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

Tue Aug 4 23:03:14 PDT 2009

  Thanks to all of you for your most excellent help ...

  Sadly, I need some more.

  I implemented Dave's GLOBALS_SECTION below and added the missing semicolon.
  Everything now compiles (with lots of warnings of the form 
/usr/local/src/admb/include/admodel.h:2256: warning: extra qualification ‘param_init_bounded_matrix_vector::’ on member ‘allocate’ 
but compiles nonetheless).

   Now when I run it the result is ... (drum roll) ...

error reading parameters from file ./mccoypred3.pin

  I have traced the error through the C++ code to the point where
it tries to allocate the random effects vector -- here's a code snippet

  h.allocate(0.0,1.0,1,"h");
  g.allocate("g");
  sigma_c.allocate(0.00001,1.0,1,"sigma_c");
  cout << "before u.allocate\n";   // BMB added
  u.allocate(1,nblock,"u");
  cout << "after u.allocate\n";    // BMB added

  I have also confirmed that this is where it runs into trouble using gdb,
although I can't seem to step into the "u.allocate" line to see what's going
on within ...  nblock seems to be read OK (it is used earlier to define the
bounds of the random effects design matrix Z)

  mccoypred3.pin looks like this:

-----------------
# "mccoypred3.pin" produced by pin_write() from ADMButils; Wed Aug  5 01:13:15 2009
# c 
 0.506685 

# d 
 4.08488 

# h 
 0.008644705 

# g 
 3.019749 

# sigma_c 
 0.1 
--------------------

mccoypred3.dat: 

# "mccoypred3.dat" produced by dat_write() from ADMButils; Wed Aug  5 01:55:40 2009
# nobs 
 10 

# nblock 
 6 

# killed 
 3 0 1 3 1 0 1 2 2 2 

# size 
 8.830333 9.155 7.991667 8.000833 8.140333 8.711167 16.5265 14.1508 18.8445 14.87525 

# initial 
 6 6 6 6 6 6 6 6 6 6 

# Z 
1 0 0 0 0 0
0 1 0 0 0 0
0 0 1 0 0 0
0 0 0 1 0 0
0 0 0 0 1 0
0 0 0 0 0 1
1 0 0 0 0 0
0 1 0 0 0 0
0 0 1 0 0 0
0 0 0 1 0 0

  The parameter section of the tpl file is:

PARAMETER_SECTION

  init_number c
  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

  I have tried the random_effects_vector line both as it appears above and as

  random_effects_vector u(1,nblock,1)

to indicate activation in phase 1  and 

  random_effects_vector u(1,nblock,2)

to indicate activation in phase 2 ...

   More suggestions?    Probably something else boneheaded here.
  As always, suggestions for diagnosis/further information/pointers to relevant locations
in TFM will be appreciated.

  I have attached a mini-version of the dat file plus tpl, pin file -- but don't know
if the attachments will make it through to the list or not.

  (I feel like I'm burning all the karma accumulated on the R-help list ...)

  cheers
    Ben Bolker

________________________________________
From: users-bounces at admb-project.org [users-bounces at admb-project.org] On Behalf Of dave fournier [otter at otter-rsch.com]
Sent: Tuesday, August 04, 2009 4:35 PM
To: users at admb-project.org
Subject: Re: [ADMB Users] "xxglobal.tmp" error with tpl2rem

>This line is causing problems in your other model (mccoypred3.tpl);

>prob = 1/(1/(elem_prod(cvec,elem_prod(pow(size/d,g),exp(-size/d))))
>+h*initial);

>specifically, it does not like "pow(size/d,g)" because there does not
>appear to be a function that takes the arguments:
>'bolker2.cpp:123: error: no matching function for call to
>‘pow(data_vector&, df1b2_init_number&)’

>You might want to re-write this function in log-space to get it to
>work properly.

It is difficult to diagnose these missing function problems if you are
not familiar with the package. A lot of the admb functions were not
extended (yet?) to the RE package. anyway what is missing is the function

  df1b2vector pow(const df1b2vector& v,const df1b2variable & x)

Probably the most elegant work around is to define your own
and put it in the GLOBALS_SECTION to make it available.

stick this at the bottom of the tpl file.

GLOBALS_SECTION

  #include <admodel.h>
  #include <df1b2fun.h>
  #include <adrndeff.h>

  df1b2vector pow(const df1b2vector& v,const df1b2variable & x)
  {
    int mmin=v.indexmin();
    int mmax=v.indexmax();
    df1b2vector tmp(mmin,mmax);
    for (int i=mmin;i<=mmax;i++)
    {
      tmp(i)=pow(v(i),x);
    }
    return tmp;
  }

Also in the version of the program I saw you are missing a ";"
in the line above.

--
David A. Fournier
P.O. Box 2040,
Sidney, B.C. V8l 3S3
Canada
Phone/FAX 250-655-3364
http://otter-rsch.com
_______________________________________________
Users mailing list
Users at admb-project.org
http://lists.admb-project.org/mailman/listinfo/users
-------------- next part --------------
A non-text attachment was scrubbed...
Name: mccoypred3.pin
Type: application/octet-stream
Size: 172 bytes
Desc: mccoypred3.pin
URL: <http://lists.admb-project.org/pipermail/users/attachments/20090805/62592208/attachment.obj>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: mccoypred3.dat
Type: application/data-file
Size: 401 bytes
Desc: mccoypred3.dat
URL: <http://lists.admb-project.org/pipermail/users/attachments/20090805/62592208/attachment.bin>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: mccoypred3.tpl
Type: application/octet-stream
Size: 1508 bytes
Desc: mccoypred3.tpl
URL: <http://lists.admb-project.org/pipermail/users/attachments/20090805/62592208/attachment-0001.obj>