How to train your multi bottom-up tree transducer
|
|
- Jessie Gibson
- 5 years ago
- Views:
Transcription
1 How to trin your multi bottom-up tree trnsducer Andres Mletti Universität tuttgrt Institute for Nturl Lnguge Processing Azenbergstrße 7074 tuttgrt Germny Abstrct The locl multi bottom-up tree trnsducer is introduced nd relted to the (non-contiguous) synchronous tree sequence substitution grmmr. It is then shown how to obtin weighted locl multi bottom-up tree trnsducer from bilingul nd biprsed corpus. Finlly the problem of non-preservtion of regulrity is ddressed. Three properties tht ensure preservtion re introduced nd it is discussed how to djust the rule extrction process such tht they re utomticlly fulfilled. Introduction A (forml) trnsltion model is t the core of every mchine trnsltion system. Predominntly sttisticl processes re used to instntite the forml model nd derive specific trnsltion device. Brown et l. (990) discuss utomticlly trinble trnsltion models in their seminl pper. However the IBM models of Brown et l. (993) re stringbsed in the sense tht they bse the trnsltion decision on the words nd their surrounding context. Contrry in the field of syntx-bsed mchine trnsltion the trnsltion models hve full ccess to the syntx of the sentences nd cn bse their decision on it. A good exposition to both fields is presented in (Knight 007). In this pper we del exclusively with syntxbsed trnsltion models such s synchronous tree substitution grmmrs (TG) multi bottom-up tree trnsducers (MBOT) nd synchronous tree-sequence substitution grmmrs (TG). Ching (006) gives good introduction to TG which originte from the syntx-directed trnsltion schemes of Aho nd Ullmn (97). Roughly speking n TG hs rules in which two linked nonterminls re replced (t the sme time) by two corresponding trees contining terminl nd nonterminl symbols. In ddition the nonterminls in the two replcement trees re linked which cretes new linked nonterminls to which further rules cn be pplied. Henceforth we refer to these two trees s input nd output tree. MBOT hve been introduced in (Arnold nd Duchet 98; Lilin 98) nd re slightly more expressive thn TG. Roughly speking they llow one replcement input tree nd severl output trees in single rule. This chnge nd the presence of sttes yields mny lgorithmiclly dvntgeous properties such s closure under composition efficient binriztion nd efficient input nd output restriction [see (Mletti 00)]. Finlly TG which hve been derived from rtionl tree reltions (Roult 997) hve been discussed by Zhng et l. (008) Zhng et l. (008b) nd un et l. (009). They re even more expressive thn the locl vrint of the multi bottom-up tree trnsducer (LMBOT) tht we introduce here nd cn hve severl input nd output trees in single rule. In this contribution we restrict MBOT to form tht is prticulrly relevnt in mchine trnsltion. We drop the generl stte behvior of MBOT nd replce it by the common loclity tests tht re lso present in TG TG nd TAG (hieber nd chbes 990; hieber 007). The obtined device is the locl MBOT (LMBOT). Mletti (00) rgued the lgorithmicl dvntges of MBOT over TG nd proposed MBOT s n implementtion lterntive for TG. In prticulr the trining procedure would trin TG; i.e. it would not utilize the dditionl expressive power 85 Proceedings of the 49th Annul Meeting of the Assocition for Computtionl Linguistics pges Portlnd Oregon June c 0 Assocition for Computtionl Linguistics
2 of MBOT. However Zhng et l. (008b) nd un et l. (009) demonstrte tht the dditionl expressivity gined from non-contiguous rules gretly improves the trnsltion qulity. In this contribution we ddress this seprtion nd investigte trining procedure for LMBOT tht llows non-contiguous frgments while preserving the lgorithmic dvntges of MBOT. To this end we introduce rule extrction nd weight trining method for LMBOT tht is bsed on the corresponding procedures for TG nd TG. However generl LMBOT cn be too expressive in the sense tht they llow trnsltions tht do not preserve regulrity. Preservtion of regulrity is n importnt property for efficient representtions nd efficient lgorithms [see (My et l. 00)]. Consequently we present 3 properties tht ensure tht n LMBOT preserves regulrity. In ddition we shortly discuss how these properties could be enforced in the rule extrction procedure. Nottion The set of nonnegtive integers is N. We write [k] for the set {i i k}. We tret functions s specil reltions. For every reltion R A B nd A we write R() = {b B : ( b) R} R = {(b ) ( b) R} where R is clled the inverse of R. Given n lphbet Σ the set of ll words (or sequences) over Σ is Σ of which the empty word is ε. The conctention of two words u nd w is simply denoted by the juxtposition uw. The length of word w = σ σ k with σ i Σ for ll i [k] is w = k. Given i j k the (i j)- spn w[i j] of w is σ i σ i+ σ j. The set T Σ of ll Σ-trees is the smllest set T such tht σ(t) T for ll σ Σ nd t T. We generlly use bold-fce chrcters (like t) for sequences nd we refer to their elements using subscripts (like t i ). Consequently tree t consists of lbeled root node σ followed by sequence t of its children. To improve redbility we sometimes write sequence t t k s t... t k. The positions pos(t) N of tree t = σ(t) re inductively defined by pos(t) = {ε} pos(t) where pos(t) = {ip p pos(t i )}. i t Note tht this yields n undesirble difference between pos(t) nd pos(t) but it will lwys be cler from the context whether we refer to single tree or sequence. Note tht positions re ordered vi the (stndrd) lexicogrphic ordering. Let t T Σ nd p pos(t). The lbel of t t position p is t(p) nd the subtree rooted t position p is t p. Formlly they re defined by { σ if p = ε t(p) = t(ip) = t i (p) t(p) otherwise { t if p = ε t p = t ip = t i p t p otherwise for ll t = σ(t) nd i t. As demonstrted these notions re lso used for sequences. A position p pos(t) is lef (in t) if p / pos(t). Given subset NT Σ we let NT (t) = {p pos(t) t(p) NT p lef in t}. Lter NT will be the set of nonterminls so tht the elements of NT (t) will be the lef nonterminls of t. We extend the notion to sequences t by NT (t) = {ip p NT (t i )}. i t We lso need substitution tht replces subtrees. Let p... p n pos(t) be pirwise incomprble positions nd t... t n T Σ. Then t[p i t i i n] denotes the tree tht is obtined from t by replcing (in prllel) the subtrees t p i by t i for every i [k]. Finlly let us recll regulr tree lnguges. A finite tree utomton M is tuple (Q Σ δ F ) such tht Q is finite set δ Q Σ Q is finite reltion nd F Q. We extend δ to mpping δ : T Σ Q by δ(σ(t)) = {q (q σ q) δ i [ t ]: q i δ(t i )} for every σ Σ nd t TΣ. The finite tree utomton M recognizes the tree lnguge L(M) = {t T Σ δ(t) F }. A tree lnguge L T Σ is regulr if there exists finite tree utomton M such tht L = L(M). 86
3 VBD signed PV twl -OBJ DET-NN AltwqyE -BJ PV -OBJ -BJ Figure : mple LMBOT rules. 3 The model In this section we recll prticulr multi bottomup tree trnsducers which hve been introduced by Arnold nd Duchet (98) nd Lilin (98). A detiled (nd English) presenttion of the generl model cn be found in Engelfriet et l. (009) nd Mletti (00). Using the nomenclture of Engelfriet et l. (009) we recll vrint of liner nd nondeleting extended multi bottom-up tree trnsducers (MBOT) here. Occsionlly we will refer to generl MBOT which differ from the locl vrint discussed here becuse they hve explicit sttes. Throughout the rticle we ssume sets Σ nd of input nd output symbols respectively. Moreover let NT Σ be the set of designted nonterminl symbols. Finlly we void weights in the forml development to keep it simple. It is strightforwrd to dd weights to our model. Essentilly the model works on pirs t u consisting of n input tree t T Σ nd sequence u T of output trees. Ech such pir is clled pre-trnsltion nd the rnk rk( t u ) the pre-trnsltion t u is u. In other words the rnk of pre-trnsltion equls the number of output trees stored in it. Given pre-trnsltion t u T Σ T k nd i [k] we cll u i the i th trnsltion of t. An lignment for the pre-trnsltion t u is n injective mpping ψ : NT (u) NT (t) N such tht (p j) ψ( NT (u)) for every (p i) ψ( NT (u)) nd j [i]. In other words n lignment should request ech trnsltion of prticulr subtree t most once nd if it requests the i th trnsltion then it should lso request ll previous trnsltions. Definition A locl multi bottom-up tree trnsducer (LMBOT) is finite set R of rules such tht every rule written l ψ r contins pre-trnsltion l r nd n lignment ψ for it. The component l is the left-hnd side r is the right-hnd side nd ψ is the lignment of rule l ψ r R. The rules of n LMBOT re similr to the rules of n TG (synchronous tree substitution grmmr) of Eisner (003) nd hieber (004) but right-hnd sides of LMBOT contin sequence of trees insted of just single tree s in n TG. In ddition the lignments in n TG rule re bijective between lef nonterminls wheres our model permits multiple lignments to single lef nonterminl in the left-hnd side. A model tht is even more powerful thn LMBOT is the non-contiguous version of TG (synchronous tree-sequence substitution grmmr) of Zhng et l. (008) Zhng et l. (008b) nd un et l. (009) which llows sequences of trees on both sides of rules [see lso (Roult 997)]. Figure displys smple rules of n LMBOT using grphicl representtion of the trees nd the lignment. Next we define the semntics of n LMBOT R. To void difficulties we explicitly exclude rules like l ψ r where l NT or r NT ; i.e. rules where the left- or right-hnd side re only lef nonterminls. We first define the trditionl bottom-up semntics. Let ρ = l ψ r R be rule nd p NT (l). The p-rnk rk(ρ p) of ρ is rk(ρ p) = {i N (p i) ψ( NT (r))}. Definition The set τ(r) of pre-trnsltions of n LMBOT R is inductively defined to be the smllest set such tht: If ρ = l ψ r R is rule t p u p τ(r) is pre-trnsltion of R for every p NT (l) nd rk(ρ p) = rk( t p u p ) l(p) = t p (ε) nd Actully difficulties rise only in the weighted setting. 87
4 IN for N erbi PREP En NN-PROP rbya... VBD signed IN for PV N erbi PV -OBJ DET-NN PREP twl AltwqyE En NN-PROP rbya -OBJ... VBD twl signed IN DET-NN PREP for N AltwqyE En NN-PROP erbi rbya Figure : Top left: () Initil pre-trnsltion; Top right: (b) Pre-trnsltion obtined from the left rule of Fig. nd (); Bottom: (c) Pre-trnsltion obtined from the right rule of Fig. nd (b). r(p ) = u p (i) with ψ(p ) = (p i) for every p NT (r) then t u τ(r) where t = l[p t p p NT (l)] nd u = r[p (u p ) i p ψ (p i)]. In plin words ech nonterminl lef p in the left-hnd side of rule ρ cn be replced by the input tree t of pre-trnsltion t u whose root is lbeled by the sme nonterminl. In ddition the rnk rk(ρ p) of the replced nonterminl should mtch the rnk rk( t u ) of the pre-trnsltion nd the nonterminls in the right-hnd side tht re ligned to p should be replced by the trnsltion tht the lignment requests provided tht the nonterminl mtches with the root symbol of the requested trnsltion. The min benefit of the bottomup semntics is tht it works exclusively on pretrnsltions. The process is illustrted in Figure. Using the clssicl bottom-up semntics we simply obtin the following theorem by Mletti (00) becuse the MBOT constructed there is in fct n LMBOT. Theorem 3 For every TG n equivlent LMBOT cn be constructed in liner time which in turn yields prticulr MBOT in liner time. Finlly we wnt to relte LMBOT to the TG of un et l. (009). To this end we lso introduce the top-down semntics for LMBOT. As expected both semntics coincide. The top-down semntics is introduced using rule compositions which will ply n importnt rule lter on. Definition 4 The set R k of k-fold composed rules is inductively defined s follows: R = R nd l ϕ s R k+ for ll ρ = l ψ r R nd ρ p = l p ψp r p R k such tht rk(ρ p) = rk( l p r p ) l(p) = l p (ε) nd r(p ) = r p (i) with ψ(p ) = (p i) for every p NT (l) nd p NT (r) where l = l[p l p p NT (l)] s = r[p (r p ) i p ψ (p i)] nd ϕ(p p) = p ψ p (ip) for ll positions p ψ (p i) nd ip NT (r p ). The rule closure R of R is R = i Ri. The top-down pre-trnsltion of R is τ t (R) = { l r l ψ r R NT (l) = }. 88
5 b b b Figure 3: Composed rule. b The composition of the rules which is illustrted in Figure 3 in the second item of Definition 4 could lso be represented s ρ(ρ... ρ k ) where ρ... ρ k is n enumertion of the rules {ρ p p NT (l)} used in the item. The following theorem is esy to prove. Theorem 5 The bottom-up nd top-down semntics coincide; i.e. τ(r) = τ t (R). Ching (005) nd Grehl et l. (008) rgue tht TG hve sufficient expressive power for syntxbsed mchine trnsltion but Zhng et l. (008) show tht the dditionl expressive power of treesequences helps the trnsltion process. This is mostly due to the fct tht smller (nd less specific) rules cn be extrcted from bi-prsed word-ligned trining dt. A detiled overview tht focusses on TG is presented by Knight (007). Theorem 6 For every LMBOT n equivlent TG cn be constructed in liner time. 4 Rule extrction nd trining In this section we will show how to utomticlly obtin n LMBOT from bi-prsed word-ligned prllel corpus. Essentilly the process hs two steps: rule extrction nd trining. In the rule extrction step n (unweighted) LMBOT is extrcted from the corpus. The rule weights re then set in the trining procedure. The two min inspirtions for our rule extrction re the corresponding procedures for TG (Glley et l. 004; Grehl et l. 008) nd for TG (un et l. 009). TG re lwys contiguous in both the left- nd right-hnd side which mens tht they (completely) cover single spn of input or output words. On the contrry TG rules cn be noncontiguous on both sides but the extrction procedure of un et l. (009) only extrcts rules tht re contiguous on the left- or right-hnd side. We cn djust its st phse tht extrcts rules with (potentilly) non-contiguous right-hnd sides. The djustment is necessry becuse LMBOT rules cnnot hve (contiguous) tree sequences in their left-hnd sides. Overll the rule extrction process is sketched in Algorithm. Algorithm Rule extrction for LMBOT Require: word-ligned tree pir (t u) Return: LMBOT rules R such tht (t u) τ(r) while there exists mximl non-lef node p pos(t) nd miniml p... p k pos(u) such tht t p nd (u p... u pk ) hve consistent lignment (i.e. no lignments from within t p to lef outside (u p... u pk ) nd vice vers) do : dd rule ρ = t p ψ (u p... u pk ) to R with the nonterminl lignments ψ // excise rule ρ from (t u) 4: t t[p t(p)] u u[p i u(p i ) i {... k}] 6: estblish lignments ccording to position end while The requirement tht we cn only hve one input tree in LMBOT rules indeed might cuse the extrction of bigger nd less useful rules (when compred to the corresponding TG rules) s demonstrted in (un et l. 009). However the stricter rule shpe preserves the good lgorithmic properties of LMBOT. The more powerful TG rules cn cuse nonclosure under composition (Roult 997; Rdmcher 008) nd prsing to be less efficient. Figure 4 shows n exmple of biprsed ligned prllel text. According to the method of Glley et l. (004) we cn extrct the (miniml) TG rule displyed in Figure 5. Using the more liberl formt of LMBOT rules we cn decompose the TG rule of Figure 5 further into the rules displyed in Figure. The method of un et l. (009) would lso extrct the rule displyed in Figure 6. Let us reconsider Figures nd. Let ρ be the top left rule of Figure nd ρ nd ρ 3 be the 89
6 -BJ NML N VBD JJ N Voislv signed IN ugoslv President for N erbi AltwqyE DET-NN twl rbya En NN-PROP Alr}ys AlywgwslAfy fwyslaf PREP DET-NN DET-ADJ NN-PROP PV -OBJ -BJ Figure 4: Biprsed ligned prllel text. -BJ VBD signed PV twl -OBJ DET-NN AltwqyE Figure 5: Miniml TG rule. -BJ left nd right rule of Figure respectively. We cn represent the lower pre-trnsltion of Figure by ρ 3 ( ρ (ρ )) where ρ (ρ ) represents the upper right pre-trnsltion of Figure. If we nme ll rules of R then we cn represent ech pretrnsltion of τ(r) symboliclly by tree contining rule nmes. uch trees contining rule nmes re often clled derivtion trees. Overll we obtin the following result for which detils cn be found in (Arnold nd Duchet 98). Theorem 7 The set D(R) is regulr tree lnguge for every LMBOT R nd the set of derivtions is lso regulr for every MBOT. VBD signed IN for PV twl DET-NN PREP En AltwqyE Figure 6: mple TG rule. Moreover using the input nd output product constructions of Mletti (00) we obtin tht even the set D tu (R) of derivtions for specific input tree t nd output tree u is regulr. ince D tu (R) is regulr we cn compute the inside nd outside weight of ech (weighted) rule of R following the method of Grehl et l. (008). imilrly we cn djust the trining procedure of Grehl et l. (008) which yields tht we cn utomticlly obtin weighted LMBOT from bi-prsed prllel corpus. Detils on the run-time cn be found in (Grehl et l. 008). 5 Preservtion of regulrity Clerly LMBOT re not symmetric. Although the bckwrds ppliction of n LMBOT preserves regulrity this property does not hold for forwrd ppliction. We will focus on forwrd ppliction here. Given set T of pre-trnsltions nd tree lnguge 830
7 L T Σ we let T c (L) = {u i (u... u k ) T (L) i [k]} which collects ll trnsltions of input trees in L. We sy tht T preserves regulrity if T c (L) is regulr for every regulr tree lnguge L T Σ. Correspondingly n LMBOT R preserves regulrity if its set τ(r) of pre-trnsltions preserves regulrity. As mentioned n LMBOT does not necessrily preserve regulrity. The rules of n LMBOT hve only lignments between the left-hnd side (input tree) nd the right-hnd side (output tree) which re lso clled inter-tree lignments. However severl lignments to single nonterminl in the left-hnd side cn trnsitively relte two different nonterminls in the output side nd thus simulte n intrtree lignment. For exmple the right rule of Figure reltes PV nd n -OBJ node to single node in the left-hnd side. This could led to n intr-tree lignment (synchroniztion) between the PV nd -OBJ nodes in the right-hnd side. Figure 7 displys the rules R of n LMBOT tht does not preserve regulrity. This cn esily be seen on the lef (word) lnguges becuse the LMBOT cn trnslte the word x to ny element of L = {wcwc w { b} }. Clerly this word lnguge L is not context-free. ince the lef lnguge of every regulr tree lnguge is context-free nd regulr tree lnguges re closed under intersection (needed to single out the trnsltions tht hve the symbol t the root) this lso proves tht τ(r) c (T Σ ) is not regulr. ince T Σ is regulr this proves tht the LMBOT does not preserve regulrity. Preservtion of regulrity is n importnt property for number of trnsltion model mnipultions. For exmple the bucket-brigde nd the on-the-fly method for the efficient inference described in (My et l. 00) essentilly build on it. Moreover regulr tree grmmr (i.e. representtion of regulr tree lnguge) is n efficient representtion. More complex representtions such s context-free tree grmmrs [see e.g. (Fujiyoshi 004)] hve worse lgorithmic properties (e.g. more complex prsing nd problemtic intersection). In this section we investigte three syntctic restrictions on the set R of rules tht gurntees tht the obtined LMBOT preserves regulrity. Then we shortly discuss how to djust the rule extrction lgorithm so tht the extrcted rules utomticlly hve these property. First we quickly recll the notion of composed rules from Definition 4 becuse it will ply n essentil role in ll three properties. Figure 3 shows composition of two rules from Figure 7. Mind tht R might not contin ll rules of R but it contins ll those without lef nonterminls. Definition 8 An LMBOT R is finitely collpsing if there is n N such tht ψ : NT (r) NT (l) {} for every rule l ψ r R n. The following sttement follows from more generl result of Roult (997) which we will introduce with our second property. Theorem 9 Every finitely collpsing LMBOT preserves regulrity. Often the simple condition finitely collpsing is fulfilled fter rule extrction. In ddition it is utomticlly fulfilled in n LMBOT tht ws obtined from n TG using Theorem 3. It cn lso be ensured in the rule extrction process by introducing collpsing points for output symbols tht cn pper recursively in the corpus. For exmple we could enforce tht ll extrcted rules for cluse-level output symbols (ssuming tht there is no recursion not involving cluse-level output symbols) should hve only output tree in the right-hnd side. However finitely collpsing is rther strict property. Finitely collpsing LMBOT hve only slightly more expressive power thn TG. In fct they could be clled TG with input desynchroniztion. This is due to the fct tht the lignment in composed rules estblishes n injective reltion between lef nonterminls (s in n TG) but it need not be bijective. Consequently there cn be lef nonterminls in the left-hnd side tht hve no ligned lef nonterminl in the right-hnd side. In this sense those lef nonterminls re desynchronized. This feture is illustrted in Figure 8 nd such n LMBOT cn compute the trnsformtion {(t ) t T Σ } which cnnot be computed by n TG (ssuming tht T Σ is suitbly rich). Thus TG with input desynchroniztion re more expressive thn TG but they still compute clss of trnsformtions tht is not closed under composition. 83
8 x c c b b Figure 7: Output subtree synchroniztion (intr-tree). t 3 Figure 8: Finitely collpsing LMBOT. Theorem 0 For every TG we cn construct n equivlent finitely collpsing LMBOT in liner time. Moreover finitely collpsing LMBOT re strictly more expressive thn TG. Next we investigte weker property by Roult (997) tht still ensures preservtion of regulrity. Definition An LMBOT R hs finite synchroniztion if there is n N such tht for every rule l ψ r R n nd p NT (l) there exists i N with ψ ({p} N) {iw w N }. In plin terms multiple lignments to single lef nonterminl t p in the left-hnd side re llowed but ll lef nonterminls of the right-hnd side tht re ligned to p must be in the sme tree. Clerly n LMBOT with finite synchroniztion is finitely collpsing. Roult (997) investigted this restriction in the context of rtionl tree reltions which re generliztion of our LMBOT. Roult (997) shows tht finite synchroniztion cn be decided. The next theorem follows from the results of Roult (997). Theorem Every LMBOT with finite synchroniztion preserves regulrity. MBOT cn compute rbitrry compositions of TG (Mletti 00). However this no longer remins true for MBOT (or LMBOT) with finite synchroniztion. In Figure 9 we illustrte trnsltion tht cn be computed by composition of two TG but tht cnnot be computed by n MBOT (or LMBOT) with finite synchroniztion. Intuitively when processing the chin of s of the trnsformtion depicted in Figure 9 the first nd second suc- This ssumes strightforwrd generliztion of the finite synchroniztion property for MBOT.. t t Z t t t 3 Figure 9: Trnsformtion tht cnnot be computed by n MBOT with finite synchroniztion. cessor of the Z -node t the root on the output side must be ligned to the -chin. This is necessry becuse those two mentioned subtrees must reproduce t nd t from the end of the -chin. We omit the forml proof here but obtin the following sttement. Theorem 3 For every TG we cn construct n equivlent LMBOT with finite synchroniztion in liner time. LMBOT nd MBOT with finite synchroniztion re strictly more expressive thn TG nd compute clsses tht re not closed under composition. Agin it is strightforwrd to djust the rule extrction lgorithm by the introduction of synchroniztion points (for exmple for cluse level output symbols). We cn simply require tht rules extrcted for those selected output symbols fulfill the condition mentioned in Definition. Finlly we introduce n even weker version. Definition 4 An LMBOT R is copy-free if there is n N such tht for every rule l ψ r R n nd p NT (l) we hve (i) ψ ({p} N) N or (ii) ψ ({p} N) {iw w N } for n i N. Intuitively copy-free LMBOT hs rules whose right hnd sides my use ll lef nonterminls tht re ligned to given lef nonterminl in the lefthnd side directly t the root (of one of the trees 83
9 Z Figure 0: Composed rule tht is not copy-free. in the right-hnd side forest) or group ll those lef nonterminls in single tree in the forest. Clerly the LMBOT of Figure 7 is not copy-free becuse the second rule composes with itself (see Figure 0) to rule tht does not fulfill the copy-free condition. Theorem 5 Every copy-free LMBOT preserves regulrity. Proof sketch: Let n be the integer of Definition 4. We replce the LMBOT with rules R by the equivlent LMBOT M with rules R n. Then ll rules hve the form required in Definition 4. Moreover let L T Σ be regulr tree lnguge. Then we cn construct the input product of τ(m) with L. In this wy we obtin n MBOT M whose rules still fulfill the requirements (dpted for MBOT) of Definition 4 becuse the input product does not chnge the structure of the rules (it only modifies the stte behvior). Consequently we only need to show tht the rnge of the MBOT M is regulr. This cn be chieved using decomposition into relbeling which clerly preserves regulrity nd deterministic finite-copying top-down tree trnsducer (Engelfriet et l. 980; Engelfriet 98). Figure shows some relevnt rules of copyfree LMBOT tht computes the trnsformtion of Figure 9. Clerly copy-free LMBOT re more generl thn LMBOT with finite synchroniztion so we gin cn obtin copy-free LMBOT from TG. In ddition we cn djust the rule extrction process using synchroniztion points s for LMBOT with finite synchroniztion using the restrictions of Definition 4. Theorem 6 For every TG we cn construct n equivlent copy-free LMBOT in liner time. Figure : Copy-free LMBOT for the trnsformtion of Figure 9. Copy-free LMBOT re strictly more expressive thn LMBOT with finite synchroniztion. 6 Conclusion We hve introduced simple restriction of multi bottom-up tree trnsducers. It bstrcts from the generl stte behvior of the generl model nd only uses the loclity tests tht re lso present in TG TG nd TAG. Next we introduced rule extrction procedure nd corresponding rule weight trining procedure for our LMBOT. However LMBOT llow trnsltions tht do not preserve regulrity which is n importnt property for efficient lgorithms. We presented 3 properties tht ensure tht regulrity is preserved. In ddition we shortly discussed how these properties could be enforced in the presented rule extrction procedure. Acknowledgements The uthor grtefully cknowledges the support by KEVIN KNIGHT who provided the inspirtion nd the dt. JONATHAN MA helped in mny fruitful discussions. The uthor ws finncilly supported by the Germn Reserch Foundtion (DFG) grnt MA / 4959 /
10 References Alfred V. Aho nd Jeffrey D. Ullmn. 97. The Theory of Prsing Trnsltion nd Compiling. Prentice Hll. André Arnold nd Mx Duchet. 98. Morphismes et bimorphismes d rbres. Theoret. Comput. ci. 0(): Peter F. Brown John Cocke tephen A. Dell Pietr Vincent J. Dell Pietr Fredrick Jelinek John D. Lfferty Robert L. Mercer nd Pul. Roossin A sttisticl pproch to mchine trnsltion. Computtionl Linguistics 6(): Peter F. Brown tephen A. Dell Pietr Vincent J. Dell Pietr nd Robert L. Mercer Mthemtics of sttisticl mchine trnsltion: Prmeter estimtion. Computtionl Linguistics 9():63 3. Dvid Ching A hierrchicl phrse-bsed model for sttisticl mchine trnsltion. In Proc. ACL pges Assocition for Computtionl Linguistics. Dvid Ching An introduction to synchronous grmmrs. In Proc. ACL. Assocition for Computtionl Linguistics. Prt of tutoril given with Kevin Knight. Json Eisner impler nd more generl minimiztion for weighted finite-stte utomt. In Proc. NAACL pges Assocition for Computtionl Linguistics. Joost Engelfriet Grzegorz Rozenberg nd Gior lutzki Tree trnsducers L systems nd two-wy mchines. J. Comput. ystem ci. 0():50 0. Joost Engelfriet Eric Lilin nd Andres Mletti Composition nd decomposition of extended multi bottom-up tree trnsducers. Act Inform. 46(8): Joost Engelfriet. 98. The copying power of one-stte tree trnsducers. J. Comput. ystem ci. 5(3): Akio Fujiyoshi Restrictions on mondic contextfree tree grmmrs. In Proc. CoLing pges Assocition for Computtionl Linguistics. Michel Glley Mrk Hopkins Kevin Knight nd Dniel Mrcu Wht s in trnsltion rule? In Proc. HLT-NAACL pges Assocition for Computtionl Linguistics. Jonthn Grehl Kevin Knight nd Jonthn My Trining tree trnsducers. Computtionl Linguistics 34(3): Kevin Knight Cpturing prcticl nturl lnguge trnsformtions. Mchine Trnsltion (): 33. Eric Lilin. 98. Propriétés de clôture d une extension de trnsducteurs d rbres déterministes. In Proc. CAAP volume of LNC pges pringer. Andres Mletti. 00. Why synchronous tree substitution grmmrs? In Proc. NAACL pges Assocition for Computtionl Linguistics. Jonthn My Kevin Knight nd Heiko Vogler. 00. Efficient inference through cscdes of weighted tree trnsducers. In Proc. ACL pges Assocition for Computtionl Linguistics. Frnk G. Rdmcher An utomt theoretic pproch to rtionl tree reltions. In Proc. OFEM volume 490 of LNC pges pringer. Jen-Clude Roult Rtionl tree reltions. Bull. Belg. Mth. oc. imon tevin 4(): turt M. hieber nd ves chbes ynchronous tree-djoining grmmrs. In Proc. CoLing volume 3 pges Assocition for Computtionl Linguistics. turt M. hieber ynchronous grmmrs s tree trnsducers. In Proc. TAG+7 pges Vncouver BC Cnd. imon Frser University. turt M. hieber Probbilistic synchronous treedjoining grmmrs for mchine trnsltion: The rgument from bilingul dictionries. In Proc. T pges Assocition for Computtionl Linguistics. Jun un Min Zhng nd Chew Lim Tn A noncontiguous tree sequence lignment-bsed model for sttisticl mchine trnsltion. In Proc. ACL pges Assocition for Computtionl Linguistics. Min Zhng Hongfei Jing Aiti Aw Hizhou Li Chew Lim Tn nd heng Li A tree sequence lignment-bsed tree-to-tree trnsltion model. In Proc. ACL pges Assocition for Computtionl Linguistics. Min Zhng Hongfei Jing Hizhou Li Aiti Aw nd heng Li. 008b. Grmmr comprison study for trnsltionl equivlence modeling nd sttisticl mchine trnsltion. In Proc. CoLing pges Assocition for Computtionl Linguistics. 834
Why Synchronous Tree Substitution Grammars?
Why ynchronous Tree ubstitution Grmmrs? Andres Mletti Universitt Rovir i Virgili, Deprtment de Filologies Romàniques Avingud de Ctluny 35, 43002 Trrgon, pin ndres.mlettiurv.ct Abstrct ynchronous tree substitution
More informationTREE AUTOMATA AND TREE GRAMMARS
TREE AUTOMATA AND TREE GRAMMARS rxiv:1510.02036v1 [cs.fl] 7 Oct 2015 by Joost Engelfriet DAIMI FN-10 April 1975 Institute of Mthemtics, University of Arhus DEPARTMENT OF COMPUTER SCIENCE Ny Munkegde, 8000
More informationFinite Automata Theory and Formal Languages TMV027/DIT321 LP4 2018
Finite Automt Theory nd Forml Lnguges TMV027/DIT321 LP4 2018 Lecture 10 An Bove April 23rd 2018 Recp: Regulr Lnguges We cn convert between FA nd RE; Hence both FA nd RE ccept/generte regulr lnguges; More
More informationAnatomy of a Deterministic Finite Automaton. Deterministic Finite Automata. A machine so simple that you can understand it in less than one minute
Victor Admchik Dnny Sletor Gret Theoreticl Ides In Computer Science CS 5-25 Spring 2 Lecture 2 Mr 3, 2 Crnegie Mellon University Deterministic Finite Automt Finite Automt A mchine so simple tht you cn
More informationCMPSCI 250: Introduction to Computation. Lecture #31: What DFA s Can and Can t Do David Mix Barrington 9 April 2014
CMPSCI 250: Introduction to Computtion Lecture #31: Wht DFA s Cn nd Cn t Do Dvid Mix Brrington 9 April 2014 Wht DFA s Cn nd Cn t Do Deterministic Finite Automt Forml Definition of DFA s Exmples of DFA
More informationMore on automata. Michael George. March 24 April 7, 2014
More on utomt Michel George Mrch 24 April 7, 2014 1 Automt constructions Now tht we hve forml model of mchine, it is useful to mke some generl constructions. 1.1 DFA Union / Product construction Suppose
More information1.4 Nonregular Languages
74 1.4 Nonregulr Lnguges The number of forml lnguges over ny lphbet (= decision/recognition problems) is uncountble On the other hnd, the number of regulr expressions (= strings) is countble Hence, ll
More informationIntermediate Math Circles Wednesday, November 14, 2018 Finite Automata II. Nickolas Rollick a b b. a b 4
Intermedite Mth Circles Wednesdy, Novemer 14, 2018 Finite Automt II Nickols Rollick nrollick@uwterloo.c Regulr Lnguges Lst time, we were introduced to the ide of DFA (deterministic finite utomton), one
More informationCS 275 Automata and Formal Language Theory
CS 275 Automt nd Forml Lnguge Theory Course Notes Prt II: The Recognition Problem (II) Chpter II.6.: Push Down Automt Remrk: This mteril is no longer tught nd not directly exm relevnt Anton Setzer (Bsed
More informationClosure Properties of Regular Languages
Closure Properties of Regulr Lnguges Regulr lnguges re closed under mny set opertions. Let L 1 nd L 2 e regulr lnguges. (1) L 1 L 2 (the union) is regulr. (2) L 1 L 2 (the conctention) is regulr. (3) L
More informationAdvanced Calculus: MATH 410 Notes on Integrals and Integrability Professor David Levermore 17 October 2004
Advnced Clculus: MATH 410 Notes on Integrls nd Integrbility Professor Dvid Levermore 17 October 2004 1. Definite Integrls In this section we revisit the definite integrl tht you were introduced to when
More information1.3 Regular Expressions
56 1.3 Regulr xpressions These hve n importnt role in describing ptterns in serching for strings in mny pplictions (e.g. wk, grep, Perl,...) All regulr expressions of lphbet re 1.Ønd re regulr expressions,
More informationCS 275 Automata and Formal Language Theory
CS 275 Automt nd Forml Lnguge Theory Course Notes Prt II: The Recognition Problem (II) Chpter II.5.: Properties of Context Free Grmmrs (14) Anton Setzer (Bsed on book drft by J. V. Tucker nd K. Stephenson)
More informationFormal languages, automata, and theory of computation
Mälrdlen University TEN1 DVA337 2015 School of Innovtion, Design nd Engineering Forml lnguges, utomt, nd theory of computtion Thursdy, Novemer 5, 14:10-18:30 Techer: Dniel Hedin, phone 021-107052 The exm
More informationNFAs and Regular Expressions. NFA-ε, continued. Recall. Last class: Today: Fun:
CMPU 240 Lnguge Theory nd Computtion Spring 2019 NFAs nd Regulr Expressions Lst clss: Introduced nondeterministic finite utomt with -trnsitions Tody: Prove n NFA- is no more powerful thn n NFA Introduce
More informationFABER Formal Languages, Automata and Models of Computation
DVA337 FABER Forml Lnguges, Automt nd Models of Computtion Lecture 5 chool of Innovtion, Design nd Engineering Mälrdlen University 2015 1 Recp of lecture 4 y definition suset construction DFA NFA stte
More informationFinite Automata. Informatics 2A: Lecture 3. John Longley. 22 September School of Informatics University of Edinburgh
Lnguges nd Automt Finite Automt Informtics 2A: Lecture 3 John Longley School of Informtics University of Edinburgh jrl@inf.ed.c.uk 22 September 2017 1 / 30 Lnguges nd Automt 1 Lnguges nd Automt Wht is
More informationHandout: Natural deduction for first order logic
MATH 457 Introduction to Mthemticl Logic Spring 2016 Dr Json Rute Hndout: Nturl deduction for first order logic We will extend our nturl deduction rules for sententil logic to first order logic These notes
More informationAUTOMATA AND LANGUAGES. Definition 1.5: Finite Automaton
25. Finite Automt AUTOMATA AND LANGUAGES A system of computtion tht only hs finite numer of possile sttes cn e modeled using finite utomton A finite utomton is often illustrted s stte digrm d d d. d q
More informationNew Expansion and Infinite Series
Interntionl Mthemticl Forum, Vol. 9, 204, no. 22, 06-073 HIKARI Ltd, www.m-hikri.com http://dx.doi.org/0.2988/imf.204.4502 New Expnsion nd Infinite Series Diyun Zhng College of Computer Nnjing University
More informationCoalgebra, Lecture 15: Equations for Deterministic Automata
Colger, Lecture 15: Equtions for Deterministic Automt Julin Slmnc (nd Jurrin Rot) Decemer 19, 2016 In this lecture, we will study the concept of equtions for deterministic utomt. The notes re self contined
More informationTutorial Automata and formal Languages
Tutoril Automt nd forml Lnguges Notes for to the tutoril in the summer term 2017 Sestin Küpper, Christine Mik 8. August 2017 1 Introduction: Nottions nd sic Definitions At the eginning of the tutoril we
More informationStrong Bisimulation. Overview. References. Actions Labeled transition system Transition semantics Simulation Bisimulation
Strong Bisimultion Overview Actions Lbeled trnsition system Trnsition semntics Simultion Bisimultion References Robin Milner, Communiction nd Concurrency Robin Milner, Communicting nd Mobil Systems 32
More informationMinimal DFA. minimal DFA for L starting from any other
Miniml DFA Among the mny DFAs ccepting the sme regulr lnguge L, there is exctly one (up to renming of sttes) which hs the smllest possile numer of sttes. Moreover, it is possile to otin tht miniml DFA
More informationHow to train your multi bottom-up tree transducer
How to train your multi bottom-up tree transducer Andreas Maletti Universität tuttgart Institute for Natural Language Processing tuttgart, Germany andreas.maletti@ims.uni-stuttgart.de Portland, OR June
More informationReview of Calculus, cont d
Jim Lmbers MAT 460 Fll Semester 2009-10 Lecture 3 Notes These notes correspond to Section 1.1 in the text. Review of Clculus, cont d Riemnn Sums nd the Definite Integrl There re mny cses in which some
More informationp-adic Egyptian Fractions
p-adic Egyptin Frctions Contents 1 Introduction 1 2 Trditionl Egyptin Frctions nd Greedy Algorithm 2 3 Set-up 3 4 p-greedy Algorithm 5 5 p-egyptin Trditionl 10 6 Conclusion 1 Introduction An Egyptin frction
More informationTheory of Computation Regular Languages. (NTU EE) Regular Languages Fall / 38
Theory of Computtion Regulr Lnguges (NTU EE) Regulr Lnguges Fll 2017 1 / 38 Schemtic of Finite Automt control 0 0 1 0 1 1 1 0 Figure: Schemtic of Finite Automt A finite utomton hs finite set of control
More information1. For each of the following theorems, give a two or three sentence sketch of how the proof goes or why it is not true.
York University CSE 2 Unit 3. DFA Clsses Converting etween DFA, NFA, Regulr Expressions, nd Extended Regulr Expressions Instructor: Jeff Edmonds Don t chet y looking t these nswers premturely.. For ech
More informationChapter 2 Finite Automata
Chpter 2 Finite Automt 28 2.1 Introduction Finite utomt: first model of the notion of effective procedure. (They lso hve mny other pplictions). The concept of finite utomton cn e derived y exmining wht
More informationCS 275 Automata and Formal Language Theory
CS 275 utomt nd Forml Lnguge Theory Course Notes Prt II: The Recognition Prolem (II) Chpter II.5.: Properties of Context Free Grmmrs (14) nton Setzer (Bsed on ook drft y J. V. Tucker nd K. Stephenson)
More informationFrobenius numbers of generalized Fibonacci semigroups
Frobenius numbers of generlized Fiboncci semigroups Gretchen L. Mtthews 1 Deprtment of Mthemticl Sciences, Clemson University, Clemson, SC 29634-0975, USA gmtthe@clemson.edu Received:, Accepted:, Published:
More informationHarvard University Computer Science 121 Midterm October 23, 2012
Hrvrd University Computer Science 121 Midterm Octoer 23, 2012 This is closed-ook exmintion. You my use ny result from lecture, Sipser, prolem sets, or section, s long s you quote it clerly. The lphet is
More informationChapter Five: Nondeterministic Finite Automata. Formal Language, chapter 5, slide 1
Chpter Five: Nondeterministic Finite Automt Forml Lnguge, chpter 5, slide 1 1 A DFA hs exctly one trnsition from every stte on every symol in the lphet. By relxing this requirement we get relted ut more
More informationW. We shall do so one by one, starting with I 1, and we shall do it greedily, trying
Vitli covers 1 Definition. A Vitli cover of set E R is set V of closed intervls with positive length so tht, for every δ > 0 nd every x E, there is some I V with λ(i ) < δ nd x I. 2 Lemm (Vitli covering)
More informationCMSC 330: Organization of Programming Languages. DFAs, and NFAs, and Regexps (Oh my!)
CMSC 330: Orgniztion of Progrmming Lnguges DFAs, nd NFAs, nd Regexps (Oh my!) CMSC330 Spring 2018 Types of Finite Automt Deterministic Finite Automt (DFA) Exctly one sequence of steps for ech string All
More informationParse trees, ambiguity, and Chomsky normal form
Prse trees, miguity, nd Chomsky norml form In this lecture we will discuss few importnt notions connected with contextfree grmmrs, including prse trees, miguity, nd specil form for context-free grmmrs
More informationChapter 4 Contravariance, Covariance, and Spacetime Diagrams
Chpter 4 Contrvrince, Covrince, nd Spcetime Digrms 4. The Components of Vector in Skewed Coordintes We hve seen in Chpter 3; figure 3.9, tht in order to show inertil motion tht is consistent with the Lorentz
More informationFinite Automata-cont d
Automt Theory nd Forml Lnguges Professor Leslie Lnder Lecture # 6 Finite Automt-cont d The Pumping Lemm WEB SITE: http://ingwe.inghmton.edu/ ~lnder/cs573.html Septemer 18, 2000 Exmple 1 Consider L = {ww
More informationTheory of Computation Regular Languages
Theory of Computtion Regulr Lnguges Bow-Yw Wng Acdemi Sinic Spring 2012 Bow-Yw Wng (Acdemi Sinic) Regulr Lnguges Spring 2012 1 / 38 Schemtic of Finite Automt control 0 0 1 0 1 1 1 0 Figure: Schemtic of
More informationThe University of Nottingham SCHOOL OF COMPUTER SCIENCE A LEVEL 2 MODULE, SPRING SEMESTER LANGUAGES AND COMPUTATION ANSWERS
The University of Nottinghm SCHOOL OF COMPUTER SCIENCE LEVEL 2 MODULE, SPRING SEMESTER 2016 2017 LNGUGES ND COMPUTTION NSWERS Time llowed TWO hours Cndidtes my complete the front cover of their nswer ook
More informationFarey Fractions. Rickard Fernström. U.U.D.M. Project Report 2017:24. Department of Mathematics Uppsala University
U.U.D.M. Project Report 07:4 Frey Frctions Rickrd Fernström Exmensrete i mtemtik, 5 hp Hledre: Andres Strömergsson Exmintor: Jörgen Östensson Juni 07 Deprtment of Mthemtics Uppsl University Frey Frctions
More informationLecture 3: Equivalence Relations
Mthcmp Crsh Course Instructor: Pdric Brtlett Lecture 3: Equivlence Reltions Week 1 Mthcmp 2014 In our lst three tlks of this clss, we shift the focus of our tlks from proof techniques to proof concepts
More informationNon-Deterministic Finite Automata. Fall 2018 Costas Busch - RPI 1
Non-Deterministic Finite Automt Fll 2018 Costs Busch - RPI 1 Nondeterministic Finite Automton (NFA) Alphbet ={} q q2 1 q 0 q 3 Fll 2018 Costs Busch - RPI 2 Nondeterministic Finite Automton (NFA) Alphbet
More informationA tutorial on sequential functions
A tutoril on sequentil functions Jen-Éric Pin LIAFA, CNRS nd University Pris 7 30 Jnury 2006, CWI, Amsterdm Outline (1) Sequentil functions (2) A chrcteriztion of sequentil trnsducers (3) Miniml sequentil
More informationFinite Automata. Informatics 2A: Lecture 3. Mary Cryan. 21 September School of Informatics University of Edinburgh
Finite Automt Informtics 2A: Lecture 3 Mry Cryn School of Informtics University of Edinburgh mcryn@inf.ed.c.uk 21 September 2018 1 / 30 Lnguges nd Automt Wht is lnguge? Finite utomt: recp Some forml definitions
More informationGoals: Determine how to calculate the area described by a function. Define the definite integral. Explore the relationship between the definite
Unit #8 : The Integrl Gols: Determine how to clculte the re described by function. Define the definite integrl. Eplore the reltionship between the definite integrl nd re. Eplore wys to estimte the definite
More informationDuality # Second iteration for HW problem. Recall our LP example problem we have been working on, in equality form, is given below.
Dulity #. Second itertion for HW problem Recll our LP emple problem we hve been working on, in equlity form, is given below.,,,, 8 m F which, when written in slightly different form, is 8 F Recll tht we
More information7.2 The Definite Integral
7.2 The Definite Integrl the definite integrl In the previous section, it ws found tht if function f is continuous nd nonnegtive, then the re under the grph of f on [, b] is given by F (b) F (), where
More informationHow to simulate Turing machines by invertible one-dimensional cellular automata
How to simulte Turing mchines by invertible one-dimensionl cellulr utomt Jen-Christophe Dubcq Déprtement de Mthémtiques et d Informtique, École Normle Supérieure de Lyon, 46, llée d Itlie, 69364 Lyon Cedex
More information12.1 Nondeterminism Nondeterministic Finite Automata. a a b ε. CS125 Lecture 12 Fall 2016
CS125 Lecture 12 Fll 2016 12.1 Nondeterminism The ide of nondeterministic computtions is to llow our lgorithms to mke guesses, nd only require tht they ccept when the guesses re correct. For exmple, simple
More informationTHE EXISTENCE-UNIQUENESS THEOREM FOR FIRST-ORDER DIFFERENTIAL EQUATIONS.
THE EXISTENCE-UNIQUENESS THEOREM FOR FIRST-ORDER DIFFERENTIAL EQUATIONS RADON ROSBOROUGH https://intuitiveexplntionscom/picrd-lindelof-theorem/ This document is proof of the existence-uniqueness theorem
More informationMath 1B, lecture 4: Error bounds for numerical methods
Mth B, lecture 4: Error bounds for numericl methods Nthn Pflueger 4 September 0 Introduction The five numericl methods descried in the previous lecture ll operte by the sme principle: they pproximte the
More informationMath 4310 Solutions to homework 1 Due 9/1/16
Mth 4310 Solutions to homework 1 Due 9/1/16 1. Use the Eucliden lgorithm to find the following gretest common divisors. () gcd(252, 180) = 36 (b) gcd(513, 187) = 1 (c) gcd(7684, 4148) = 68 252 = 180 1
More informationApplicable Analysis and Discrete Mathematics available online at
Applicble Anlysis nd Discrete Mthemtics vilble online t http://pefmth.etf.rs Appl. Anl. Discrete Mth. 4 (2010), 23 31. doi:10.2298/aadm100201012k NUMERICAL ANALYSIS MEETS NUMBER THEORY: USING ROOTFINDING
More information3 Regular expressions
3 Regulr expressions Given n lphet Σ lnguge is set of words L Σ. So fr we were le to descrie lnguges either y using set theory (i.e. enumertion or comprehension) or y n utomton. In this section we shll
More informationLecture 08: Feb. 08, 2019
4CS4-6:Theory of Computtion(Closure on Reg. Lngs., regex to NDFA, DFA to regex) Prof. K.R. Chowdhry Lecture 08: Fe. 08, 2019 : Professor of CS Disclimer: These notes hve not een sujected to the usul scrutiny
More information1.9 C 2 inner variations
46 CHAPTER 1. INDIRECT METHODS 1.9 C 2 inner vritions So fr, we hve restricted ttention to liner vritions. These re vritions of the form vx; ǫ = ux + ǫφx where φ is in some liner perturbtion clss P, for
More informationEuler, Ioachimescu and the trapezium rule. G.J.O. Jameson (Math. Gazette 96 (2012), )
Euler, Iochimescu nd the trpezium rule G.J.O. Jmeson (Mth. Gzette 96 (0), 36 4) The following results were estblished in recent Gzette rticle [, Theorems, 3, 4]. Given > 0 nd 0 < s
More informationCS 314 Principles of Programming Languages
C 314 Principles of Progrmming Lnguges Lecture 6: LL(1) Prsing Zheng (Eddy) Zhng Rutgers University Ferury 5, 2018 Clss Informtion Homework 2 due tomorrow. Homework 3 will e posted erly next week. 2 Top
More informationUniversitaireWiskundeCompetitie. Problem 2005/4-A We have k=1. Show that for every q Q satisfying 0 < q < 1, there exists a finite subset K N so that
Problemen/UWC NAW 5/7 nr juni 006 47 Problemen/UWC UniversitireWiskundeCompetitie Edition 005/4 For Session 005/4 we received submissions from Peter Vndendriessche, Vldislv Frnk, Arne Smeets, Jn vn de
More informationRefined interfaces for compositional verification
Refined interfces for compositionl verifiction Frédéric Lng INRI Rhône-lpes http://www.inrilpes.fr/vsy Motivtion Enumertive verifiction of concurrent systems Prllel composition of synchronous processes
More informationNotes on length and conformal metrics
Notes on length nd conforml metrics We recll how to mesure the Eucliden distnce of n rc in the plne. Let α : [, b] R 2 be smooth (C ) rc. Tht is α(t) (x(t), y(t)) where x(t) nd y(t) re smooth rel vlued
More informationThe Regulated and Riemann Integrals
Chpter 1 The Regulted nd Riemnn Integrls 1.1 Introduction We will consider severl different pproches to defining the definite integrl f(x) dx of function f(x). These definitions will ll ssign the sme vlue
More information1 Structural induction
Discrete Structures Prelim 2 smple questions Solutions CS2800 Questions selected for Spring 2018 1 Structurl induction 1. We define set S of functions from Z to Z inductively s follows: Rule 1. For ny
More informationNon Deterministic Automata. Linz: Nondeterministic Finite Accepters, page 51
Non Deterministic Automt Linz: Nondeterministic Finite Accepters, pge 51 1 Nondeterministic Finite Accepter (NFA) Alphbet ={} q 1 q2 q 0 q 3 2 Nondeterministic Finite Accepter (NFA) Alphbet ={} Two choices
More information1 Nondeterministic Finite Automata
1 Nondeterministic Finite Automt Suppose in life, whenever you hd choice, you could try oth possiilities nd live your life. At the end, you would go ck nd choose the one tht worked out the est. Then you
More informationBernoulli Numbers Jeff Morton
Bernoulli Numbers Jeff Morton. We re interested in the opertor e t k d k t k, which is to sy k tk. Applying this to some function f E to get e t f d k k tk d k f f + d k k tk dk f, we note tht since f
More information1 From NFA to regular expression
Note 1: How to convert DFA/NFA to regulr expression Version: 1.0 S/EE 374, Fll 2017 Septemer 11, 2017 In this note, we show tht ny DFA cn e converted into regulr expression. Our construction would work
More information1 Structural induction, finite automata, regular expressions
Discrete Structures Prelim 2 smple uestions s CS2800 Questions selected for spring 2017 1 Structurl induction, finite utomt, regulr expressions 1. We define set S of functions from Z to Z inductively s
More informationUNIT 1 FUNCTIONS AND THEIR INVERSES Lesson 1.4: Logarithmic Functions as Inverses Instruction
Lesson : Logrithmic Functions s Inverses Prerequisite Skills This lesson requires the use of the following skills: determining the dependent nd independent vribles in n exponentil function bsed on dt from
More informationUnit #9 : Definite Integral Properties; Fundamental Theorem of Calculus
Unit #9 : Definite Integrl Properties; Fundmentl Theorem of Clculus Gols: Identify properties of definite integrls Define odd nd even functions, nd reltionship to integrl vlues Introduce the Fundmentl
More information5. (±±) Λ = fw j w is string of even lengthg [ 00 = f11,00g 7. (11 [ 00)± Λ = fw j w egins with either 11 or 00g 8. (0 [ ffl)1 Λ = 01 Λ [ 1 Λ 9.
Regulr Expressions, Pumping Lemm, Right Liner Grmmrs Ling 106 Mrch 25, 2002 1 Regulr Expressions A regulr expression descries or genertes lnguge: it is kind of shorthnd for listing the memers of lnguge.
More informationJim Lambers MAT 169 Fall Semester Lecture 4 Notes
Jim Lmbers MAT 169 Fll Semester 2009-10 Lecture 4 Notes These notes correspond to Section 8.2 in the text. Series Wht is Series? An infinte series, usully referred to simply s series, is n sum of ll of
More informationLecture 3 ( ) (translated and slightly adapted from lecture notes by Martin Klazar)
Lecture 3 (5.3.2018) (trnslted nd slightly dpted from lecture notes by Mrtin Klzr) Riemnn integrl Now we define precisely the concept of the re, in prticulr, the re of figure U(, b, f) under the grph of
More informationTypes of Finite Automata. CMSC 330: Organization of Programming Languages. Comparing DFAs and NFAs. Comparing DFAs and NFAs (cont.) Finite Automata 2
CMSC 330: Orgniztion of Progrmming Lnguges Finite Automt 2 Types of Finite Automt Deterministic Finite Automt () Exctly one sequence of steps for ech string All exmples so fr Nondeterministic Finite Automt
More informationContext-Free Grammars and Languages
Context-Free Grmmrs nd Lnguges (Bsed on Hopcroft, Motwni nd Ullmn (2007) & Cohen (1997)) Introduction Consider n exmple sentence: A smll ct ets the fish English grmmr hs rules for constructing sentences;
More informationAssignment 1 Automata, Languages, and Computability. 1 Finite State Automata and Regular Languages
Deprtment of Computer Science, Austrlin Ntionl University COMP2600 Forml Methods for Softwre Engineering Semester 2, 206 Assignment Automt, Lnguges, nd Computility Smple Solutions Finite Stte Automt nd
More informationNondeterminism and Nodeterministic Automata
Nondeterminism nd Nodeterministic Automt 61 Nondeterminism nd Nondeterministic Automt The computtionl mchine models tht we lerned in the clss re deterministic in the sense tht the next move is uniquely
More informationZero-Sum Magic Graphs and Their Null Sets
Zero-Sum Mgic Grphs nd Their Null Sets Ebrhim Slehi Deprtment of Mthemticl Sciences University of Nevd Ls Vegs Ls Vegs, NV 89154-4020. ebrhim.slehi@unlv.edu Abstrct For ny h N, grph G = (V, E) is sid to
More informationSpeech Recognition Lecture 2: Finite Automata and Finite-State Transducers. Mehryar Mohri Courant Institute and Google Research
Speech Recognition Lecture 2: Finite Automt nd Finite-Stte Trnsducers Mehryr Mohri Cournt Institute nd Google Reserch mohri@cims.nyu.com Preliminries Finite lphet Σ, empty string. Set of ll strings over
More informationReversals of Signal-Posterior Monotonicity for Any Bounded Prior
Reversls of Signl-Posterior Monotonicity for Any Bounded Prior Christopher P. Chmbers Pul J. Hely Abstrct Pul Milgrom (The Bell Journl of Economics, 12(2): 380 391) showed tht if the strict monotone likelihood
More informationIntroduction to Group Theory
Introduction to Group Theory Let G be n rbitrry set of elements, typiclly denoted s, b, c,, tht is, let G = {, b, c, }. A binry opertion in G is rule tht ssocites with ech ordered pir (,b) of elements
More informationLecture 09: Myhill-Nerode Theorem
CS 373: Theory of Computtion Mdhusudn Prthsrthy Lecture 09: Myhill-Nerode Theorem 16 Ferury 2010 In this lecture, we will see tht every lnguge hs unique miniml DFA We will see this fct from two perspectives
More information12.1 Nondeterminism Nondeterministic Finite Automata. a a b ε. CS125 Lecture 12 Fall 2014
CS125 Lecture 12 Fll 2014 12.1 Nondeterminism The ide of nondeterministic computtions is to llow our lgorithms to mke guesses, nd only require tht they ccept when the guesses re correct. For exmple, simple
More informationSelf-similarity and symmetries of Pascal s triangles and simplices mod p
Sn Jose Stte University SJSU ScholrWorks Fculty Publictions Mthemtics nd Sttistics Februry 2004 Self-similrity nd symmetries of Pscl s tringles nd simplices mod p Richrd P. Kubelk Sn Jose Stte University,
More informationStreamed Validation of XML Documents
Preliminries DTD Document Type Definition References Jnury 29, 2009 Preliminries DTD Document Type Definition References Structure Preliminries Unrnked Trees Recognizble Lnguges DTD Document Type Definition
More informationThe First Fundamental Theorem of Calculus. If f(x) is continuous on [a, b] and F (x) is any antiderivative. f(x) dx = F (b) F (a).
The Fundmentl Theorems of Clculus Mth 4, Section 0, Spring 009 We now know enough bout definite integrls to give precise formultions of the Fundmentl Theorems of Clculus. We will lso look t some bsic emples
More informationIntegral points on the rational curve
Integrl points on the rtionl curve y x bx c x ;, b, c integers. Konstntine Zeltor Mthemtics University of Wisconsin - Mrinette 750 W. Byshore Street Mrinette, WI 5443-453 Also: Konstntine Zeltor P.O. Box
More informationTypes of Finite Automata. CMSC 330: Organization of Programming Languages. Comparing DFAs and NFAs. NFA for (a b)*abb.
CMSC 330: Orgniztion of Progrmming Lnguges Finite Automt 2 Types of Finite Automt Deterministic Finite Automt () Exctly one sequence of steps for ech string All exmples so fr Nondeterministic Finite Automt
More informationMath 360: A primitive integral and elementary functions
Mth 360: A primitive integrl nd elementry functions D. DeTurck University of Pennsylvni October 16, 2017 D. DeTurck Mth 360 001 2017C: Integrl/functions 1 / 32 Setup for the integrl prtitions Definition:
More informationCS375: Logic and Theory of Computing
CS375: Logic nd Theory of Computing Fuhu (Frnk) Cheng Deprtment of Computer Science University of Kentucky 1 Tble of Contents: Week 1: Preliminries (set lgebr, reltions, functions) (red Chpters 1-4) Weeks
More informationDesigning finite automata II
Designing finite utomt II Prolem: Design DFA A such tht L(A) consists of ll strings of nd which re of length 3n, for n = 0, 1, 2, (1) Determine wht to rememer out the input string Assign stte to ech of
More informationCOMPUTER SCIENCE TRIPOS
CST.2011.2.1 COMPUTER SCIENCE TRIPOS Prt IA Tuesdy 7 June 2011 1.30 to 4.30 COMPUTER SCIENCE Pper 2 Answer one question from ech of Sections A, B nd C, nd two questions from Section D. Submit the nswers
More informationState Minimization for DFAs
Stte Minimiztion for DFAs Red K & S 2.7 Do Homework 10. Consider: Stte Minimiztion 4 5 Is this miniml mchine? Step (1): Get rid of unrechle sttes. Stte Minimiztion 6, Stte is unrechle. Step (2): Get rid
More informationChapter 0. What is the Lebesgue integral about?
Chpter 0. Wht is the Lebesgue integrl bout? The pln is to hve tutoril sheet ech week, most often on Fridy, (to be done during the clss) where you will try to get used to the ides introduced in the previous
More informationFormal Languages and Automata
Moile Computing nd Softwre Engineering p. 1/5 Forml Lnguges nd Automt Chpter 2 Finite Automt Chun-Ming Liu cmliu@csie.ntut.edu.tw Deprtment of Computer Science nd Informtion Engineering Ntionl Tipei University
More informationBases for Vector Spaces
Bses for Vector Spces 2-26-25 A set is independent if, roughly speking, there is no redundncy in the set: You cn t uild ny vector in the set s liner comintion of the others A set spns if you cn uild everything
More informationCMSC 330: Organization of Programming Languages
CMSC 330: Orgniztion of Progrmming Lnguges Finite Automt 2 CMSC 330 1 Types of Finite Automt Deterministic Finite Automt (DFA) Exctly one sequence of steps for ech string All exmples so fr Nondeterministic
More informationFor convenience, we rewrite m2 s m2 = m m m ; where m is repeted m times. Since xyz = m m m nd jxyj»m, we hve tht the string y is substring of the fir
CSCI 2400 Models of Computtion, Section 3 Solutions to Homework 4 Problem 1. ll the solutions below refer to the Pumping Lemm of Theorem 4.8, pge 119. () L = f n b l k : k n + lg Let's ssume for contrdiction
More information