[ADMB Users] adpool unit size error

[Robert H Brown]

Dear ADMB Users,

I am trying to fit a fairly simple 2-piece linear growth curve to some 
longitudinal data with different curves in different subsets of the data 
as defined by the design matrix.  Hope to extend to random effects mixed 
model.  The tpl file compiles without error but when I try to run the 
executable I get the message   "error in adpool object  you must set the 
unit size".  This message is written by the adpool.cpp.  There is no 
object name, just a blank.   Not sure what this means or what to do.  
Any suggestions would be appreciated. tpl file is attached.

Thank you.

Bob Brown
-------------- next part --------------

DATA_SECTION

  init_int n			// Number of patients
  init_matrix obs(1,n,1,18)	// Subject number, trtpn, fmgrp and Average weekly pain scores
  init_matrix design(1,n,1,9)  // Design matrix
  vector op(1,n)
  vector ds(1,n)
  
PARAMETER_SECTION

  init_number acta1
  init_number actb1
  init_number acta2
  init_number actb2 
  init_number acta3
  init_number actb3
  init_number acta4
  init_number actb4 
  init_number actc
  init_number plaa1
  init_number plab1
  init_number plaa2
  init_number plab2 
  init_number plaa3
  init_number plab3
  init_number plaa4
  init_number plab4 
  init_number plac
  init_number s			
  init_number d1
  init_number d2	

  objective_function_value g			// Joint log-likelihood

TOP_OF_MAIN_SECTION

  // arrmblsize = 40000000;
  // gradient_structure::set_MAX_NVAR_OFFSET(450000);

GLOBALS_SECTION

 #include <fvar.hpp>
 #include <df1b2fun.h>
 
PRELIMINARY_CALCS_SECTION
  op = column(obs,17);
  ds = column(obs,18);
  g = 0;

		  
PROCEDURE_SECTION

  dvariable s2, va2, vb2, mu, x, z, q, beta, a, b, c ;                     // Local variables 

  s2 = s*s;
  beta = 0.90;
  
  for(int i=1;i<=n;i++)                     
  { 
    value(a) = value(design(i,2)*acta1 + design(i,3)*acta2 + design(i,4)*acta3 + design(i,5)*acta4 +
        design(i,6)*plaa1 + design(i,7)*plaa2 + design(i,8)*plaa3 + design(i,9)*plaa4);
    value(b) = value(design(i,2)*actb1 + design(i,3)*actb2 + design(i,4)*actb3 + design(i,5)*actb4 +
        design(i,6)*plab1 + design(i,7)*plab2 + design(i,8)*plab3 + design(i,9)*plab4);
    value(c) = value((design(i,2) + design(i,3) + design(i,4) + design(i,5))*actc
      + (design(i,6) + design(i,7) + design(i,8) + design(i,9))*plac);
    for(int j=1;j<=13;j++)                     
    { x = j-1;
	  z = x - c;
      mu = d1*op(i) + d2*ds(i) + a + b*x;
      if (value(z) <= value(-beta)) {
        q = 0; 
		}
      if ((value(z) > value(-beta)) & (value(z) < value(beta))) {
        q = (z+beta)*(z+beta)/(4*beta); 
		}
      if (value(z) >= value(beta)) {
        q = z; 
		}
	  mu = mu - b*q;	
	  if (value(obs(i,3+j)) > -99.0) {
	     g -= (0.5*log(s2) + 0.5*(obs(i,3+j) - mu)*(obs(i,3+j) - mu)/s2);  
		  } 
    }
  }
  g *= -1;