summaryrefslogtreecommitdiff
path: root/build
diff options
context:
space:
mode:
Diffstat (limited to 'build')
-rw-r--r--build/Scilab/intfmincon.sci51
-rw-r--r--build/cpp/cpp_intfmincon.cpp4
-rw-r--r--build/cpp/sci_iofunc.cpp3
-rw-r--r--build/cpp/sci_minconTMINLP.cpp139
4 files changed, 101 insertions, 96 deletions
diff --git a/build/Scilab/intfmincon.sci b/build/Scilab/intfmincon.sci
index 9cf9116..0d6cf6d 100644
--- a/build/Scilab/intfmincon.sci
+++ b/build/Scilab/intfmincon.sci
@@ -207,14 +207,14 @@ function [xopt,fopt,exitflag,gradient,hessian] = intfmincon (varargin)
//To check for correct size of A(3rd paramter)
if(size(A,2)~=nbVar & size(A,2)~=0) then
- errmsg = msprintf(gettext("%s: Expected Matrix of size (No of linear inequality constraints X No of Variables) or an Empty Matrix for Linear Inequality Constraint coefficient Matrix A"), solver_name);
+ errmsg = msprintf(gettext("%s: Expected Matrix of size (No of linear inequality constraints X No of Variables) or an Empty Matrix for Linear Inequality Constraint coefficient Matrix A"), intfmincon);
error(errmsg);
end
nbConInEq=size(A,"r");
//To check for the correct size of Aeq (5th paramter)
if(size(Aeq,2)~=nbVar & size(Aeq,2)~=0) then
- errmsg = msprintf(gettext("%s: Expected Matrix of size (No of linear equality constraints X No of Variables) or an Empty Matrix for Linear Equality Constraint coefficient Matrix Aeq"), solver_name);
+ errmsg = msprintf(gettext("%s: Expected Matrix of size (No of linear equality constraints X No of Variables) or an Empty Matrix for Linear Equality Constraint coefficient Matrix Aeq"), intfmincon);
error(errmsg);
end
nbConEq=size(Aeq,"r");
@@ -340,18 +340,18 @@ options = list('integertolerance',1d-06,'maxnodes',2147483647,'cputime',1d10,'al
if (type(nlc) == 13 | type(nlc) == 11) then
if(execstr('[sample_c,sample_ceq] = nlc(x0)','errcatch')==21) then
- errmsg = msprintf(gettext("%s: Non-Linear Constraint function and x0 did not match"), solver_name);
+ errmsg = msprintf(gettext("%s: Non-Linear Constraint function and x0 did not match"), intfmincon);
error(errmsg);
end
numNlic = size(sample_c,"*");
numNlec = size(sample_ceq,"*");
- numNlc = no_nlic + no_nlec;
+ numNlc = numNlic + numNlec;
end
/////////////// Creating conLb and conUb ////////////////////////
- conLb = [repmat(-%inf,nbConInEq,1);beq;repmat(-%inf,numNlic,1);repmat(0,numNlic,1);]
- conUb = [b;beq;repmat(0,numNlic,1);repmat(0,numNlic,1);]
+ conLb = [repmat(-%inf,numNlic,1);repmat(0,numNlec,1);repmat(-%inf,nbConInEq,1);beq;]
+ conUb = [repmat(0,numNlic,1);repmat(0,numNlec,1);b;beq;]
//Converting the User defined Objective function into Required form (Error Detectable)
function [y,check] = _f(x)
@@ -386,17 +386,13 @@ options = list('integertolerance',1d-06,'maxnodes',2147483647,'cputime',1d10,'al
endfunction
function [y,check] = _addnlc(x)
- x= x(:);
- c=[]
- ceq = [];
+ x= x(:)
try
- if(type(nlc) == 13 & numNlc~=0) then
- {
- [c,ceq] = nlc(x);
- }
+ if((type(nlc) == 13 | type(nlc) == 11) & numNlc~=0) then
+ [c,ceq]=nlc(x)
end
ylin = [A*x;Aeq*x];
- y = [ylin;c(:);ceq(:)];
+ y = [c(:);ceq(:);ylin(:);];
[y,check] = checkIsreal(y)
catch
y=0;
@@ -404,11 +400,11 @@ options = list('integertolerance',1d-06,'maxnodes',2147483647,'cputime',1d10,'al
end
endfunction
- //Defining an inbuilt Objective gradient function
+ //Defining an inbuilt jacobian of constraints function
function [dy,check] = _gradnlc(x)
if (options(16) =="on") then
try
- [y,dy]=_addnlc(x)
+ [y,dy]=nlc(x)
[dy,check] = checkIsreal(dy)
catch
dy = 0;
@@ -427,6 +423,7 @@ options = list('integertolerance',1d-06,'maxnodes',2147483647,'cputime',1d10,'al
//Defining a function to calculate Hessian if the respective user entry is OFF
function [hessy,check]=_gradhess(x,obj_factor,lambda)
+ x=x(:);
if (type(options(14)) == "function") then
try
[obj,dy,hessy] = fun(x,obj_factor,lambda)
@@ -437,17 +434,15 @@ options = list('integertolerance',1d-06,'maxnodes',2147483647,'cputime',1d10,'al
end
else
try
- [dy,hessfy]=numderivative(_f,x,%eps^(1/3),1,"blockmat");
- hessny = []
- if(type(nlc) == 13 & numNlc~=0) then
- {
- [dy,hessny] = numderivative(_addnlc,x,%eps^(1/3),1,"blockmat");
- }
+ [dy,hessfy]=numderivative(_f,x)
+ hessfy = matrix(hessfy,nbVar,nbVar)
+ if((type(nlc) == 13 | type(nlc) == 11) & numNlc~=0) then
+ [dy,hessny]=numderivative(nlc,x)
end
hessianc = []
- for i = 1:numNlc
- hessianc = hessianc + lambda(i)*hessny((i-1)*nbVar+1:nbVar*i,:)
- end
+ for i = 1:numNlc
+ hessianc = hessianc + lambda(i)*matrix(hessny(i,:),nbVar,nbVar)
+ end
hessy = obj_factor*hessfy + hessianc;
[hessy,check] = checkIsreal(hessy)
catch
@@ -457,7 +452,7 @@ options = list('integertolerance',1d-06,'maxnodes',2147483647,'cputime',1d10,'al
end
endfunction
- intconsize = size(intcon,"*");
+ intconsize = size(intcon,"*")
[xopt,fopt,exitflag] = inter_fmincon(_f,_gradf,_addnlc,_gradnlc,_gradhess,x0,lb,ub,conLb,conUb,intcon,options,nbConInEq+nbConEq);
@@ -466,7 +461,7 @@ options = list('integertolerance',1d-06,'maxnodes',2147483647,'cputime',1d10,'al
gradient = [];
hessian = [];
else
- [ gradient, hessian] = numderivative(_f, xopt, [], [], "blockmat");
+ [ gradient, hessian] = numderivative(_f, xopt)
end
//To print output message
@@ -495,7 +490,7 @@ function [y, check] = checkIsreal(x)
y = 0
check=1;
else
- y = x;
+ y = x;
check=0;
end
endfunction
diff --git a/build/cpp/cpp_intfmincon.cpp b/build/cpp/cpp_intfmincon.cpp
index 50270cf..d921128 100644
--- a/build/cpp/cpp_intfmincon.cpp
+++ b/build/cpp/cpp_intfmincon.cpp
@@ -121,6 +121,7 @@ int cpp_intfmincon(char *fname)
BonminSetup bonmin;
bonmin.initializeOptionsAndJournalist();
bonmin.options()->SetStringValue("mu_oracle","loqo");
+ bonmin.options()->SetIntegerValue("bonmin.print_level",5);
bonmin.options()->SetNumericValue("bonmin.integer_tolerance", *integertolerance);
bonmin.options()->SetIntegerValue("bonmin.node_limit", (int)*maxnodes);
bonmin.options()->SetNumericValue("bonmin.time_limit", *cputime);
@@ -136,13 +137,10 @@ int cpp_intfmincon(char *fname)
bb(bonmin);//process parameter file using Ipopt and do branch and bound using Cbc
}
catch(TNLPSolver::UnsolvedError *E) {
- Scierror(999, "\nIpopt has failed to solve the problem!\n");
}
catch(OsiTMINLPInterface::SimpleError &E) {
- Scierror(999, "\nFailed to solve a problem!\n");
}
catch(CoinError &E) {
- Scierror(999, "\nFailed to solve a problem!\n");
}
rstatus=tminlp->returnStatus();
diff --git a/build/cpp/sci_iofunc.cpp b/build/cpp/sci_iofunc.cpp
index 259f7c3..f05839c 100644
--- a/build/cpp/sci_iofunc.cpp
+++ b/build/cpp/sci_iofunc.cpp
@@ -258,8 +258,7 @@ bool getHessFromScilab(int n,int numConstr_,char name[], double *x,double *obj,d
if(getDoubleMatrixFromScilab(posFirstElementOnStackForSF, &x_rows, &x_cols, dest))
{
sciprint("No results ");
- return 1;
-
+ return 1;
}
}
return 0;
diff --git a/build/cpp/sci_minconTMINLP.cpp b/build/cpp/sci_minconTMINLP.cpp
index ac688d4..7885083 100644
--- a/build/cpp/sci_minconTMINLP.cpp
+++ b/build/cpp/sci_minconTMINLP.cpp
@@ -27,7 +27,7 @@ extern "C"
using namespace Ipopt;
using namespace Bonmin;
-#define DEBUG 0
+//#define DEBUG 0
minconTMINLP::~minconTMINLP()
{
@@ -37,6 +37,9 @@ minconTMINLP::~minconTMINLP()
// Set the type of every variable - CONTINUOUS or INTEGER
bool minconTMINLP::get_variables_types(Index n, VariableType* var_types)
{
+ #ifdef DEBUG
+ sciprint("Code is in get_variables_types\n");
+ #endif
n = numVars_;
for(int i=0; i < n; i++)
var_types[i] = CONTINUOUS;
@@ -48,6 +51,9 @@ bool minconTMINLP::get_variables_types(Index n, VariableType* var_types)
// The linearity of the variables - LINEAR or NON_LINEAR
bool minconTMINLP::get_variables_linearity(Index n, Ipopt::TNLP::LinearityType* var_types)
{
+ #ifdef DEBUG
+ sciprint("Code is in get_variables_linearity\n");
+ #endif
for(int i=0;i<n;i++)
{
var_types[i] = Ipopt::TNLP::NON_LINEAR;
@@ -57,6 +63,10 @@ bool minconTMINLP::get_variables_linearity(Index n, Ipopt::TNLP::LinearityType*
// The linearity of the constraints - LINEAR or NON_LINEAR
bool minconTMINLP::get_constraints_linearity(Index m, Ipopt::TNLP::LinearityType* const_types)
{
+
+ #ifdef DEBUG
+ sciprint("Code is in get_constraints_linearity\n");
+ #endif
for(int i=0;i<numLC_;i++)
{
const_types[i] = Ipopt::TNLP::LINEAR;
@@ -71,6 +81,9 @@ bool minconTMINLP::get_constraints_linearity(Index m, Ipopt::TNLP::LinearityType
//get NLP info such as number of variables,constraints,no.of elements in jacobian and hessian to allocate memory
bool minconTMINLP::get_nlp_info(Index& n, Index& m, Index& nnz_jac_g, Index& nnz_h_lag, TNLP::IndexStyleEnum& index_style)
{
+ #ifdef DEBUG
+ sciprint("Code is in get_nlp_info\n");
+ #endif
n=numVars_; // Number of variables
m=numCons_; // Number of constraints
nnz_jac_g = n*m; // No. of elements in Jacobian of constraints
@@ -99,6 +112,57 @@ bool minconTMINLP::get_bounds_info(Index n, Number* x_l, Number* x_u, Index m, N
return true;
}
+// This method sets initial values for required vectors . For now we are assuming 0 to all values.
+bool minconTMINLP::get_starting_point(Index n, bool init_x, Number* x,bool init_z, Number* z_L, Number* z_U,Index m, bool init_lambda,Number* lambda)
+{
+ assert(init_x == true);
+ assert(init_z == false);
+ assert(init_lambda == false);
+ if (init_x == true)
+ { //we need to set initial values for vector x
+ for (Index var=0;var<n;var++)
+ {x[var]=x0_[var];}//initialize with 0.
+ }
+ return true;
+}
+
+//get value of objective function at vector x
+bool minconTMINLP::eval_f(Index n, const Number* x, bool new_x, Number& obj_value)
+{
+ #ifdef DEBUG
+ sciprint("Code is eval_f\n");
+ #endif
+ char name[20]="_f";
+ Number *obj;
+ if (getFunctionFromScilab(n,name,x, 7, 1,2,&obj))
+ {
+ return false;
+ }
+ obj_value = *obj;
+ return true;
+}
+
+//get value of gradient of objective function at vector x.
+bool minconTMINLP::eval_grad_f(Index n, const Number* x, bool new_x, Number* grad_f)
+{
+ #ifdef DEBUG
+ sciprint("Code is in eval_grad_f\n");
+ #endif
+ char name[20]="_gradf";
+ Number *resg;
+ if (getFunctionFromScilab(n,name,x, 7, 1,2,&resg))
+ {
+ return false;
+ }
+
+ Index i;
+ for(i=0;i<numVars_;i++)
+ {
+ grad_f[i]=resg[i];
+ }
+ return true;
+}
+
// return the value of the constraints: g(x)
bool minconTMINLP::eval_g(Index n, const Number* x, bool new_x, Index m, Number* g)
{
@@ -145,7 +209,7 @@ bool minconTMINLP::eval_jac_g(Index n, const Number* x, bool new_x,Index m, Inde
else
{
unsigned int i,j,idx=0;
- for(int i=0;i<m;i++)
+ for(i=0;i<m;i++)
for(j=0;j<n;j++)
{
iRow[idx]=i;
@@ -182,58 +246,6 @@ bool minconTMINLP::eval_jac_g(Index n, const Number* x, bool new_x,Index m, Inde
return true;
}
-//get value of objective function at vector x
-bool minconTMINLP::eval_f(Index n, const Number* x, bool new_x, Number& obj_value)
-{
- #ifdef DEBUG
- sciprint("Code is eval_f\n");
- #endif
- char name[20]="_f";
- Number *obj;
- if (getFunctionFromScilab(n,name,x, 7, 1,2,&obj))
- {
- return false;
- }
- obj_value = *obj;
- return true;
-}
-
-//get value of gradient of objective function at vector x.
-bool minconTMINLP::eval_grad_f(Index n, const Number* x, bool new_x, Number* grad_f)
-{
- #ifdef DEBUG
- sciprint("Code is in eval_grad_f\n");
- #endif
- char name[20]="_gradf";
- Number *resg;
- if (getFunctionFromScilab(n,name,x, 7, 1,2,&resg))
- {
- return false;
- }
-
- Index i;
- for(i=0;i<numVars_;i++)
- {
- grad_f[i]=resg[i];
- }
-
- return true;
-}
-
-// This method sets initial values for required vectors . For now we are assuming 0 to all values.
-bool minconTMINLP::get_starting_point(Index n, bool init_x, Number* x,bool init_z, Number* z_L, Number* z_U,Index m, bool init_lambda,Number* lambda)
-{
- assert(init_x == true);
- assert(init_z == false);
- assert(init_lambda == false);
- if (init_x == true)
- { //we need to set initial values for vector x
- for (Index var=0;var<n;var++)
- {x[var]=x0_[var];}//initialize with 0.
- }
- return true;
-}
-
/*
* Return either the sparsity structure of the Hessian of the Lagrangian,
* or the values of the Hessian of the Lagrangian for the given values for
@@ -251,26 +263,27 @@ bool minconTMINLP::eval_h(Index n, const Number* x, bool new_x,Number obj_factor
Index idx=0;
for (Index row = 0; row < numVars_; row++)
{
- for (Index col = 0; col <= row; col++)
- { iRow[idx] = row;
+ for (Index col = 0; col < numVars_; col++)
+ {
+ iRow[idx] = row;
jCol[idx] = col;
idx++;
}
}
- }
+ }
else
{ char name[20]="_gradhess";
- Number *resh;
- if (getHessFromScilab(n,m,name,x, &obj_factor, lambda, 7, 3,2,&resh))
+ Number *resCh;
+ if (getHessFromScilab(n,m,name,x, &obj_factor, lambda, 7, 3,2,&resCh))
{
return false;
}
Index index=0;
for (Index row=0;row < numVars_ ;++row)
{
- for (Index col=0; col <= row; ++col)
+ for (Index col=0; col < numVars_; ++col)
{
- values[index++]=(resh[numVars_*row+col]);
+ values[index++]=resCh[numVars_*row+col];
}
}
}
@@ -281,18 +294,18 @@ void minconTMINLP::finalize_solution(SolverReturn status,Index n, const Number*
{
#ifdef DEBUG
sciprint("Code is in finalize_solution\n");
+ sciprint("%d",status);
#endif
finalObjVal_ = obj_value;
status_ = status;
if(status==0 ||status== 3)
{
- finalX_ = (double*)malloc(sizeof(double) * numVars_ * 1);
+ finalX_ = (double*)malloc(sizeof(double) * numVars_*1);
for (Index i=0; i<numVars_; i++)
{
finalX_[i] = x[i];
}
}
-
}
const double * minconTMINLP::getX()
generated by cgit (git 2.34.1) at 2024-12-22 04:42:21 +0000