PORTFOLIO OPTIMIZATION WITH JUMPS AND UNOBSERVABLE INTENSITY PROCESS

Size: px
Start display at page:

Download "PORTFOLIO OPTIMIZATION WITH JUMPS AND UNOBSERVABLE INTENSITY PROCESS"

Transcription

1 PORTFOLIO OPTIMIZATION WITH JUMPS AND UNOBSERVABLE INTENSITY PROCESS Nicole Bäuerle Insiue for Mahemaical Sochasics, Universiy of Karlsruhe, Germany Ulrich Rieder Deparmen of Opimizaion and Operaions Research, Universiy of Ulm, Germany We consider a financial marke wih one bond and one sock. The dynamics of he sock price process allow jumps which occur according o a Markov-modulaed Poisson process. We assume ha here is an invesor who is only able o observe he sock price process and no he driving Markov chain. The invesor s aim is o maximize he expeced uiliy of erminal wealh. Using a classical resul from filer heory i is possible o reduce his problem wih parial observaion o one wih complee observaion. Wih he help of a generalized Hamilon- Jacobi-Bellman equaion where we replace he derivaive by Clarke s generalized gradien, we idenify an opimal porfolio sraegy. Finally, we discuss some special cases of his model and prove several properies of he opimal porfolio sraegy. In paricular we derive bounds and discuss he influence of uncerainy on he opimal porfolio sraegy. Key words: jump-diffusion process, filering, uiliy maximizaion, sochasic conrol, generalized HJB equaion, opimal porfolio sraegies, Bayesian conrol, sochasic comparison 1 Inroducion We consider an incomplee financial marke wih one bond and one sock. The sock price process allows for jumps of a random heigh where he jump ime poins are generaed by a Markov-modulaed Poisson process. There is an invesor who wans o maximize his uiliy from erminal wealh and who is only able o observe he sock price process. In paricular he is no informed abou he sae of he coninuous-ime Markov chain which drives he jump inensiy. Such a model is also called a Hidden Markov Model. For a general reamen of such models see e.g. Ellio e al. (1994). A model wih unobservable inensiy process is naural, since jumps in he sock price process are ofen generaed by various exernal evens whose impac on he sock marke Acknowledgemen: We are graeful o an anonymous referee for careful reading. Address correspondence o Nicole Bäuerle, Insiue for Mahemaical Sochasics, Universiy of Karlsruhe, D Karlsruhe, Germany, baeuerle@soch.uni-karlsruhe.de

2 canno compleely be analyzed. We are only able o draw some conclusions abou he jump inensiy from he observaion of he sock prices. Also i is more appropriae o allow a sochasically varying jump inensiy since a deerminisic jump inensiy seems only o be realisic for a shor period of ime. A coninuous-ime Markov chain can model he changing condiions which give rise o a changing jump behavior. This underlying Markov chain can be inerpreed as an environmen process which collecs facors which are relevan for he sock price dynamics like echnical progress, poliical siuaions, law or naural caasrophes. There is an exensive lieraure on porfolio opimizaion wih parial observaion as well as on porfolio opimizaion wih disconinuous sock price processes. In his paper we will rea hese wo aspecs in one model. Mos papers on problems wih parial observaion deal wih he case of an unobserved (sochasic) appreciaion rae process (µ ). Lakner (1995, 1998) for example reas he case where he appreciaion rae follows a linear Gaussian model. The mos recen papers by Honda (23), Sass and Haussmann (24), Haussmann and Sass (24) and Rieder and Bäuerle (25) consider a Hidden Markov Model for (µ ). We refer he reader o hese papers for a recen survey on financial models wih parial observaion. Of course i would be more realisic o assume ha boh he appreciaion rae and he jump inensiy depend on he hidden Markov chain bu his seems o be oo challenging a he momen. For a risk averse invesor i migh be more imporan o model an unobserved jump inensiy han an unobserved appreciaion rae, since poenial losses due o jumps can be much higher. In order o solve hese problems he usual echnique is o use he well-esablished filer heory o reduce he sochasic conrol problem wih parial observaion o one wih complee observaion. I is hen possible o solve his problem eiher wih sochasic conrol mehods or via he maringale approach (which is mosly done in case of a complee marke in he lieraure). On he oher hand here exis several papers on porfolio opimizaion problems wih disconinuous sock price processes, in paricular in he case where he price is modelled wih he help of a Lévy process. Empirical work has shown ha logreurns are in general no normally disribued and ha sock price models should conain a jump componen. In Framsad e al. (1999), he auhors deal wih he problem of opimal consumpion and porfolio selecion in a model where he sock price follows a geomeric Lévy process. They assume a power uiliy and solve he problem explicily by showing ha he value funcion is a classical soluion of he associaed Hamilon-Jacobi-Bellman equaion. Benh e al. (21) consider a similar quesion in he case ha he sock price is given by an exponenial Lévy process. They have o use he noion of a consrained viscosiy soluion o characerize heir soluion. As we will see laer, our price process canno be wrien as a funcional of a Lévy process. Imporan applicaions of opimizaion problems wih jumps are well-known in risk heory and insurance mahemaics (see e.g. Hipp and Plum (23), Schmidli (22)). Bu in hese papers he inensiy process is always observable. In his paper we combine he jump diffusion model wih an unknown jump inensiy. Our main conribuions are a non-sandard approach o solve he sochasic conrol problem by a generalized Hamilon-Jacobi-Bellman (HJB) equaion which migh be ineresing for oher porfolio problems as well and moreover, a sudy of he influence of uncerainy on he opimal porfolio sraegy. The ouline of he paper is as follows. In Secion 2 we give a precise mahemaical formulaion of our model and define he opimizaion problem. In Secion 3 we show how we can use filering heory o reduce he problem o one wih complee observaion. The reduced marke model we end up is no complee. Thus, he maringale approach canno be 2

3 applied direcly. In he case of a logarihmic uiliy funcion, i is shown in secion 4 ha he porfolio opimizaion problem can be solved (as usual) raher easily by a pahwise opimizaion. Secion 5 deals wih he power uiliy. Here we use he heory of sochasic conrol o solve he problem explicily. The value funcion is no a classical soluion of he corresponding Hamilon-Jacobi-Bellman equaion, however, we can characerize he value funcion as a soluion of a generalized HJB equaion where we use he Clarke generalized gradien. This is possible since he value funcion can be shown o be locally Lipschiz-coninuous and almos convex. This approach is non-sandard and has he advanage ha he opimal porfolio sraegy can be given raher explicily. Mos ineresing is he fac ha he expression for he opimal porfolio sraegy includes he value funcion bu no derivaive of i. Secion 5.1 conains he main resuls. In Secion 5.2 we deal wih some special cases and derive some imporan properies and comparison resuls for he opimal porfolio sraegy. In paricular we highligh he role of uncerainy in our model. I urns ou ha adding more uncerainy by jumps reduces he invesmen in he sock for all invesors wih power uiliy. Moreover, we look a he Bayesian model, i.e. when he jump inensiy is consan bu unobservable. We are able o derive bounds and compare he opimal porfolio sraegy in his case wih he sraegy we obain in a model where he consan jump inensiy is equal o he esimaed one. The comparison depends on wheher jumps go upwards or downwards and on he parameer of he power uiliy funcion. In his case furher uncerainy does no auomaically lead o a smaller invesmen in he sock. Some auxiliary resuls which are needed for he proof of our main heorems are given in Secion 6. Secion 7 finally conains he proofs of our main heorems in he case of a power uiliy. 2 The Model We consider a financial marke wih one bond and one risky asse. More precisely le (Ω, F, F = {F, T }, P ) be a filered probabiliy space. T > is a fixed ime horizon. The bond price process (B ) evolves according o db = rb d wih ineres rae r > and he sock price process evolves according o ) ds = S (µd + σdw + yn(d, dy), where σ >, µ IR are given consans and (W ) is a Brownian moion w.r.. F. N is he random couning measure of a Markov-modulaed compound Poisson process. Tha means N is consruced as follows: le us denoe by (Y ) a coninuous-ime Markov chain wih sae space {e 1,..., e d } where e k is he k-h uni vecor in IR d and (Y ) has he generaor Q = (q ij ). q ij is he inensiy of geing from sae e i in sae e j. Furher le us denoe by T =, T 1, T 2,... he jump ime poins of a Poisson process wih F-inensiy (λ ) := (λ Y ) where λ = (λ 1,..., λ d ) IR d +, i.e. as long as Y = e j, jumps arrive a rae λ j. Finally we assume ha (δ n ) is a sequence of independen and idenically disribued random variables bounded from above wih δ n > 1. The probabiliy disribuion of δ n is denoed by Q. Then we have N = n 1 ɛ (Tn,δ n) 3

4 where ɛ x is he one poin measure in x. δ n is he relaive jump heigh of he sock a ime T n. Noe ha he resricion δ n > 1 guaranees ha he sock price says posiive. The compound Poisson process which describes he jumps is hen obained by yn(ds, dy) = The economic inerpreaion of (Y ) is some kind of environmen process which collecs facors which are relevan for he sock price dynamics like e.g. echnical progress, poliical siuaions, laws or naural caasrophes. These facors change sochasically over ime. All processes are adaped w.r.. F and (W ) and N are independen as well as (δ n ) and (Y ). In wha follows we assume an invesor who is only able o observe he sock price process and who knows he disribuion of Y. This means ha he invesor is no informed abou he inensiy wih which he sock price process jumps. Of course i is more realisic o assume ha he appreciaion rae of he sock price also depends on he unobservable environmen process (Y ), however his case is much more challenging. The case of unobservable Markovmodulaed appreciaion rae has been invesigaed in Honda (23), Sass and Haussmann (24), Haussmann and Sass (24) and Rieder and Bäuerle (25) among ohers. Le F S = (F S ) be he filraion generaed by he sock price process (S ). Our aim is o solve he opimizaion problem our invesor faces, when he ries o find porfolio sraegies ha maximize he expeced uiliy from erminal wealh. We resric ourselves o self-financing porfolio sraegies and denoe by π [, 1] he fracion of he wealh invesed in he sock a ime. The resricion of he fracion o [, 1] ( no shor-sellings ) guaranees a posiive wealh process which is reasonable for logarihmic and power uiliy. Due o he jumps of he sock price a violaion of his resricions may lead o a negaive wealh wih posiive probabiliy. The process π = (π ) is called porfolio sraegy. An admissible porfolio sraegy has o be F S -predicable and akes values in [, 1]. Thus, we inroduce he se U[, T ] := {π = (π s ) s T π s [, 1] for all s [, T ], π is F S predicable.}. The wealh process under an admissible porfolio sraegy π U[, T ] is given by ) d X π = X ((r π + (µ r)π )d + σπ dw + π yn(d, dy). We assume ha X π = x is he given iniial wealh. Le U : IR + IR be an increasing, concave uiliy funcion. Then we define he value funcions for π U[, T ], [, T ], x > by [ Ṽ π (, x) := E,x U( X ] T π ) F S N n=1 δ n. Ṽ (, x) := sup Ṽ π (, x) π U[,T ] where he expecaion is aken w.r.. he probabiliy measure P,x wih X π = x. Noe ha Ṽ π (, x) and Ṽ (, x) are random variables, in paricular Ṽπ(, x) is F S -measurable. Moreover, Ṽ π (, x ) and Ṽ (, x ) depend on he disribuion of Y which is fixed. A porfolio sraegy π U[, T ] is opimal if Ṽ (, x ) = Ṽπ (, x ). 4

5 We have chosen he appreciaion rae and he volailiy o be consan. All he analysis which follows can be done in a similar way if hey are modelled by bounded, deerminisic (observable) processes. 3 The Reducion We can reduce he conrol problem o one wih complee observaion. This procedure is classical. The idea is o updae our belief abou he disribuion of he environmen sae Y coninuously and make i par of our sae space. Noe ha only he jump ime poins of he sock price conain relevan informaion for esimaing he environmen sae and hus he jump inensiy. This coninuous esimaion is done by he so-called Wonham filer. We proceed as in Brémaud (1981) p. 94 ff. Define p k () = P (Y = e k F S ), k = 1,..., d and p = (p 1 (),..., p d ()). p k () is he probabiliy ha he environmen process is in sae k a ime, given ha we have observed he sock price process unil ime. The process (p ) is called filer process. Recall form secion 2 ha he Markov-modulaed Poisson process is given by N(ds, dy) and couns he number of jumps in he sock price unil ime. This couning process has F-inensiy (λ ) = (λ Y ) which is equivalen o η := N(ds, dy) λ s ds being an F-maringale. The following saemens hold: Lemma 3.1: There exiss an F S maringale (ˆη ) such ha a) he filer processes p k () saisfy for dp k () = j ( λk q jk p j ()d + p k ( ) ˆλ ) dˆη ˆλ wih ˆλ := d k=1 λ k p k () = E[λ F S ]. b) λ d + dη = ˆλ d + dˆη. c) (W ) and (ˆη ) are independen. Par b) of Lemma 3.1 saes ha he Markov-modulaed Poisson process N(ds, dy) admis an F S -inensiy (ˆλ ) and can be compensaed in order o obain an F S -maringale ˆη := N(ds, dy) ˆλ s ds. 5

6 In wha follows we also need he compensaed random measure Noe ha ˆM(d, dy) = N(d, dy) ˆλ dq(dy). f(s, y)n(ds, dy) f(s, y)ˆλ s dsq(dy) is a maringale for arbirary f whenever he inegrals exis. The conrol model wih complee observaion is now characerized for π U[, T ] by he following d+1-dimensional sae process: ( ( ) dx π = X π r + π (µ r) + π ˆλ yq(dy) )d + σπ dw + π y ˆM(d, dy) X π = x dp k () = ( λk q jk p j ()d + p k ( ) ˆλ ) dˆη j ˆλ p k () = P (Y = k), k = 1,..., d where P (Y = k), k = 1,..., d, is he given disribuion of Y. A soluion of he sochasic differenial equaion for he wealh process is given by { X π = x exp (r + (µ r)π s 1 } 2 σ2 πs)ds 2 + σπ s dw s + ln(1 + π s y)n(ds, dy) { = x exp (r + (µ r)π s 1 2 σ2 πs 2 + ˆλ ) s ln(1 + π s y)q(dy) ds + σπ s dw s + ln(1 + π s y) ˆM(ds, dy) } By d we denoe he probabiliy simplex in IR d. The value funcions in he reduced model are for π U[, T ] and p d, [, T ], x > defined by V π (, x, p) := E,x,p [U(X π T )]. V (, x, p) := sup V π (, x, p) π U[,T ] where E,x,p is he condiional expecaion, given X π = x, p = p. The reduced model now solves our original problem. This is ofen aken for graned, however i has o be proved formally. The nex heorem saes ha he filer conains he necessary informaion in order o solve our original problem. Insead of he whole hisory F S i is sufficien o know p. More precisely, Ṽ (, x) depends on he hisory F S only hrough p. Theorem 3.2: For all π U[, T ] i holds ha V π (, x, p ) = Ṽπ(, x) and V (, x, p ) = Ṽ (, x) for all x >, [, T ]. Proof: From Lemma 3.1 and he sochasic differenial equaion for he wealh process i follows ha X T π = Xπ T a.s. for all π U[, T ] which obviously yields he saemen. In he reduced model, all processes are F S -adaped and admi an F S -inensiy respecively. 6

7 Therefore, we can solve his problem by sochasic conrol echniques. The following properies are easily derived. Lemma 3.3: a) For all π U[, T ], p d and x > we have d V π (, x, p) = p j V π (, x, e j ). j=1 b) The mapping p V (, x, p) is convex for all [, T ] and x >. Proof: Par a) is obained by condiioning. For b) le p, q d be wo iniial disribuions and α [, 1]. Then V (, x, αp + (1 α)q) = sup α V π (, x, e j )p j + (1 α) V π (, x, e j )q j π U[,T ] j j α sup V π (, x, e j )p j + (1 α) sup V π (, x, e j )q j π U[,T ] j π U[,T ] j = αv (, x, p) + (1 α)v (, x, q). 4 Logarihmic Uiliy In his secion we briefly summarize he resuls in he case of a logarihmic uiliy funcion U(x) = log(x). This is always he easies case. For π U[, T ] we obain from he explici soluion for X π V π (, x, p) = log(x) + h π (, p) where [ T h π (, p) = E,p r + (µ r)π s 1 2 σ2 πs 2 + ˆλ s log(1 + π s y)q(dy)ds ]. Noe ha h π does no depend on x. Obviously we obain he following resul: Lemma 4.1: a) For all [, T ], x >, p d we have V (, x, p) = log(x) + h(, p), where h(, p) = sup π U[,T ] h π (, p). 7

8 b) Suppose ha for all p d, u (p) maximizes u r + (µ r)u 1 2 σ2 u 2 + λ p log(1 + yu)q(dy) on [, 1] hen π = (π ) U[, T ] wih π = u (p ) is an opimal porfolio sraegy for he given porfolio problem. Noe ha π depends on F S only hrough p. I is easy o show ha in he case of complee observaion, i.e. when we know ha he sae of he Markov chain is for example e i, he opimal porfolio sraegy would be o inves a consan fracion u of he wealh in he sock, where u is he maximizer of u r + (µ r)u 1 2 σ2 u 2 + λ i log(1 + yu)q(dy) on [, 1]. Par b) of Lemma 4.1 shows ha he so-called cerainy equivalence principle holds, i.e. he unknown inensiy λ is replaced by he esimae ˆλ = E[λ F S ] in he opimal porfolio sraegy (cf. Kuwana (1991)). This means ha uncerainy abou he jump inensiy does no change he opimal porfolio sraegy in his case. The siuaion is compleely differen in he case of a power uiliy funcion as we will see in secion Power Uiliy In his secion we assume ha he uiliy funcion is given by U(x) = 1 γ xγ for γ < 1, γ. The value funcion under sraegy π U[, T ] is herefore where V π (, x, p) = 1 γ xγ g π (, p), [ { g π (, p) = E,p T exp γ(r + (µ r)π s 1 T 2 σ2 πs)ds 2 + γσπ s dw s T +γ ln(1 + π s y)n(ds, dy)} ]. Noe ha g π does no depend on x. If we define (1) hen i obviously holds ha g(, p) := sup g π (, p) π U[,T ] V (, x, p) = 1 γ xγ g(, p). 8

9 5.1 The Soluion of he Porfolio Opimizaion Problem In his secion we summarize he main resuls. We use a sochasic conrol approach o solve he problem. Unforunaely i is no clear wheher he value funcion is coninuously differeniable in p and and we hus are no able o obain a classical soluion for he associaed Hamilon- Jacobi-Bellman (HJB) equaion. The sandard way would hen be o show ha he value funcion is he unique viscosiy soluion of he HJB equaion. However his ype of soluion is quie weak and he uniqueness proof can be hard. In our seing i is possible o show ha he value funcion is locally Lipschiz-coninuous and hus almos everywhere differeniable. This is much more han coninuiy which is required for he viscosiy soluion. Therefore we decided o pursue a differen approach by considering a generalized HJB equaion where he classical derivaive is replaced by a generalized derivaive. A similar approach has been used by Davis (1993) for piecewise deerminisic models. We can show ha he value funcion is he unique soluion of he generalized HJB equaion and he maximizer yields an opimal porfolio sraegy. In his secion we only presen he resuls, proofs are posponed o secion 7. For he analysis, i is imporan o noe ha (p ) is a piecewise deerminisic process wih jumps appearing according o he F S -inensiy (ˆλ ). We denoe by φ k (, p ) = p k () + j q jk p j (s) p k (s)(λ k ˆλ s )ds, k = 1,..., d and φ(, p ) = (φ 1 (, p ),..., φ d (, p )) he evoluion of he filer beween jumps and by ( λ1 p 1 J(p) = λ p,..., λ ) dp d λ p he new sae of he filer direcly afer a jump from sae p. Moreover, we le S d be he inerior of he probabiliy simplex d. In order o obain a reasonable model we assume now ha all saes of he Markov chain Y communicae. Thus, he filer process p will for > always say in S d. Le us inroduce he following operaor L which acs on funcions v : [, T ] S d IR and u [, 1] Lv(, p, u) := v(, p)(r + (µ r)u (γ 1)σ2 u 2 ) ( ) + λ p v(, J(p)) (1 + yu) γ Q(dy) v(, p). γ In order o moivae he HJB equaion of his problem, we give some heurisic argumens. For his purpose, suppose ha he value funcion V is sufficienly differeniable. An applicaion of Io s Lemma gives: T T V (T, XT π, p T ) = V (, x, p) + V (s, Xs π, p s )ds + V x (s, Xs π, p s )dxs π d T + V pk (s, Xs π, p s )dp k (s) + 1 T V xx (s, Xs π, p s )σ 2( ) 2π X π 2 s k=1 2 s ds + [V (s, Xs π, p s ) V (s, Xs, π p s )] V x (s, Xs π, p s ) Xs π <s d V pk (s, Xs π, p s ) p k (s). k=1 9

10 I can be shown ha V (, X π, p ) is a maringale under he opimal porfolio sraegy and a supermaringale under any admissible sraegy. Thus, he drif erms in he preceding equaion have o be zero. Moreover, plugging in he form V (, x, p) = 1 γ xγ g(, p) yields as an opimaliy condiion: = 1 γ g (, p) + g(, p) (r + π(µ r) (γ 1)σ2 π 2) ( ) g(, J(p)) (1 + yπ) γ Q(dy) g(, p) + λ p γ + 1 d ( g pk (, p) γ k=1 j ) q jk p j p k (λ k λ p). However, since he value funcion (in paricular g defined in equaion (1)) is probably no differeniable w.r.. g k we replace he gradien by he Clarke generalized gradien. The resuling generalized Hamilon-Jacobi-Bellman equaion for our problem hen reads as follows = sup {Lg(, p, u)} + u [,1] { 1 sup θ g(,p) γ θ + 1 γ d ( )} θ k q jk p j p k (λ k λ p) wih boundary condiion g(t, p) = 1 for all p S d. The se g(, p) IR d+1 denoes he Clarke generalized gradien (see Clarke (1983)). This is a weaker noion for differeniabiliy which is defined as follows: le f : IR d IR be a locally Lipschiz coninuous funcion. For x, y IR d he upper generalized direcional derivaive of f a x in direcion y is defined by f (x; y) := lim sup z x,ε k=1 f(z + εy) f(z). ε The Clarke generalized gradien of f a x is now defined by he se f(x) := {θ IR n f (x; y) θy for all y IR d }. f(x) is a non-empy, convex, compac subse of IR d and if f is differeniable a x, hen f(x) := { f(x)}. Moreover, since f is locally Lipschiz coninuous, i is almos everywhere differeniable and we can find for every poin x IR d sequences of poins x n IR d such ha lim n x n = x and f is differeniable a x n. f(x) can hen be wrien as he closed convex hull of exising limis of sequences f(x n ), i.e. f(x) := co{lim sup f(x n ) n Our firs resul is a verificaion heorem: j lim x n = x}. n Theorem 5.1: Suppose here exiss a bounded funcion v : [, T ] S d IR + such ha for all p S d, v(, φ(, p)) is absoluely coninuous, v(t, p) = 1 and v saisfies he generalized HJB equaion. Furher assume ha u is a maximizer of he generalized HJB equaion, i.e. for all [, T ] and p S d, u (, p) maximizes u Lv(, p, u) on [, 1]. 1

11 Then V (, x, p) = 1 γ xγ v(, p) and he sraegy π = (π ) U[, T ] wih π := u (, p ) is an opimal feedback sraegy for he given porfolio problem. Noe ha π depends on F S only hrough p. The nex heorem saes he exisence of a soluion of he generalized HJB equaion. Theorem 5.2: The value funcion of our problem is given by V (, x, p) = 1 γ xγ g(, p) wih g defined by (1) above and g saisfies he generalized HJB equaion = sup {Lg(, p, u)} + u [,1] { 1 sup θ g(,p) γ θ + 1 γ d ( )} θ k q jk p j p k (λ k λ p) wih boundary condiion g(t, p) = 1 for all p S d. Moreover, π from Theorem 5.1 (wih v replaced by g) is an opimal porfolio sraegy. k=1 j 5.2 Special Cases and Properies of he Opimal Porfolio Sraegy In his secion we invesigae he opimal porfolio sraegy in some special cases in greaer deail and esablish some ineresing properies. In paricular we discuss he influence of uncerainy on he opimal porfolio sraegy. A) Jumps occur wih known and consan inensiy Suppose ha δ n δ ( 1, ) is deerminisic and ha he jumps in he sock price process occur wih known consan inensiy λ >, i.e. λ = λ 1 =... = λ d. This model is similar o he seup invesigaed in Øksendal and Sulem (24) and Framsad e al. (1999). In his case i is opimal o inves a consan fracion u δ (λ) (independen of ime) of he wealh in he sock. Specializing our HJB equaion (noe ha J(p) = p in his case), i is easy o see ha u δ (λ) is he maximum poin of he mapping u (µ r)u (γ 1)σ2 u 2 + λ γ (1 + δu)γ on [, 1]. In his case i can also be shown ha he value funcion is a classical soluion of he HJB equaion. In wha follows we wan o compare he opimal fracions which are invesed in he sock in differen models. In paricular we highligh he role of uncerainy. For his ask he following simple lemma is useful: Lemma 5.3: Le f, h : [, 1] IR be coninuous funcions and suppose ha h is increasing. If we denoe u f := argmax {f(u) u [, 1]} u f+h := argmax {f(u) + h(u) u [, 1]} 11

12 hen u f u f+h. Throughou he paper we use increasing and decreasing in he non-sric sense. A direc implicaion of he previous lemma is Lemma 5.4: If δ <, hen λ u δ (λ) is decreasing and if δ >, hen λ u δ (λ) is increasing. Of course his resul is no surprising. If we have downward jumps, we inves less in he sock if he jump inensiy increases. In order o invesigae he influence of furher uncerainy we have o add a jump maringale o he sock price o keep he expeced drif unchanged. Thus, suppose for a momen ha he sock price process evolves according o ds = S (ˆµd + σdw + δdη ), where η = N(ds, dy) λ. If we se ˆµ = µ+λδ we obain a sochasic differenial equaion for he sock price in he form given in secion 1. In he case wihou jumps (δ = ), we know ha he opimal fracion maximizes u (ˆµ r)u (γ 1)σ2 u 2 on [, 1]. In he case wih jumps (δ ), we know ha he opimal fracion maximizes u (ˆµ r)u (γ 1)σ2 u 2 + λ γ (1 + δu)γ λδu on [, 1]. Thus we obain he following comparison resul: Theorem 5.5: In he previous model we have u u δ(λ). Proof: In view of Lemma 5.3 i is sufficien o show ha h(u) := λ γ (1 + δu)γ λδu is decreasing for all δ > 1, δ. This can be done by showing ha h (u). Theorem 5.5 means ha he opimal fracion invesed in he sock in he model wih furher uncerainy coming from jumps is always less or equal o he opimal fracion in he model wihou jumps. Noe ha he expeced drif of he sock remains he same in boh scenarios. Since we have a risk averse invesor such a resul is no unexpeced. However also noe ha he saemen is rue for all γ < 1, γ. In B) we will observe a differen behavior. B) Jumps occur wih unknown and consan inensiy - he Bayesian case Suppose ha δ n δ ( 1, ) is deerminisic and ha he jumps in he sock price process occur wih unknown consan inensiy λ >. We assume ha λ can be one of he possible 12

13 values λ 1... λ d and ha he iniial probabiliy p S d for he values is given. Thus, we have a Bayesian conrol problem wih an unknown parameer. This is a special case of our model, if we formally se he inensiy marix of he Markov chain (Y ) o zero, i.e. Q = and he Markov chain says in he iniial sae. If we define p k () = P (Y = e k F S ) = P (λ = λ k F S ) and p = (p 1 (),..., p d ()), hen he following equaion holds ( λk p k () = p k () + p k (s ) ˆλ ) s dˆη s ˆλ s where (ˆη ) is defined as in Lemma 3.1. The opimal fracion π invesed in he sock depends on he ime and he esimae p, i.e π = u δ (, p ) and u δ maximizes u (µ r)u (γ 1)σ2 u 2 + λ p γ g(, J(p)) (1 + δu) γ on [, 1]. g(, p) I is possible o compare he opimal porfolio sraegy of his scenario wih he previous case A) of complee observaion. Theorem 5.6: The opimal fracion u δ (, p) invesed in he sock has he following properies: a) If δ < (downward jumps) i holds for all (, p) [, T ] S d ha u δ(λ d ) u δ(, p) u δ(λ 1 ). If δ > (upward jumps) he inequaliies are reversed. b) If δγ < i holds for all (, p) [, T ] S d ha If δγ > he inequaliy is reversed. u δ(λ p) u δ(, p). Proof: a) Suppose δ <. In view of Lemma 5.3 i suffices o show Recall ha λ p g(, J(p)) λ 1 g(, p) and λ p g(, J(p)) λ d g(, p). Now suppose π U[, T ] is fix. [, T ] and j. We obain d g(, p) = sup g π (, p) = sup p j g π (, e j ). π U[,T ] π U[,T ] j=1 Noe ha due o he definiion g π (, e j ) for all d d λ p g π (, J(p)) = p j λ j g π (, e j ) λ 1 p j g π (, e j ) = λ 1 g π (, p). j=1 j=1 Taking he supremum over all π U[, T ] hen yields he firs inequaliy. The case δ > and he second saemen obviously follows similarly. 13

14 b) Suppose δ < and < γ < 1. In view of Lemma 5.3 i suffices o show g(, J(p)) g(, p). Noe ha if p = e j, he couning process of jumps is simply a Poisson process wih inensiy λ j. Moreover, i is well-known ha if λ ˆλ >, hen a Poisson process wih inensiy λ pahwise sochasically dominaes a Poisson process wih inensiy ˆλ (see e.g. Sec in Müller and Soyan (22)). Thus, under an arbirary fixed π U[, T ] we have X π s ˆXπ where s is he usual sochasic order. Thus, he value funcion is decreasing in λ and we obain for all [, T ]. Thus, i follows ha g π (, e 1 )... g π (, e d ), d d d p j g π (, e j ) p j λ j p j λ j g π (, e j ) j=1 j=1 j=1 where his inequaliy is derived by applying he following general inequaliy (cf. Mironovic e al. (1993)): le α 1... α d and β 1... β d be real numbers and p 1,..., p d, d j=1 p j = 1. Then d d d p j α j p j β j p j α j β j. j=1 Taking he supremum over all π U[, T ] hen yields he saemen. j=1 j=1 Please noe ha in he case γ < we obain he inequaliy g π (, e 1 )... g π (, e d ), for all [, T ] since he value funcion is negaive. analogously. The case δ > can be shown Par a) of Theorem 5.6 means ha he opimal fracion which is invesed in he sock is bounded by he smalles and larges invesed fracion in he models wih known inensiy λ 1 and λ d. Par b) of his heorem is mos ineresing. For example in he case of downward jumps δ < and γ (, 1), he opimal fracion invesed in he sock in he model wih unknown jump inensiy in sae (, p) is larger han in he model wih known (average) inensiy λ p. Though our invesor is risk averse, his is a siuaion where more uncerainy leads o a higher invesmen in he risky sock. If γ < he siuaion is vice versa. An economic explanaion is ha he degree of risk aversion changes wih γ. From he Arrow-Pra absolue risk aversion coefficien which is U (x) U (x) = (1 γ) 1 x in he case of he power uiliy U(x) = 1 γ xγ, we see ha he risk aversion decreases wih γ for all wealh levels. If γ we obain he logarihmic uiliy case and we know from secion 4 ha here he opimal fracions invesed coincide, i.e. u δ (λ p) = u δ (, p). In paricular if γ (, 1) he invesor is less risk averse. A similar resul has been obained for a model wih unobservable appreciaion rae in Rieder and Bäuerle (25). 14

15 6 Auxiliary Resuls In his secion we summarize some resuls which are imporan for he proofs of our main heorems. Lemma 6.1 summarizes imporan properies of he funcion g defined in (1) which is par of he value funcion V. Lemma 6.1: Le g be defined by (1) in Secion 5. a) p g(, p) is convex for all [, T ]. b) g(, p) is decreasing (increasing) for all p S d if < γ < 1, (γ < ). c) g(, p) is bounded on [, T ] S d. d) g(, p) is locally Lipschiz-coninuous for all p S d. e) g(, φ(, p)) is locally Lipschiz-coninuous for all p S d. Proof: a) follows from Lemma 3.3 b). b) This is equivalen o showing ha V (, x, p) is decreasing. Bu his is clear since because of r > we ge a posiive reward over a small ime inerval by puing all he money in he sock. c) For γ < he saemen is obvious due o par b) and he fac ha g(, p) and g(t, p) = 1. For γ (, 1) i suffices o show ha g(, p) is bounded on S d. I is convenien o inroduce a new measure Q π by dq π = L π T dp, where π U[, T ] and Lπ is a soluion of he sochasic differenial equaion ( ) dl π = L π γσπ dw + ((1 + yπ ) γ 1) ˆM(d, dy) where ˆM(d, dy) := N(d, dy) ˆλ dq(dy) is he compensaed random measure defined before. The soluion is given by { T ( L π T = exp 1 2 γ2 σ 2 πs 2 ˆλ ) T s ((1 + yπ s ) γ 1)Q(dy) ds + γσπ s dw s T +γ I is easy o see ha for π U[, T ] [ g π (, p) = E,p Q π ln(1 + yπ s )N(ds, dy) { T ( exp γ r + (µ r)π s + 1 ) 2 (γ 1)σ2 πs 2 +ˆλ s ((1 + yπ s ) γ 1)Q(dy)ds} ]. Since π, δ n and ˆλ = λ p are bounded, i follows from his equaion ha g π (, p) is bounded on [, T ] S d and he bound is independen of π. }. 15

16 d) In his par we make he dependence of g on he ime horizon explici by wriing g π,t (, p). Firs noe he following: if π U[, T ] we define ˆπ by ˆπ s = π +s for s [, T ] which implies ha g π,t (, p) = gˆπ,t (, p). Now le 1 < 2 T. Then here exiss for every ε > a sraegy π U[, T 1 ] wih g( 1, p) g( 2, p) g π,t 1 (, p) g π,t 2 (, p) + ε π[ T { 1 K E,p Q exp T 2 γ ( r + (µ r)π s + 1 ) 2 (γ 1)σ2 πs 2 ] } +ˆλ s ((1 + yπ s ) γ 1)Q(dy)ds 1 + ε T 1 K 1 E,p Q π γ 2r + µ + σ 2 + λ ((1 + yπ s ) γ + 1)Q(dy)ds + ε T 2 K ε where λ = max k λ k. This implies he saemen if we le ε. Noe ha K 2 can be chosen independen of p and π. e) Le 1 < 2 T. Then g( 2, φ( 2, p)) g( 1, φ( 1, p)) = g( 2, φ( 2, p)) g( 2, φ( 1, p)) + g( 2, φ( 1, p)) g( 1, φ( 1, p)) g( 2, φ( 2, p)) g( 2, φ( 1, p)) + g( 2, φ( 1, p)) g( 1, φ( 1, p)). Since g is convex in p i is also locally Lipschiz-coninuous in p S d wih a module K 3 which can be chosen independen of (see e.g. Sec. 1 in Rockafellar (197)). Therefore we obain g( 2, φ( 2, p)) g( 1, φ( 1, p)) K 3 φ( 2, p) φ( 1, p) + K Bu obviously φ(, p) is also locally Lipschiz-coninuous wih a module independen of p which yields he resul. Recall ha = T < T 1 < T 2 <... are he jump ime poins of he Markov-modulaed Poisson process. Since g(, φ(, p)) is locally Lipschiz-coninuous, here exiss a funcion Dg(s, p s ) such ha g(t i, p Ti ) g(t i 1, p Ti 1 ) = Ti T i 1 Dg(s, p s )ds. Almos everywhere on he ime inerval [, T ], he derivaive of g(s, p s ) w.r.. s exiss and we can choose Dg(s, p s ) = g (s, p s ) + k g p k (s, p s ) φ k (s, p). Le us define he operaor Hv(, p, u) := Lv(, p, u) + 1 Dv(, p) γ 16

17 for all funcions v : [, T ] S d IR where he righ-hand side is well defined. Noe ha he HJB equaion can be wrien as = sup {Hv(, p, u)} = sup {Lv(, p, u)} + 1 Dv(, p) u [,1] u [,1] γ a hose poins (, p) where v is differeniable. Lemma 6.2: Suppose ha π U[, T ] is an arbirary sraegy. The value funcion V saisfies he following sochasic differenial equaion dv (, X π, p ) = (X π ) γ Hg(, p, π )d + dη π, where (η π ) is an F S -maringale wih zero expecaion. Proof: Le π U[, T ] be arbirary. Using Io s Lemma we can verify ha Z π := (X π ) γ saisfies he following sochasic differenial equaion dz π = γz π (r + (µ r)π (γ 1)σ2 π 2 )d +γz π σπ dw + Z π ((1 + yπ ) γ 1)N(d, dy). Moreover, since g(, φ(, p)) is absoluely coninuous, we can wrie g(, p ) = g(, p ) + Dg(s, p s )ds + g(s, p s ) g(s, p s ). <s Since V (, X π, p ) = 1 γ Zπ g(, p ), he produc rule implies Le us define V (, X π, p ) = V (, x, p) + + g(s, p s )Zs π σπ s dw s + η π,1 := 1 γ 1 γ + 1 γ <s <s g(s, p s )Z π s (r + (µ r)π s (γ 1)σ2 π 2 s)ds Zs π Z π s g(s, p s ) Z π s g(s, p s ). Z π s g(s, p s ) Z π s g(s, p s ) Zs πˆλ s (g(s, J(p s )) 1 γ Dg(s, p s)ds ) (1 + yπ s ) γ Q(dy) g(s, p s ) ds. According o Brémaud p.27, T8, (η π,1 ) is an F S maringale, since [ ] T E Zs π g(s, p s ) Zs g(s, π p s ) λ p s ds <. 17

18 Moreover, due o our boundedness condiions η π,2 := g(s, p s )Zs π σπ s dw s is also an F S -maringale. If we define now η π := η π,1 + η π,2 he saemen follows. Theorem 6.3: Bellman equaion Le τ [, T ] be an F S -sopping ime. Then V (, x, p) = sup E,x,p [V (τ, Xτ π, p τ )]. π U[,T ] The proof of he Bellman equaion follows he usual recipe and we skip i here. 7 Proofs for he Resuls in Secion 5.1 In his secion we provide he proofs of he Verificaion Theorem and he fac ha V is a soluion of he generalized HJB equaion. Proof of Theorem 5.1: Le π U[, T ] be an arbirary sraegy. Then we obain for Z π := (X π ) γ and G(, x, p) := 1 γ xγ v(, p) as in Lemma 6.2 T G(T, XT π, p T ) = G(, x, p) + Zs π Hv(s, p s, π s )ds + ηt π η π where (η π ) is an F S -maringale wih zero expecaion. A hose poins where v(s, p s ) is differeniable we have Hv(s, p s, π s ) since v saisfies he generalized HJB equaion. Moreover, s v(s, p s ) is almos everywhere differeniable which yields Thus, we obain T Z π s Hv(s, p s, π s )ds. G(T, X π T, p T ) G(, x, p) + η π T η π. Taking he condiional expecaion on boh sides (noe ha G(T, XT π, p T ) = U(XT π )) yields: E,x,p [U(X π T )] G(, x, p). Taking he supremum over all admissible sraegies gives V (, x, p) G(, x, p). Nex, noe ha he maximum poins of he HJB equaion rivially exis. If we use π we obain T Z π s Hv(s, p s, π s)ds = 18

19 and he resul follows. Proof of Theorem 5.2: Le τ be he ime of he firs jump of he sock price process (S ) afer ime. From he Bellman equaion (Theorem 6.3) we obain for every sraegy π U[, T ] and < T : V (, x, p) E,x,p [ V (τ, X π τ, p τ )]. From Lemma 6.2 we know ha for Z π = (X π ) γ τ V (τ, Xτ π, p τ ) = V (, x, p) + Zs π Hg(s, p s, π s )ds + ητ π ηπ. Insering his equaion in he preceding inequaliy yields [ ] τ E,x,p Zs π Hg(s, p s, π s )ds. Le π be now a fixed sraegy wih π s u [, 1] for s [, + ε), ε >. Thus, we ge [ ] 1 τ lim E,x,p Zs π Hg(s, p s, π s )ds = [ ] 1 = lim E,x,p Z π s Hg(s, p s, π s )ds < τ P ( < τ) + [ 1 τ ] + lim E,x,p Zs π Hg(s, p s, π s )ds τ P ( τ) Since P (τ ) 1 e λ( ) for, where λ = max k λ k, we obain a hose poins (, p) where g is differeniable Z π Hg(, p, u). From he definiion we see ha Z π > which yields Hg(, p, u). Le now (, p) be an arbirary poin, where g migh no be differeniable. We know ha g(, p) = co{lim sup n g( n, p n ), n } which means by definiion ha every θ g(, p) is a convex combinaion of θ m = lim sup n g( m n, p m n ) for sequences m n, along which g is differeniable. Since g is coninuous, we obain Lg(, p; u) + 1 γ θm + 1 d ( θ m k q jk p j p k (λ k γ ˆλ) ) k=1 j which yields he same inequaliy wih θ. Finally since u and θ are arbirary, we obain sup {Lg(, p; u)} + u [,1] { 1 sup θ g(,p) γ θ + 1 γ d ( θ k q jk p j p k (λ k ˆλ) )}. k=1 j On he oher hand, for ε > and < < T wih > small enough here exiss a sraegy π ε, U[, T ] wih [ ] V (, x, p) ε( ) E,x,p V (τ, Xτ πε,, p τ ). 19

20 Again wih Lemma 6.2 we obain ε( ) E,x,p [ τ Z πε, s Hg(s, p s, πs ε, )ds ]. Thus, we ge [ 1 τ ε E,x,p Zs πε, Hg(s, p s, π s ε, )ds [ 1 τ E,x,p Zs πε, sup Hg(s, p s, u)ds u [,1] ] ]. Denoe now by u () he maximum poin of u Lg(, φ(, p), u) on [, 1]. Since u () is coninuous we obain a hose poins (, p) where g is differeniable and since ε > is arbirary ε x γ sup Hg(, p, u) u [,1] sup Hg(, p, u). u [,1] The analysis if g is no differeniable a (, p) follows in he same way as before by using he convexiy of g(, p). Alogeher i follows ha V saisfies he generalized HJB equaion. References [1] Benh, F.E., K.H. Karlsen and K. Reikvam (21): Opimal porfolio selecion wih consumpion and nonlinear inegro-differenial equaions wih gradien consrain: a viscosiy soluion approach, Finance Soch. 5, [2] Brémaud, P. (1981): Poin processes and queues. Springer-Verlag, New York. [3] Clarke, F.H. (1983): Opimizaion and nonsmooh analysis. John Wiley & Sons, New York. [4] Davis, M. H. A. (1993): Markov models and opimizaion. Chapman & Hall, London. [5] Ellio, R.J., L. Aggoun and J. B. Moore (1994): Hidden Markov models: esimaion and conrol. Springer-Verlag, New York. [6] Framsad, N.C., B. Øksendal and A. Sulem (1999): Opimal consumpion and porfolio in a jump diffusion marke, in: Shiryaev, A. e al. (eds.) Workshop on mahemaical finance. Paris: INRIA, 9-2. [7] Haussmann U.G. and J. Sass (24): Opimal erminal wealh under parial informaion for HMM sock reurns, in Mahemaics of Finance (Conemp. Mah. 351), AMS, Providence, [8] Honda, T. (23): Opimal porfolio choice for unobservable and regime-swiching mean reurns, J. Econ. Dyn. Conr. 28,

21 [9] Hipp C. and M. Plum (23): Opimal invesmen for invesors wih sae dependen income, and for insurers. Finance Soch., 7, [1] Kuwana, Y. (1991): Cerainy equivalence and logarihmic uiliies in consumpion /invesmen problems. Mahem. Finance, 5, [11] Lakner, P. (1995): Uiliy maximizaion wih parial informaion. Sochasic Process. Appl., 56, [12] Lakner, P. (1998): Opimal rading sraegy for an invesor: he case of parial informaion. Sochasic Process. Appl., 76, [13] Mironovic, D.S., J. E. Pecaric and A. M. Fink (1993):, Classical and new inequaliies in analysis. Kluwer Academic Publishers, Amserdam. [14] Müller, A. and D. Soyan (22): Comparison mehods for sochasic models and risks. Wiley& Sons, Chicheser. [15] Øksendal, B. and A. Sulem (25): Applied sochasic conrol of jump diffusions. Springer- Verlag, Berlin. [16] Rieder, U. and N. Bäuerle (25): Porfolio Opimizaion wih unobservable Markovmodulaed drif process, J. Appl. Probab., 42, [17] Rockafellar, R.T. (197): Convex analysis. Princeon Universiy Press. [18] Sass, J. and U. G. Haussmann (24): Opimizing he erminal wealh under parial informaion: he drif process as a coninuous ime Markov chain, Finance Soch., 8, [19] Schmidli, H. (22): On minimizing he ruin probabiliy by invesmen and reinsurance, Ann. Appl. Probab., 12,

Utility maximization in incomplete markets

Utility maximization in incomplete markets Uiliy maximizaion in incomplee markes Marcel Ladkau 27.1.29 Conens 1 Inroducion and general seings 2 1.1 Marke model....................................... 2 1.2 Trading sraegy.....................................

More information

Optimal Investment under Dynamic Risk Constraints and Partial Information

Optimal Investment under Dynamic Risk Constraints and Partial Information Opimal Invesmen under Dynamic Risk Consrains and Parial Informaion Wolfgang Puschögl Johann Radon Insiue for Compuaional and Applied Mahemaics (RICAM) Ausrian Academy of Sciences www.ricam.oeaw.ac.a 2

More information

An Introduction to Backward Stochastic Differential Equations (BSDEs) PIMS Summer School 2016 in Mathematical Finance.

An Introduction to Backward Stochastic Differential Equations (BSDEs) PIMS Summer School 2016 in Mathematical Finance. 1 An Inroducion o Backward Sochasic Differenial Equaions (BSDEs) PIMS Summer School 2016 in Mahemaical Finance June 25, 2016 Chrisoph Frei cfrei@ualbera.ca This inroducion is based on Touzi [14], Bouchard

More information

6. Stochastic calculus with jump processes

6. Stochastic calculus with jump processes A) Trading sraegies (1/3) Marke wih d asses S = (S 1,, S d ) A rading sraegy can be modelled wih a vecor φ describing he quaniies invesed in each asse a each insan : φ = (φ 1,, φ d ) The value a of a porfolio

More information

Hamilton- J acobi Equation: Weak S olution We continue the study of the Hamilton-Jacobi equation:

Hamilton- J acobi Equation: Weak S olution We continue the study of the Hamilton-Jacobi equation: M ah 5 7 Fall 9 L ecure O c. 4, 9 ) Hamilon- J acobi Equaion: Weak S oluion We coninue he sudy of he Hamilon-Jacobi equaion: We have shown ha u + H D u) = R n, ) ; u = g R n { = }. ). In general we canno

More information

f(s)dw Solution 1. Approximate f by piece-wise constant left-continuous non-random functions f n such that (f(s) f n (s)) 2 ds 0.

f(s)dw Solution 1. Approximate f by piece-wise constant left-continuous non-random functions f n such that (f(s) f n (s)) 2 ds 0. Advanced Financial Models Example shee 3 - Michaelmas 217 Michael Tehranchi Problem 1. Le f : [, R be a coninuous (non-random funcion and W a Brownian moion, and le σ 2 = f(s 2 ds and assume σ 2

More information

The Optimal Stopping Time for Selling an Asset When It Is Uncertain Whether the Price Process Is Increasing or Decreasing When the Horizon Is Infinite

The Optimal Stopping Time for Selling an Asset When It Is Uncertain Whether the Price Process Is Increasing or Decreasing When the Horizon Is Infinite American Journal of Operaions Research, 08, 8, 8-9 hp://wwwscirporg/journal/ajor ISSN Online: 60-8849 ISSN Prin: 60-8830 The Opimal Sopping Time for Selling an Asse When I Is Uncerain Wheher he Price Process

More information

Cash Flow Valuation Mode Lin Discrete Time

Cash Flow Valuation Mode Lin Discrete Time IOSR Journal of Mahemaics (IOSR-JM) e-issn: 2278-5728,p-ISSN: 2319-765X, 6, Issue 6 (May. - Jun. 2013), PP 35-41 Cash Flow Valuaion Mode Lin Discree Time Olayiwola. M. A. and Oni, N. O. Deparmen of Mahemaics

More information

An Introduction to Malliavin calculus and its applications

An Introduction to Malliavin calculus and its applications An Inroducion o Malliavin calculus and is applicaions Lecure 5: Smoohness of he densiy and Hörmander s heorem David Nualar Deparmen of Mahemaics Kansas Universiy Universiy of Wyoming Summer School 214

More information

Essential Microeconomics : OPTIMAL CONTROL 1. Consider the following class of optimization problems

Essential Microeconomics : OPTIMAL CONTROL 1. Consider the following class of optimization problems Essenial Microeconomics -- 6.5: OPIMAL CONROL Consider he following class of opimizaion problems Max{ U( k, x) + U+ ( k+ ) k+ k F( k, x)}. { x, k+ } = In he language of conrol heory, he vecor k is he vecor

More information

On a Fractional Stochastic Landau-Ginzburg Equation

On a Fractional Stochastic Landau-Ginzburg Equation Applied Mahemaical Sciences, Vol. 4, 1, no. 7, 317-35 On a Fracional Sochasic Landau-Ginzburg Equaion Nguyen Tien Dung Deparmen of Mahemaics, FPT Universiy 15B Pham Hung Sree, Hanoi, Vienam dungn@fp.edu.vn

More information

Optimal Investment Strategy Insurance Company

Optimal Investment Strategy Insurance Company Opimal Invesmen Sraegy for a Non-Life Insurance Company Łukasz Delong Warsaw School of Economics Insiue of Economerics Division of Probabilisic Mehods Probabiliy space Ω I P F I I I he filraion saisfies

More information

Portfolio optimization for a large investor under partial information and price impact

Portfolio optimization for a large investor under partial information and price impact Mah Meh Oper Res DOI 1.17/s186-17-589-x Porfolio opimizaion for a large invesor under parial informaion and price impac Zehra Eksi 1 Hyejin Ku Received: 4 Ocober 16 / Acceped: 1 April 17 Springer-Verlag

More information

Chapter 2. First Order Scalar Equations

Chapter 2. First Order Scalar Equations Chaper. Firs Order Scalar Equaions We sar our sudy of differenial equaions in he same way he pioneers in his field did. We show paricular echniques o solve paricular ypes of firs order differenial equaions.

More information

(MS, ) Problem 1

(MS, ) Problem 1 MS, 7.6.4) AKTUAREKSAMEN KONTROL I FINANSIERING OG LIVSFORSIKRING ved Københavns Universie Sommer 24 Skriflig prøve den 4. juni 24 kl..-4.. All wrien aids are allowed. The wo problems of oally 3 quesions

More information

arxiv: v1 [math.pr] 19 Feb 2011

arxiv: v1 [math.pr] 19 Feb 2011 A NOTE ON FELLER SEMIGROUPS AND RESOLVENTS VADIM KOSTRYKIN, JÜRGEN POTTHOFF, AND ROBERT SCHRADER ABSTRACT. Various equivalen condiions for a semigroup or a resolven generaed by a Markov process o be of

More information

Backward stochastic dynamics on a filtered probability space

Backward stochastic dynamics on a filtered probability space Backward sochasic dynamics on a filered probabiliy space Gechun Liang Oxford-Man Insiue, Universiy of Oxford based on join work wih Terry Lyons and Zhongmin Qian Page 1 of 15 gliang@oxford-man.ox.ac.uk

More information

The Asymptotic Behavior of Nonoscillatory Solutions of Some Nonlinear Dynamic Equations on Time Scales

The Asymptotic Behavior of Nonoscillatory Solutions of Some Nonlinear Dynamic Equations on Time Scales Advances in Dynamical Sysems and Applicaions. ISSN 0973-5321 Volume 1 Number 1 (2006, pp. 103 112 c Research India Publicaions hp://www.ripublicaion.com/adsa.hm The Asympoic Behavior of Nonoscillaory Soluions

More information

Matrix Versions of Some Refinements of the Arithmetic-Geometric Mean Inequality

Matrix Versions of Some Refinements of the Arithmetic-Geometric Mean Inequality Marix Versions of Some Refinemens of he Arihmeic-Geomeric Mean Inequaliy Bao Qi Feng and Andrew Tonge Absrac. We esablish marix versions of refinemens due o Alzer ], Carwrigh and Field 4], and Mercer 5]

More information

BU Macro BU Macro Fall 2008, Lecture 4

BU Macro BU Macro Fall 2008, Lecture 4 Dynamic Programming BU Macro 2008 Lecure 4 1 Ouline 1. Cerainy opimizaion problem used o illusrae: a. Resricions on exogenous variables b. Value funcion c. Policy funcion d. The Bellman equaion and an

More information

and Applications Alexander Steinicke University of Graz Vienna Seminar in Mathematical Finance and Probability,

and Applications Alexander Steinicke University of Graz Vienna Seminar in Mathematical Finance and Probability, Backward Sochasic Differenial Equaions and Applicaions Alexander Seinicke Universiy of Graz Vienna Seminar in Mahemaical Finance and Probabiliy, 6-20-2017 1 / 31 1 Wha is a BSDE? SDEs - he differenial

More information

RANDOM LAGRANGE MULTIPLIERS AND TRANSVERSALITY

RANDOM LAGRANGE MULTIPLIERS AND TRANSVERSALITY ECO 504 Spring 2006 Chris Sims RANDOM LAGRANGE MULTIPLIERS AND TRANSVERSALITY 1. INTRODUCTION Lagrange muliplier mehods are sandard fare in elemenary calculus courses, and hey play a cenral role in economic

More information

Optimality Conditions for Unconstrained Problems

Optimality Conditions for Unconstrained Problems 62 CHAPTER 6 Opimaliy Condiions for Unconsrained Problems 1 Unconsrained Opimizaion 11 Exisence Consider he problem of minimizing he funcion f : R n R where f is coninuous on all of R n : P min f(x) x

More information

Optimal Investment, Consumption and Retirement Decision with Disutility and Borrowing Constraints

Optimal Investment, Consumption and Retirement Decision with Disutility and Borrowing Constraints Opimal Invesmen, Consumpion and Reiremen Decision wih Disuiliy and Borrowing Consrains Yong Hyun Shin Join Work wih Byung Hwa Lim(KAIST) June 29 July 3, 29 Yong Hyun Shin (KIAS) Workshop on Sochasic Analysis

More information

SZG Macro 2011 Lecture 3: Dynamic Programming. SZG macro 2011 lecture 3 1

SZG Macro 2011 Lecture 3: Dynamic Programming. SZG macro 2011 lecture 3 1 SZG Macro 2011 Lecure 3: Dynamic Programming SZG macro 2011 lecure 3 1 Background Our previous discussion of opimal consumpion over ime and of opimal capial accumulaion sugges sudying he general decision

More information

arxiv: v1 [math.fa] 9 Dec 2018

arxiv: v1 [math.fa] 9 Dec 2018 AN INVERSE FUNCTION THEOREM CONVERSE arxiv:1812.03561v1 [mah.fa] 9 Dec 2018 JIMMIE LAWSON Absrac. We esablish he following converse of he well-known inverse funcion heorem. Le g : U V and f : V U be inverse

More information

E β t log (C t ) + M t M t 1. = Y t + B t 1 P t. B t 0 (3) v t = P tc t M t Question 1. Find the FOC s for an optimum in the agent s problem.

E β t log (C t ) + M t M t 1. = Y t + B t 1 P t. B t 0 (3) v t = P tc t M t Question 1. Find the FOC s for an optimum in the agent s problem. Noes, M. Krause.. Problem Se 9: Exercise on FTPL Same model as in paper and lecure, only ha one-period govenmen bonds are replaced by consols, which are bonds ha pay one dollar forever. I has curren marke

More information

t 2 B F x,t n dsdt t u x,t dxdt

t 2 B F x,t n dsdt t u x,t dxdt Evoluion Equaions For 0, fixed, le U U0, where U denoes a bounded open se in R n.suppose ha U is filled wih a maerial in which a conaminan is being ranspored by various means including diffusion and convecion.

More information

Lecture 10: The Poincaré Inequality in Euclidean space

Lecture 10: The Poincaré Inequality in Euclidean space Deparmens of Mahemaics Monana Sae Universiy Fall 215 Prof. Kevin Wildrick n inroducion o non-smooh analysis and geomery Lecure 1: The Poincaré Inequaliy in Euclidean space 1. Wha is he Poincaré inequaliy?

More information

Physics 127b: Statistical Mechanics. Fokker-Planck Equation. Time Evolution

Physics 127b: Statistical Mechanics. Fokker-Planck Equation. Time Evolution Physics 7b: Saisical Mechanics Fokker-Planck Equaion The Langevin equaion approach o he evoluion of he velociy disribuion for he Brownian paricle migh leave you uncomforable. A more formal reamen of his

More information

LECTURE 1: GENERALIZED RAY KNIGHT THEOREM FOR FINITE MARKOV CHAINS

LECTURE 1: GENERALIZED RAY KNIGHT THEOREM FOR FINITE MARKOV CHAINS LECTURE : GENERALIZED RAY KNIGHT THEOREM FOR FINITE MARKOV CHAINS We will work wih a coninuous ime reversible Markov chain X on a finie conneced sae space, wih generaor Lf(x = y q x,yf(y. (Recall ha q

More information

T L. t=1. Proof of Lemma 1. Using the marginal cost accounting in Equation(4) and standard arguments. t )+Π RB. t )+K 1(Q RB

T L. t=1. Proof of Lemma 1. Using the marginal cost accounting in Equation(4) and standard arguments. t )+Π RB. t )+K 1(Q RB Elecronic Companion EC.1. Proofs of Technical Lemmas and Theorems LEMMA 1. Le C(RB) be he oal cos incurred by he RB policy. Then we have, T L E[C(RB)] 3 E[Z RB ]. (EC.1) Proof of Lemma 1. Using he marginal

More information

Hamilton- J acobi Equation: Explicit Formulas In this lecture we try to apply the method of characteristics to the Hamilton-Jacobi equation: u t

Hamilton- J acobi Equation: Explicit Formulas In this lecture we try to apply the method of characteristics to the Hamilton-Jacobi equation: u t M ah 5 2 7 Fall 2 0 0 9 L ecure 1 0 O c. 7, 2 0 0 9 Hamilon- J acobi Equaion: Explici Formulas In his lecure we ry o apply he mehod of characerisics o he Hamilon-Jacobi equaion: u + H D u, x = 0 in R n

More information

Homogenization of random Hamilton Jacobi Bellman Equations

Homogenization of random Hamilton Jacobi Bellman Equations Probabiliy, Geomery and Inegrable Sysems MSRI Publicaions Volume 55, 28 Homogenizaion of random Hamilon Jacobi Bellman Equaions S. R. SRINIVASA VARADHAN ABSTRACT. We consider nonlinear parabolic equaions

More information

Optimal Consumption and Investment Portfolio in Jump markets. Optimal Consumption and Portfolio of Investment in a Financial Market with Jumps

Optimal Consumption and Investment Portfolio in Jump markets. Optimal Consumption and Portfolio of Investment in a Financial Market with Jumps Opimal Consumpion and Invesmen Porfolio in Jump markes Opimal Consumpion and Porfolio of Invesmen in a Financial Marke wih Jumps Gan Jin Lingnan (Universiy) College, China Insiue of Economic ransformaion

More information

Singular control of SPDEs and backward stochastic partial diffe. reflection

Singular control of SPDEs and backward stochastic partial diffe. reflection Singular conrol of SPDEs and backward sochasic parial differenial equaions wih reflecion Universiy of Mancheser Join work wih Bern Øksendal and Agnès Sulem Singular conrol of SPDEs and backward sochasic

More information

Dual control Monte-Carlo method for tight bounds of value function in regime switching utility maximization

Dual control Monte-Carlo method for tight bounds of value function in regime switching utility maximization Dual conrol Mone-Carlo mehod for igh bounds of value funcion in regime swiching uiliy maximizaion Jingang Ma, Wenyuan Li and Harry Zheng Absrac In his paper we sudy he dual conrol approach for he opimal

More information

On Measuring Pro-Poor Growth. 1. On Various Ways of Measuring Pro-Poor Growth: A Short Review of the Literature

On Measuring Pro-Poor Growth. 1. On Various Ways of Measuring Pro-Poor Growth: A Short Review of the Literature On Measuring Pro-Poor Growh 1. On Various Ways of Measuring Pro-Poor Growh: A Shor eview of he Lieraure During he pas en years or so here have been various suggesions concerning he way one should check

More information

Finish reading Chapter 2 of Spivak, rereading earlier sections as necessary. handout and fill in some missing details!

Finish reading Chapter 2 of Spivak, rereading earlier sections as necessary. handout and fill in some missing details! MAT 257, Handou 6: Ocober 7-2, 20. I. Assignmen. Finish reading Chaper 2 of Spiva, rereading earlier secions as necessary. handou and fill in some missing deails! II. Higher derivaives. Also, read his

More information

Algorithmic Trading: Optimal Control PIMS Summer School

Algorithmic Trading: Optimal Control PIMS Summer School Algorihmic Trading: Opimal Conrol PIMS Summer School Sebasian Jaimungal, U. Torono Álvaro Carea,U. Oxford many hanks o José Penalva,(U. Carlos III) Luhui Gan (U. Torono) Ryan Donnelly (Swiss Finance Insiue,

More information

Vanishing Viscosity Method. There are another instructive and perhaps more natural discontinuous solutions of the conservation law

Vanishing Viscosity Method. There are another instructive and perhaps more natural discontinuous solutions of the conservation law Vanishing Viscosiy Mehod. There are anoher insrucive and perhaps more naural disconinuous soluions of he conservaion law (1 u +(q(u x 0, he so called vanishing viscosiy mehod. This mehod consiss in viewing

More information

Undetermined coefficients for local fractional differential equations

Undetermined coefficients for local fractional differential equations Available online a www.isr-publicaions.com/jmcs J. Mah. Compuer Sci. 16 (2016), 140 146 Research Aricle Undeermined coefficiens for local fracional differenial equaions Roshdi Khalil a,, Mohammed Al Horani

More information

Stochastic Modelling in Finance - Solutions to sheet 8

Stochastic Modelling in Finance - Solutions to sheet 8 Sochasic Modelling in Finance - Soluions o shee 8 8.1 The price of a defaulable asse can be modeled as ds S = µ d + σ dw dn where µ, σ are consans, (W ) is a sandard Brownian moion and (N ) is a one jump

More information

Math 334 Fall 2011 Homework 11 Solutions

Math 334 Fall 2011 Homework 11 Solutions Dec. 2, 2 Mah 334 Fall 2 Homework Soluions Basic Problem. Transform he following iniial value problem ino an iniial value problem for a sysem: u + p()u + q() u g(), u() u, u () v. () Soluion. Le v u. Then

More information

Vehicle Arrival Models : Headway

Vehicle Arrival Models : Headway Chaper 12 Vehicle Arrival Models : Headway 12.1 Inroducion Modelling arrival of vehicle a secion of road is an imporan sep in raffic flow modelling. I has imporan applicaion in raffic flow simulaion where

More information

Online Appendix to Solution Methods for Models with Rare Disasters

Online Appendix to Solution Methods for Models with Rare Disasters Online Appendix o Soluion Mehods for Models wih Rare Disasers Jesús Fernández-Villaverde and Oren Levinal In his Online Appendix, we presen he Euler condiions of he model, we develop he pricing Calvo block,

More information

CONTROL SYSTEMS, ROBOTICS AND AUTOMATION Vol. XI Control of Stochastic Systems - P.R. Kumar

CONTROL SYSTEMS, ROBOTICS AND AUTOMATION Vol. XI Control of Stochastic Systems - P.R. Kumar CONROL OF SOCHASIC SYSEMS P.R. Kumar Deparmen of Elecrical and Compuer Engineering, and Coordinaed Science Laboraory, Universiy of Illinois, Urbana-Champaign, USA. Keywords: Markov chains, ransiion probabiliies,

More information

23.2. Representing Periodic Functions by Fourier Series. Introduction. Prerequisites. Learning Outcomes

23.2. Representing Periodic Functions by Fourier Series. Introduction. Prerequisites. Learning Outcomes Represening Periodic Funcions by Fourier Series 3. Inroducion In his Secion we show how a periodic funcion can be expressed as a series of sines and cosines. We begin by obaining some sandard inegrals

More information

Generalized Snell envelope and BSDE With Two general Reflecting Barriers

Generalized Snell envelope and BSDE With Two general Reflecting Barriers 1/22 Generalized Snell envelope and BSDE Wih Two general Reflecing Barriers EL HASSAN ESSAKY Cadi ayyad Universiy Poly-disciplinary Faculy Safi Work in progress wih : M. Hassani and Y. Ouknine Iasi, July

More information

Chapter 14 Wiener Processes and Itô s Lemma. Options, Futures, and Other Derivatives, 9th Edition, Copyright John C. Hull

Chapter 14 Wiener Processes and Itô s Lemma. Options, Futures, and Other Derivatives, 9th Edition, Copyright John C. Hull Chaper 14 Wiener Processes and Iô s Lemma Copyrigh John C. Hull 014 1 Sochasic Processes! Describes he way in which a variable such as a sock price, exchange rae or ineres rae changes hrough ime! Incorporaes

More information

A Primal-Dual Type Algorithm with the O(1/t) Convergence Rate for Large Scale Constrained Convex Programs

A Primal-Dual Type Algorithm with the O(1/t) Convergence Rate for Large Scale Constrained Convex Programs PROC. IEEE CONFERENCE ON DECISION AND CONTROL, 06 A Primal-Dual Type Algorihm wih he O(/) Convergence Rae for Large Scale Consrained Convex Programs Hao Yu and Michael J. Neely Absrac This paper considers

More information

Section 3.5 Nonhomogeneous Equations; Method of Undetermined Coefficients

Section 3.5 Nonhomogeneous Equations; Method of Undetermined Coefficients Secion 3.5 Nonhomogeneous Equaions; Mehod of Undeermined Coefficiens Key Terms/Ideas: Linear Differenial operaor Nonlinear operaor Second order homogeneous DE Second order nonhomogeneous DE Soluion o homogeneous

More information

Inventory Analysis and Management. Multi-Period Stochastic Models: Optimality of (s, S) Policy for K-Convex Objective Functions

Inventory Analysis and Management. Multi-Period Stochastic Models: Optimality of (s, S) Policy for K-Convex Objective Functions Muli-Period Sochasic Models: Opimali of (s, S) Polic for -Convex Objecive Funcions Consider a seing similar o he N-sage newsvendor problem excep ha now here is a fixed re-ordering cos (> 0) for each (re-)order.

More information

6.2 Transforms of Derivatives and Integrals.

6.2 Transforms of Derivatives and Integrals. SEC. 6.2 Transforms of Derivaives and Inegrals. ODEs 2 3 33 39 23. Change of scale. If l( f ()) F(s) and c is any 33 45 APPLICATION OF s-shifting posiive consan, show ha l( f (c)) F(s>c)>c (Hin: In Probs.

More information

Solutions from Chapter 9.1 and 9.2

Solutions from Chapter 9.1 and 9.2 Soluions from Chaper 9 and 92 Secion 9 Problem # This basically boils down o an exercise in he chain rule from calculus We are looking for soluions of he form: u( x) = f( k x c) where k x R 3 and k is

More information

Solutions to Assignment 1

Solutions to Assignment 1 MA 2326 Differenial Equaions Insrucor: Peronela Radu Friday, February 8, 203 Soluions o Assignmen. Find he general soluions of he following ODEs: (a) 2 x = an x Soluion: I is a separable equaion as we

More information

Expert Advice for Amateurs

Expert Advice for Amateurs Exper Advice for Amaeurs Ernes K. Lai Online Appendix - Exisence of Equilibria The analysis in his secion is performed under more general payoff funcions. Wihou aking an explici form, he payoffs of he

More information

Course Notes for EE227C (Spring 2018): Convex Optimization and Approximation

Course Notes for EE227C (Spring 2018): Convex Optimization and Approximation Course Noes for EE7C Spring 018: Convex Opimizaion and Approximaion Insrucor: Moriz Hard Email: hard+ee7c@berkeley.edu Graduae Insrucor: Max Simchowiz Email: msimchow+ee7c@berkeley.edu Ocober 15, 018 3

More information

15. Vector Valued Functions

15. Vector Valued Functions 1. Vecor Valued Funcions Up o his poin, we have presened vecors wih consan componens, for example, 1, and,,4. However, we can allow he componens of a vecor o be funcions of a common variable. For example,

More information

Ann. Funct. Anal. 2 (2011), no. 2, A nnals of F unctional A nalysis ISSN: (electronic) URL:

Ann. Funct. Anal. 2 (2011), no. 2, A nnals of F unctional A nalysis ISSN: (electronic) URL: Ann. Func. Anal. 2 2011, no. 2, 34 41 A nnals of F uncional A nalysis ISSN: 2008-8752 elecronic URL: www.emis.de/journals/afa/ CLASSIFICAION OF POSIIVE SOLUIONS OF NONLINEAR SYSEMS OF VOLERRA INEGRAL EQUAIONS

More information

arxiv: v1 [math.ca] 15 Nov 2016

arxiv: v1 [math.ca] 15 Nov 2016 arxiv:6.599v [mah.ca] 5 Nov 26 Counerexamples on Jumarie s hree basic fracional calculus formulae for non-differeniable coninuous funcions Cheng-shi Liu Deparmen of Mahemaics Norheas Peroleum Universiy

More information

MODULE 3 FUNCTION OF A RANDOM VARIABLE AND ITS DISTRIBUTION LECTURES PROBABILITY DISTRIBUTION OF A FUNCTION OF A RANDOM VARIABLE

MODULE 3 FUNCTION OF A RANDOM VARIABLE AND ITS DISTRIBUTION LECTURES PROBABILITY DISTRIBUTION OF A FUNCTION OF A RANDOM VARIABLE Topics MODULE 3 FUNCTION OF A RANDOM VARIABLE AND ITS DISTRIBUTION LECTURES 2-6 3. FUNCTION OF A RANDOM VARIABLE 3.2 PROBABILITY DISTRIBUTION OF A FUNCTION OF A RANDOM VARIABLE 3.3 EXPECTATION AND MOMENTS

More information

Example on p. 157

Example on p. 157 Example 2.5.3. Le where BV [, 1] = Example 2.5.3. on p. 157 { g : [, 1] C g() =, g() = g( + ) [, 1), var (g) = sup g( j+1 ) g( j ) he supremum is aken over all he pariions of [, 1] (1) : = < 1 < < n =

More information

Hamilton Jacobi equations

Hamilton Jacobi equations Hamilon Jacobi equaions Inoducion o PDE The rigorous suff from Evans, mosly. We discuss firs u + H( u = 0, (1 where H(p is convex, and superlinear a infiniy, H(p lim p p = + This by comes by inegraion

More information

Approximation Algorithms for Unique Games via Orthogonal Separators

Approximation Algorithms for Unique Games via Orthogonal Separators Approximaion Algorihms for Unique Games via Orhogonal Separaors Lecure noes by Konsanin Makarychev. Lecure noes are based on he papers [CMM06a, CMM06b, LM4]. Unique Games In hese lecure noes, we define

More information

Examples of Dynamic Programming Problems

Examples of Dynamic Programming Problems M.I.T. 5.450-Fall 00 Sloan School of Managemen Professor Leonid Kogan Examples of Dynamic Programming Problems Problem A given quaniy X of a single resource is o be allocaed opimally among N producion

More information

Existence of positive solution for a third-order three-point BVP with sign-changing Green s function

Existence of positive solution for a third-order three-point BVP with sign-changing Green s function Elecronic Journal of Qualiaive Theory of Differenial Equaions 13, No. 3, 1-11; hp://www.mah.u-szeged.hu/ejqde/ Exisence of posiive soluion for a hird-order hree-poin BVP wih sign-changing Green s funcion

More information

GMM - Generalized Method of Moments

GMM - Generalized Method of Moments GMM - Generalized Mehod of Momens Conens GMM esimaion, shor inroducion 2 GMM inuiion: Maching momens 2 3 General overview of GMM esimaion. 3 3. Weighing marix...........................................

More information

di Bernardo, M. (1995). A purely adaptive controller to synchronize and control chaotic systems.

di Bernardo, M. (1995). A purely adaptive controller to synchronize and control chaotic systems. di ernardo, M. (995). A purely adapive conroller o synchronize and conrol chaoic sysems. hps://doi.org/.6/375-96(96)8-x Early version, also known as pre-prin Link o published version (if available):.6/375-96(96)8-x

More information

Solutions Problem Set 3 Macro II (14.452)

Solutions Problem Set 3 Macro II (14.452) Soluions Problem Se 3 Macro II (14.452) Francisco A. Gallego 04/27/2005 1 Q heory of invesmen in coninuous ime and no uncerainy Consider he in nie horizon model of a rm facing adjusmen coss o invesmen.

More information

A general continuous auction system in presence of insiders

A general continuous auction system in presence of insiders A general coninuous aucion sysem in presence of insiders José M. Corcuera (based on join work wih G. DiNunno, G. Farkas and B. Oksendal) Faculy of Mahemaics Universiy of Barcelona BCAM, Basque Cener for

More information

International Journal of Scientific & Engineering Research, Volume 4, Issue 10, October ISSN

International Journal of Scientific & Engineering Research, Volume 4, Issue 10, October ISSN Inernaional Journal of Scienific & Engineering Research, Volume 4, Issue 10, Ocober-2013 900 FUZZY MEAN RESIDUAL LIFE ORDERING OF FUZZY RANDOM VARIABLES J. EARNEST LAZARUS PIRIYAKUMAR 1, A. YAMUNA 2 1.

More information

Uniqueness of solutions to quadratic BSDEs. BSDEs with convex generators and unbounded terminal conditions

Uniqueness of solutions to quadratic BSDEs. BSDEs with convex generators and unbounded terminal conditions Recalls and basic resuls on BSDEs Uniqueness resul Links wih PDEs On he uniqueness of soluions o quadraic BSDEs wih convex generaors and unbounded erminal condiions IRMAR, Universié Rennes 1 Châeau de

More information

2. Nonlinear Conservation Law Equations

2. Nonlinear Conservation Law Equations . Nonlinear Conservaion Law Equaions One of he clear lessons learned over recen years in sudying nonlinear parial differenial equaions is ha i is generally no wise o ry o aack a general class of nonlinear

More information

Lecture 2 October ε-approximation of 2-player zero-sum games

Lecture 2 October ε-approximation of 2-player zero-sum games Opimizaion II Winer 009/10 Lecurer: Khaled Elbassioni Lecure Ocober 19 1 ε-approximaion of -player zero-sum games In his lecure we give a randomized ficiious play algorihm for obaining an approximae soluion

More information

The expectation value of the field operator.

The expectation value of the field operator. The expecaion value of he field operaor. Dan Solomon Universiy of Illinois Chicago, IL dsolom@uic.edu June, 04 Absrac. Much of he mahemaical developmen of quanum field heory has been in suppor of deermining

More information

Appendix 14.1 The optimal control problem and its solution using

Appendix 14.1 The optimal control problem and its solution using 1 Appendix 14.1 he opimal conrol problem and is soluion using he maximum principle NOE: Many occurrences of f, x, u, and in his file (in equaions or as whole words in ex) are purposefully in bold in order

More information

Dynamic Portfolio Optimization with a Defaultable Security and Regime-Switching

Dynamic Portfolio Optimization with a Defaultable Security and Regime-Switching Dynamic Porfolio Opimizaion wih a Defaulable Securiy and Regime-Swiching Agosino Capponi José E. Figueroa-López Absrac We consider a porfolio opimizaion problem in a defaulable marke wih finiely-many economical

More information

Fractional Method of Characteristics for Fractional Partial Differential Equations

Fractional Method of Characteristics for Fractional Partial Differential Equations Fracional Mehod of Characerisics for Fracional Parial Differenial Equaions Guo-cheng Wu* Modern Teile Insiue, Donghua Universiy, 188 Yan-an ilu Road, Shanghai 51, PR China Absrac The mehod of characerisics

More information

MATH 5720: Gradient Methods Hung Phan, UMass Lowell October 4, 2018

MATH 5720: Gradient Methods Hung Phan, UMass Lowell October 4, 2018 MATH 5720: Gradien Mehods Hung Phan, UMass Lowell Ocober 4, 208 Descen Direcion Mehods Consider he problem min { f(x) x R n}. The general descen direcions mehod is x k+ = x k + k d k where x k is he curren

More information

10. State Space Methods

10. State Space Methods . Sae Space Mehods. Inroducion Sae space modelling was briefly inroduced in chaper. Here more coverage is provided of sae space mehods before some of heir uses in conrol sysem design are covered in he

More information

t is a basis for the solution space to this system, then the matrix having these solutions as columns, t x 1 t, x 2 t,... x n t x 2 t...

t is a basis for the solution space to this system, then the matrix having these solutions as columns, t x 1 t, x 2 t,... x n t x 2 t... Mah 228- Fri Mar 24 5.6 Marix exponenials and linear sysems: The analogy beween firs order sysems of linear differenial equaions (Chaper 5) and scalar linear differenial equaions (Chaper ) is much sronger

More information

Monotonic Solutions of a Class of Quadratic Singular Integral Equations of Volterra type

Monotonic Solutions of a Class of Quadratic Singular Integral Equations of Volterra type In. J. Conemp. Mah. Sci., Vol. 2, 27, no. 2, 89-2 Monoonic Soluions of a Class of Quadraic Singular Inegral Equaions of Volerra ype Mahmoud M. El Borai Deparmen of Mahemaics, Faculy of Science, Alexandria

More information

Existence and uniqueness of solution for multidimensional BSDE with local conditions on the coefficient

Existence and uniqueness of solution for multidimensional BSDE with local conditions on the coefficient 1/34 Exisence and uniqueness of soluion for mulidimensional BSDE wih local condiions on he coefficien EL HASSAN ESSAKY Cadi Ayyad Universiy Mulidisciplinary Faculy Safi, Morocco ITN Roscof, March 18-23,

More information

4 Sequences of measurable functions

4 Sequences of measurable functions 4 Sequences of measurable funcions 1. Le (Ω, A, µ) be a measure space (complee, afer a possible applicaion of he compleion heorem). In his chaper we invesigae relaions beween various (nonequivalen) convergences

More information

IMPLICIT AND INVERSE FUNCTION THEOREMS PAUL SCHRIMPF 1 OCTOBER 25, 2013

IMPLICIT AND INVERSE FUNCTION THEOREMS PAUL SCHRIMPF 1 OCTOBER 25, 2013 IMPLICI AND INVERSE FUNCION HEOREMS PAUL SCHRIMPF 1 OCOBER 25, 213 UNIVERSIY OF BRIISH COLUMBIA ECONOMICS 526 We have exensively sudied how o solve sysems of linear equaions. We know how o check wheher

More information

1 Review of Zero-Sum Games

1 Review of Zero-Sum Games COS 5: heoreical Machine Learning Lecurer: Rob Schapire Lecure #23 Scribe: Eugene Brevdo April 30, 2008 Review of Zero-Sum Games Las ime we inroduced a mahemaical model for wo player zero-sum games. Any

More information

Mean-Variance Hedging for General Claims

Mean-Variance Hedging for General Claims Projekbereich B Discussion Paper No. B 167 Mean-Variance Hedging for General Claims by Marin Schweizer ) Ocober 199 ) Financial suppor by Deusche Forschungsgemeinschaf, Sonderforschungsbereich 33 a he

More information

Math 10B: Mock Mid II. April 13, 2016

Math 10B: Mock Mid II. April 13, 2016 Name: Soluions Mah 10B: Mock Mid II April 13, 016 1. ( poins) Sae, wih jusificaion, wheher he following saemens are rue or false. (a) If a 3 3 marix A saisfies A 3 A = 0, hen i canno be inverible. True.

More information

Quadratic and Superquadratic BSDEs and Related PDEs

Quadratic and Superquadratic BSDEs and Related PDEs Quadraic and Superquadraic BSDEs and Relaed PDEs Ying Hu IRMAR, Universié Rennes 1, FRANCE hp://perso.univ-rennes1.fr/ying.hu/ ITN Marie Curie Workshop "Sochasic Conrol and Finance" Roscoff, March 21 Ying

More information

KEY. Math 334 Midterm I Fall 2008 sections 001 and 003 Instructor: Scott Glasgow

KEY. Math 334 Midterm I Fall 2008 sections 001 and 003 Instructor: Scott Glasgow 1 KEY Mah 4 Miderm I Fall 8 secions 1 and Insrucor: Sco Glasgow Please do NOT wrie on his eam. No credi will be given for such work. Raher wrie in a blue book, or on our own paper, preferabl engineering

More information

Convergence of the Neumann series in higher norms

Convergence of the Neumann series in higher norms Convergence of he Neumann series in higher norms Charles L. Epsein Deparmen of Mahemaics, Universiy of Pennsylvania Version 1.0 Augus 1, 003 Absrac Naural condiions on an operaor A are given so ha he Neumann

More information

Chapter 6. Systems of First Order Linear Differential Equations

Chapter 6. Systems of First Order Linear Differential Equations Chaper 6 Sysems of Firs Order Linear Differenial Equaions We will only discuss firs order sysems However higher order sysems may be made ino firs order sysems by a rick shown below We will have a sligh

More information

FINM 6900 Finance Theory

FINM 6900 Finance Theory FINM 6900 Finance Theory Universiy of Queensland Lecure Noe 4 The Lucas Model 1. Inroducion In his lecure we consider a simple endowmen economy in which an unspecified number of raional invesors rade asses

More information

13.3 Term structure models

13.3 Term structure models 13.3 Term srucure models 13.3.1 Expecaions hypohesis model - Simples "model" a) shor rae b) expecaions o ge oher prices Resul: y () = 1 h +1 δ = φ( δ)+ε +1 f () = E (y +1) (1) =δ + φ( δ) f (3) = E (y +)

More information

CHARACTERIZATION OF REARRANGEMENT INVARIANT SPACES WITH FIXED POINTS FOR THE HARDY LITTLEWOOD MAXIMAL OPERATOR

CHARACTERIZATION OF REARRANGEMENT INVARIANT SPACES WITH FIXED POINTS FOR THE HARDY LITTLEWOOD MAXIMAL OPERATOR Annales Academiæ Scieniarum Fennicæ Mahemaica Volumen 31, 2006, 39 46 CHARACTERIZATION OF REARRANGEMENT INVARIANT SPACES WITH FIXED POINTS FOR THE HARDY LITTLEWOOD MAXIMAL OPERATOR Joaquim Marín and Javier

More information

Games Against Nature

Games Against Nature Advanced Course in Machine Learning Spring 2010 Games Agains Naure Handous are joinly prepared by Shie Mannor and Shai Shalev-Shwarz In he previous lecures we alked abou expers in differen seups and analyzed

More information

Empirical Process Theory

Empirical Process Theory Empirical Process heory 4.384 ime Series Analysis, Fall 27 Reciaion by Paul Schrimpf Supplemenary o lecures given by Anna Mikusheva Ocober 7, 28 Reciaion 7 Empirical Process heory Le x be a real-valued

More information

Predator - Prey Model Trajectories and the nonlinear conservation law

Predator - Prey Model Trajectories and the nonlinear conservation law Predaor - Prey Model Trajecories and he nonlinear conservaion law James K. Peerson Deparmen of Biological Sciences and Deparmen of Mahemaical Sciences Clemson Universiy Ocober 28, 213 Ouline Drawing Trajecories

More information

Matlab and Python programming: how to get started

Matlab and Python programming: how to get started Malab and Pyhon programming: how o ge sared Equipping readers he skills o wrie programs o explore complex sysems and discover ineresing paerns from big daa is one of he main goals of his book. In his chaper,

More information