Behavioural Differential Equations and Coinduction for Binary Trees

Size: px
Start display at page:

Download "Behavioural Differential Equations and Coinduction for Binary Trees"

Transcription

1 Behviourl Differentil Equtions nd Coinduction for Binry Trees Alexndr Silv nd Jn Rutten,2 Centrum voor Wiskunde en Informtic (CWI) 2 Vrije Universiteit Amsterdm (VUA) {ms,jnr}@cwi.nl Abstrct. We study the set T A of infinite binry trees with nodes lbelled in semiring A from colgebric perspective. We present coinductive definition nd proof principles bsed on the fct tht T A crries finl colgebr structure. By viewing trees s forml power series, we develop clculus where definitions re presented s behviourl differentil equtions. We present generl formt for these equtions tht gurntees the existence nd uniqueness of solutions. Although techniclly not very difficult, the resulting frmework hs surprisingly nice pplictions, which is illustrted by vrious concrete exmples. Introduction Infinite dt structures re often used to model problems nd computing solutions for them. Therefore, resoning tools for such structures hve become more nd more relevnt. Colgebric techniques turned out to be suited for proving nd deriving properties of infinite systems. In [6], coinductive clculus of forml power series ws developed. In close nlogy to clssicl nlysis, the definitions were presented s behviourl differentil equtions nd properties were proved in clcultionl (nd very nturl) wy. This pproch hs shown to be quite effective for resoning bout strems [6,7] nd it seems worthwhile to explore its effectiveness for other dt structures s well. In this pper, we shll tke colgebric perspective on clssicl dt structure infinite binry trees, nd develop similr clculus, using the fct tht binry trees re prticulr cse of forml power series. The contributions of the present pper re: coinductive clculus, bsed on the notion of derivtive, to define nd to reson bout trees nd functions on trees; set of illustrtive exmples nd properties tht show the usefulness nd expressiveness of such clculus; the formultion of generl formt tht gurntees the existence nd uniqueness of solutions of behviourl differentil equtions. Infinite trees rise in severl forms in other res. Forml tree series (functions from trees to n rbitrry semiring) hve been studied in [3], closely relted to distributive Σ-lgebrs. The work presented in this pper is completely different since we re concerned with infinite binry trees nd not with forml series over trees. In [5], infinite trees pper s generlistions of infinite words nd n extensive study of tree utomt nd topologicl spects of trees is mde. We hve not yet ddressed the reltion of our work with utomt theory. Here we emphsize coinductive definition nd proof principles for defining nd resoning bout (functions on) trees. At the end of the pper, in Section 6, the novelty of our pproch is discussed further. Also severl directions for further pplictions re mentioned there. 2 Trees nd coinduction We introduce the set T A of infinite node-lbelled binry trees, show tht T A stisfies coinduction proof principle nd illustrte its usefulness. Prtilly supported by the Fundção pr Ciênci e Tecnologi, Portugl, under grnt number SFRH/BD/27482/26.

2 The set T A of infinite node-lbelled binry trees, where to ech node is ssigned vlue in A, is the finl colgebr for the functor F X = X A X nd cn be formlly defined by: T A = {t t : {L, R} A} The set T A crries finl colgebr structure consisting of the following function: l, i, r : T A T A A T A t λw.t(lw), t(ε), λw.t(rw) where l nd r return the left nd right subtrees, respectively, nd i gives the lbel of the root node of the tree. Here, ε denotes the empty word nd, for b {L, R}, bw denotes the word resulting from prefixing the word w with the letter b. These definitions of the set T A nd the respective colgebr mp my not seem obvious. The follow resoning justifies its correctness: It is well known from the literture [4] tht the finl system for the functor G(X) = A X B is (A B, π), where π : A B A (A B ) B π(φ) = φ(ε), λb v. φ(bv) The functor F is isomorphic to H(X) = A X 2. Therefore, the set A 2 is the finl colgebr for the functor F. Considering 2 = {L, R} we cn derive the definition of l, i, r from the one presented bove for π. The fct tht T A is finl colgebr mens tht for ny rbitrry colgebr lt, o, rt : U U A U, there exists unique f : U T A, such tht the following digrm commutes: lt,o,rt U U A U!f f id A f T A l,i,r T A A T A The existence prt of this sttement gives us coinductive definition principle. Every triplet of functions lt : U U, o : U A nd rt : U U defines function h : U T A, such tht: i(h(x)) = o(x) l(h(x)) = h(lt(x)) r(h(x)) = h(rt(x)) We will see more generl formultion of this principle in section 3, where the right hnd side of the bove equtions will be more generl. Tking A = R we present the definition of the elementwise sum s n exmple. d b e f c g + u s v r w t By the definition principle presented bove, tking o( σ, τ ) = i(σ) + i(τ), lt( σ, τ ) = l(σ), l(τ) nd rt( σ, τ ) = r(σ), r(τ) there is unique function + : T R T R T R stisfying: i(σ + τ) = i(σ) + i(τ) l(σ + τ) = l(σ) + l(τ) r(σ + τ) = r(σ) + r(τ) Note tht in the first eqution bove we re using + to represent both the sum of trees nd the sum of rel numbers. Now tht we hve explined the forml definition for the set T A nd how one cn uniquely define functions into T A, nother importnt question is still to be nswered: how do we prove equlity on T A? In order to prove tht two trees σ nd τ re equl it is necessry nd sufficient to prove x = d+u b+s e+v +r w {L,R} σ(w) = τ(w) () The use of induction on w (prove tht σ(ε) = τ(ε) nd tht whenever σ(w) = τ(w) holds then σ(w) = τ(w) lso holds, for {L, R}) clerly is correct method to estblish the vlidity of (). However, we will often encounter exmples where there is not generl formul for σ(w) nd τ(w). Insted, we tke colgebric perspective on T A nd use the coinduction proof principle in order to estblish equlities. This proof principle is bsed on f+w c+t g+x

3 the notion of bisimultion. A bisimultion on T A is reltion S T A T A such tht, for ll σ nd τ in T A, (σ, τ) S σ(ε) = τ(ε) (l(σ), l(τ)) S (r(σ), r(τ)) S We will write σ τ whenever there exists bisimultion tht contins (σ, τ). The reltion, clled the bisimilrity reltion, is the union of ll bisimultions (one cn esily check tht the union of bisimultion is itself bisimultion). The following theorem expresses the proof principle mentioned bove. Theorem (Coinduction). For ll trees σ nd τ in T A, if σ τ then σ = τ. Proof. Consider two trees σ nd τ in T A nd let S T A T A be bisimultion reltion which contins the pir (σ, τ). The equlity σ(w) = τ(w) now follows by induction on the length of w. We hve tht σ(ε) = τ(ε), becuse S is bisimultion. If w = Lw, then σ(lw ) = l(σ)(w ) (Definition of l) = l(τ)(w ) (S is bisimultion nd induction hypothesis) = τ(lw ) (Definition of l) Similrly, if w = Rw, then σ(rw ) = r(σ)(w ) = r(τ)(w ) = τ(rw ). Therefore, for ll w {L, R}, σ(w) = τ(w). This proves σ = τ. Thus, in order to prove tht two trees re equl, it is sufficient to show tht they re bisimilr. We shll see severl exmples of proofs by coinduction below. As first simple exmple, let us prove tht the pointwise sum for trees of rel numbers defined before is commuttive. Let S = { σ + τ, τ + σ σ, τ T R }. Since i(σ + τ) = i(σ) + i(τ) = i(τ + σ) nd l(σ + τ) = l(σ) + l(τ) S l(τ) + l(σ) = l(τ + σ) r(σ + τ) = r(σ) + r(τ) S r(τ) + r(σ) = r(τ + σ) for ny σ nd τ, S is bisimultion reltion on T R. The commuttivity property follows by coinduction. 3 Behviourl differentil equtions In this section, we shll view trees s forml power series. Following [6], coinductive definitions of opertors into T A nd constnt trees then tke the shpe of so-clled behviourl differentil equtions. We shll prove theorem gurnteeing the existence of unique solution for lrge fmily of systems of behviourl differentil equtions. Forml power series re functions σ : X k from the set of words over n lphbet X to semiring k. If A is semiring, T A, s defined in section 2, is the set of ll forml power series over the lphbet {L, R} with coefficients in A. In ccordnce with the generl notion of derivtive of forml power series [6] we shll write σ L for l(σ) nd σ R for r(σ). We will often refer to σ L, σ R nd σ(ε) s the left derivtive, right derivtive nd initil vlue of σ. Following [6], we will develop coinductive clculus of infinite binry trees. From now on coinductive definitions will hve the shpe of behviourl differentil equtions. Let us illustrte this pproch by simple exmple the coinductive definition of tree, clled one, decorted with s in every node. A forml definition of this tree consists the following behviourl differentil equtions: one L = one one(ε) = one R = one

4 The fct tht there exists unique tree tht stisfies the bove equtions will follow from the theorem below, which presents generl formt for behviourl differentil equtions gurnteeing the existence nd uniqueness of solutions. Behviourl differentil equtions will be used not just to define single constnt trees but lso functions on trees. We shll see exmples below. Before we present the min result of this section, we need one more definition. We wnt to be ble to view ny element n A s tree (which we will denote by [n]): n The tree [n] is formlly defined s [n](ε) = n [n](w) = w ε Next we present syntx describing the formt of behviourl differentil equtions tht we will consider. Let Σ be set of function symbols, ech with n rity r(f) for f Σ. (As usul we cll f constnt if r(f) =.) Let X = {x, x 2,...} be set of (syntctic) vribles, nd let X L = {x L x X}, X R = {x R x X}, [X(ε)] = {[x(ε)] x X} nd X(ε) = {x(ε) x X} be sets of nottionl vrints of them. The vribles x X will ply the role of plce holders for trees τ T A. Vribles x L, x R, nd [x(ε)] will then ct s plce holders for the corresponding trees τ L, τ R nd [τ(ε)] in T A, while x(ε) (without the squre brckets) will correspond to τ s initil vlue τ(ε) A. We cll behviourl differentil eqution for function symbol f Σ with rity r = r(f) well-formed if it is of the form f ( x,..., x r ) L = t ( f ( x f ( x,..., x r ) R = t,..., x r ) ) (ε) = c(x (ε),..., x r(ε)) 2 where c : A r A is given function, nd where t nd t 2 re terms built out of function symbols in Σ nd vribles in {x,..., x r} nd their corresponding nottionl vrints in X L, X R nd [X(ε)]. A well-formed system of equtions for Σ will then consist of one wellformed eqution for ech f Σ. A solution of such system of equtions is set of tree functions Σ = { f : (TA) r T A f Σ} stisfying, for ll f Σ with rity r nd for ll τ,..., τ r T A, f(τ,..., τ r) (ε) = c(τ (ε),..., τ r(ε)) nd f(τ,..., τ r) L = t nd f(τ,..., τ r) where the tree t T A (nd similrly for t 2) is obtined from the term t by replcing (ll occurrences of) x i by τ i, (x i) L by (τ i) L, (x i) R by (τ i) R, nd [x i(ε)] by [τ i(ε)], for ll i =,..., r, nd ll function symbols g Σ by their corresponding function g. Theorem 2. Let Σ be set of function symbols. Every well-formed system of behviourl differentil equtions for Σ hs precisely one solution of tree functions Σ. Proof. Plese see ppendix A. Let us illustrte the generlity of this theorem by mentioning few exmples of systems of differentil equtions tht stisfy the formt bove. As first exmple, tke Σ = {one} consisting of single constnt symbol (with rity ) nd X =. We observe tht the differentil equtions for one mentioned t the beginning of this section stisfy the formt of the theorem. For second exmple, let Σ = {+, } with rities r(+) = r( ) = 2 nd let X = {σ, τ}. Consider the following equtions: (σ + τ) L = σ L + τ L (σ + τ) R = σ R + τ R (σ + τ)(ε) = σ(ε) + τ(ε) R = t 2 differentil equtions initil vlue (σ τ) L = (σ L τ) + ([σ(ε)] τ L) (σ τ)(ε) = σ(ε) τ(ε) (σ τ) R = (σ R τ) + ([σ(ε)] τ R)

5 These equtions define the opertions of sum nd convolution product of trees, to be further discussed in Section 4. Note tht the right-hnd side of the eqution for (σ τ) L (nd similrly for (σ τ) R) is good illustrtion of the generl formt: it is built from the functions + nd, pplied to ( subset of) the vribles on the left (τ), their derivtives (σ L nd τ L), nd their initil vlues viewed s trees ([σ(ε)]). Clerly there re mny interesting interesting instnces of well-formed differentil equtions. Note, however, tht the formt does impose certin restrictions. The min point is tht in the right-hnd sides of the equtions, only single L nd R derivtives re llowed. The following is n exmple of system of equtions tht is not well-formed nd tht does not hve unique solution. Let Σ = {f}, with rity r(f) =, nd let X = {σ}. The equtions for f re f(σ) L = f(f(σ LL)) f(σ)(ε) = σ(ε) f(σ) R = [] Both g(σ) = [σ(ε)] + (L [σ LL(ε)]) nd h(σ) = [σ(ε)] + (L [σ LL(ε)] + L 2 ( L) ) re solutions. All the exmples of systems of behviourl differentil equtions tht will pper in the rest of this document fit into the formt of Theorem 2. Therefore, we will omit proofs of the existence nd uniqueness of solutions for those systems. In the next section we will define opertors on trees, bsed on some generl opertors on forml power series [6]. 4 Tree clculus In this section, we present opertors on trees, nmely sum, convolution product nd inverse, nd stte some elementry properties, which we will prove using coinduction. The sum of two trees is defined s the unique opertor stisfying: (σ + τ) L = σ L + τ L (σ + τ) R = σ R + τ R (σ + τ)(ε) = σ(ε) + τ(ε) Note tht this is generlistion of the sum on trees of rel numbers defined in section 2 nd tht gin we re overloding the use of + to represent both sum on trees nd sum on the elements of the semiring. Sum stisfies some desired properties, esily proved by coinduction, such s commuttivity or ssocitivity: Theorem 3. For ll σ, τ nd ρ in T A, σ + = σ, σ +τ = τ +σ nd σ +(τ +ρ) = (σ +τ)+ρ. Here, we re using s shorthnd for []. We shll use this convention (for ll n A) throughout this document. We define the convolution product of two trees s the unique opertion stisfying: (σ τ) L = (σ L τ) + (σ(ε) τ L) (σ τ)(ε) = σ(ε) τ(ε) (σ τ) R = (σ R τ) + (σ(ε) τ R) Note tht in the bove definition we use for representing both multipliction on trees nd on the elements of the semiring. Following the convention mentioned bove σ(ε) τ L nd σ(ε) τ R re shorthnds for [σ(ε)] τ L nd [σ(ε)] τ R. We shll lso use the stndrd convention of writing στ for σ τ. The generl formul to compute the vlue of σ ccording to pth given by the word w {L, R} is: (σ τ)(w) = X σ(u) τ(v) w=u v where denotes word conctention. To give the reder some intuition bout this opertion we will give concrete exmple. Tke A to be the Boolen semiring B = {, }, with opertions + = nd =. Then, T A corresponds to the lnguges over two letter lphbet, nd in this cse the tree product opertor coincides with lnguge conctention. The following theorem sttes some fmilir properties of this opertor.

6 Theorem 4. For ll σ, τ, ρ in T A nd, b in A, σ = σ = σ, σ = σ =, σ (τ + ρ) = (σ τ) + (σ ρ), σ (τ ρ) = (σ τ) ρ nd [] [b] = [ b]. Proof. An exercise in coinduction. In [7], these properties re proved for strems. Note tht the convolution product is not commuttive. Before we present the inverse opertion, let us introduce two (very useful) constnts, which we shll cll left constnt L nd right constnt R. They will hve n importnt role in the tree clculus tht we re bout to develop nd will turn out to hve interesting properties when intercting with the product opertion. The left constnt L is tree with s in every node except in the root of the left brnch where it hs : L = It is formlly defined s L(w) = if w = L L(w) = otherwise Similrly, the right constnt R is only different from in the root of its right brnch: R = nd is defined s R(w) = if w = R R(w) = otherwise These constnts hve interesting properties when multiplied by n rbitrry tree. L σ produces tree whose root nd right subtrees re null nd the left brnch is σ: X d b c = e f g d f b c Dully, R σ produces tree whose root nd left subtrees re null nd the right brnch is σ: X s p q r = t u As before, if we see L nd σ s lnguges nd the product s conctention, we cn gin some intuition on the mening of this opertion. L σ will prefix every word of σ with the letter L, mening tht no word strting by R will be n element of L σ, nd thus, L σ hs null right brnch. Similr for R σ. As we pointed out before, the product opertion is not commuttive. For exmple, σ L L σ nd σ R R σ. In fct, multiplying tree σ on the right with L or R is interesting in itself. For instnce, σ L stisfies j σ(u) w = ul (σ L)(w) = otherwise which corresponds to the following trnsformtion: b c X = d e f g Similrly, σ R produces the following tree: b c X = d e f g Agin, if you interpret these opertions in the lnguge setting, wht is being constructed is the lnguge tht hs ll words of the form ul nd ur, respectively, such tht σ(u). v b d d e e b e s q g c t f f p u r v g c g

7 We define the inverse of tree s the unique opertor stisfying: (σ ) L = σ(ε) ( σ L (σ )) (σ ) R = σ(ε) ( σ R (σ )) σ (ε) = σ(ε) We re using σ L nd σ R s shorthnds for [ ] σ L nd [ ] σ R, respectively. In this definition, we need to require A to be ring, in order to hve dditive inverses. Moreover, the tree σ is supposed to hve lso multiplictive inverse for its initil vlue. The inverse of tree hs the usul properties: Theorem 5. For ll σ nd τ in T A: σ is the unique tree s.t. σ σ = (2) (σ τ) = τ σ (3) Proof. For the existence prt of (2), note tht. (σ σ )(ε) = σ(ε) σ(ε) = 2. (σ σ ) L = (σ L σ ) + (σ(ε) (σ(ε) ( σ L σ ))) = 3. (σ σ ) R = (σ R σ ) + (σ(ε) (σ(ε) ( σ R σ ))) = So, by uniqueness (using the behviourl differentil equtions tht define ) we hve proved tht σ σ =. Now, for the uniqueness prt of (2), suppose tht there is tree τ such tht σ τ =. We shll prove tht τ = σ. Note tht from the equlity σ τ = we derive tht. τ(ε) = σ(ε) 2. τ L = σ(ε) ( σ L τ) 3. τ R = σ(ε) ( σ R τ) Thus, by uniqueness of solutions for systems of behviourl differentil equtions, τ = σ. For (3), note tht (σ τ) τ σ = σ (τ τ ) σ =. Therefore, using the uniqueness property of (2), (σ τ) = τ σ. 5 Applictions of tree clculus We will illustrte the usefulness of our clculus by looking t series of interesting exmples. In order to compute closed formule for trees we will be using the following identity: σ TA σ = σ(ε) + (L σ L) + (R σ R) (4) which cn be esily proved by coinduction. We will now show how to use this identity to construct the closed formul for tree. Recll our first system of behviourl differentil equtions: one L = one one(ε) = one R = one There we sw tht the unique solution for this system ws the tree with s in every node. Alterntively, we cn compute the solution using (4) s follows. one = one(ε) + (L one L) + (R one R) one = + (L one) + (R one) ( L R)one = one = ( L R) Therefore, one cn be represented by the (very compct) closed formul 3 ( L R). Let us see few more exmples. From now on we will work with A = R, where we hve the extr property: [n] σ = σ [n], for ll n R nd σ T R. 3 Note the similrity of this closed formul with the one obtined for the strem (,,...) in [7]: ( X).

8 The tree where every node t level k is lbelled with the vlue 2 k, clled pow, is defined by the following system: We proceed s before, pplying (4): pow L = 2 pow pow(ε) = pow R = 2 pow pow = pow(ε) + (L pow L) + (R pow R) pow = + (2L pow) + (2R pow) ( 2L 2R)pow = pow = ( 2L 2R) which gives us nice closed formul for pow 4. The tree with the nturl numbers is represented by the following system of differentil equtions: Applying identity (4): nt L = nt + pow nt R = nt + (2 pow) nt(ε) = nt = nt(ε) + (L nt L) + (R nt R) nt = + (L (nt + pow)) + (R (nt + 2pow)) ( L R)nt = + L( 2L 2R) + 2R( 2L 2R) ( L R)nt = ( L) ( 2L 2R) nt = ( L R) ( L) ( 2L 2R) The Thue-Morse sequence [] cn be obtined by tking the prities of the counts of s in the binry representtion of non-negtive integers. Alterntively, it cn be defined by the repeted ppliction of the substitution mp { ; }:... We cn encode this substitution mp in binry tree, clled thue, which t ech level k will hve the first 2 k digits of the Thue-Morse sequence: In this exmple, we tke for A the Boolen ring 2 = {, } (where + = ). The following system of differentil equtions defines thue: thue L = thue thue R = thue + one thue(ε) = 4 Agin, there is strong similrity with strems: the closed formul for the strem (, 2, 4, 8,...) is ( 2X)

9 Note tht thue + one equls the (elementwise) complement of thue. Applying (4) to thue, we clculte: thue = (L thue) + (R (thue + one)) ( L R) thue = R one thue = ( L R) R one which then leds to the following pretty formul for thue: thue = one R one Let us present nother exmple substitution opertion, which given two trees σ nd τ, replces the left subtree of σ by τ. σ(ε), τ σ σ L R = subst ( ) τ It is esy to see tht the equtions tht define this opertion re: Then, we pply (4) nd we reson: σ(ε) subst(σ, τ) L = τ subst(σ, τ)(ε) = σ(ε) subst(σ, τ) R = σ R subst(σ, τ) = σ(ε) + (L τ) + (R σ R) subst(σ, τ) = σ (L σ L) + (L τ) subst(σ, τ) = σ L(σ L τ) Note tht in the second step, we pplied identity (4) to σ. Moreover, remrk tht the finl closed formul for subst(σ, τ) gives us the lgorithm to compute the substitution: σ(ε), τ σ σ L R = subst ( ) τ σ(ε) σ R σ R - σ L σ(ε) σ R + τ We cn now wonder how to define more generl substitution opertion tht hs n rbitrry pth P {L, R} + s n extr rgument nd replces the subtree of σ given by this pth by τ. It seems obvious to define it s subst(σ, τ, P ) = σ P (σ P τ) where, in the right hnd side, P = 2... n is interpreted s 2... n nd the derivtive σ P is defined s j σδ P = δ σ P = (σ δ ) P P = δ.p with δ being either L or R. Let us check tht our intuition is correct. First, we present the definition for this opertion: differentil equtions 8 initil vlue < τ P = δ subst(σ, τ, P ) δ = subst(σ δ, τ, P ) P = δ.p subst(σ, τ, P )(ε) = σ(ε) : σ δ P = δ.p where δ δ. Now, observe tht R = { subst(σ, τ, P ), σ P (σ P τ) σ, τ T R, P {L, R} + } { σ, σ σ T R } is bisimultion reltion becuse:. (σ P (σ P τ))(ε) = σ(ε) = subst(σ, τ, P )(ε)

10 2. For δ {L, R}, (σ P (σ P τ)) δ = σ δ P δ (σ P τ) 8 < τ P = δ = σ δ P ((σ δ ) P τ) P = δ.p : σ δ P = δ.p 8 < τ P = δ R subst(σ δ, τ, P ) P = δ.p : σ δ P = δ.p = subst(σ, τ, P ) δ Therefore, by Theorem, subst(σ, τ, P ) = σ P (σ P τ). Using this formul we cn now prove properties bout this opertion. For instnce, one would expect tht subst(σ, σ P, P ) = σ nd lso subst(subst(σ, τ, P ), σ P, P ) = σ. The first equlity follows esily: subst(σ, σ P, P ) = σ P (σ P σ P ) = σ. For the second one we hve: subst(subst(σ, τ, P ), σ P, P ) = subst(σ P (σ P τ), σ P, P ) (Definition of subst) = σ P (σ P τ) P ((σ P (σ P τ)) P σ P ) (Definition of subst) = σ P (σ P τ) P (τ σ P ) ((σ P (σ P τ)) P = σ P σ P + τ = τ) = σ Remrk tht this opertion is stndrd exmple in introductory courses on lgorithms nd dt structures. It is often presented either s recursive expression (very much in the style of our differentil equtions) or s contrived itertive procedure. This exmple shows tht our compct formule constitute cler wy of presenting lgorithms nd tht they cn be used to eliminte recursion. Moreover, the differentil equtions re directly implementble lgorithms (in functionl progrmming) nd our clculus provides systemtic wy of resoning bout such progrms. 6 Discussion We hve modelled binry trees s forml power series nd, using the fct tht the ltter constitute finl colgebr, this hs enbled us to pply some colgebric resoning. Techniclly, none of this is very difficult. Rther, it is n ppliction of well known colgebric insights. As is the cse with mny of such pplictions, it hs the flvour of n exercise. At the sme time, the result contins severl new elements tht hve surprised us. Although techniclly Theorem 2 is n esy extension of similr such theorem for strems, the resulting formt for differentil equtions for trees is surprisingly generl nd useful. It hs llowed us to define vrious non-trivil trees by mens of simple differentil equtions, nd to compute rther plesnt closed formule for them. We hve lso illustrted tht bsed on this, coinduction is convenient proof method for trees. As n ppliction, ll of this is new, to the best of our knowledge. (Forml tree series, which hve been studied extensively, my seem to be closely relted but re not: here we re deling with differentil equtions tht chrcterise single trees.) In ddition to the illustrtions of the present differentil clculus for trees, we see vrious directions for further pplictions: (i) The connection with (vrious types of) utomt nd the finl colgebr T A of binry trees needs further study. For instnce, every Moore utomton with input in 2 = {L, R} nd output in A hs miniml representtion in T A. Detils of these utomt (nd on weighted vrints of them) re not complicted but hve not been worked out here becuse of lck of spce. (ii) The closed formul tht we hve obtined for the (binry tree representing the) Thue-Morse sequence suggests possible use of coinduction nd differentil equtions in the re of utomtic sequences [2]. Typiclly, utomtic sequences re represented by utomt. The present clculus seems n interesting lterntive, in which properties such s lgebricity of sequences cn be derived from the tree differentil equtions tht define them. (iii) Finlly, the closed formule tht we obtin for tree substitution suggest mny further pplictions of our tree clculus to (functionl) progrms on trees, including the nlysis of their complexity.

11 References. J.-P. Allouche nd J. Shllit. The ubiquitous Prouhet-Thue-Morse sequence. In C. Ding, T. Helleseth, nd N. H., editors, Sequences nd their pplictions, Proceedings of SETA 98, pges 6. Springer Verlg, J.-P. Allouche nd J. Shllit. Automtic sequences: theory, pplictions, generliztions. Cmbridge University Press, Z. Ésik nd W. Kuich. Forml tree series. Journl of Automt, Lnguges nd Combintorics, 8(2):29 285, E. G. Mnes nd M. A. Arbib. Algebric pproches to progrm semntics. Springer- Verlg New York, Inc., New York, NY, USA, D. Perrin nd J.-E. Pin. Infinite Words, volume 4 of Pure nd Applied Mthemtics. Elsevier, 24. ISBN J. J. M. M. Rutten. Behviourl differentil equtions: coinductive clculus of strems, utomt, nd power series. Theor. Comput. Sci., 38(-3): 53, J. J. M. M. Rutten. A coinductive clculus of strems. Mthemticl Structures in Computer Science, 5():93 47, 25. A Proof of theorem 2 Proof (Proof of theorem 2). Consider well-formed system of differentil equtions for Σ, s defined bove. We define set T of terms t by the following syntx: t ::= τ (τ T A) f(t,..., t r(f) ) (f Σ) where for every tree τ T A the set T contins corresponding term, denoted by τ, nd where new terms re constructed by (syntctic) composition of function symbols from Σ with the pproprite number of rgument terms. Next we turn T into n F -colgebr by defining function l, o, r : T (T A T ) by induction on the structure of terms, s follows. First we define o : T A by o(τ) = τ(ε) o ` f(t,... t r(f) ) = c ` o(t ),..., o(t r(f) ) (where c is the function used in the equtions for f). Next we define l : T T nd r : T T by l(τ) = τ L nd r(τ) = τ R, nd by l ( f(t,... t r) ) = t nd l ( f(t,... t r) ) = t 2 Here the terms t nd t 2 re obtined from the terms t nd t 2 used in the equtions for f, by replcing (every occurrence of) x i by t i, (x i) L by l(t i), (x i) R by r(t i), nd [x i(ε)] by [o(t)], for i =,..., r. Becuse T A is finl F -colgebr, there exists unique homomorphism h : T T A. We cn use it to define tree functions f : (T A) r T A, for every f Σ, by putting, for ll τ,..., τ r T A, f(τ,..., τ r) = h ` f(τ,..., τ r) This gives us set Σ of tree functions. One cn prove tht it is solution of our system of differentil equtions by coinduction, using the fcts tht h(τ) = τ, for ll τ T A, nd h ( f(t,..., t r) ) = f ( h(t ),..., h(t r) ) for ll f Σ nd t i T. This solution is unique becuse, by finlity of T A, the homomorphism h is.

Roadmap of This Lecture

Roadmap of This Lecture Reltionl Model Rodmp of This Lecture Structure of Reltionl Dtbses Fundmentl Reltionl-Algebr-Opertions Additionl Reltionl-Algebr-Opertions Extended Reltionl-Algebr-Opertions Null Vlues Modifiction of the

More information

9.3. Regular Languages

9.3. Regular Languages 9.3. REGULAR LANGUAGES 139 9.3. Regulr Lnguges 9.3.1. Properties of Regulr Lnguges. Recll tht regulr lnguge is the lnguge ssocited to regulr grmmr, i.e., grmmr G = (N, T, P, σ) in which every production

More information

Research Article Existence of Positive Solution to Second-Order Three-Point BVPs on Time Scales

Research Article Existence of Positive Solution to Second-Order Three-Point BVPs on Time Scales Hindwi Publishing Corportion Boundry Vlue Problems Volume 2009, Article ID 685040, 6 pges doi:10.1155/2009/685040 Reserch Article Existence of Positive Solution to Second-Order hree-point BVPs on ime Scles

More information

CH 71 COMPLETING THE SQUARE INTRODUCTION FACTORING PERFECT SQUARE TRINOMIALS

CH 71 COMPLETING THE SQUARE INTRODUCTION FACTORING PERFECT SQUARE TRINOMIALS CH 7 COMPLETING THE SQUARE INTRODUCTION I t s now time to py our dues regrding the Qudrtic Formul. Wht, you my sk, does this men? It mens tht the formul ws merely given to you once or twice in this course,

More information

Addition and Subtraction

Addition and Subtraction Addition nd Subtrction Nme: Dte: Definition: rtionl expression A rtionl expression is n lgebric expression in frction form, with polynomils in the numertor nd denomintor such tht t lest one vrible ppers

More information

Chapter 2: Relational Model. Chapter 2: Relational Model

Chapter 2: Relational Model. Chapter 2: Relational Model Chpter : Reltionl Model Dtbse System Concepts, 5 th Ed. See www.db-book.com for conditions on re-use Chpter : Reltionl Model Structure of Reltionl Dtbses Fundmentl Reltionl-Algebr-Opertions Additionl Reltionl-Algebr-Opertions

More information

164 CHAPTER 2. VECTOR FUNCTIONS

164 CHAPTER 2. VECTOR FUNCTIONS 164 CHAPTER. VECTOR FUNCTIONS.4 Curvture.4.1 Definitions nd Exmples The notion of curvture mesures how shrply curve bends. We would expect the curvture to be 0 for stright line, to be very smll for curves

More information

Chapter55. Algebraic expansion and simplification

Chapter55. Algebraic expansion and simplification Chpter55 Algebric expnsion nd simplifiction Contents: A The distributive lw B The product ( + b)(c + d) C Difference of two squres D Perfect squres expnsion E Further expnsion F The binomil expnsion 88

More information

A Fuzzy Inventory Model With Lot Size Dependent Carrying / Holding Cost

A Fuzzy Inventory Model With Lot Size Dependent Carrying / Holding Cost IOSR Journl of Mthemtics (IOSR-JM e-issn: 78-578,p-ISSN: 9-765X, Volume 7, Issue 6 (Sep. - Oct. 0, PP 06-0 www.iosrournls.org A Fuzzy Inventory Model With Lot Size Dependent Crrying / olding Cost P. Prvthi,

More information

UNIT 7 SINGLE SAMPLING PLANS

UNIT 7 SINGLE SAMPLING PLANS UNIT 7 SINGLE SAMPLING PLANS Structure 7. Introduction Objectives 7. Single Smpling Pln 7.3 Operting Chrcteristics (OC) Curve 7.4 Producer s Risk nd Consumer s Risk 7.5 Averge Outgoing Qulity (AOQ) 7.6

More information

3: Inventory management

3: Inventory management INSE6300 Ji Yun Yu 3: Inventory mngement Concordi Februry 9, 2016 Supply chin mngement is bout mking sequence of decisions over sequence of time steps, fter mking observtions t ech of these time steps.

More information

A ppendix to. I soquants. Producing at Least Cost. Chapter

A ppendix to. I soquants. Producing at Least Cost. Chapter A ppendix to Chpter 0 Producing t est Cost This ppendix descries set of useful tools for studying firm s long-run production nd costs. The tools re isoqunts nd isocost lines. I soqunts FIGURE A0. SHOWS

More information

Cache CPI and DFAs and NFAs. CS230 Tutorial 10

Cache CPI and DFAs and NFAs. CS230 Tutorial 10 Cche CPI nd DFAs nd NFAs CS230 Tutoril 10 Multi-Level Cche: Clculting CPI When memory ccess is ttempted, wht re the possible results? ccess miss miss CPU L1 Cche L2 Cche Memory L1 cche hit L2 cche hit

More information

3/1/2016. Intermediate Microeconomics W3211. Lecture 7: The Endowment Economy. Today s Aims. The Story So Far. An Endowment Economy.

3/1/2016. Intermediate Microeconomics W3211. Lecture 7: The Endowment Economy. Today s Aims. The Story So Far. An Endowment Economy. 1 Intermedite Microeconomics W3211 Lecture 7: The Endowment Economy Introduction Columbi University, Spring 2016 Mrk Den: mrk.den@columbi.edu 2 The Story So Fr. 3 Tody s Aims 4 Remember: the course hd

More information

INF 4130 Exercise set 4

INF 4130 Exercise set 4 INF 4130 Exercise set 4 Exercise 1 List the order in which we extrct the nodes from the Live Set queue when we do redth first serch of the following grph (tree) with the Live Set implemented s LIFO queue.

More information

JFE Online Appendix: The QUAD Method

JFE Online Appendix: The QUAD Method JFE Online Appendix: The QUAD Method Prt of the QUAD technique is the use of qudrture for numericl solution of option pricing problems. Andricopoulos et l. (00, 007 use qudrture s the only computtionl

More information

On the Complexity of Computing the Justification Status of an Argument

On the Complexity of Computing the Justification Status of an Argument On the Complexity of Computing the Justifiction Sttus of n Argument dbi Reserch Seminr, Vienn Wolfgng Dvořák Institute of Informtion Systems, Vienn University of Technology Oct 13, 2011 Supported by the

More information

3. Argumentation Frameworks

3. Argumentation Frameworks 3. Argumenttion Frmeworks Argumenttion current hot topic in AI. Historiclly more recent thn other pproches discussed here. Bsic ide: to construct cceptble set(s) of beliefs from given KB: 1 construct rguments

More information

UNinhabited aerial vehicles (UAVs) are becoming increasingly

UNinhabited aerial vehicles (UAVs) are becoming increasingly A Process Algebr Genetic Algorithm Sertc Krmn Tl Shim Emilio Frzzoli Abstrct A genetic lgorithm tht utilizes process lgebr for coding of solution chromosomes nd for defining evolutionry bsed opertors is

More information

The Okun curve is non-linear

The Okun curve is non-linear Economics Letters 70 (00) 53 57 www.elsevier.com/ locte/ econbse The Okun curve is non-liner Mtti Viren * Deprtment of Economics, 004 University of Turku, Turku, Finlnd Received 5 My 999; ccepted 0 April

More information

Outline. CSE 326: Data Structures. Priority Queues Leftist Heaps & Skew Heaps. Announcements. New Heap Operation: Merge

Outline. CSE 326: Data Structures. Priority Queues Leftist Heaps & Skew Heaps. Announcements. New Heap Operation: Merge CSE 26: Dt Structures Priority Queues Leftist Heps & Skew Heps Outline Announcements Leftist Heps & Skew Heps Reding: Weiss, Ch. 6 Hl Perkins Spring 2 Lectures 6 & 4//2 4//2 2 Announcements Written HW

More information

Time Scales: From Nabla Calculus to Delta Calculus and Vice Versa via Duality

Time Scales: From Nabla Calculus to Delta Calculus and Vice Versa via Duality Interntionl Journl of Difference Equtions ISSN 0973-6069, Volume 5, Number 1, pp. 25 40 (2010) http://cmpus.mst.edu/ijde Time Scles: From Nbl Clculus to Delt Clculus nd Vice Vers vi Dulity M. Cristin Cputo

More information

(a) by substituting u = x + 10 and applying the result on page 869 on the text, (b) integrating by parts with u = ln(x + 10), dv = dx, v = x, and

(a) by substituting u = x + 10 and applying the result on page 869 on the text, (b) integrating by parts with u = ln(x + 10), dv = dx, v = x, and Supplementry Questions for HP Chpter 5. Derive the formul ln( + 0) d = ( + 0) ln( + 0) + C in three wys: () by substituting u = + 0 nd pplying the result on pge 869 on the tet, (b) integrting by prts with

More information

Young differential equations with power type nonlinearities

Young differential equations with power type nonlinearities Avilble online t www.sciencedirect.com ScienceDirect Stochstic Processes nd their Applictions 127 (217) 342 367 www.elsevier.com/locte/sp Young differentil equtions with power type nonlinerities Jorge

More information

Rational Equity Bubbles

Rational Equity Bubbles ANNALS OF ECONOMICS AND FINANCE 14-2(A), 513 529 (2013) Rtionl Equity Bubbles Ge Zhou * College of Economics, Zhejing University Acdemy of Finncil Reserch, Zhejing University E-mil: flhszh@gmil.com This

More information

PRICING CONVERTIBLE BONDS WITH KNOWN INTEREST RATE. Jong Heon Kim

PRICING CONVERTIBLE BONDS WITH KNOWN INTEREST RATE. Jong Heon Kim Kngweon-Kyungki Mth. Jour. 14 2006, No. 2, pp. 185 202 PRICING CONVERTIBLE BONDS WITH KNOWN INTEREST RATE Jong Heon Kim Abstrct. In this pper, using the Blck-Scholes nlysis, we will derive the prtil differentil

More information

The Market Approach to Valuing Businesses (Second Edition)

The Market Approach to Valuing Businesses (Second Edition) BV: Cse Anlysis Completed Trnsction & Guideline Public Comprble MARKET APPROACH The Mrket Approch to Vluing Businesses (Second Edition) Shnnon P. Prtt This mteril is reproduced from The Mrket Approch to

More information

USE OF MATHEMATICAL EXPRESSIONS CHALLENGES IN ECONOMICS CLASSROOMS. Ashita Raveendran, Lecturer, DESSH, NCERT, New Delhi

USE OF MATHEMATICAL EXPRESSIONS CHALLENGES IN ECONOMICS CLASSROOMS. Ashita Raveendran, Lecturer, DESSH, NCERT, New Delhi USE OF MATHEMATICAL EXPRESSIONS CHALLENGES IN ECONOMICS CLASSROOMS Ashit Rveendrn, Lecturer, DESSH, NCERT, New Delhi Mthemtics definition s Science of order (White hed (1929)) points on to its fetures

More information

What is Monte Carlo Simulation? Monte Carlo Simulation

What is Monte Carlo Simulation? Monte Carlo Simulation Wht is Monte Crlo Simultion? Monte Crlo methods re widely used clss of computtionl lgorithms for simulting the ehvior of vrious physicl nd mthemticl systems, nd for other computtions. Monte Crlo lgorithm

More information

Database System Concepts, 5 th Ed. Silberschatz, Korth and Sudarshan See for conditions on re-use

Database System Concepts, 5 th Ed. Silberschatz, Korth and Sudarshan See  for conditions on re-use Chpter 2: Reltionl Model Dtbse System Concepts, 5 th Ed. See www.db-book.com for conditions on re-use Bnking Exmple brnch (brnch_nme, brnch_city, ssets) customer (customer_nme, customer_street, customer_city)

More information

A Closer Look at Bond Risk: Duration

A Closer Look at Bond Risk: Duration W E B E X T E S I O 4C A Closer Look t Bond Risk: Durtion This Extension explins how to mnge the risk of bond portfolio using the concept of durtion. BOD RISK In our discussion of bond vlution in Chpter

More information

Burrows-Wheeler Transform and FM Index

Burrows-Wheeler Transform and FM Index Burrows-Wheeler Trnsform nd M Index Ben ngmed You re free to use these slides. If you do, plese sign the guestbook (www.lngmed-lb.org/teching-mterils), or emil me (ben.lngmed@gmil.com) nd tell me briefly

More information

Do We Really Need Gaussian Filters for Feature Detection? (Supplementary Material)

Do We Really Need Gaussian Filters for Feature Detection? (Supplementary Material) Do We Relly Need Gussin Filters for Feture Detection? (Supplementry Mteril) Lee-Kng Liu, Stnley H. Chn nd Truong Nguyen Februry 5, 0 This document is supplementry mteril to the pper submitted to EUSIPCO

More information

ASYMMETRIC SWITCHING COSTS CAN IMPROVE THE PREDICTIVE POWER OF SHY S MODEL

ASYMMETRIC SWITCHING COSTS CAN IMPROVE THE PREDICTIVE POWER OF SHY S MODEL Document de trvil 2012-14 / April 2012 ASYMMETRIC SWITCHIG COSTS CA IMPROVE THE PREDICTIVE POWER OF SHY S MODEL Evens Slies OFCE-Sciences-Po Asymmetric switching costs cn improve the predictive power of

More information

Problem Set for Chapter 3: Simple Regression Analysis ECO382 Econometrics Queens College K.Matsuda

Problem Set for Chapter 3: Simple Regression Analysis ECO382 Econometrics Queens College K.Matsuda Problem Set for Chpter 3 Simple Regression Anlysis ECO382 Econometrics Queens College K.Mtsud Excel Assignments You re required to hnd in these Excel Assignments by the due Mtsud specifies. Legibility

More information

Option exercise with temptation

Option exercise with temptation Economic Theory 2008) 34: 473 501 DOI 10.1007/s00199-006-0194-3 RESEARCH ARTICLE Jinjun Mio Option exercise with tempttion Received: 25 Jnury 2006 / Revised: 5 December 2006 / Published online: 10 Jnury

More information

Reinforcement Learning. CS 188: Artificial Intelligence Fall Grid World. Markov Decision Processes. What is Markov about MDPs?

Reinforcement Learning. CS 188: Artificial Intelligence Fall Grid World. Markov Decision Processes. What is Markov about MDPs? CS 188: Artificil Intelligence Fll 2010 Lecture 9: MDP 9/2/2010 Reinforcement Lerning [DEMOS] Bic ide: Receive feedbck in the form of rewrd Agent utility i defined by the rewrd function Mut (lern to) ct

More information

Today s Outline. One More Operation. Priority Queues. New Operation: Merge. Leftist Heaps. Priority Queues. Admin: Priority Queues

Today s Outline. One More Operation. Priority Queues. New Operation: Merge. Leftist Heaps. Priority Queues. Admin: Priority Queues Tody s Outline Priority Queues CSE Dt Structures & Algorithms Ruth Anderson Spring 4// Admin: HW # due this Thursdy / t :9pm Printouts due Fridy in lecture. Priority Queues Leftist Heps Skew Heps 4// One

More information

Problem Set 4 - Solutions. Suppose when Russia opens to trade, it imports automobiles, a capital-intensive good.

Problem Set 4 - Solutions. Suppose when Russia opens to trade, it imports automobiles, a capital-intensive good. roblem Set 4 - Solutions uestion Suppose when ussi opens to trde, it imports utomobiles, cpitl-intensive good. ) According to the Heckscher-Ohlin theorem, is ussi cpitl bundnt or lbor bundnt? Briefly explin.

More information

MARKET POWER AND MISREPRESENTATION

MARKET POWER AND MISREPRESENTATION MARKET POWER AND MISREPRESENTATION MICROECONOMICS Principles nd Anlysis Frnk Cowell Note: the detil in slides mrked * cn only e seen if you run the slideshow July 2017 1 Introduction Presenttion concerns

More information

Open Space Allocation and Travel Costs

Open Space Allocation and Travel Costs Open Spce Alloction nd Trvel Costs By Kent Kovcs Deprtment of Agriculturl nd Resource Economics University of Cliforni, Dvis kovcs@priml.ucdvis.edu Pper prepred for presenttion t the Americn Agriculturl

More information

Math F412: Homework 4 Solutions February 20, κ I = s α κ α

Math F412: Homework 4 Solutions February 20, κ I = s α κ α All prts of this homework to be completed in Mple should be done in single worksheet. You cn submit either the worksheet by emil or printout of it with your homework. 1. Opre 1.4.1 Let α be not-necessrily

More information

CS 188 Introduction to Artificial Intelligence Fall 2018 Note 4

CS 188 Introduction to Artificial Intelligence Fall 2018 Note 4 CS 188 Introduction to Artificil Intelligence Fll 2018 Note 4 These lecture notes re hevily bsed on notes originlly written by Nikhil Shrm. Non-Deterministic Serch Picture runner, coming to the end of

More information

Arithmetic and Geometric Sequences

Arithmetic and Geometric Sequences Arithmetic nd Geometric Sequences A sequence is list of numbers or objects, clled terms, in certin order. In n rithmetic sequence, the difference between one term nd the next is lwys the sme. This difference

More information

1. Detailed information about the Appellant s and Respondent s personal information including mobile no. and -id are to be furnished.

1. Detailed information about the Appellant s and Respondent s personal information including mobile no. and  -id are to be furnished. Revised Form 36 nd Form 36A for filing ppel nd cross objection respectively before income tx ppellte tribunl (Notifiction No. 72 dted 23.10.2018) Bckground CBDT issued drft notifiction vide press relese

More information

Chapter 3: The Reinforcement Learning Problem. The Agent'Environment Interface. Getting the Degree of Abstraction Right. The Agent Learns a Policy

Chapter 3: The Reinforcement Learning Problem. The Agent'Environment Interface. Getting the Degree of Abstraction Right. The Agent Learns a Policy Chpter 3: The Reinforcement Lerning Problem The Agent'Environment Interfce Objectives of this chpter: describe the RL problem we will be studying for the reminder of the course present idelized form of

More information

Topics Covered: Rules of indices Expanding bracketed expressions Factorisation of simple algebraic expressions

Topics Covered: Rules of indices Expanding bracketed expressions Factorisation of simple algebraic expressions Algebr Workshop: Algebr Topics Covered: Rules of indices Expnding brcketed expressions Fctoristion of siple lgebric expressions Powers A power is used when nuber is ultiplied by itself severl ties. For

More information

Measuring Search Trees

Measuring Search Trees Mesuring Serch Trees Christin Bessiere 1, Bruno Znuttini 2, nd Cèsr Fernández 3 1 LIRMM-CNRS, Montpellier, Frnce 2 GREYC, Cen, Frnce 3 Univ. de Lleid, Lleid, Spin Astrct. The SAT nd CSP communities mke

More information

MATH 236 ELAC MATH DEPARTMENT FALL 2017 SHORT ANSWER. Write the word or phrase that best completes each statement or answers the question.

MATH 236 ELAC MATH DEPARTMENT FALL 2017 SHORT ANSWER. Write the word or phrase that best completes each statement or answers the question. MATH 236 ELAC MATH DEPARTMENT FALL 2017 TEST 1 REVIEW SHORT ANSWER. Write the word or phrse tht best completes ech sttement or nswers the question. 1) The supply nd demnd equtions for certin product re

More information

THE FINAL PROOF SUPPORTING THE TURNOVER FORMULA.

THE FINAL PROOF SUPPORTING THE TURNOVER FORMULA. THE FINAL PROOF SUPPORTING THE TURNOVER FORMULA. I would like to thnk Aris for his mthemticl contriutions nd his swet which hs enled deeper understnding of the turnover formul to emerge. His contriution

More information

UNIVERSITY OF NOTTINGHAM. Discussion Papers in Economics BERTRAND VS. COURNOT COMPETITION IN ASYMMETRIC DUOPOLY: THE ROLE OF LICENSING

UNIVERSITY OF NOTTINGHAM. Discussion Papers in Economics BERTRAND VS. COURNOT COMPETITION IN ASYMMETRIC DUOPOLY: THE ROLE OF LICENSING UNIVERSITY OF NOTTINGHAM Discussion Ppers in Economics Discussion Pper No. 0/0 BERTRAND VS. COURNOT COMPETITION IN ASYMMETRIC DUOPOLY: THE ROLE OF LICENSING by Arijit Mukherjee April 00 DP 0/0 ISSN 160-48

More information

Synchronizing Asynchronous IOCO

Synchronizing Asynchronous IOCO Synchronizing Asynchronous IOCO Ned Noroozi 1,2, Rmtin Khosrvi 3, MohmmdRez Mousvi 1, nd Tim Willemse 1 1 Eindhoven University of Technology, Eindhoven, The Netherlnds 2 Fnp Corportion (IT Subsidiry of

More information

DYNAMIC PROGRAMMING REINFORCEMENT LEARNING. COGS 202 : Week 7 Presentation

DYNAMIC PROGRAMMING REINFORCEMENT LEARNING. COGS 202 : Week 7 Presentation DYNAMIC PROGRAMMING REINFORCEMENT LEARNING COGS 202 : Week 7 Preenttion OUTLINE Recp (Stte Vlue nd Action Vlue function) Computtion in MDP Dynmic Progrmming (DP) Policy Evlution Policy Improvement Policy

More information

The Saga of Reduced Factorizations of Elements of the Symmetric Group

The Saga of Reduced Factorizations of Elements of the Symmetric Group Topics in Algebric Combintorics LECTURE NOTES April, 00 The Sg of Reduced Fctoriztions of Elements of the Symmetric Group by A. M. Grsi Abstrct These notes cover the contents of series of lectures in Topics

More information

Trigonometry - Activity 21 General Triangle Solution: Given three sides.

Trigonometry - Activity 21 General Triangle Solution: Given three sides. Nme: lss: p 43 Mths Helper Plus Resoure Set. opyright 003 rue. Vughn, Tehers hoie Softwre Trigonometry - tivity 1 Generl Tringle Solution: Given three sides. When the three side lengths '', '' nd '' of

More information

Optimal firm's policy under lead time- and price-dependent demand: interest of customers rejection policy

Optimal firm's policy under lead time- and price-dependent demand: interest of customers rejection policy Optiml firm's policy under led time- nd price-dependent demnd: interest of customers rejection policy Abduh Syid Albn Université Grenoble Alpes, G-SCOP, F-38000 Grenoble, Frnce bduh-syid.lbn@grenoble-inp.org

More information

A portfolio approach to the optimal funding of pensions

A portfolio approach to the optimal funding of pensions Economics Letters 69 (000) 01 06 www.elsevier.com/ locte/ econbse A portfolio pproch to the optiml funding of pensions Jysri Dutt, Sndeep Kpur *, J. Michel Orszg b, b Fculty of Economics University of

More information

Drawing Finite State Machines in L A TEX and TikZ A Tutorial

Drawing Finite State Machines in L A TEX and TikZ A Tutorial Drwing Finite Stte Mchines in L A TEX nd TikZ A Tutoril Styki Sikdr nd Dvid Ching ssikdr@nd.edu Version 3 Jnury 7, 208 Introduction L A TEX (pronounced ly-tek) is n open-source, multipltform document preprtion

More information

Tble 1: Syntx of LOTOS/T+ E :: stop (untimed dedlock) j exit (successful termintion) j ; E (ction prex, untimed) j [P (t; x)]; E (ction prex, timed) j

Tble 1: Syntx of LOTOS/T+ E :: stop (untimed dedlock) j exit (successful termintion) j ; E (ction prex, untimed) j [P (t; x)]; E (ction prex, timed) j Protocol Synthesis from Timed nd Structured Specictions Akio Nkt Teruo Higshino Kenichi Tniguchi Dept. of Informtion nd Computer Sciences, Osk University Toyonk, Osk 56, Jpn Abstrct In this pper, we propose

More information

OPEN BUDGET QUESTIONNAIRE SOUTH AFRICA

OPEN BUDGET QUESTIONNAIRE SOUTH AFRICA Interntionl Budget Prtnership OPEN BUDGET QUESTIONNAIRE SOUTH AFRICA September 28, 2007 Interntionl Budget Prtnership Center on Budget nd Policy Priorities 820 First Street, NE Suite 510 Wshington, DC

More information

Announcements. CS 188: Artificial Intelligence Fall Reinforcement Learning. Markov Decision Processes. Example Optimal Policies.

Announcements. CS 188: Artificial Intelligence Fall Reinforcement Learning. Markov Decision Processes. Example Optimal Policies. CS 188: Artificil Intelligence Fll 2008 Lecture 9: MDP 9/25/2008 Announcement Homework olution / review eion: Mondy 9/29, 7-9pm in 2050 Vlley LSB Tuedy 9/0, 6-8pm in 10 Evn Check web for detil Cover W1-2,

More information

Complete the table below to show the fixed, variable and total costs. In the final column, calculate the profit or loss made by J Kane.

Complete the table below to show the fixed, variable and total costs. In the final column, calculate the profit or loss made by J Kane. Tsk 1 J Kne sells mchinery to the frm industry. His fixed costs re 10,000 nd ech mchine costs him 400 to buy. He sells them t 600 nd is trying to work out his profit or loss t vrious levels of sles. He

More information

Menu costs, firm size and price rigidity

Menu costs, firm size and price rigidity Economics Letters 66 (2000) 59 63 www.elsevier.com/ locte/ econbse Menu costs, firm size nd price rigidity Robert A. Buckle *, John A. Crlson, b School of Economics nd Finnce, Victori University of Wellington,

More information

Unifying Synchronous Tree-Adjoining Grammars and Tree Transducers via Bimorphisms

Unifying Synchronous Tree-Adjoining Grammars and Tree Transducers via Bimorphisms Unifying ynchronous Tree-Adjoining Grmmrs nd Tree Trnsducers vi Bimorphisms turt M. hieer Division of Engineering nd Applied ciences Hrvrd University Cmridge, MA, UA shieer@des.hrvrd.edu turt M. hieer.

More information

On Moments of Folded and Truncated Multivariate Normal Distributions

On Moments of Folded and Truncated Multivariate Normal Distributions On Moments of Folded nd Truncted Multivrite Norml Distributions Rymond Kn Rotmn School of Mngement, University of Toronto 05 St. George Street, Toronto, Ontrio M5S 3E6, Cnd (E-mil: kn@chss.utoronto.c)

More information

Technical Report Global Leader Dry Bulk Derivatives

Technical Report Global Leader Dry Bulk Derivatives Soybens Mrch 17 - Weekly Soybens Mrch 17 - Dily Weekly Close US$ 1,054 ½ RSI 59 MACD Bullish The hisgrm is widening S1 US$ 1,016 ½ S2 US$ 993 R1 US$ 1,071 R2 US$ 1,096 Dily Close US$ 1,030 RSI 60 MACD

More information

The Combinatorial Seller s Bid Double Auction: An Asymptotically Efficient Market Mechanism*

The Combinatorial Seller s Bid Double Auction: An Asymptotically Efficient Market Mechanism* The Combintoril Seller s Bid Double Auction: An Asymptoticlly Efficient Mret Mechnism* Rhul Jin IBM Wtson Reserch Hwthorne, NY rhul.jin@us.ibm.com Prvin Vriy EECS Deprtment University of Cliforni, Bereley

More information

Pushdown Automata. Courtesy: Costas Busch RPI

Pushdown Automata. Courtesy: Costas Busch RPI Pushdown Automt Courtesy: Costs Busch RPI Pushdown Automt Pushdown Automt Pushdown Automt Pushdown Automt Pushdown Automt Pushdown Automt Non-Determinism:NPDA PDAs re non-deterministic: non-deterministic

More information

PSAS: Government transfers what you need to know

PSAS: Government transfers what you need to know PSAS: Government trnsfers wht you need to know Ferury 2018 Overview This summry will provide users with n understnding of the significnt recognition, presenttion nd disclosure requirements of the stndrd.

More information

Exhibit A Covered Employee Notification of Rights Materials Regarding Allied Managed Care Incorporated Allied Managed Care MPN MPN ID # 2360

Exhibit A Covered Employee Notification of Rights Materials Regarding Allied Managed Care Incorporated Allied Managed Care MPN MPN ID # 2360 Covered Notifiction of Rights Mterils Regrding Allied Mnged Cre Incorported Allied Mnged Cre MPN This pmphlet contins importnt informtion bout your medicl cre in cse of workrelted injmy or illness You

More information

The Combinatorial Seller s Bid Double Auction: An Asymptotically Efficient Market Mechanism*

The Combinatorial Seller s Bid Double Auction: An Asymptotically Efficient Market Mechanism* The Combintoril Seller s Bid Double Auction: An Asymptoticlly Efficient Mret Mechnism* Rhul Jin nd Prvin Vriy EECS Deprtment University of Cliforni, Bereley (rjin,vriy)@eecs.bereley.edu We consider the

More information

OPEN BUDGET QUESTIONNAIRE UKRAINE

OPEN BUDGET QUESTIONNAIRE UKRAINE Interntionl Budget Prtnership OPEN BUDGET QUESTIONNAIRE UKRAINE September 28, 2007 Interntionl Budget Prtnership Center on Budget nd Policy Priorities 820 First Street, NE Suite 510 Wshington, DC 20002

More information

Chapter 4. Profit and Bayesian Optimality

Chapter 4. Profit and Bayesian Optimality Chpter 4 Profit nd Byesin Optimlity In this chpter we consider the objective of profit. The objective of profit mximiztion dds significnt new chllenge over the previously considered objective of socil

More information

CHAPTER-IV PRE-TEST ESTIMATOR OF REGRESSION COEFFICIENTS: PERFORMANCE UNDER LINEX LOSS FUNCTION

CHAPTER-IV PRE-TEST ESTIMATOR OF REGRESSION COEFFICIENTS: PERFORMANCE UNDER LINEX LOSS FUNCTION CHAPTER-IV PRE-TEST ESTIMATOR OF REGRESSION COEFFICIENTS: PERFORMANCE UNDER LINEX LOSS FUNCTION 4.1 INTRODUCTION It hs lredy been demonstrted tht the restricted lest squres estimtor is more efficient thn

More information

Outline. CS 188: Artificial Intelligence Spring Speeding Up Game Tree Search. Minimax Example. Alpha-Beta Pruning. Pruning

Outline. CS 188: Artificial Intelligence Spring Speeding Up Game Tree Search. Minimax Example. Alpha-Beta Pruning. Pruning CS 188: Artificil Intelligence Spring 2011 Lecture 8: Gme, MDP 2/14/2010 Pieter Abbeel UC Berkeley Mny lide dpted from Dn Klein Outline Zero-um determinitic two plyer gme Minimx Evlution function for non-terminl

More information

Interacting with mathematics in Key Stage 3. Year 9 proportional reasoning: mini-pack

Interacting with mathematics in Key Stage 3. Year 9 proportional reasoning: mini-pack Intercting with mthemtics in Key Stge Yer 9 proportionl resoning: mini-pck Intercting with mthemtics Yer 9 proportionl resoning: mini-pck Crown copyright 00 Contents Yer 9 proportionl resoning: smple unit

More information

Inequality and the GB2 income distribution

Inequality and the GB2 income distribution Working Pper Series Inequlity nd the GB2 income distribution Stephen P. Jenkins ECINEQ WP 2007 73 ECINEC 2007-73 July 2007 www.ecineq.org Inequlity nd the GB2 income distribution Stephen P. Jenkins* University

More information

Math 205 Elementary Algebra Fall 2010 Final Exam Study Guide

Math 205 Elementary Algebra Fall 2010 Final Exam Study Guide Mth 0 Elementr Algebr Fll 00 Finl Em Stud Guide The em is on Tuesd, December th from :0m :0m. You re llowed scientific clcultor nd " b " inde crd for notes. On our inde crd be sure to write n formuls ou

More information

Technical Report Global Leader Dry Bulk Derivatives. FIS Technical - Grains And Ferts. Highlights:

Technical Report Global Leader Dry Bulk Derivatives. FIS Technical - Grains And Ferts. Highlights: Technicl Report Technicl Anlyst FIS Technicl - Grins And Ferts Edwrd Hutn 44 20 7090 1120 Edwrdh@freightinvesr.com Highlights: SOY The weekly schstic is wrning slowing momentum in the mrket. USD 966 ¼

More information

Notes on the BENCHOP implementations for the COS method

Notes on the BENCHOP implementations for the COS method Notes on the BENCHOP implementtions for the COS method M. J. uijter C. W. Oosterlee Mrch 29, 2015 Abstrct This text describes the COS method nd its implementtion for the BENCHOP-project. 1 Fourier cosine

More information

Gridworld Values V* Gridworld: Q*

Gridworld Values V* Gridworld: Q* CS 188: Artificil Intelligence Mrkov Deciion Procee II Intructor: Dn Klein nd Pieter Abbeel --- Univerity of Cliforni, Berkeley [Thee lide were creted by Dn Klein nd Pieter Abbeel for CS188 Intro to AI

More information

International Monopoly under Uncertainty

International Monopoly under Uncertainty Interntionl Monopoly under Uncertinty Henry Ary University of Grnd Astrct A domestic monopolistic firm hs the option to service foreign mrket through export or y setting up plnt in the host country under

More information

checks are tax current income.

checks are tax current income. Humn Short Term Disbility Pln Wht is Disbility Insurnce? An esy explntion is; Disbility Insurnce is protection for your pycheck. Imgine if you were suddenly disbled, unble to work, due to n ccident or

More information

Characterizing Higher-Order Ross More Risk Aversion by Comparison of Risk Compensation

Characterizing Higher-Order Ross More Risk Aversion by Comparison of Risk Compensation Chrcterizing Higher-Order Ross More Risk Aversion by Comprison of Risk Compenstion Guoqing Tin Yougong Tin b,c Deprtment of Economics, Texs A&M University, College Sttion, TX77843, USA b School of Economics,

More information

production for Community & Culture Project Reference e 2 design episodes Bogotá: Building a Sustainable City and Affordable Green Housing.

production for Community & Culture Project Reference e 2 design episodes Bogotá: Building a Sustainable City and Affordable Green Housing. Community & Culture Project Reference e 2 design episodes Bogotá: Building Sustinble City nd Affordble Green Housing. 1) Red the bckground essy nd discussion questions for e 2 design episodes Bogotá: Building

More information

OPEN BUDGET QUESTIONNAIRE MACEDONIA

OPEN BUDGET QUESTIONNAIRE MACEDONIA Interntionl Budget Prtnership OPEN BUDGET QUESTIONNAIRE MACEDONIA September 28, 2007 Interntionl Budget Prtnership Center on Budget nd Policy Priorities 820 First Street, NE Suite 510 Wshington, DC 20002

More information

1 Manipulation for binary voters

1 Manipulation for binary voters STAT 206A: Soil Choie nd Networks Fll 2010 Mnipultion nd GS Theorem Otoer 21 Leturer: Elhnn Mossel Srie: Kristen Woyh In this leture we over mnipultion y single voter: whether single voter n lie out his

More information

Access your online resources today at

Access your online resources today at 978--07-670- - CmbridgeMths: NSW Syllbus for the Austrlin Curriculum: Yer 0: Stte./. Access your online resources tody t www.cmbridge.edu.u/go. Log in to your existing Cmbridge GO user ccount or crete

More information

The Effects of Taxation on Income-Producing Crimes with Variable. Leisure Time

The Effects of Taxation on Income-Producing Crimes with Variable. Leisure Time The Effects of Txtion on Income-Producing Crimes with Vrible Leisure Time Avrhm D. Tbbch I. INTRODUCTION The existing literture on the effects of txtion on income-producing crimes lys clim to severl importnt

More information

Recap: MDPs. CS 188: Artificial Intelligence Fall Optimal Utilities. The Bellman Equations. Value Estimates. Practice: Computing Actions

Recap: MDPs. CS 188: Artificial Intelligence Fall Optimal Utilities. The Bellman Equations. Value Estimates. Practice: Computing Actions CS 188: Artificil Intelligence Fll 2008 Lecture 10: MDP 9/30/2008 Dn Klein UC Berkeley Recp: MDP Mrkov deciion procee: Stte S Action A Trnition P(,) (or T(,, )) Rewrd R(,, ) (nd dicount γ) Strt tte 0 Quntitie:

More information

Asymptotic Stability of a Rate Control System. with Communication Delays

Asymptotic Stability of a Rate Control System. with Communication Delays Asymptotic Stbility of Rte Control System with Communiction Delys Richrd J. L nd Priy Rnjn University of Mrylnd, College Prk {hyongl, priy}@eng.umd.edu 1 Abstrct We study the issue of symptotic stbility

More information

Name Date. Find the LCM of the numbers using the two methods shown above.

Name Date. Find the LCM of the numbers using the two methods shown above. Lest Common Multiple Multiples tht re shred by two or more numbers re clled common multiples. The lest of the common multiples is clled the lest common multiple (LCM). There re severl different wys to

More information

Technical Appendix. The Behavior of Growth Mixture Models Under Nonnormality: A Monte Carlo Analysis

Technical Appendix. The Behavior of Growth Mixture Models Under Nonnormality: A Monte Carlo Analysis Monte Crlo Technicl Appendix 1 Technicl Appendix The Behvior of Growth Mixture Models Under Nonnormlity: A Monte Crlo Anlysis Dniel J. Buer & Ptrick J. Currn 10/11/2002 These results re presented s compnion

More information

PSAKUIJIR Vol. 4 No. 2 (July-December 2015)

PSAKUIJIR Vol. 4 No. 2 (July-December 2015) Resonble Concession Period for Build Operte Trnsfer Contrct Projects: A Cse Study of Theun-Hiboun Hydropower Dm Project nd Ntionl Rod No. 14 A Project Pnysith Vorsing * nd Dr.Sounthone Phommsone ** Abstrct

More information

A Sharper Ratio: A General Measure for Correctly Ranking Non-Normal Investment Risks

A Sharper Ratio: A General Measure for Correctly Ranking Non-Normal Investment Risks A Shrper Rtio: A Generl Mesure for Correctly Rnking on-orml Investment Risks Kent Smetters Xingtn Zhng This Version: Februry 3, 2014 Abstrct While the Shrpe rtio is still the dominnt mesure for rnking

More information

Problem Set 2 Suggested Solutions

Problem Set 2 Suggested Solutions 4.472 Prolem Set 2 Suggested Solutions Reecc Zrutskie Question : First find the chnge in the cpitl stock, k, tht will occur when the OLG economy moves to the new stedy stte fter the government imposes

More information

OPEN BUDGET QUESTIONNAIRE FRANCE

OPEN BUDGET QUESTIONNAIRE FRANCE Interntionl Budget Project OPEN BUDGET QUESTIONNAIRE FRANCE October 2005 Interntionl Budget Project Center on Budget nd Policy Priorities 820 First Street, NE Suite 510 Wshington, DC 20002 www.interntionlbudget.org

More information

Controlling a population of identical MDP

Controlling a population of identical MDP Controlling popultion of identicl MDP Nthlie Bertrnd Inri Rennes ongoing work with Miheer Dewskr (CMI), Blise Genest (IRISA) nd Hugo Gimert (LBRI) Trends nd Chllenges in Quntittive Verifiction Mysore,

More information

Buckling of Stiffened Panels 1 overall buckling vs plate buckling PCCB Panel Collapse Combined Buckling

Buckling of Stiffened Panels 1 overall buckling vs plate buckling PCCB Panel Collapse Combined Buckling Buckling of Stiffened Pnels overll uckling vs plte uckling PCCB Pnel Collpse Comined Buckling Vrious estimtes hve een developed to determine the minimum size stiffener to insure the plte uckles while the

More information

A Static Model for Voting on Social Security

A Static Model for Voting on Social Security A Sttic Model for Voting on Socil Security Henning Bohn Deprtment of Economics University of Cliforni t Snt Brbr Snt Brbr, CA 93106, USA; nd CESifo Phone: 1-805-893-4532; Fx: 1-805-893-8830. E-mil: bohn@econ.ucsb.edu

More information