Model-based clustering of longitudinal data. McNicholas, Paul D.; Murphy, Thomas Brendan. Canadian Journal of Statistics, 38 (1):

Size: px
Start display at page:

Download "Model-based clustering of longitudinal data. McNicholas, Paul D.; Murphy, Thomas Brendan. Canadian Journal of Statistics, 38 (1): 153-168"

Transcription

1 Povided by the autho(s) and Univesity College Dublin Libay in accodance with publishe policies Please cite the published vesion when available Title Model-based clusteing of longitudinal data Autho(s) McNicholas, Paul D; Muphy, Thomas Bendan Publication Date Publication infomation Canadian Jounal of Statistics, 8 (1): Publishe Wiley Link to publishe's vesion This item's ecod/moe infomation Rights This is the autho's vesion of the following aticle: "Modelbased clusteing of longitudinal data" published in The Canadian Jounal of Statistics Vol 4, No 4, 2006, available at DOI Downloaded T15:59:06Z Some ights eseved Fo moe infomation, please see the item ecod link above

2 The Canadian Jounal of Statistics Vol 4, No 4, 2006, Pages???-??? La evue canadienne de statistique Model-based clusteing of longitudinal data Paul D McNICHOLAS and Thomas Bendan MURPHY Key wods and phases: Cholesky decomposition; longitudinal data; mixtue models; model-based clusteing; time couse data; yeast spoulation MSC 2000: Pimay 62H0; seconday 62P10 Abstact: A new family of mixtue models fo the model-based clusteing of longitudinal data is intoduced The covaiance stuctues of eight membes of this new family of models ae given and the associated maximum likelihood estimates fo the paametes ae deived via expectation-maximization (EM) algoithms The Bayesian infomation citeion is used fo model selection and a convegence citeion based on Aitken s acceleation is used to detemine convegence of these EM algoithms This new family of models is applied to yeast spoulation time couse data, whee the models give good clusteing pefomance Futhe constaints ae then imposed on the decomposition to allow a deepe investigation of coelation stuctue of the yeast data These constaints geatly extend this new family of models, with the addition of many pasimonious models Title in Fench: we can supply this Résumé : A new family of mixtue models fo the model-based clusteing of longitudinal data is intoduced The covaiance stuctues of eight membes of this new family of models ae given and the associated maximum likelihood estimates fo the paametes ae deived via expectation-maximization (EM) algoithms The Bayesian infomation citeion is used fo model selection and a convegence citeion based on Aitken s acceleation is used to detemine convegence of these EM algoithms This new family of models is applied to yeast spoulation time couse data, whee the models give good clusteing pefomance Futhe constaints ae then imposed on the decomposition to allow a deepe investigation of coelation stuctue of the yeast data These constaints geatly extend this new family of models, with the addition of many pasimonious models 1 INTRODUCTION Longitudinal data aise when measuements ae taken on each subject at a numbe of points in time The esulting insight into behaviou ove time sepaates longitudinal data fom othe types of data Howeve, modelling longitudinal data equies special consideations; in paticula, the coelation between measuements on each subject must be taken into account Subjects in longitudinal studies, o panel studies, ae often consideed to be independent, but this is not always the case Conside, fo example, data on the weights of calves on one of two diffeent methods of contolling intestinal paasites (Kenwad, 1987) Diggle et al (1994) and Eveitt (1995) pesent a vaiety of methods that can be used to analyze these data and, in geneal, to analyze longitudinal data with known goups Due to the typically pospective natue of longitudinal data studies, goup membeships will usually be known a pioi Howeve, situations do aise whee goup membe- 1

3 ships ae not known and even whee the pupose of the analysis is to find goups, o clustes, in the data One such situation aises when the pupose of the study is to find goups of genes with simila activation pattens ove time An example of this is the study that was conducted by Chu et al (1998) to investigate the behaviou of yeast spoulation data ove time The esulting data ae analyzed in Section using the model-based clusteing technique that is intoduced in this wok Model-based clusteing is a technique fo clusteing data though the imposition of a mixtue modelling famewok A Gaussian mixtue model is most fequently used and its density is of the fom G f(x) = π g φ(x µ g, Σ g ), whee π g is the pobability of membeship of goup g and φ(x µ g, Σ g ) is the density of a multivaiate Gaussian distibution with mean µ g and covaiance Σ g Banfield and Raftey (199), Celeux and Govaet (1995) and Faley and Raftey (1998, 2002) exploited an eigenvalue decomposition of the goup covaiance matices to give a wide ange of pasimonious covaiance stuctues This wok culminated in the MCLUST family of models, which consists of ten mixtue models that aise fom the imposition of constaints on the goup covaiance matix Σ g = λ g H g A g H g, whee λ g is a constant, H g is a matix of eigenvectos of Σ g and A g is a diagonal matix with enties popotional to the eigenvalues of Σ g Details of the constaints that can be imposed ae summaized in Faley and Raftey (2006, Table 1) MCLUST is the most well-established model-based clusteing technique within the liteatue, which is patly due to the mclust package (Faley and Raftey, 200) that is available within the R softwae (R Development Coe Team, 2009) Bouveyon et al (2007) intoduced a family of mixtue models specifically fo the analysis of high-dimensional data and McNicholas and Muphy (2008) developed a family of pasimonious Gaussian mixtue models that is closely elated to the mixtue of facto analyzes model (Ghahamani and Hinton, 1997; McLachlan et al, 200) In all of these cases, the classical appoach to model-based clusteing is taken, whee each altenative covaiance stuctue coesponds to a membe of the family of mixtue models Howeve, some non-classical appoaches have been taken to the model-based clusteing of longitudinal data De la Cuz-Mesía et al (2008) use a mixtue of non-linea hieachical models The modelling paadigm that they popose, which is essentially an extension of Paule and Laid (2000), makes each component density subject-specific and the only modelling of the component covaiance matix that they engage in is the imposition of the isotopic constaint Although classical model-based clusteing continues to extend into new application aeas, none of the models that ae cuently available have a covaiance stuctue specifically designed fo the analysis of longitudinal data The aim of this pape is to intoduce a family of mixtue models with a covaiance stuctue specifically designed fo the model-based clusteing of longitudinal data Since the outcome vaiable x is ecoded in a time odeed manne, a covaiance stuctue that explicitly accounts fo the elationship between measuements at diffeent time points is necessay Pouahmadi (1999, 2000) exploited the fact that covaiance matix Σ of a andom vaiable can be decomposed using the elation T ΣT = D, whee T is a unique lowe tiangula matix with diagonal elements 1 and D is a unique diagonal matix with stictly positive enties This elation is known as the modified Cholesky decomposition and it was used by Kzanowski et al (1995) in a disciminant analysis application The modified Cholesky decomposition may equivalently be expessed in the fom Σ 1 = T D 1 T, which is convenient when modelling the covaiance of a multivaiate Gaussian distibution The values of T and D have intepetations as genealized autoegessive paametes and innovation vaiances, espectively (Pouahmadi, 1999) so that the 2

4 linea least-squaes pedicto of Y t, based on Y t 1,,Y 1, can be witten t 1 Ŷ t = µ t + ( φ ts )(Y s µ s )+ d t ɛ t, (1) s=1 whee ɛ t N (0, 1), the φ ts ae the (sub-diagonal) elements of T and the d t ae the diagonal elements of D Pan and MacKenzie (200) exploited the modified Cholesky decomposition to jointly model the mean and covaiance in longitudinal studies Pouahmadi et al (2007) developed a method of simultaneously modelling seveal covaiance matices using this decomposition; this wok gives an altenative to common pincipal components analysis (Fluy, 1988) fo longitudinal data In Section 2, we develop a model-based clusteing famewok fo longitudinal data by using Gaussian mixtue models whee the modified Cholesky decomposition of the goup covaiance matices ae constained in ode to give pasimonious models The mixtue models ae fitted using an EM algoithm (Dempste et al, 1977), as outlined in Section 22 The models ae applied to time couse gene expession data in Section, whee they exhibit good clusteing pefomance In Section 4, the stuctue of the lowe tiangula matix is exploited to extend this family of models to allow fo situations whee only autocoelations up to lag d ae equied This extension of the family of models gives ise to moe pasimonious models The extended family of models is then applied to a data set on the weight of ats on one of thee diffeent dietay supplements, whee one of the extended models is chosen The esults of this wok ae summaized in Section 5 2 GAUSSIAN MIXTURE MODELS WITH CHOLESKY-DECOMPOSED COVARIANCE STRUCTURE 21 The model We assume a Gaussian mixtue model, with a modified Cholesky-decomposed covaiance stuctue, fo each mixtue component Theefoe, the density of an obsevation x i in goup g is given by 1 f(x i µ g,t g,d g )= { (2π)p D g exp 1 } 2 (x i µ g ) T g D 1 g T g(x i µ g ), whee T g is the p p lowe tiangula matix and D g is the p p diagonal matix that follow fom the modified Cholesky decomposition of Σ g Now, thee is the option to constain the T g o the D g to be equal acoss goups and thee is also the option to impose the isotopic constaint D g = δ g I p (cf Tipping and Bishop, 1999), which leads to a family of eight Gaussian mixtue models Each membe of this family, along with thei espective nomenclatue and numbe of covaiance paametes, is given in Table 1 The nomenclatue is quite intuitive; fo example, the VEA model has vaiable autoegessive stuctue and equal, anisotopic noise acoss goups Constaining the T g to be equal acoss goups suggests that the coelation stuctue of the longitudinally ecoded data values is the same fo all of the goups In this context, the coelation stuctue eflects the autoegessive elationship between time points as outlined in Equation 1 Imposing the constaint that the D g ae equal acoss goups suggests that the vaiability at each time point is the same fo each goup and imposing the isotopic constaint D g = δ g I p suggests that the vaiability is the same at all time points Fo each given data set, any of the eight combinations of these constaints given in Table 1 might be most appopiate Two of the models given in Table 1, EEA and VVA, ae equivalent, fom a clusteing viewpoint, to models that aleady exist within the MCLUST famewok Howeve, the MCLUST covaiance stuctue does not explicitly account fo the longitudinal coelation stuctue and so the models

5 Table 1: The nomenclatue, covaiance stuctue and numbe of covaiance paametes fo each model Id Model T g D g D g Numbe of Covaiance Paametes 1 EEA Equal Equal Anisotopic p(p 1)/2+p 2 VVA Vaiable Vaiable Anisotopic G[p(p 1)/2] + Gp VEA Vaiable Equal Anisotopic G[p(p 1)/2] + p 4 EVA Equal Vaiable Anisotopic p(p 1)/2+Gp 5 VVI Vaiable Vaiable Isotopic G[p(p 1)/2] + G 6 VEI Vaiable Equal Isotopic G[p(p 1)/2] EVI Equal Vaiable Isotopic p(p 1)/2+G 8 EEI Equal Equal Isotopic p(p 1)/2 + 1 intoduced heein ae moe natual fo longitudinal data Futhe, these models will give infomation about the natue of the covaiance stuctue specifically, egading the autoegessive stuctue and the innovation vaiances that will not aise fom MCLUST 22 Model fitting The models ae fitted using an EM algoithm The missing data ae taken to be the goup membeship labels, which we denote z, whee z ig = 1 if obsevation i is in goup g and z ig =0 othewise Combining the missing data z with the known data x, gives the complete-data (x, z) The complete-data likelihood fo the mixtue model is given by L c (π g,µ g,t g,d g )= n i=1 G [π g f(x i µ g,t g,d g )] zig, and the expected value of the complete-data log-likelihood fo the mixtue model is Q(π g,µ g,t g,d g )= G n g log π g np 2 log 2π G whee the z ig have been eplaced by thei expected values ẑ ig = n g 2 log D g ˆπ g f(x i ˆµ g, ˆT g, ˆD g ) G h=1 ˆπ hf(x i ˆµ h, ˆT h, ˆD h ), G n g 2 t{ T g S g T gd g 1 }, (2) n g = n i=1 ẑig and S g = (1/n g ) n i=1 ẑig(x i µ g )(x i µ g ) Now, maximising Q with espect to π g and µ g gives ˆµ g = n i=1 ẑigx i / n i=1 ẑig and ˆπ g = n g /n, espectively The paamete estimates fo T g and D g ae also deived by maximizing Q and these depend on the constaints (Table 1) used in the model The paamete estimates fom the M-step of the VVI model ae deived in Section 2 Aside fom the EVA and EVI models, estimates fo the paametes of the othe models ae aived at in a simila fashion and ae available fom the authos upon equest The deivation of estimates fo the EVA model ae given in the Appendix; the EVI estimation pocedue is simila to the EVA pocedue 4

6 2 Paamete Estimates fo the VVI Model Imposing the constaint D g = δ g I p and diffeentiating Equation 2 with espect to T g and δg 1 espectively gives the following scoe functions S 1 (T g,δ g )= Q(T g,δ g ) T g S 2 (T g,δ g )= Q(T g,δ g ) δg 1 ) = n g δ g T g S g = n g ( T g Sg + S g 2δ g = n g ( pδg t { T g S g T }) g 2 Only the lowe tiangula pat of T g needs to be estimated, so we need to solve the system of equations given by the lowe tiangle of S 1 ( ˆT g,δ g ) = 0 Using the notation of Pouahmadi et al (2007), let φ (g) ij epesent those elements of T g that ae to be estimated, so that T g = φ (g) φ (g) 1 φ (g) φ (g) p 1,2 φ (g) p 1,p φ (g) p1 φ (g) p2 φ (g) p,p 2 φ (g) p,p 1 1 φ (g) p 1,1, () and wite S 1 (T g,δ g ) S 1 (Φ g,δ g ), whee Φ g = {φ (g) ij } fo i > j and i, j {1,, p} Also, let LT{ } denote the lowe tiangula pat of a matix Now, solving LT { S 1 (ˆΦ g,δ g ) } = 0 fo ˆΦ g leads to a total of p 1 systems of linea equations and the solution to each of these equations can be witten 1 2, 1 = , ,2 1, 1 2, 2 1, , 1 fo =2,, p Solving diag{s 2 (ˆΦ g, ˆδ g )} = 0 fo ˆδ g, gives ˆδ g = (1/p) t { ˆTg S g ˆT g } ANALYSES 1 Convegence citeion The Aitken acceleation was used to povide an asymptotic estimate of the log-likelihood at each iteation and this estimate was then used to detemine the convegence of each EM algoithm The Aitken acceleation at iteation m is given by a (m) = l(m+1) l (m) l (m) l (m 1), whee l (m+1), l (m) and l (m 1) ae the log-likelihood values fom iteations m + 1, m and m 1 espectively Then, the asymptotic estimate of the log-likelihood at iteation m + 1 is given by l (m+1) = l (m) a (m) (l(m+1) l (m) ), 5

7 (Böhning et al, 1994) and the EM algoithm can be said to have conveged when l (m+1) l (m+1) <ɛ (Lindsay, 1995) o when l (m+1) l (m) <ɛ (McNicholas et al, 2010) The latte citeion has the advantage that it is necessaily at least as stict as the lack of pogess l (m+1) l (m) <ɛ 2 Model selection The Bayesian infomation citeion (BIC; Schwatz, 1978) is used to select the best membe of this family of Gaussian mixtue models The BIC can be witten BIC =2l(x, ˆθ) ρ log N, whee l(x, ˆθ) is the maximized log-likelihood, ˆθ is the maximum likelihood estimate of θ, ρ is the numbe of fee paametes in the model and N is the numbe of obsevations The use of the BIC fo mixtue model selection is justified by Keibin (1998, 2000), who shows that it gives consistent estimates of the numbe of components in a mixtue model, unde cetain egulatoy conditions Futhemoe, Faley and Raftey (1998, 2002) and McNicholas and Muphy (2008) give pactical evidence that the BIC is effective as a model selection citeion fo Gaussian mixtue models Yeast Spoulation Time Couse Data Spoulation is a pocess by which diploid cells of budding yeast give ise to haploid cells Chu et al (1998) measue changes in gene expession duing spoulation using 97% of yeast genes; thei study used 6118 gene expessions that wee measued ove seven time points t {0, 05, 20, 50, 70, 90, 115} The ole of clusteing in such time couse analyses is impotant since the objective is to find goups of genes that expess similaly ove the couse of the expeiment Genes with simila expessions ae said to be co-expessed Cetain genes ae known to have specific functions and when othe genes ae found to co-expess with these genes, new insight can be gained into thei function Chu et al (1998) eliminated about 80% of the genes pio to thei analysis by focusing on the genes that showed obvious changes in expession and esticting themselves to genes that wee induced (showed inceased expession) Mitchell (1994) had peviously suggested that thee wee fou tempoal classes of these genes Chu et al (1998) contended that fou goups was not sufficient to epesent the divesity of the obseved expession pattens and they chose seven tempoal pattens that seemed appopiate These pattens wee chosen by eye A total of thity genes (Table 2) wee selected as epesentative of these seven pattens; these ae known as model expession pofiles Then the emaining genes wee clusteed into the seven goups based on thei coelation with these model pofiles Wakefield et al (200) used a fou-stage Bayesian hieachical model to analyze this time couse data Befoe applying the hieachical model, Wakefield et al (200) used Bayes factos to educe the numbe of genes fom 6118 to 1104, conceding that thei modelling paadigm may be computationally pohibitive fo a lage numbe of genes They found that the numbe of tempoal classes was pobably between 11 and 14, with G = 12 being the most pobable On compaison of thei G = 12 component model to the model pofiles (40 genes) of Chu et al (1998), they concluded that thee model offeed new insights into co-expession Note that, while Chu et al (1998) consideed seven time points, Wakefield et al (200) consideed six time points, dopping t = 0 and and taking the values at each othe time elative to time t = 0 fo each gene In ou analyses of these data, all seven time points wee consideed and no genes wee deleted As is common, the negative logaithm, base 2, of each obsevation was taken and then each time point was standadized to have mean zeo and vaiance one pio to the analysis Then the novel model-based clusteing technique intoduced heein was applied to the these data fo G =1,, 20 using five andom stating values fo ẑ fo each of the eight models and each value of G The BIC fo each model and each of the 20 values of G is depicted in Figue 1; aside fom a few of the ealie G values, fo which the VVA model was chosen, the EVA model was selected Fom Figue 1, it is appaent that this family of models suggest that the tue numbe of goups in the spoulation data is somewhee in the ealy to mid teens Moe pecisely, the best model 6

8 BIC G Figue 1: A plot of BIC values vesus numbe of goups fo all eight models was a EVA model with G = 1 and a BIC of Selection of the EVA model indicates that, while the autoegessive stuctue of the data, as suggested by the T matix, is the same acoss goups (tempoal pattens), the innovation vaiances diffe both between tempoal pattens and between times A coss tabulation (Table 2) of the cluste membeship and the model pofiles of Chu et al (1998) shows that thee is some coespondence fo these 40 genes; the Late goup coesponds pefectly and some of the late goups ae vey simila This is what one would expect since genes that wee induced late would be the easiest to spot by eye Howeve, thee ae notable diffeences in estimated co-expessions fo the ealie goups Note that thee ae only eight columns in Table 2, despite the fact that the best model in ou analysis had G = 1 goups This is because ou model is based on all 6,118 genes and the 40 model pofiles of Chu et al (1998) appeaed in just eight of ou 1 goups Table 2: Fequencies of the 40 genes selected as model pofiles by Chu et al (1988), coss-tabulated by the seven tempoal pattens suggested by Chu et al (1998) and by ou tempoal pattens (A H) Goup A Goup B Goup C Goup D Goup E Goup F Goup G Goup H Metabolic 4 2 Ealy 1 5 Ealy Ealy-Mid 4 1 Middle 4 Mid-Late Late 4 7

9 This analysis pesents new insight into these data by poviding 1 distinct tempoal pattens, based all 6118 genes This insight is quite diffeent fom the esults of Wakefield et al (200), who split 1104 of the 6118 genes into twelve tempoal pattens The lagest goup in ou analysis had 180 genes and so it is cetainly not the case that ou goups ae just the twelve of Wakefield et al (200) plus a noise goup containing the 5014 genes that Wakefield et al (200) deleted As mentioned in Section 1, MCLUST is the most well-established Gaussian model-based clusteing technique within the liteatue In ode to illustate the usefulness of the novel technique intoduced heein, elative to existing methods, the MCLUST family of models was also used to analyze these data The data was pepocessed in the same fashion, by taking negative logaithms and standadizing, and the mclust softwae fo R was used All ten MCLUST models wee un fo G =1,, 20 and the best model was a VVV model with six components This model, which is equivalent to the VVA model fom Table 1, had a BIC of , which is notably less than the BIC fo the G = 1 component EVA model ( ) Fom Table, it is appaent that thee is no coespondence between the model pofiles of Chu et al (1998) and the MCLUST tempoal pattens As mentioned ealie, one would expect to see a coespondence in some of the late model pofiles The coespondence between the MCLUST esults and the method intoduced heein is given in Table 4: it is appaent that ou EVA model is not simply splitting the MCLUST goups but is suggesting a substantially diffeent suctue Table : Fequencies of the 40 genes selected as model pofiles by Chu et al (1988), coss-tabulated by the tempoal pattens suggested by Chu et al (1998) and by the MCLUST pattens (I IV) Goup I Goup II Goup III Goup IV Metabolic 6 Ealy 1 5 Ealy Ealy-Mid 5 Middle 4 Mid-Late 1 2 Late 4 4 CONSTRAINING SUB-DIAGONALS 41 Intoduction Duing the analysis of the yeast spoulation data it was noted that many of the values below the fist sub-diagonal of the estimated T matix (Equation 4) wee small In fact, all but one of the elements below the fist sub-diagonal could be consideed small; T = (4) 8

10 Table 4: Fequencies of all 6,118 genes, coss-tabulated by ou tempoal pattens (A M) and by the MCLUST pattens (I VI) A B C D E F G H I J K L M Goup I Goup II Goup II Goup IV Goup V Goup VI While it is difficult to detemine if a value in T is small without taking the values of the D g in context, it led to the notion of setting vaious sub-diagonals of T g to zeo and hence the possibility of a moe pasimonious class of models This constained coelation stuctue emoves any autocoelation ove lage time lags; that is, T g constained to d sub-diagonals implies an ode d autoegessive stuctue within the famewok of Equation 1 In this section we deive paamete estimates when cetain sub-diagonals of T g ae set to zeo 42 Constaints & nomenclatue We constain the elements of T g to be zeo below a given numbe of sub-diagonals The notation V 1 VA is used to denote the VVA model whee the elements of T g ae zeo below the fist subdiagonal, V 2 VA denotes the VVA model whee the elements of T g ae zeo below the second sub-diagonal, and so foth Note that, although not used heein, models whee all sub-diagonal elements ae zeo, such as V 0 VA, ae equivalent to the diagonal MCLUST models Woking out paamete estimates when only the fist sub-diagonal of T g is non-zeo is tivial Fo example, fom the computations of Section 2 it is clea that the paamete estimates fo T g (g) in the M-step of the V 1 VI model will be ˆφ, 1 = s(g), 1 /s(g) 1, 1, fo =2,, p 4 Paamete estimates fo the V 2 VA and V d VA models Diffeentiating Equation 2 with espect to T g and Dg 1, espectively, gives the scoe functions S 1 (T g,d g )= n g Dg 1T gs g and S 2 (T g,d g )=n g /2 ( D g T g S g T g) Using the familia notation, we can wite T g as in Equation but with zeos below the second sub diagonal The notation SD { } is { used heetofoe to denote the fist sub-diagonals of a matix Now, solving SD 2 S1 (ˆΦ g,d g ) } =0 fo ˆΦ g leads to a total of p 1 systems of linea equations, all but one of which is 2 2 This one (g) exception is 1 1, which gives the familia solution ˆφ 21 = s(g) 21 /s(g) 11 The solutions in the 2 2 cases ae given by ( ˆφ(g), 2, 1 ) = ( 2, 2 2, 1 1, 2 1, 1 ) 1 (, 2, 1 fo =,, p and solving diag { S 2 (ˆΦ g, ˆD g ) } = 0 fo ˆD g, gives ˆD g = diag { ˆTg S g ˆT g } ), 9

11 The paamete estimates fo T g and D g can be genealized to the V d VA case Using the same notation, the ˆΦ g ae given by, d, (d 1), 1 = d, d d, (d 1) d, 1 (d 1), d 1, d (d 1), (d 1) 1, (d 1) (d 1), 1 1, 1 1, d, (d 1), 1 fo =2,, p and, once again, ˆDg = diag { ˆTg S g ˆT g } Paamete estimates in the othe cases fo this geneal constaint ae simila except in the EVA and EVI cases Estimates fo the EVA case ae given in the Appendix and those fo the EVI case ae vey simila 44 Application to yeast spoulation data The eight models intoduced heein (Table 1) wee applied to the yeast spoulation data with all elements of T g below the fist and second sub-diagonals, espectively, set to zeo (d =1,d = 2) The esult of this analysis was that the EVA model, with full T (d = 6), was still the best model The best of these two constained models fo G = 1 was the E 1 VA model, with BIC= The E d VA model was then fitted to these data fo d =, 4, 5 The best of these models was the E 5 VA with BIC Theefoe, the best model was still the full EVA model Inteestingly, the full model being bette than the E 5 VA model indicates that coelation pesists acoss all time points including time zeo; as peviously mentioned, this time point was dopped in the analysis of Wakefield et al (200) 45 Application to ats data In ode to show that a model with constaints imposed on the sub-diagonals of T g will sometimes be selected, the following analysis is pesented Data on the body weights of ats on one of thee diffeent dietay supplements wee souced fom the nlme package (Pinheio et al, 2008) fo the R softwae These data wee published in Cowde and Hand (1990) and have been analyzed many times: see Hand and Cowde (1996) and Haslett (1997) fo examples They ae used solely fo illustative puposes hee and what follows is not intended to be an in-depth analysis of these data Fo one thing, we make no attempt to model the component means which one may do by allowing fo a systematic tend, fo example A total of 16 ats wee put on one of thee diffeent diets; eight ats wee on Diet 1, fou wee put on Diet 2 and fou on Diet Weights wee fist ecoded afte a settling-in peiod and then weekly fo a peiod of nine weeks An exta measuement was taken at 44 days to help gauge the effect of a teatment that occued duing the sixth week These 11 measuements can be seen on the time seies plot in Figue 2 Fom this figue, it is clea that the ats ae gouped by weight, with two exceptions: a heavie at on Diet 2 and a lighte at on Diet Although the tue diets ae known, we teat this as a clusteing poblem and so assume no pio knowledge of component membeship The eight models in Table 1 wee fitted to these data fo G =1, 2,, 6 The best model was an EEA model with G = 5 This model put the two exceptions into goups on thei own and all othe ats wee coectly classified Selection of an EEA model suggests that the covaiance stuctue is the same fo each goup and so the classification is effectively based on the mean Futhe, the fact that the isotopic constaint was not imposed implies that the innovation vaiance is not the same at each time point To illustate that a model with constaints imposed on the sub-diagonals of T g will sometimes be selected, the E d EA models wee fitted to these data fo d =1, 2,, 10, whee the E 10 EA model is equivalent to the full EEA model The BIC fo each model is given in Table 5 and the, 10

12 Weight (g) Time (days) Figue 2: Time seies fo each at, by diet: Diet 1 (solid lines), Diet 2 (dashed) and Diet (dotted) best model was the E 8 EA model, which has a T matix with eight non-zeo sub-diagonals The estimated cluste membeships wee identical to those fo the EEA model Table 5: BIC values fo the E i EA models fitted to the ats data fo i =1, 2,, 10 Model BIC Model BIC E 1EA E 6EA 527 E 2EA E 7EA 5691 E EA E 8EA E 4EA 5047 E 9EA E 5EA E 10EA SUMMARY A new model-based clusteing technique, using Gaussian mixtue models, has been intoduced fo the analysis of longitudinal data This family of mixtue models follows the classical appoach whee each membe of the family has diffeent constaints imposed on the modified Cholesky-decomposed covaiance stuctue Initially, eight membes of this family wee given and the associated maximum likelihood estimates fo thei paametes wee deived using an EM algoithm These models povided new insight when applied to yeast spoulation time couse data Futhemoe, by constaining cetain sub-diagonals of T g to be zeo, the numbe of membes of this family of models was geatly inceased, including membes with moe pasimonious covaiance stuctues The family of models offes much scope fo futhe expansion by constaining subsets of the T matix, which captue the coelation stuctue of the data Initial extensions constaining sub- 11

13 diagonals of the T matix wee given in this wok, focusing on adjacent sub-diagonals The exta models that wee obtained as a esult of this extension whee shown to be useful when applied to a well known data set on the weight of ats Futue wok will focus on constaining diffeent combinations of the sub-diagonals of T thee ae 2 p 1 possible combinations and on the incopoation of missing data into the modelling famewok APPENDIX Paamete estimates fo the EVA model when T g is not constained Fo the EVA model, T g = T and so, fom Equation 2, the expected value of the complete-data log-likelihood Q can be witten Q(T, D g )=C G n g 2 log D g espectively gives the following scoe func- Diffeentiating Equation 5 with espect to T g and Dg 1 tions, S 1 (T, D g )= Q(T, D g) T = G G n g 2 t { TS g T Dg 1 } (5) n g Dg 1 TS g, and S 2 (T, D g )= Q(T, D g) Dg 1 = n g 2 (D g TS g T ) Now, solving LT { S 1 ( ˆT, D g ) } S 1 (ˆΦ,D g ) = 0 fo ˆΦ leads again to a total of p 1 systems of linea equations The solution fo the fist ow of the lowe tiangle of T g is, G n g [ 11 ˆφ 21 ˆd (g) 22 ] + s(g) 21 ˆd (g) 22 G n g =0, and so ˆφ21 = [ G n g Fo convenience, we intoduce the notation κ ij m = [ G π g ij [ 21 ˆd (g) ˆd (g) 22 ] [ G π g ] = [ G π g 21 ˆd (g) ˆd (g) 22 ] ] (6) ] (g) / ˆd mm, so that Equation 6 can be witten ˆφ 21 = κ 21 2 /κ 11 2 Now, solving the second ow means solving the linea system ( κ 11 κ 21 κ 12 κ 22 )( ) ( ˆφ1 κ 1 = ˆφ 2 κ 2, 1 ), and so, ( ˆφ1 ˆφ 2 ) ( κ 11 = κ 21 κ 12 κ 22 It follows that the solution to the ( 1)st system of equations is given by ˆφ 1 κ 11 κ 21 κ 1,1 1 ˆφ 2 = κ 12 κ 22 κ 1,2 κ 1, 1 κ 2, 2 κ 1, 1 κ 1 κ 2 κ, 1 ) 1 ( κ 1 κ 2 ), (7) fo =2,, p Note that κ ij m = κ ji m and so the ( 1) ( 1) matix in Equation 7 is symmetic Solving the second scoe function, diag { S 2 ( ˆT g, ˆD g ) } = 0, gives ˆD g = diag { ˆTSg ˆT } Paamete estimates when sub-diagonals of T g ae constained In most of the cases whee subdiagonals ae set to zeo, the solutions ae vey simila to two and d-sub-diagonal cases detailed in Section 4, and so they ae not given hee Howeve, in the EVA and EVI cases, the solutions ae a little moe involved than the othe cases and so the deivations of the paametes in the M-step fo the E 2 VA and E d VA cases ae povided in full The coesponding estimates fo the EVI model 12

14 ae simila Recall that diffeentiating Equation 5 with espect to T g and Dg 1 espectively gives the following scoe functions, S 1 (T, D g )= Q(T, D g) T = G n g Dg 1 TS g, and S 2 (T, D g )= Q(T, D g) Dg 1 = n g 2 (D g TS g T ) { Fo model E 2 VA, solving SD 2 S1 ( ˆT, D g ) } = 0 fo ˆΦ leads again to a total of p 1 systems of linea equations, all but one of which is 1 1 The solution fo the 1 1 system is ˆφ 21 = κ 21 2 /κ11 2 and solving the emaining systems gives the solution ( ) ( ˆφ, 2 κ 2, 2 = ˆφ, 1 κ 2, 1 κ 1, 2 κ 1, 1 ) 1 ( κ, 2 κ, 1 ), fo =,, p Now, these estimates can be extended to the E d VA case as follows;, d, (d 1), 1 = κ d, d κ d, (d 1) κ d, 1 κ (d 1), d κ 1, d κ (d 1), (d 1) κ 1, (d 1) κ (d 1), 1 κ 1, 1 1 κ, d κ, (d 1) κ, 1, fo =2,, p Fo any d, solving diag { S 2 ( ˆT g, ˆD g ) } = 0 fo ˆD g, gives ˆD g = diag { ˆTSg ˆT } ACKNOWLEDGEMENTS This eseach was suppoted by a Discovey Gant fom the Natual Sciences and Engineeing Reseach Council of Canada and a Basic Reseach Gant fom Science Foundation Ieland The authos ae gateful to an Associate Edito and two efeees fo thei helpful and insightful suggestions REFERENCES J D Banfield & A E Raftey (199) Model-based Gaussian and non-gaussian clusteing Biometics, 49, D Böhning, E Dietz, R Schaub, P Schlattmann & B Lindsay (1994) The distibution of the likelihood atio fo mixtues of densities fom the one-paamete exponential family Annals of the Institute of Statistical Mathematics, 46, 7 88 C Bouveyon, S Giad & C Schmid (2007) High-dimensional data clusteing Computational Statistics and Data Analysis, 52, G Celeux & G Govaet (1995) Gaussian pasimonious clusteing models Patten Recognition, 28, S Chu, J DeRisi, M Eisen, J Mulholland, D Botstein, P Bown & I Heskowitz (1998) The tansciptional pogam of spoulation in budding yeast Science, 282, M J Cowde and D J Hand (1990) Analysis of Repeated Measues Chapman and Hall, London R De la Cuz-Mesía, F A Quintana & G Mashall (2008) Model-based clusteing fo longitudinal data Computational Statistics and Data Analysis, 52,

15 A P Dempste, N M Laid & D B Rubin (1977) Maximum likelihood fom incomplete data via the EM algoithm Jounal of the Royal Statistical Society Seies B, 9, 1 8 P J Diggle, K-Y Liang & S L Zege (1994) Analysis of Longitudinal Data Oxfod Univesity Pess, Oxfod B Eveitt (1995) The analysis of epeated measues: A pactical eview with examples The Statistician, 44, B Fluy (1988) Common Pincipal Components and Related Multivaiate Models Wiley, New Yok C Faley & A E Raftey (1998) How many clustes? Which clusteing methods? Answes via modelbased cluste analysis The Compute Jounal, 41, C Faley & A E Raftey (2002) Model-based clusteing, disciminant analysis, and density estimation Jounal of the Ameican Statistical Association, 97, C Faley & A E Raftey (200) Enhanced softwae fo model-based clusteing, density estimation, and disciminant analysis: MCLUST Jounal of Classification, 20, C Faley & A E Raftey (2006) MCLUST vesion fo R: Nomal mixtue modeling and model-based clusteing Technical Repot 504, Depatment of Statistics, Univesity of Washington Z Ghahamani & G E Hinton (1997) The EM algoithm fo facto analyzes Technical Repot CRG- TR-96-1, Univesity Of Toonto D J Hand & M J Cowde (1996) Pactical Longitudinal Data Analysis Chapman and Hall, London J Haslett (1997) Conditional expectations and esidual analysis fo the linea models Applied Stochastic Models and Data Analysis, 1, M G Kenwad (1987) A method fo compaing pofiles of epeated measuements Jounal of the Royal Statistical Society Seies C, 6, C Keibin (1998) Estimation consistante de l ode de modèles de mélange Comptes Rendus de l Académie des Sciences Séie I Mathématique, 26, C Keibin (2000) Consistent estimation of the ode of mixtue models Sankhyā The Indian Jounal of Statistics Seies A, 62, W J Kzanowski, P Jonathan, W V McCathy & M R Thomas (1995) Disciminant analysis with singula covaiance matices: Methods and applications to spectoscopic data Jounal of the Royal Statistical Society Seies C, 44, B G Lindsay (1995) Mixtue Models: Theoy, Geomety and Applications, volume 5 of NSF-CBMS Regional Confeence Seies in Pobability and Statistics Institute of Mathematical Statistics, Haywad, Califonia G J McLachlan, D Peel & R W Bean (200) Modelling high-dimensional data by mixtues of facto analyzes Computational Statistics and Data Analysis, 41, P D McNicholas & T B Muphy (2008) Pasimonious Gaussian mixtue models Statistics and Computing, 18, P D McNicholas, T B Muphy, A F McDaid & D Fost (2010) Seial and paallel implementations of model-based clusteing via pasimonious Gaussian mixtue models Computational Statistics and Data Analysis In pess, doi: /jcsda

16 A P Mitchell (1994) Contol of meiotic gene expession in sacchaomyces ceevisiae Micobiology and Molecula Biology Reviews, 58, J Pan & G MacKenzie (200) On modelling mean-covaiance stuctues in longitudinal studies Biometika, 90, D K Paule & N M Laid (2000) A mixtue model fo longitudinal data with application to assessment of noncompliance Biometics, 56, J Pinheio, D Bates, S DebRoy, D Saka & the R Coe team (2008) nlme: Linea and Nonlinea Mixed Effects Models R package vesion 1-89 M Pouahmadi (1999) Joint mean-covaiance models with applications to longitudinal data: Unconstained paameteisation Biometika, 86, M Pouahmadi (2000) Maximum likelihood estimation of genealised linea models fo multivaiate nomal covaiance matix Biometika, 87, M Pouahmadi, M Daniels & T Pak (2007) Simultaneous modelling of the Cholesky decomposition of seveal covaiance matices Jounal of Multivaiate Analysis, 98, R Development Coe Team (2009) R: A Language and Envionment fo Statistical Computing R Foundation fo Statistical Computing, Vienna, Austia G Schwatz (1978) Estimating the dimension of a model Annals of Statistics, 6, 1 8 T E Tipping & C M Bishop (1999) Mixtues of pobabilistic pincipal component analyses Neual Computation, 11, J C Wakefield, C Zhou & S G Self (200) Modelling gene expession ove time: Cuve clusteing with infomative pio distibutions In J M Benado, M J Bayai, J O Bege, A P Dawid, D Heckeman, A F M Smith & M West, editos, Bayesian Statistics, volume 7, pages Oxfod Univesity Pess, Oxfod Received??? Accepted??? Paul D McNICHOLAS: pmcnicho@uoguelphca Depatment of Mathematics & Statistics, Univesity of Guelph Guelph, Ontaio Canada, N1G 2W1 Thomas Bendan MURPHY: bendanmuphy@ucdie School of Mathematical Sciences, Univesity College Dublin Belfield, Dublin 4 Ieland 15

Chapter 3 Savings, Present Value and Ricardian Equivalence

Chapter 3 Savings, Present Value and Ricardian Equivalence Chapte 3 Savings, Pesent Value and Ricadian Equivalence Chapte Oveview In the pevious chapte we studied the decision of households to supply hous to the labo maket. This decision was a static decision,

More information

STUDENT RESPONSE TO ANNUITY FORMULA DERIVATION

STUDENT RESPONSE TO ANNUITY FORMULA DERIVATION Page 1 STUDENT RESPONSE TO ANNUITY FORMULA DERIVATION C. Alan Blaylock, Hendeson State Univesity ABSTRACT This pape pesents an intuitive appoach to deiving annuity fomulas fo classoom use and attempts

More information

AN IMPLEMENTATION OF BINARY AND FLOATING POINT CHROMOSOME REPRESENTATION IN GENETIC ALGORITHM

AN IMPLEMENTATION OF BINARY AND FLOATING POINT CHROMOSOME REPRESENTATION IN GENETIC ALGORITHM AN IMPLEMENTATION OF BINARY AND FLOATING POINT CHROMOSOME REPRESENTATION IN GENETIC ALGORITHM Main Golub Faculty of Electical Engineeing and Computing, Univesity of Zageb Depatment of Electonics, Micoelectonics,

More information

Financing Terms in the EOQ Model

Financing Terms in the EOQ Model Financing Tems in the EOQ Model Habone W. Stuat, J. Columbia Business School New Yok, NY 1007 hws7@columbia.edu August 6, 004 1 Intoduction This note discusses two tems that ae often omitted fom the standad

More information

ON THE (Q, R) POLICY IN PRODUCTION-INVENTORY SYSTEMS

ON THE (Q, R) POLICY IN PRODUCTION-INVENTORY SYSTEMS ON THE R POLICY IN PRODUCTION-INVENTORY SYSTEMS Saifallah Benjaafa and Joon-Seok Kim Depatment of Mechanical Engineeing Univesity of Minnesota Minneapolis MN 55455 Abstact We conside a poduction-inventoy

More information

An Introduction to Omega

An Introduction to Omega An Intoduction to Omega Con Keating and William F. Shadwick These distibutions have the same mean and vaiance. Ae you indiffeent to thei isk-ewad chaacteistics? The Finance Development Cente 2002 1 Fom

More information

Software Engineering and Development

Software Engineering and Development I T H E A 67 Softwae Engineeing and Development SOFTWARE DEVELOPMENT PROCESS DYNAMICS MODELING AS STATE MACHINE Leonid Lyubchyk, Vasyl Soloshchuk Abstact: Softwae development pocess modeling is gaining

More information

The transport performance evaluation system building of logistics enterprises

The transport performance evaluation system building of logistics enterprises Jounal of Industial Engineeing and Management JIEM, 213 6(4): 194-114 Online ISSN: 213-953 Pint ISSN: 213-8423 http://dx.doi.og/1.3926/jiem.784 The tanspot pefomance evaluation system building of logistics

More information

Ilona V. Tregub, ScD., Professor

Ilona V. Tregub, ScD., Professor Investment Potfolio Fomation fo the Pension Fund of Russia Ilona V. egub, ScD., Pofesso Mathematical Modeling of Economic Pocesses Depatment he Financial Univesity unde the Govenment of the Russian Fedeation

More information

Questions & Answers Chapter 10 Software Reliability Prediction, Allocation and Demonstration Testing

Questions & Answers Chapter 10 Software Reliability Prediction, Allocation and Demonstration Testing M13914 Questions & Answes Chapte 10 Softwae Reliability Pediction, Allocation and Demonstation Testing 1. Homewok: How to deive the fomula of failue ate estimate. λ = χ α,+ t When the failue times follow

More information

Chris J. Skinner The probability of identification: applying ideas from forensic statistics to disclosure risk assessment

Chris J. Skinner The probability of identification: applying ideas from forensic statistics to disclosure risk assessment Chis J. Skinne The pobability of identification: applying ideas fom foensic statistics to disclosue isk assessment Aticle (Accepted vesion) (Refeeed) Oiginal citation: Skinne, Chis J. (2007) The pobability

More information

MATHEMATICAL SIMULATION OF MASS SPECTRUM

MATHEMATICAL SIMULATION OF MASS SPECTRUM MATHEMATICA SIMUATION OF MASS SPECTUM.Beánek, J.Knížek, Z. Pulpán 3, M. Hubálek 4, V. Novák Univesity of South Bohemia, Ceske Budejovice, Chales Univesity, Hadec Kalove, 3 Univesity of Hadec Kalove, Hadec

More information

An Efficient Group Key Agreement Protocol for Ad hoc Networks

An Efficient Group Key Agreement Protocol for Ad hoc Networks An Efficient Goup Key Ageement Potocol fo Ad hoc Netwoks Daniel Augot, Raghav haska, Valéie Issany and Daniele Sacchetti INRIA Rocquencout 78153 Le Chesnay Fance {Daniel.Augot, Raghav.haska, Valéie.Issany,

More information

INITIAL MARGIN CALCULATION ON DERIVATIVE MARKETS OPTION VALUATION FORMULAS

INITIAL MARGIN CALCULATION ON DERIVATIVE MARKETS OPTION VALUATION FORMULAS INITIAL MARGIN CALCULATION ON DERIVATIVE MARKETS OPTION VALUATION FORMULAS Vesion:.0 Date: June 0 Disclaime This document is solely intended as infomation fo cleaing membes and othes who ae inteested in

More information

est using the formula I = Prt, where I is the interest earned, P is the principal, r is the interest rate, and t is the time in years.

est using the formula I = Prt, where I is the interest earned, P is the principal, r is the interest rate, and t is the time in years. 9.2 Inteest Objectives 1. Undestand the simple inteest fomula. 2. Use the compound inteest fomula to find futue value. 3. Solve the compound inteest fomula fo diffeent unknowns, such as the pesent value,

More information

Concept and Experiences on using a Wiki-based System for Software-related Seminar Papers

Concept and Experiences on using a Wiki-based System for Software-related Seminar Papers Concept and Expeiences on using a Wiki-based System fo Softwae-elated Semina Papes Dominik Fanke and Stefan Kowalewski RWTH Aachen Univesity, 52074 Aachen, Gemany, {fanke, kowalewski}@embedded.wth-aachen.de,

More information

YARN PROPERTIES MEASUREMENT: AN OPTICAL APPROACH

YARN PROPERTIES MEASUREMENT: AN OPTICAL APPROACH nd INTERNATIONAL TEXTILE, CLOTHING & ESIGN CONFERENCE Magic Wold of Textiles Octobe 03 d to 06 th 004, UBROVNIK, CROATIA YARN PROPERTIES MEASUREMENT: AN OPTICAL APPROACH Jana VOBOROVA; Ashish GARG; Bohuslav

More information

VISCOSITY OF BIO-DIESEL FUELS

VISCOSITY OF BIO-DIESEL FUELS VISCOSITY OF BIO-DIESEL FUELS One of the key assumptions fo ideal gases is that the motion of a given paticle is independent of any othe paticles in the system. With this assumption in place, one can use

More information

Supplementary Material for EpiDiff

Supplementary Material for EpiDiff Supplementay Mateial fo EpiDiff Supplementay Text S1. Pocessing of aw chomatin modification data In ode to obtain the chomatin modification levels in each of the egions submitted by the use QDCMR module

More information

Reduced Pattern Training Based on Task Decomposition Using Pattern Distributor

Reduced Pattern Training Based on Task Decomposition Using Pattern Distributor > PNN05-P762 < Reduced Patten Taining Based on Task Decomposition Using Patten Distibuto Sheng-Uei Guan, Chunyu Bao, and TseNgee Neo Abstact Task Decomposition with Patten Distibuto (PD) is a new task

More information

30 H. N. CHIU 1. INTRODUCTION. Recherche opérationnelle/operations Research

30 H. N. CHIU 1. INTRODUCTION. Recherche opérationnelle/operations Research RAIRO Rech. Opé. (vol. 33, n 1, 1999, pp. 29-45) A GOOD APPROXIMATION OF THE INVENTORY LEVEL IN A(Q ) PERISHABLE INVENTORY SYSTEM (*) by Huan Neng CHIU ( 1 ) Communicated by Shunji OSAKI Abstact. This

More information

A framework for the selection of enterprise resource planning (ERP) system based on fuzzy decision making methods

A framework for the selection of enterprise resource planning (ERP) system based on fuzzy decision making methods A famewok fo the selection of entepise esouce planning (ERP) system based on fuzzy decision making methods Omid Golshan Tafti M.s student in Industial Management, Univesity of Yazd Omidgolshan87@yahoo.com

More information

Efficient Redundancy Techniques for Latency Reduction in Cloud Systems

Efficient Redundancy Techniques for Latency Reduction in Cloud Systems Efficient Redundancy Techniques fo Latency Reduction in Cloud Systems 1 Gaui Joshi, Emina Soljanin, and Gegoy Wonell Abstact In cloud computing systems, assigning a task to multiple seves and waiting fo

More information

The Binomial Distribution

The Binomial Distribution The Binomial Distibution A. It would be vey tedious if, evey time we had a slightly diffeent poblem, we had to detemine the pobability distibutions fom scatch. Luckily, thee ae enough similaities between

More information

Tracking/Fusion and Deghosting with Doppler Frequency from Two Passive Acoustic Sensors

Tracking/Fusion and Deghosting with Doppler Frequency from Two Passive Acoustic Sensors Tacking/Fusion and Deghosting with Dopple Fequency fom Two Passive Acoustic Sensos Rong Yang, Gee Wah Ng DSO National Laboatoies 2 Science Pak Dive Singapoe 11823 Emails: yong@dso.og.sg, ngeewah@dso.og.sg

More information

Cloud Service Reliability: Modeling and Analysis

Cloud Service Reliability: Modeling and Analysis Cloud Sevice eliability: Modeling and Analysis Yuan-Shun Dai * a c, Bo Yang b, Jack Dongaa a, Gewei Zhang c a Innovative Computing Laboatoy, Depatment of Electical Engineeing & Compute Science, Univesity

More information

UNIT CIRCLE TRIGONOMETRY

UNIT CIRCLE TRIGONOMETRY UNIT CIRCLE TRIGONOMETRY The Unit Cicle is the cicle centeed at the oigin with adius unit (hence, the unit cicle. The equation of this cicle is + =. A diagam of the unit cicle is shown below: + = - - -

More information

Statistics and Data Analysis

Statistics and Data Analysis Pape 274-25 An Extension to SAS/OR fo Decision System Suppot Ali Emouznead Highe Education Funding Council fo England, Nothavon house, Coldhabou Lane, Bistol, BS16 1QD U.K. ABSTRACT This pape exploes the

More information

arxiv:1110.2612v1 [q-fin.st] 12 Oct 2011

arxiv:1110.2612v1 [q-fin.st] 12 Oct 2011 Maket inefficiency identified by both single and multiple cuency tends T.Toká 1, and D. Hováth 1, 1 Sos Reseach a.s., Stojáenská 3, 040 01 Košice, Slovak Republic Abstact axiv:1110.2612v1 [q-fin.st] 12

More information

METHODOLOGICAL APPROACH TO STRATEGIC PERFORMANCE OPTIMIZATION

METHODOLOGICAL APPROACH TO STRATEGIC PERFORMANCE OPTIMIZATION ETHODOOGICA APPOACH TO STATEGIC PEFOANCE OPTIIZATION ao Hell * Stjepan Vidačić ** Željo Gaača *** eceived: 4. 07. 2009 Peliminay communication Accepted: 5. 0. 2009 UDC 65.02.4 This pape pesents a matix

More information

HEALTHCARE INTEGRATION BASED ON CLOUD COMPUTING

HEALTHCARE INTEGRATION BASED ON CLOUD COMPUTING U.P.B. Sci. Bull., Seies C, Vol. 77, Iss. 2, 2015 ISSN 2286-3540 HEALTHCARE INTEGRATION BASED ON CLOUD COMPUTING Roxana MARCU 1, Dan POPESCU 2, Iulian DANILĂ 3 A high numbe of infomation systems ae available

More information

CONCEPTUAL FRAMEWORK FOR DEVELOPING AND VERIFICATION OF ATTRIBUTION MODELS. ARITHMETIC ATTRIBUTION MODELS

CONCEPTUAL FRAMEWORK FOR DEVELOPING AND VERIFICATION OF ATTRIBUTION MODELS. ARITHMETIC ATTRIBUTION MODELS CONCEPUAL FAMEOK FO DEVELOPING AND VEIFICAION OF AIBUION MODELS. AIHMEIC AIBUION MODELS Yui K. Shestopaloff, is Diecto of eseach & Deelopment at SegmentSoft Inc. He is a Docto of Sciences and has a Ph.D.

More information

Peer-to-Peer File Sharing Game using Correlated Equilibrium

Peer-to-Peer File Sharing Game using Correlated Equilibrium Pee-to-Pee File Shaing Game using Coelated Equilibium Beibei Wang, Zhu Han, and K. J. Ray Liu Depatment of Electical and Compute Engineeing and Institute fo Systems Reseach, Univesity of Mayland, College

More information

A Capacitated Commodity Trading Model with Market Power

A Capacitated Commodity Trading Model with Market Power A Capacitated Commodity Tading Model with Maket Powe Victo Matínez-de-Albéniz Josep Maia Vendell Simón IESE Business School, Univesity of Navaa, Av. Peason 1, 08034 Bacelona, Spain VAlbeniz@iese.edu JMVendell@iese.edu

More information

Questions for Review. By buying bonds This period you save s, next period you get s(1+r)

Questions for Review. By buying bonds This period you save s, next period you get s(1+r) MACROECONOMICS 2006 Week 5 Semina Questions Questions fo Review 1. How do consumes save in the two-peiod model? By buying bonds This peiod you save s, next peiod you get s() 2. What is the slope of a consume

More information

Uncertain Version Control in Open Collaborative Editing of Tree-Structured Documents

Uncertain Version Control in Open Collaborative Editing of Tree-Structured Documents Uncetain Vesion Contol in Open Collaboative Editing of Tee-Stuctued Documents M. Lamine Ba Institut Mines Télécom; Télécom PaisTech; LTCI Pais, Fance mouhamadou.ba@ telecom-paistech.f Talel Abdessalem

More information

Comparing Availability of Various Rack Power Redundancy Configurations

Comparing Availability of Various Rack Power Redundancy Configurations Compaing Availability of Vaious Rack Powe Redundancy Configuations By Victo Avela White Pape #48 Executive Summay Tansfe switches and dual-path powe distibution to IT equipment ae used to enhance the availability

More information

Financial Planning and Risk-return profiles

Financial Planning and Risk-return profiles Financial Planning and Risk-etun pofiles Stefan Gaf, Alexande Kling und Jochen Russ Pepint Seies: 2010-16 Fakultät fü Mathematik und Witschaftswissenschaften UNIERSITÄT ULM Financial Planning and Risk-etun

More information

The impact of migration on the provision. of UK public services (SRG.10.039.4) Final Report. December 2011

The impact of migration on the provision. of UK public services (SRG.10.039.4) Final Report. December 2011 The impact of migation on the povision of UK public sevices (SRG.10.039.4) Final Repot Decembe 2011 The obustness The obustness of the analysis of the is analysis the esponsibility is the esponsibility

More information

The Predictive Power of Dividend Yields for Stock Returns: Risk Pricing or Mispricing?

The Predictive Power of Dividend Yields for Stock Returns: Risk Pricing or Mispricing? The Pedictive Powe of Dividend Yields fo Stock Retuns: Risk Picing o Mispicing? Glenn Boyle Depatment of Economics and Finance Univesity of Cantebuy Yanhui Li Depatment of Economics and Finance Univesity

More information

Risk Sensitive Portfolio Management With Cox-Ingersoll-Ross Interest Rates: the HJB Equation

Risk Sensitive Portfolio Management With Cox-Ingersoll-Ross Interest Rates: the HJB Equation Risk Sensitive Potfolio Management With Cox-Ingesoll-Ross Inteest Rates: the HJB Equation Tomasz R. Bielecki Depatment of Mathematics, The Notheasten Illinois Univesity 55 Noth St. Louis Avenue, Chicago,

More information

An Analysis of Manufacturer Benefits under Vendor Managed Systems

An Analysis of Manufacturer Benefits under Vendor Managed Systems An Analysis of Manufactue Benefits unde Vendo Managed Systems Seçil Savaşaneil Depatment of Industial Engineeing, Middle East Technical Univesity, 06531, Ankaa, TURKEY secil@ie.metu.edu.t Nesim Ekip 1

More information

Strength Analysis and Optimization Design about the key parts of the Robot

Strength Analysis and Optimization Design about the key parts of the Robot Intenational Jounal of Reseach in Engineeing and Science (IJRES) ISSN (Online): 2320-9364, ISSN (Pint): 2320-9356 www.ijes.og Volume 3 Issue 3 ǁ Mach 2015 ǁ PP.25-29 Stength Analysis and Optimization Design

More information

STABILITY ANALYSIS IN MILLING BASED ON OPERATIONAL MODAL DATA 1. INTRODUCTION

STABILITY ANALYSIS IN MILLING BASED ON OPERATIONAL MODAL DATA 1. INTRODUCTION Jounal of Machine Engineeing, Vol. 11, No. 4, 211 Batosz POWALKA 1 Macin CHODZKO 1 Kzysztof JEMIELNIAK 2 milling, chatte, opeational modal analysis STABILITY ANALYSIS IN MILLING BASED ON OPERATIONAL MODAL

More information

Comparing Availability of Various Rack Power Redundancy Configurations

Comparing Availability of Various Rack Power Redundancy Configurations Compaing Availability of Vaious Rack Powe Redundancy Configuations White Pape 48 Revision by Victo Avela > Executive summay Tansfe switches and dual-path powe distibution to IT equipment ae used to enhance

More information

Modeling and Verifying a Price Model for Congestion Control in Computer Networks Using PROMELA/SPIN

Modeling and Verifying a Price Model for Congestion Control in Computer Networks Using PROMELA/SPIN Modeling and Veifying a Pice Model fo Congestion Contol in Compute Netwoks Using PROMELA/SPIN Clement Yuen and Wei Tjioe Depatment of Compute Science Univesity of Toonto 1 King s College Road, Toonto,

More information

An Epidemic Model of Mobile Phone Virus

An Epidemic Model of Mobile Phone Virus An Epidemic Model of Mobile Phone Vius Hui Zheng, Dong Li, Zhuo Gao 3 Netwok Reseach Cente, Tsinghua Univesity, P. R. China zh@tsinghua.edu.cn School of Compute Science and Technology, Huazhong Univesity

More information

How To Find The Optimal Stategy For Buying Life Insuance

How To Find The Optimal Stategy For Buying Life Insuance Life Insuance Puchasing to Reach a Bequest Ehan Bayakta Depatment of Mathematics, Univesity of Michigan Ann Abo, Michigan, USA, 48109 S. David Pomislow Depatment of Mathematics, Yok Univesity Toonto, Ontaio,

More information

Research on Risk Assessment of the Transformer Based on Life Cycle Cost

Research on Risk Assessment of the Transformer Based on Life Cycle Cost ntenational Jounal of Smat Gid and lean Enegy eseach on isk Assessment of the Tansfome Based on Life ycle ost Hui Zhou a, Guowei Wu a, Weiwei Pan a, Yunhe Hou b, hong Wang b * a Zhejiang Electic Powe opoation,

More information

Channel selection in e-commerce age: A strategic analysis of co-op advertising models

Channel selection in e-commerce age: A strategic analysis of co-op advertising models Jounal of Industial Engineeing and Management JIEM, 013 6(1):89-103 Online ISSN: 013-0953 Pint ISSN: 013-843 http://dx.doi.og/10.396/jiem.664 Channel selection in e-commece age: A stategic analysis of

More information

Timing Synchronization in High Mobility OFDM Systems

Timing Synchronization in High Mobility OFDM Systems Timing Synchonization in High Mobility OFDM Systems Yasamin Mostofi Depatment of Electical Engineeing Stanfod Univesity Stanfod, CA 94305, USA Email: yasi@wieless.stanfod.edu Donald C. Cox Depatment of

More information

SUPPORT VECTOR MACHINE FOR BANDWIDTH ANALYSIS OF SLOTTED MICROSTRIP ANTENNA

SUPPORT VECTOR MACHINE FOR BANDWIDTH ANALYSIS OF SLOTTED MICROSTRIP ANTENNA Intenational Jounal of Compute Science, Systems Engineeing and Infomation Technology, 4(), 20, pp. 67-7 SUPPORT VECTOR MACHIE FOR BADWIDTH AALYSIS OF SLOTTED MICROSTRIP ATEA Venmathi A.R. & Vanitha L.

More information

Semipartial (Part) and Partial Correlation

Semipartial (Part) and Partial Correlation Semipatial (Pat) and Patial Coelation his discussion boows heavily fom Applied Multiple egession/coelation Analysis fo the Behavioal Sciences, by Jacob and Paticia Cohen (975 edition; thee is also an updated

More information

Strategic Asset Allocation and the Role of Alternative Investments

Strategic Asset Allocation and the Role of Alternative Investments Stategic Asset Allocation and the Role of Altenative Investments DOUGLAS CUMMING *, LARS HELGE HAß, DENIS SCHWEIZER Abstact We intoduce a famewok fo stategic asset allocation with altenative investments.

More information

An application of stochastic programming in solving capacity allocation and migration planning problem under uncertainty

An application of stochastic programming in solving capacity allocation and migration planning problem under uncertainty An application of stochastic pogamming in solving capacity allocation and migation planning poblem unde uncetainty Yin-Yann Chen * and Hsiao-Yao Fan Depatment of Industial Management, National Fomosa Univesity,

More information

Database Management Systems

Database Management Systems Contents Database Management Systems (COP 5725) D. Makus Schneide Depatment of Compute & Infomation Science & Engineeing (CISE) Database Systems Reseach & Development Cente Couse Syllabus 1 Sping 2012

More information

MULTIPLE SOLUTIONS OF THE PRESCRIBED MEAN CURVATURE EQUATION

MULTIPLE SOLUTIONS OF THE PRESCRIBED MEAN CURVATURE EQUATION MULTIPLE SOLUTIONS OF THE PRESCRIBED MEAN CURVATURE EQUATION K.C. CHANG AND TAN ZHANG In memoy of Pofesso S.S. Chen Abstact. We combine heat flow method with Mose theoy, supe- and subsolution method with

More information

Review Graph based Online Store Review Spammer Detection

Review Graph based Online Store Review Spammer Detection Review Gaph based Online Stoe Review Spamme Detection Guan Wang, Sihong Xie, Bing Liu, Philip S. Yu Univesity of Illinois at Chicago Chicago, USA gwang26@uic.edu sxie6@uic.edu liub@uic.edu psyu@uic.edu

More information

9:6.4 Sample Questions/Requests for Managing Underwriter Candidates

9:6.4 Sample Questions/Requests for Managing Underwriter Candidates 9:6.4 INITIAL PUBLIC OFFERINGS 9:6.4 Sample Questions/Requests fo Managing Undewite Candidates Recent IPO Expeience Please povide a list of all completed o withdawn IPOs in which you fim has paticipated

More information

Valuation of Floating Rate Bonds 1

Valuation of Floating Rate Bonds 1 Valuation of Floating Rate onds 1 Joge uz Lopez us 316: Deivative Secuities his note explains how to value plain vanilla floating ate bonds. he pupose of this note is to link the concepts that you leaned

More information

Mining Relatedness Graphs for Data Integration

Mining Relatedness Graphs for Data Integration Mining Relatedness Gaphs fo Data Integation Jeemy T. Engle (jtengle@indiana.edu) Ying Feng (yingfeng@indiana.edu) Robet L. Goldstone (goldsto@indiana.edu) Indiana Univesity Bloomington, IN. 47405 USA Abstact

More information

Data Center Demand Response: Avoiding the Coincident Peak via Workload Shifting and Local Generation

Data Center Demand Response: Avoiding the Coincident Peak via Workload Shifting and Local Generation (213) 1 28 Data Cente Demand Response: Avoiding the Coincident Peak via Wokload Shifting and Local Geneation Zhenhua Liu 1, Adam Wieman 1, Yuan Chen 2, Benjamin Razon 1, Niangjun Chen 1 1 Califonia Institute

More information

Application of the VISEVA demand generation software to Berlin using publicly available behavioral data

Application of the VISEVA demand generation software to Berlin using publicly available behavioral data Justen, Beuck, Nagel 1 Application of the VISEVA demand geneation softwae to Belin using publicly available behavioal data Submission date: 15-Nov-06 Wods: 5973 Figues and tables: 6 ( = 1500 Wods) Total:

More information

Things to Remember. r Complete all of the sections on the Retirement Benefit Options form that apply to your request.

Things to Remember. r Complete all of the sections on the Retirement Benefit Options form that apply to your request. Retiement Benefit 1 Things to Remembe Complete all of the sections on the Retiement Benefit fom that apply to you equest. If this is an initial equest, and not a change in a cuent distibution, emembe to

More information

Instituto Superior Técnico Av. Rovisco Pais, 1 1049-001 Lisboa E-mail: virginia.infante@ist.utl.pt

Instituto Superior Técnico Av. Rovisco Pais, 1 1049-001 Lisboa E-mail: virginia.infante@ist.utl.pt FATIGUE LIFE TIME PREDICTIO OF POAF EPSILO TB-30 AIRCRAFT - PART I: IMPLEMETATIO OF DIFERET CYCLE COUTIG METHODS TO PREDICT THE ACCUMULATED DAMAGE B. A. S. Seano 1, V. I. M.. Infante 2, B. S. D. Maado

More information

Converting knowledge Into Practice

Converting knowledge Into Practice Conveting knowledge Into Pactice Boke Nightmae srs Tend Ride By Vladimi Ribakov Ceato of Pips Caie 20 of June 2010 2 0 1 0 C o p y i g h t s V l a d i m i R i b a k o v 1 Disclaime and Risk Wanings Tading

More information

Effect of Contention Window on the Performance of IEEE 802.11 WLANs

Effect of Contention Window on the Performance of IEEE 802.11 WLANs Effect of Contention Window on the Pefomance of IEEE 82.11 WLANs Yunli Chen and Dhama P. Agawal Cente fo Distibuted and Mobile Computing, Depatment of ECECS Univesity of Cincinnati, OH 45221-3 {ychen,

More information

UNIVERSIDAD DE CANTABRIA TESIS DOCTORAL

UNIVERSIDAD DE CANTABRIA TESIS DOCTORAL UNIVERSIDAD DE CANABRIA Depatamento de Ingenieía de Comunicaciones ESIS DOCORAL Cyogenic echnology in the Micowave Engineeing: Application to MIC and MMIC Vey Low Noise Amplifie Design Juan Luis Cano de

More information

Modal Characteristics study of CEM-1 Single-Layer Printed Circuit Board Using Experimental Modal Analysis

Modal Characteristics study of CEM-1 Single-Layer Printed Circuit Board Using Experimental Modal Analysis Available online at www.sciencediect.com Pocedia Engineeing 41 (2012 ) 1360 1366 Intenational Symposium on Robotics and Intelligent Sensos 2012 (IRIS 2012) Modal Chaacteistics study of CEM-1 Single-Laye

More information

PAN STABILITY TESTING OF DC CIRCUITS USING VARIATIONAL METHODS XVIII - SPETO - 1995. pod patronatem. Summary

PAN STABILITY TESTING OF DC CIRCUITS USING VARIATIONAL METHODS XVIII - SPETO - 1995. pod patronatem. Summary PCE SEMINIUM Z PODSTW ELEKTOTECHNIKI I TEOII OBWODÓW 8 - TH SEMIN ON FUNDMENTLS OF ELECTOTECHNICS ND CICUIT THEOY ZDENĚK BIOLEK SPŠE OŽNO P.., CZECH EPUBLIC DLIBO BIOLEK MILITY CDEMY, BNO, CZECH EPUBLIC

More information

Trading Volume and Serial Correlation in Stock Returns in Pakistan. Abstract

Trading Volume and Serial Correlation in Stock Returns in Pakistan. Abstract Tading Volume and Seial Coelation in Stock Retuns in Pakistan Khalid Mustafa Assistant Pofesso Depatment of Economics, Univesity of Kaachi e-mail: khalidku@yahoo.com and Mohammed Nishat Pofesso and Chaiman,

More information

Spirotechnics! September 7, 2011. Amanda Zeringue, Michael Spannuth and Amanda Zeringue Dierential Geometry Project

Spirotechnics! September 7, 2011. Amanda Zeringue, Michael Spannuth and Amanda Zeringue Dierential Geometry Project Spiotechnics! Septembe 7, 2011 Amanda Zeingue, Michael Spannuth and Amanda Zeingue Dieential Geomety Poject 1 The Beginning The geneal consensus of ou goup began with one thought: Spiogaphs ae awesome.

More information

Scheduling Hadoop Jobs to Meet Deadlines

Scheduling Hadoop Jobs to Meet Deadlines Scheduling Hadoop Jobs to Meet Deadlines Kamal Kc, Kemafo Anyanwu Depatment of Compute Science Noth Caolina State Univesity {kkc,kogan}@ncsu.edu Abstact Use constaints such as deadlines ae impotant equiements

More information

INVESTIGATION OF FLOW INSIDE AN AXIAL-FLOW PUMP OF GV IMP TYPE

INVESTIGATION OF FLOW INSIDE AN AXIAL-FLOW PUMP OF GV IMP TYPE 1 INVESTIGATION OF FLOW INSIDE AN AXIAL-FLOW PUMP OF GV IMP TYPE ANATOLIY A. YEVTUSHENKO 1, ALEXEY N. KOCHEVSKY 1, NATALYA A. FEDOTOVA 1, ALEXANDER Y. SCHELYAEV 2, VLADIMIR N. KONSHIN 2 1 Depatment of

More information

Towards Automatic Update of Access Control Policy

Towards Automatic Update of Access Control Policy Towads Automatic Update of Access Contol Policy Jinwei Hu, Yan Zhang, and Ruixuan Li Intelligent Systems Laboatoy, School of Computing and Mathematics Univesity of Westen Sydney, Sydney 1797, Austalia

More information

Gravitational Mechanics of the Mars-Phobos System: Comparing Methods of Orbital Dynamics Modeling for Exploratory Mission Planning

Gravitational Mechanics of the Mars-Phobos System: Comparing Methods of Orbital Dynamics Modeling for Exploratory Mission Planning Gavitational Mechanics of the Mas-Phobos System: Compaing Methods of Obital Dynamics Modeling fo Exploatoy Mission Planning Alfedo C. Itualde The Pennsylvania State Univesity, Univesity Pak, PA, 6802 This

More information

Pessu Behavior Analysis for Autologous Fluidations

Pessu Behavior Analysis for Autologous Fluidations EXPERIENCE OF USING A CFD CODE FOR ESTIMATING THE NOISE GENERATED BY GUSTS ALONG THE SUN- ROOF OF A CAR Liang S. Lai* 1, Geogi S. Djambazov 1, Choi -H. Lai 1, Koulis A. Peicleous 1, and Fédéic Magoulès

More information

THE CARLO ALBERTO NOTEBOOKS

THE CARLO ALBERTO NOTEBOOKS THE CARLO ALBERTO NOTEBOOKS Mean-vaiance inefficiency of CRRA and CARA utility functions fo potfolio selection in defined contibution pension schemes Woking Pape No. 108 Mach 2009 Revised, Septembe 2009)

More information

Loyalty Rewards and Gift Card Programs: Basic Actuarial Estimation Techniques

Loyalty Rewards and Gift Card Programs: Basic Actuarial Estimation Techniques Loyalty Rewads and Gift Cad Pogams: Basic Actuaial Estimation Techniques Tim A. Gault, ACAS, MAAA, Len Llaguno, FCAS, MAAA and Matin Ménad, FCAS, MAAA Abstact In this pape we establish an actuaial famewok

More information

Mean-Reverting-Ebit-Based Stock Option Evaluation: Theory and Practice

Mean-Reverting-Ebit-Based Stock Option Evaluation: Theory and Practice Jounal of Applied Finance & aning, vol. 3, no. 5, 03, 35-36 ISSN: 79-6580 pint vesion, 79-6599 online Scienpess Ltd, 03 Mean-Reveting-bit-ased Stoc Option valuation: Theoy and Pactice Hassan l Ibami Abstact

More information

The Supply of Loanable Funds: A Comment on the Misconception and Its Implications

The Supply of Loanable Funds: A Comment on the Misconception and Its Implications JOURNL OF ECONOMICS ND FINNCE EDUCTION Volume 7 Numbe 2 Winte 2008 39 The Supply of Loanable Funds: Comment on the Misconception and Its Implications. Wahhab Khandke and mena Khandke* STRCT Recently Fields-Hat

More information

Uncertainty Associated with Microbiological Analysis

Uncertainty Associated with Microbiological Analysis Appendix J STWG Pat 3 Uncetainty 7-8-06 Page 1 of 31 Uncetainty Associated with Micobiological Analysis 1. Intoduction 1.1. Thee ae only two absolute cetainties in life: death and taxes! Whateve task we

More information

Episode 401: Newton s law of universal gravitation

Episode 401: Newton s law of universal gravitation Episode 401: Newton s law of univesal gavitation This episode intoduces Newton s law of univesal gavitation fo point masses, and fo spheical masses, and gets students pactising calculations of the foce

More information

NBER WORKING PAPER SERIES FISCAL ZONING AND SALES TAXES: DO HIGHER SALES TAXES LEAD TO MORE RETAILING AND LESS MANUFACTURING?

NBER WORKING PAPER SERIES FISCAL ZONING AND SALES TAXES: DO HIGHER SALES TAXES LEAD TO MORE RETAILING AND LESS MANUFACTURING? NBER WORKING PAPER SERIES FISCAL ZONING AND SALES TAXES: DO HIGHER SALES TAXES LEAD TO MORE RETAILING AND LESS MANUFACTURING? Daia Bunes David Neumak Michelle J. White Woking Pape 16932 http://www.nbe.og/papes/w16932

More information

Over-encryption: Management of Access Control Evolution on Outsourced Data

Over-encryption: Management of Access Control Evolution on Outsourced Data Ove-encyption: Management of Access Contol Evolution on Outsouced Data Sabina De Capitani di Vimecati DTI - Univesità di Milano 26013 Cema - Italy decapita@dti.unimi.it Stefano Paaboschi DIIMM - Univesità

More information

Patent renewals and R&D incentives

Patent renewals and R&D incentives RAND Jounal of Economics Vol. 30, No., Summe 999 pp. 97 3 Patent enewals and R&D incentives Fancesca Conelli* and Mak Schankeman** In a model with moal hazad and asymmetic infomation, we show that it can

More information

Carter-Penrose diagrams and black holes

Carter-Penrose diagrams and black holes Cate-Penose diagams and black holes Ewa Felinska The basic intoduction to the method of building Penose diagams has been pesented, stating with obtaining a Penose diagam fom Minkowski space. An example

More information

The Role of Gravity in Orbital Motion

The Role of Gravity in Orbital Motion ! The Role of Gavity in Obital Motion Pat of: Inquiy Science with Datmouth Developed by: Chistophe Caoll, Depatment of Physics & Astonomy, Datmouth College Adapted fom: How Gavity Affects Obits (Ohio State

More information

Research and the Approval Process

Research and the Approval Process Reseach and the Appoval Pocess Emeic Heny y Maco Ottaviani z Febuay 2014 Abstact An agent sequentially collects infomation to obtain a pincipal s appoval, such as a phamaceutical company seeking FDA appoval

More information

Magnetic Bearing with Radial Magnetized Permanent Magnets

Magnetic Bearing with Radial Magnetized Permanent Magnets Wold Applied Sciences Jounal 23 (4): 495-499, 2013 ISSN 1818-4952 IDOSI Publications, 2013 DOI: 10.5829/idosi.wasj.2013.23.04.23080 Magnetic eaing with Radial Magnetized Pemanent Magnets Vyacheslav Evgenevich

More information

Intertemporal Macroeconomics

Intertemporal Macroeconomics Intetempoal Macoeconomics Genot Doppelhofe* May 2009 Fothcoming in J. McCombie and N. Allington (eds.), Cambidge Essays in Applied Economics, Cambidge UP This chapte eviews models of intetempoal choice

More information

Nontrivial lower bounds for the least common multiple of some finite sequences of integers

Nontrivial lower bounds for the least common multiple of some finite sequences of integers J. Numbe Theoy, 15 (007), p. 393-411. Nontivial lowe bounds fo the least common multiple of some finite sequences of integes Bai FARHI bai.fahi@gmail.com Abstact We pesent hee a method which allows to

More information

Promised Lead-Time Contracts Under Asymmetric Information

Promised Lead-Time Contracts Under Asymmetric Information OPERATIONS RESEARCH Vol. 56, No. 4, July August 28, pp. 898 915 issn 3-364X eissn 1526-5463 8 564 898 infoms doi 1.1287/ope.18.514 28 INFORMS Pomised Lead-Time Contacts Unde Asymmetic Infomation Holly

More information

MERGER SIMULATION AS A SCREENING DEVICE: SIMULATING THE EFFECTS OF THE KRAFT/CADBURY TRANSACTION

MERGER SIMULATION AS A SCREENING DEVICE: SIMULATING THE EFFECTS OF THE KRAFT/CADBURY TRANSACTION MERGER SIMULATION AS A SCREENING DEVICE: SIMULATING THE EFFECTS OF THE KRAFT/CADBURY TRANSACTION Enique Andeu, Kisten Edwads, Aleando Requeo 1,2 Novembe 2010 Abstact In this aticle we pesent a method that

More information

Contingent capital with repeated interconversion between debt and equity

Contingent capital with repeated interconversion between debt and equity Contingent capital with epeated inteconvesion between debt and equity Zhaojun Yang 1, Zhiming Zhao School of Finance and Statistics, Hunan Univesity, Changsha 410079, China Abstact We develop a new type

More information

How Much Should a Firm Borrow. Effect of tax shields. Capital Structure Theory. Capital Structure & Corporate Taxes

How Much Should a Firm Borrow. Effect of tax shields. Capital Structure Theory. Capital Structure & Corporate Taxes How Much Should a Fim Boow Chapte 19 Capital Stuctue & Copoate Taxes Financial Risk - Risk to shaeholdes esulting fom the use of debt. Financial Leveage - Incease in the vaiability of shaeholde etuns that

More information

Do Bonds Span the Fixed Income Markets? Theory and Evidence for Unspanned Stochastic Volatility

Do Bonds Span the Fixed Income Markets? Theory and Evidence for Unspanned Stochastic Volatility Do Bonds Span the Fied Income Makets? Theoy and Evidence fo Unspanned Stochastic olatility PIERRE COLLIN-DUFRESNE and ROBERT S. GOLDSTEIN July, 00 ABSTRACT Most tem stuctue models assume bond makets ae

More information

COMPLYING WITH THE DRUG-FREE SCHOOLS AND CAMPUSES REGULATIONS

COMPLYING WITH THE DRUG-FREE SCHOOLS AND CAMPUSES REGULATIONS Highe Education Cente fo Alcohol and Othe Dug Abuse and Violence Pevention Education Development Cente, Inc. 55 Chapel Steet Newton, MA 02458-1060 COMPLYING WITH THE DRUG-FREE SCHOOLS AND CAMPUSES REGULATIONS

More information

Left- and Right-Brain Preferences Profile

Left- and Right-Brain Preferences Profile Left- and Right-Bain Pefeences Pofile God gave man a total bain, and He expects us to pesent both sides of ou bains back to Him so that He can use them unde the diection of His Holy Spiit as He so desies

More information

Approximation Algorithms for Data Management in Networks

Approximation Algorithms for Data Management in Networks Appoximation Algoithms fo Data Management in Netwoks Chistof Kick Heinz Nixdof Institute and Depatment of Mathematics & Compute Science adebon Univesity Gemany kueke@upb.de Haald Räcke Heinz Nixdof Institute

More information