genmax: New version that sends down arrays instead of pointers!!

In Auburngrads/SMRD: Statistical Methods For Reliability Data

Description

# kmod (I) model code (usual list 1-5) # ————————————————————- # 0 distribution kdist with one or more parameters related to # explanatory variables # 1 same as 0 with an lfp parameter # 2 same as 0 with a doa parameter # 3 steady-state model with distribution kdist plus exponential # competing risk # 4 proportional hazards model for distribution kdist plus # explanatory variables for the power (or other) parameters # # kdist (I) distribution number (usual list 1-12) # >100 is a user specified distribution # # theta(nparm) (O) double precision vector returning ml estimates # of model parameters and values of any # fixed parameters # # thetas(nparm) (O) double precision vector returning ml # scaled estimates # of model parameters and (scaled) values of # any fixed parameters # data scaled according to centered and scaled x # # kodet(nparm) (O) integer vector of parameter codes indicating range # 0 fixed # 1 unrestricted # 2 positive # 3 0-1 # # # ifix(nparm) (I) an integer vector of 0's and 1's to indicate which # parameters are to be fixed (1 for fixed) # # nparm (O) number of model parameters # nparm is the sum of the parameters in all of the # regression relationships plus any model parameters # that do not depend on explanatory variables # If this number is needed above before calling # genmax (to allocate parameter space) it can be # gotten from a call to gensiz # the model stuff (kmod,kdist,kparv,nrvar,mrelat,nrelat) # has been set. # # npard (O) number of model parameters # npard is the number of parameters in the distributional # model (i.e., for a particular combination of the # explanatory variables # If this number is needed above before calling # genmax (to allocate parameter space) it can be # gotten from a call to gensiz # the model stuff (kmod,kdist,kparv,nrvar,mrelat,nrelat) # has been set. # # y (I) real matrix of response times . # transformed times may be returned # # ncoly (I) number of columns in the observation matrix (1 or 2) # # nrownw (I) number of rows in the data matrix # # x (I) real matrix x of explanatory variables # which must have a # column of ones in the first column (even if it # is not used). # # ***** note that upon return, x may not be in its original # form due to scaling or other standardization. # # ncolx (I) number of columns in x, not including the required col # of ones (because we think of the ones as column zero) # # codes (I) real vector of censor codes # # weight (I) real vector of observation weights # # ty (I) real matrix of truncation times # # ncolty (I) number of columns in the matrix of truncation times # (0, 1, or 2) # # tc (I) real vector of truncation codes # # kprint (I) print code 0-none 1-minimal 2-usual 3-light debug # 4 and higher provides more debug output # # # kparv (I) integer vector giving parameter # numbers having regression # relationships # # nrvar (I) integer vector. nrvar(i) gives the number of columns # of x that are in the relationship for the # parameter specified in kparv(i) # # mrelat(mnrvar,nrelat) (I) integer matrix # in which entry i,j gives the col of x # for the ith term of the jth relationship. i=1,nrvar(j), # j=1,nrelat # # nrelat (I) number of regression relationships # # mnrvar (I) integer vector. nrvar(i) gives the number of columns # # xlogl (O) double precision scaler returning value of the # log likelihood # # yhat (O) real vector (matrix) of fitted values (same size as y) # # resid (O) real vector (matrix) of residuals (same size as y) # # vcv(nparm,nparm) (O) double precision matrix returning estimated matrix of # variances and covariances of the ml estimates # # vcvs(nparm,nparm)(O) double precision matrix returning estimated # matrix of variances and covariances of the # ml scaled estimates # # r(nparm,nparm) (O) double precision matrix returning estimated # correlation matrix from the vcv matrix # # start (I–opt) optional vector of starting parameter values # (ignored if lstar=0) # # lstar (I) =1 is start values are in start # =0 if automatic start values are # to be computed velow # # ilabp (O) integer vector of parameter labels (length 8*nparm) # in this new version we copy over to stack and pass # pointers below # # ilabd (O) integer vector of distribution parmameters # (length 8*nparm) # in this new version we copy over to stack and pass # pointers below # # ier code meaning # ——————————————————————- # third digit: 0 no optimization errors detected # 1 likelihood shape caused problems with the powell alg # 2 convergence criterion not met after maximum number # or iterations # # second digit: 0 first derivatives of likelihood small # 1 first derivatives of the loglikelihood too large # # first digit: 0 estimated fisher info matrix inverted successfully # 1 estimated fisher info matrix appears to be singular # #note: all pointers sent down in the argument list are pointing to the # rs(.) stack in labeled common. # # fortran typing conventions are generally followed for typing # except that double precision is used except when data is to # be taken from the worksheet.

Usage

GENMAX(
  kmod,
  kdist,
  theta,
  thetas,
  kodet,
  ifix,
  nparm,
  npard,
  y,
  ncoly,
  nrownw,
  x,
  ncolx,
  codes,
  weight,
  ty,
  ncolty,
  tcodes,
  kprint,
  kparv,
  nrvar,
  mrelat,
  nrelat,
  mnrvar,
  xlogl,
  yhat,
  resid,
  vcvs,
  vcv,
  r,
  start,
  lstar,
  conlev,
  ilabp,
  ilabd,
  ier,
  nxd,
  intd,
  ipxcd,
  irelad,
  fstder,
  nregr,
  kcentr,
  kpoint,
  ifit,
  kgtall,
  llog,
  kmodp,
  maxit,
  pest,
  epsx,
  npardm,
  nnum,
  kparm,
  iup,
  nterd,
  maxpd,
  pfail,
  kmccde,
  nstart,
  maxmsd,
  tol,
  lsd,
  pchmax
)

Auburngrads/SMRD documentation built on Sept. 14, 2020, 2:21 a.m.