[ADMB Users] "xxglobal.tmp" error with tpl2rem

Bolker,Benjamin Michael bolker at ufl.edu
Mon Aug 3 22:09:08 PDT 2009


   I'm just beginning to use AD Model Builder.
  I have successfully (after some struggles) created a few different
likelihood models, with results matching previous model fits in R.
Now I'm trying to add a random effect.  Versions of the TPL file
below compiled OK with tpl2cpp before I added the random effect;
now, with tpl2rem, I'm getting

cat: xxalloc4.tmp: No such file or directory
cat: xxalloc5.tmp: No such file or directory
Error executing command cat xxglobal.tmp   xxhtop.tmp   header.tmp   xxalloc1.tmp   xxalloc2.tmp   xxalloc3.tmp   xxalloc4.tmp    xxalloc5.tmp    xxtopm.tmp    xxalloc6.tmp > mccoypred4.cpp

   There was an issue with similar keywords listed on the ADMB site, but it
was listed as "resolved".

  I'm running on Ubuntu Linux 9.04, with a recent (?? can't find version number ??)
version of ADMB. (Maybe 9.0.202 from the google code page?)

  I'd be happy to provide any other diagnostics/information if that will
help solve / diagnose ...

  sincerely
    Ben Bolker



-------------------------------------
DATA_SECTION

  init_int nobs               // # of observations
  init_int nblock             // # of blocks
  init_vector killed(1,nobs)  // # killed per trial
  init_vector size(1,nobs)    // size of individuals
  init_vector initial(1,nobs) // starting density (# individuals)
  init_matrix Z(1,nobs,1,nblock) // random-effects model matrix

PARAMETER_SECTION

  init_number c    // *mean* c value
  init_number d
  init_bounded_number h(0.0,1.0,1)
  init_number g
  init_bounded_number sigma_c(0.00001,1.0,1)
  random_effects_vector u(1,nblock)
  vector prob(1,nobs)
  vector cvec(1,nobs)
  objective_function_value f

PROCEDURE_SECTION
  // compute vector of c values
  cvec = c + sigma_c*(Z*u)
  // power-Ricker
  prob = 1/(1/(elem_prod(cvec,elem_prod(pow(size/d,g),exp(-size/d))))+h*initial);
  // binomial negative log-likelihood
  f -= sum( log_comb(initial,killed)+
            elem_prod(killed,log(prob))+
            elem_prod(initial-killed,log(1-prob)));



More information about the Users mailing list