Unversty of Munch and CEPR Ths verson: May 5, 1998

More information

Awell-known result in the Bayesian inventory management literature is: If lost sales are not observed, the

Awell-known result in the Bayesian inventory management literature is: If lost sales are not observed, the MANUFACTURING & SERVICE OPERATIONS MANAGEMENT Vol. 10, No. 2, Sprng 2008, pp. 236 256 ssn 1523-4614 essn 1526-5498 08 1002 0236 nforms do 10.1287/msom.1070.0165 2008 INFORMS Dynamc Inventory Management

More information

Rate Monotonic (RM) Disadvantages of cyclic. TDDB47 Real Time Systems. Lecture 2: RM & EDF. Priority-based scheduling. States of a process

Rate Monotonic (RM) Disadvantages of cyclic. TDDB47 Real Time Systems. Lecture 2: RM & EDF. Priority-based scheduling. States of a process Dsadvantages of cyclc TDDB47 Real Tme Systems Manual scheduler constructon Cannot deal wth any runtme changes What happens f we add a task to the set? Real-Tme Systems Laboratory Department of Computer

More information

PRIVATE SCHOOL CHOICE: THE EFFECTS OF RELIGIOUS AFFILIATION AND PARTICIPATION

PRIVATE SCHOOL CHOICE: THE EFFECTS OF RELIGIOUS AFFILIATION AND PARTICIPATION PRIVATE SCHOOL CHOICE: THE EFFECTS OF RELIIOUS AFFILIATION AND PARTICIPATION Danny Cohen-Zada Department of Economcs, Ben-uron Unversty, Beer-Sheva 84105, Israel Wllam Sander Department of Economcs, DePaul

More information

An Evaluation of the Extended Logistic, Simple Logistic, and Gompertz Models for Forecasting Short Lifecycle Products and Services

An Evaluation of the Extended Logistic, Simple Logistic, and Gompertz Models for Forecasting Short Lifecycle Products and Services An Evaluaton of the Extended Logstc, Smple Logstc, and Gompertz Models for Forecastng Short Lfecycle Products and Servces Charles V. Trappey a,1, Hsn-yng Wu b a Professor (Management Scence), Natonal Chao

More information

Linear Circuits Analysis. Superposition, Thevenin /Norton Equivalent circuits

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

More information

ANALYZING THE RELATIONSHIPS BETWEEN QUALITY, TIME, AND COST IN PROJECT MANAGEMENT DECISION MAKING

ANALYZING THE RELATIONSHIPS BETWEEN QUALITY, TIME, AND COST IN PROJECT MANAGEMENT DECISION MAKING ANALYZING THE RELATIONSHIPS BETWEEN QUALITY, TIME, AND COST IN PROJECT MANAGEMENT DECISION MAKING Matthew J. Lberatore, Department of Management and Operatons, Vllanova Unversty, Vllanova, PA 19085, 610-519-4390,

More information

Nordea G10 Alpha Carry Index

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

More information

Portfolio Loss Distribution

Portfolio Loss Distribution Portfolo Loss Dstrbuton Rsky assets n loan ortfolo hghly llqud assets hold-to-maturty n the bank s balance sheet Outstandngs The orton of the bank asset that has already been extended to borrowers. Commtment

More information

THE DISTRIBUTION OF LOAN PORTFOLIO VALUE * Oldrich Alfons Vasicek

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

More information

Slow Fading Channel Selection: A Restless Multi-Armed Bandit Formulation

Slow Fading Channel Selection: A Restless Multi-Armed Bandit Formulation Slow Fadng Channel Selecton: A Restless Mult-Armed Bandt Formulaton Konstantn Avrachenkov INRIA, Maestro Team BP95, 06902 Sopha Antpols, France Emal: k.avrachenkov@sopha.nra.fr Laura Cottatellucc, Lorenzo

More information

Staff Paper. Farm Savings Accounts: Examining Income Variability, Eligibility, and Benefits. Brent Gloy, Eddy LaDue, and Charles Cuykendall

Staff Paper. Farm Savings Accounts: Examining Income Variability, Eligibility, and Benefits. Brent Gloy, Eddy LaDue, and Charles Cuykendall SP 2005-02 August 2005 Staff Paper Department of Appled Economcs and Management Cornell Unversty, Ithaca, New York 14853-7801 USA Farm Savngs Accounts: Examnng Income Varablty, Elgblty, and Benefts Brent

More information

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

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

More information

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

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

More information

DEFINING %COMPLETE IN MICROSOFT PROJECT

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

More information

Implied (risk neutral) probabilities, betting odds and prediction markets

Implied (risk neutral) probabilities, betting odds and prediction markets Impled (rsk neutral) probabltes, bettng odds and predcton markets Fabrzo Caccafesta (Unversty of Rome "Tor Vergata") ABSTRACT - We show that the well known euvalence between the "fundamental theorem of

More information

An Interest-Oriented Network Evolution Mechanism for Online Communities

An Interest-Oriented Network Evolution Mechanism for Online Communities An Interest-Orented Network Evoluton Mechansm for Onlne Communtes Cahong Sun and Xaopng Yang School of Informaton, Renmn Unversty of Chna, Bejng 100872, P.R. Chna {chsun,yang}@ruc.edu.cn Abstract. Onlne

More information

Can Auto Liability Insurance Purchases Signal Risk Attitude?

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

More information

Stock market as a dynamic game with continuum of players 2

Stock market as a dynamic game with continuum of players 2 Stock market as a dynamc game wth contnuum of players Agneszka Wsznewska-Matyszkel Insttute of Appled Mathematcs and Mechancs Warsaw Unversty e-mal: agnese@mmuw.edu.pl Abstract. Ths paper contans a game-theoretc

More information

CHOLESTEROL REFERENCE METHOD LABORATORY NETWORK. Sample Stability Protocol

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

More information

PERRON FROBENIUS THEOREM

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

More information

BERNSTEIN POLYNOMIALS

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

More information

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

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

More information

Learning the Best K-th Channel for QoS Provisioning in Cognitive Networks

Learning the Best K-th Channel for QoS Provisioning in Cognitive Networks 000 001 002 003 004 005 006 007 008 009 010 011 012 013 014 015 016 017 018 019 020 021 022 023 024 025 026 027 028 029 030 031 032 033 034 035 036 037 038 039 040 041 042 043 044 045 046 047 048 049 050

More information

Information Acquisition and Transparency in Global Games

Information Acquisition and Transparency in Global Games Informaton Acquston and Transparency n Global Games Mchal Szkup y and Isabel Trevno New York Unversty Abstract We ntroduce costly nformaton acquston nto the standard global games model. In our setup agents

More information

Forecasting the Direction and Strength of Stock Market Movement

Forecasting the Direction and Strength of Stock Market Movement Forecastng the Drecton and Strength of Stock Market Movement Jngwe Chen Mng Chen Nan Ye cjngwe@stanford.edu mchen5@stanford.edu nanye@stanford.edu Abstract - Stock market s one of the most complcated systems

More information

What should (public) health insurance cover?

What should (public) health insurance cover? Journal of Health Economcs 26 (27) 251 262 What should (publc) health nsurance cover? Mchael Hoel Department of Economcs, Unversty of Oslo, P.O. Box 195 Blndern, N-317 Oslo, Norway Receved 29 Aprl 25;

More information

The Stock Market Game and the Kelly-Nash Equilibrium

The Stock Market Game and the Kelly-Nash Equilibrium The Stock Market Game and the Kelly-Nash Equlbrum Carlos Alós-Ferrer, Ana B. Ana Department of Economcs, Unversty of Venna. Hohenstaufengasse 9, A-1010 Venna, Austra. July 2003 Abstract We formulate the

More information

EXAMPLE PROBLEMS SOLVED USING THE SHARP EL-733A CALCULATOR

EXAMPLE PROBLEMS SOLVED USING THE SHARP EL-733A CALCULATOR EXAMPLE PROBLEMS SOLVED USING THE SHARP EL-733A CALCULATOR 8S CHAPTER 8 EXAMPLES EXAMPLE 8.4A THE INVESTMENT NEEDED TO REACH A PARTICULAR FUTURE VALUE What amount must you nvest now at 4% compoune monthly

More information

Marginal Returns to Education For Teachers

Marginal Returns to Education For Teachers The Onlne Journal of New Horzons n Educaton Volume 4, Issue 3 MargnalReturnstoEducatonForTeachers RamleeIsmal,MarnahAwang ABSTRACT FacultyofManagementand Economcs UnverstPenddkanSultan Idrs ramlee@fpe.ups.edu.my

More information

Section 5.4 Annuities, Present Value, and Amortization

Section 5.4 Annuities, Present Value, and Amortization Secton 5.4 Annutes, Present Value, and Amortzaton Present Value In Secton 5.2, we saw that the present value of A dollars at nterest rate per perod for n perods s the amount that must be deposted today

More information

Scale Dependence of Overconfidence in Stock Market Volatility Forecasts

Scale Dependence of Overconfidence in Stock Market Volatility Forecasts Scale Dependence of Overconfdence n Stoc Maret Volatlty Forecasts Marus Glaser, Thomas Langer, Jens Reynders, Martn Weber* June 7, 007 Abstract In ths study, we analyze whether volatlty forecasts (judgmental

More information

Solution: Let i = 10% and d = 5%. By definition, the respective forces of interest on funds A and B are. i 1 + it. S A (t) = d (1 dt) 2 1. = d 1 dt.

Solution: Let i = 10% and d = 5%. By definition, the respective forces of interest on funds A and B are. i 1 + it. S A (t) = d (1 dt) 2 1. = d 1 dt. Chapter 9 Revew problems 9.1 Interest rate measurement Example 9.1. Fund A accumulates at a smple nterest rate of 10%. Fund B accumulates at a smple dscount rate of 5%. Fnd the pont n tme at whch the forces

More information

The literature on many-server approximations provides significant simplifications toward the optimal capacity

The literature on many-server approximations provides significant simplifications toward the optimal capacity Publshed onlne ahead of prnt November 13, 2009 Copyrght: INFORMS holds copyrght to ths Artcles n Advance verson, whch s made avalable to nsttutonal subscrbers. The fle may not be posted on any other webste,

More information

Combinatorial Agency of Threshold Functions

Combinatorial Agency of Threshold Functions Combnatoral Agency of Threshold Functons Shal Jan Computer Scence Department Yale Unversty New Haven, CT 06520 shal.jan@yale.edu Davd C. Parkes School of Engneerng and Appled Scences Harvard Unversty Cambrdge,

More information

SPEE Recommended Evaluation Practice #6 Definition of Decline Curve Parameters Background:

SPEE Recommended Evaluation Practice #6 Definition of Decline Curve Parameters Background: SPEE Recommended Evaluaton Practce #6 efnton of eclne Curve Parameters Background: The producton hstores of ol and gas wells can be analyzed to estmate reserves and future ol and gas producton rates and

More information

) of the Cell class is created containing information about events associated with the cell. Events are added to the Cell instance

) of the Cell class is created containing information about events associated with the cell. Events are added to the Cell instance Calbraton Method Instances of the Cell class (one nstance for each FMS cell) contan ADC raw data and methods assocated wth each partcular FMS cell. The calbraton method ncludes event selecton (Class Cell

More information

FINANCIAL MATHEMATICS. A Practical Guide for Actuaries. and other Business Professionals

FINANCIAL MATHEMATICS. A Practical Guide for Actuaries. and other Business Professionals FINANCIAL MATHEMATICS A Practcal Gude for Actuares and other Busness Professonals Second Edton CHRIS RUCKMAN, FSA, MAAA JOE FRANCIS, FSA, MAAA, CFA Study Notes Prepared by Kevn Shand, FSA, FCIA Assstant

More information

Product-Form Stationary Distributions for Deficiency Zero Chemical Reaction Networks

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

More information

Chapter 11 Practice Problems Answers

Chapter 11 Practice Problems Answers Chapter 11 Practce Problems Answers 1. Would you be more wllng to lend to a frend f she put all of her lfe savngs nto her busness than you would f she had not done so? Why? Ths problem s ntended to make

More information

An Analysis of Central Processor Scheduling in Multiprogrammed Computer Systems

An Analysis of Central Processor Scheduling in Multiprogrammed Computer Systems STAN-CS-73-355 I SU-SE-73-013 An Analyss of Central Processor Schedulng n Multprogrammed Computer Systems (Dgest Edton) by Thomas G. Prce October 1972 Techncal Report No. 57 Reproducton n whole or n part

More information

Value Driven Load Balancing

Value Driven Load Balancing Value Drven Load Balancng Sherwn Doroud a, Esa Hyytä b,1, Mor Harchol-Balter c,2 a Tepper School of Busness, Carnege Mellon Unversty, 5000 Forbes Ave., Pttsburgh, PA 15213 b Department of Communcatons

More information

17 Capital tax competition

17 Capital tax competition 17 Captal tax competton 17.1 Introducton Governments would lke to tax a varety of transactons that ncreasngly appear to be moble across jursdctonal boundares. Ths creates one obvous problem: tax base flght.

More information

How To Calculate An Approxmaton Factor Of 1 1/E

How To Calculate An Approxmaton Factor Of 1 1/E Approxmaton algorthms for allocaton problems: Improvng the factor of 1 1/e Urel Fege Mcrosoft Research Redmond, WA 98052 urfege@mcrosoft.com Jan Vondrák Prnceton Unversty Prnceton, NJ 08540 jvondrak@gmal.com

More information

1 De nitions and Censoring

1 De nitions and Censoring De ntons and Censorng. Survval Analyss We begn by consderng smple analyses but we wll lead up to and take a look at regresson on explanatory factors., as n lnear regresson part A. The mportant d erence

More information

Project Networks With Mixed-Time Constraints

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

More information

A Lyapunov Optimization Approach to Repeated Stochastic Games

A Lyapunov Optimization Approach to Repeated Stochastic Games PROC. ALLERTON CONFERENCE ON COMMUNICATION, CONTROL, AND COMPUTING, OCT. 2013 1 A Lyapunov Optmzaton Approach to Repeated Stochastc Games Mchael J. Neely Unversty of Southern Calforna http://www-bcf.usc.edu/

More information

Laws of Electromagnetism

Laws of Electromagnetism There are four laws of electromagnetsm: Laws of Electromagnetsm The law of Bot-Savart Ampere's law Force law Faraday's law magnetc feld generated by currents n wres the effect of a current on a loop of

More information

Chapter 7: Answers to Questions and Problems

Chapter 7: Answers to Questions and Problems 19. Based on the nformaton contaned n Table 7-3 of the text, the food and apparel ndustres are most compettve and therefore probably represent the best match for the expertse of these managers. Chapter

More information

Logical Development Of Vogel s Approximation Method (LD-VAM): An Approach To Find Basic Feasible Solution Of Transportation Problem

Logical Development Of Vogel s Approximation Method (LD-VAM): An Approach To Find Basic Feasible Solution Of Transportation Problem INTERNATIONAL JOURNAL OF SCIENTIFIC & TECHNOLOGY RESEARCH VOLUME, ISSUE, FEBRUARY ISSN 77-866 Logcal Development Of Vogel s Approxmaton Method (LD- An Approach To Fnd Basc Feasble Soluton Of Transportaton

More information

Sketching Sampled Data Streams

Sketching Sampled Data Streams Sketchng Sampled Data Streams Florn Rusu, Aln Dobra CISE Department Unversty of Florda Ganesvlle, FL, USA frusu@cse.ufl.edu adobra@cse.ufl.edu Abstract Samplng s used as a unversal method to reduce the

More information

Fisher Markets and Convex Programs

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

More information

Understanding the Impact of Marketing Actions in Traditional Channels on the Internet: Evidence from a Large Scale Field Experiment

Understanding the Impact of Marketing Actions in Traditional Channels on the Internet: Evidence from a Large Scale Field Experiment A research and educaton ntatve at the MT Sloan School of Management Understandng the mpact of Marketng Actons n Tradtonal Channels on the nternet: Evdence from a Large Scale Feld Experment Paper 216 Erc

More information

12 Evolutionary Dynamics

12 Evolutionary Dynamics 12 Evolutonary Dynamcs Through the anmal and vegetable kngdoms, nature has scattered the seeds of lfe abroad wth the most profuse and lberal hand; but has been comparatvely sparng n the room and nourshment

More information

CHAPTER 14 MORE ABOUT REGRESSION

CHAPTER 14 MORE ABOUT REGRESSION CHAPTER 14 MORE ABOUT REGRESSION We learned n Chapter 5 that often a straght lne descrbes the pattern of a relatonshp between two quanttatve varables. For nstance, n Example 5.1 we explored the relatonshp

More information

Georey E. Hinton. University oftoronto. Email: zoubin@cs.toronto.edu. Technical Report CRG-TR-96-1. May 21, 1996 (revised Feb 27, 1997) Abstract

Georey E. Hinton. University oftoronto. Email: zoubin@cs.toronto.edu. Technical Report CRG-TR-96-1. May 21, 1996 (revised Feb 27, 1997) Abstract The EM Algorthm for Mxtures of Factor Analyzers Zoubn Ghahraman Georey E. Hnton Department of Computer Scence Unversty oftoronto 6 Kng's College Road Toronto, Canada M5S A4 Emal: zoubn@cs.toronto.edu Techncal

More information

In our example i = r/12 =.0825/12 At the end of the first month after your payment is received your amount in the account, the balance, is

In our example i = r/12 =.0825/12 At the end of the first month after your payment is received your amount in the account, the balance, is Payout annutes: Start wth P dollars, e.g., P = 100, 000. Over a 30 year perod you receve equal payments of A dollars at the end of each month. The amount of money left n the account, the balance, earns

More information

Housing Liquidity, Mobility and the Labour Market

Housing Liquidity, Mobility and the Labour Market Housng Lqudty, Moblty and the Labour Market Allen Head Huw Lloyd-Ells January 29, 2009 Abstract The relatonshps among geographcal moblty, unemployment and the value of owner-occuped housng are studed n

More information

Dynamic Pricing for Smart Grid with Reinforcement Learning

Dynamic Pricing for Smart Grid with Reinforcement Learning Dynamc Prcng for Smart Grd wth Renforcement Learnng Byung-Gook Km, Yu Zhang, Mhaela van der Schaar, and Jang-Won Lee Samsung Electroncs, Suwon, Korea Department of Electrcal Engneerng, UCLA, Los Angeles,

More information

Performance Analysis of Energy Consumption of Smartphone Running Mobile Hotspot Application

Performance Analysis of Energy Consumption of Smartphone Running Mobile Hotspot Application Internatonal Journal of mart Grd and lean Energy Performance Analyss of Energy onsumpton of martphone Runnng Moble Hotspot Applcaton Yun on hung a chool of Electronc Engneerng, oongsl Unversty, 511 angdo-dong,

More information

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

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

More information

How To Calculate The Prce Of An Ndex Opton

How To Calculate The Prce Of An Ndex Opton Prcng ndex optons n a multvarate Black & Scholes model Danël Lnders AFI_1383 Prcng ndex optons n a multvarate Black & Scholes model Danël Lnders Verson: October 2, 2013 1 Introducton In ths paper, we consder

More information

Forecasting the Demand of Emergency Supplies: Based on the CBR Theory and BP Neural Network

Forecasting the Demand of Emergency Supplies: Based on the CBR Theory and BP Neural Network 700 Proceedngs of the 8th Internatonal Conference on Innovaton & Management Forecastng the Demand of Emergency Supples: Based on the CBR Theory and BP Neural Network Fu Deqang, Lu Yun, L Changbng School

More information

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

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

More information

Power law distribution of dividends in horse races

Power law distribution of dividends in horse races EUROPHYSICS LETTERS 15 February 2001 Europhys. Lett., 53 (4), pp. 419 425 (2001) Power law dstrbuton of dvdends n horse races K. Park and E. Domany Department of Physcs of Complex Systems, Wezmann Insttute

More information

When Talk is Free : The Effect of Tariff Structure on Usage under Two- and Three-Part Tariffs

When Talk is Free : The Effect of Tariff Structure on Usage under Two- and Three-Part Tariffs 0 When Talk s Free : The Effect of Tarff Structure on Usage under Two- and Three-Part Tarffs Eva Ascarza Ana Lambrecht Naufel Vlcassm July 2012 (Forthcomng at Journal of Marketng Research) Eva Ascarza

More information

Coordinated Denial-of-Service Attacks in IEEE 802.22 Networks

Coordinated Denial-of-Service Attacks in IEEE 802.22 Networks Coordnated Denal-of-Servce Attacks n IEEE 82.22 Networks Y Tan Department of ECE Stevens Insttute of Technology Hoboken, NJ Emal: ytan@stevens.edu Shamk Sengupta Department of Math. & Comp. Sc. John Jay

More information

Internalization, Clearing and Settlement, and Stock Market Liquidity 1

Internalization, Clearing and Settlement, and Stock Market Liquidity 1 Internalzaton, Clearng and Settlement, and Stock Market Lqudty 1 Hans Degryse, Mark Van Achter 3, and Gunther Wuyts 4 May 010 1 We would lke to thank partcpants at semnars n Louvan, Mannhem, and Tlburg

More information

REGULAR MULTILINEAR OPERATORS ON C(K) SPACES

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

More information

Simple Interest Loans (Section 5.1) :

Simple Interest Loans (Section 5.1) : Chapter 5 Fnance The frst part of ths revew wll explan the dfferent nterest and nvestment equatons you learned n secton 5.1 through 5.4 of your textbook and go through several examples. The second part

More information

How Bad are Selfish Investments in Network Security?

How Bad are Selfish Investments in Network Security? 1 How Bad are Selfsh Investments n Networ Securty? Lbn Jang, Venat Anantharam and Jean Walrand EECS Department, Unversty of Calforna, Bereley {ljang,ananth,wlr}@eecs.bereley.edu Abstract Internet securty

More information

Availability-Based Path Selection and Network Vulnerability Assessment

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

More information

Number of Levels Cumulative Annual operating Income per year construction costs costs ($) ($) ($) 1 600,000 35,000 100,000 2 2,200,000 60,000 350,000

Number of Levels Cumulative Annual operating Income per year construction costs costs ($) ($) ($) 1 600,000 35,000 100,000 2 2,200,000 60,000 350,000 Problem Set 5 Solutons 1 MIT s consderng buldng a new car park near Kendall Square. o unversty funds are avalable (overhead rates are under pressure and the new faclty would have to pay for tself from

More information