C++ compiler help

awood11@mail.co

Hello all,

I am sure this error has come across the board before, but could not find it through searching. I am using the borland 5.5 compiler along with a program called admodel builder to run non-linear regression models. When converting my base template file which looks like this:

DATA_SECTION
init_int nobs;
init_matrix data(1,nobs,1,2)
vector SSB(1,nobs);
vector R(1,nobs);
PARAMETER_SECTION
init_number a;
init_number b;
vector pred_R(1,nobs)
objective_function_value f;
PRELIMINARY_CALCS_SECTION
// get the data out of the columns of the data matrix 
SSB=column(data,1);
R=column(data,2);
PROCEDURE_SECTION
pred_R=((a*SSB)/(1+b*SSB));
f=regression(R,pred_R);

into a cpp file I get an error saying:

"operator/ not implemented in type dvar_vector for arguments of the same type in function model_parameters::user function()"
it is referring to a problem with my division (/) operator in my function? The resulting cpp file looks like this:

#include <admodel.h>

#include <bh.htp>

model_data::model_data(void)
{
nobs.allocate("nobs");
data.allocate(1,nobs,1,2,"data");
SSB.allocate(1,nobs);
R.allocate(1,nobs);
}

model_parameters::model_parameters(int sz,int argc,char * argv[]) : 
ad_comm(argc,argv), model_data() , function_minimizer(sz)
{
initializationfunction();
a.allocate("a");
b.allocate("b");
pred_R.allocate(1,nobs,"pred_R");
#ifndef NO_AD_INITIALIZE
pred_R.initialize();
#endif
f.allocate("f");
}

void model_parameters::preliminary_calculations(void)
{
// get the data out of the columns of the data matrix 
SSB=column(data,1);
R=column(data,2);
}

void model_parameters::userfunction(void)
{
pred_R=((a*SSB)/(1+b*SSB));
f=regression(R,pred_R);
}

model_data::~model_data()
{}

model_parameters::~model_parameters()
{}

void model_parameters::report(void){}

void model_parameters::final_calcs(void){}

void model_parameters::set_runtime(void){}

#ifdef _BORLANDC_
extern unsigned _stklen=10000U;
#endif


#ifdef __ZTC__
extern unsigned int _stack=10000U;
#endif

long int arrmblsize=0;

int main(int argc,char * argv[])
{
gradient_structure::set_NO_DERIVATIVES();
gradient_structure::set_YES_SAVE_VARIABLES_VALUES();
#if defined(__GNUDOS__) || defined(DOS386) || defined(__DPMI32__) || \
defined(__MSVC32__)
if (!arrmblsize) arrmblsize=150000;
#else
if (!arrmblsize) arrmblsize=25000;
#endif
model_parameters mp(arrmblsize,argc,argv);
mp.iprint=10;
mp.preliminary_calculations();
mp.computations(argc,argv);
return 0;
}

Thanks for any help offered.

Mako

Dizz

Why don't you split the troublesome line up to confirm your suspicisons?

from:
pred_R=((a*SSB)/(1+b*SSB));

to:
x = a*SSB;
y = 1+b*SSB;
pred_R = x/y

where x,y are the correct types . If x and y are integer types then to have the problems is surprising - if not the operator needs overriding and implementing.

Dizz

awood11@mail.co

I have determined that my / operator needs to be overloaded. Can someone tell me where and what I would insert into my code to do this? Thanks,

Dizz

Have a look here (courtesy of google ) http://gethelp.devx.com/techtips/cpp_pro/10min/2001/august/10min0801.asp
As for where to do it - you'll prob have to modify the generated code by your backend.

Dizz