Scilab Home page | Wiki | Bug tracker | Forge | Mailing list archives | ATOMS | File exchange
Change language to: English - Português - 日本語 - Русский

Aide de Scilab >> Optimisation et Simulation > Optimization base > optimbase_function

# optimbase_function

Calls cost function.

### Calling Sequence

```[opt, f, index] = optimbase_function(opt, x, index)
[opt, f, c, index] = optimbase_function(opt, x, index)
[opt, f, g, index] = optimbase_function(opt, x, index)
[opt, f, g, c, gc, index] = optimbase_function(opt, x, index)```

### Argument

opt

The object of TOPTIM type (tlist). (INPUT/OUTPUT)

x

A column vector of doubles, the current point.

index

An integer. (INPUT/OUTPUT)

The index input parameter has the following meaning.

• index = 1: nothing is to be computed, the user may display messages, for example.

• index = 2: compute f.

• index = 3: compute g.

• index = 4: compute f and g.

• index = 5: compute c.

• index = 6: compute f and c.

• index = 7: compute f, g, c and gc.

The index output parameter has the following meaning .

• index > 0: everything is fine.

• index = 0: the optimization must stop.

• index < 0: one function could not be evaluated.

f

Scalar. The value of the cost function.

g

Row matrix of doubles. The gradient of the cost function.

c

Row matrix of doubles. The non-linear, positive, inequality constraints.

gc

Matrix of doubles. The gradient of the non-linear, positive, inequality constraints.

### Description

The `optimbase_function` function calls the cost function and return the required results. If a cost function additional argument is defined in current object, it is passed it to the function as the last argument.

The `-function()` option allows to configure the cost function. The cost function is used, dependind on the context, to compute the cost, the non-linear inequality positive constraints, the gradient of the function and the gradient of the non-linear inequality constraints.

The cost function can also be used to produce outputs and to terminate an optimization algorithm.

Each calling sequence of the `optimbase_function` function corresponds to a specific calling sequence of the user-provided cost function.

• If the `-withderivatives` is false and there is no non-linear constraint, the calling sequence is

`[ this , f , index ] = optimbase_function ( this , x , index )`

which corresponds to the cost functions:

`[f , index ] = costf( x , index )`
• If the `-withderivatives` is false and there are non-linear constraints, the calling sequence is

`[ this , f , c , index ] = optimbase_function ( this , x , index )`

which corresponds to the cost functions:

`[f , c , index ] = costf( x , index )`
• If the `-withderivatives` is true and there is no non-linear constraint, the calling sequence is

`[ this , f , g , index ] = optimbase_function ( this , x , index )`

which corresponds to the cost functions:

`[f , g , index ] = costf( x , index )`
• If the `-withderivatives` is true and there are non-linear constraints, the calling sequence is

`[ this , f , g , c , gc , index ] = optimbase_function ( this , x , index )`

which corresponds to the cost functions:

`[f , g , c , gc , index ] = costf( x , index )`

Each calling sequence corresponds to a particular class of algorithms, includinf for e xample

• unconstrained, derivative-free algorithms,

• unconstrained constrained, derivative-free algorithms,

• unconstrained, derivative-based algorithms,

• nonlinearily constrained, derivative-based algorithms,

• etc.

It might happen that the function requires additional arguments to be evaluated. In this case, we can use the following feature. The argument `fun` can also be the list `(f, a1, a2, ...)`. In this case `f`, the first element in the list, must bea function and must have the header:

```[ f , index ] = f ( x , index , a1 , a2 , ... )
[ f , c , index ] = f ( x , index , a1 , a2 , ... )
[ f , g , index ] = f ( x , index , a1 , a2 , ... )
[ f , g , c , gc , index ] = f ( x , index , a1 , a2 , ... )```

where the input arguments `a1, a2, ...` are automatically appended at the end of the calling sequence.

### Example : Setting up an optimization

In the following example, one searches the minimum of the 2D Rosenbrock function. One begins by defining the function "rosenbrock" which computes the Rosenbrock function. The traditionnal initial guess [-1.2 1.0] is used. The initial simplex is computed along the axes with a length equal to 0.1. The Nelder-Mead algorithm with variable simplex size is used. The verbose mode is enabled so that messages are generated during the algorithm. After the optimization is performed, the optimum is retrieved with quiery features.

```function [f, index]=rosenbrock(x, index)
f = 100*(x(2)-x(1)^2)^2 + (1-x(1))^2;
endfunction

opt = optimbase_new();
opt = optimbase_configure(opt,"-numberofvariables",2);
nbvar = optimbase_cget(opt,"-numberofvariables");
opt = optimbase_configure(opt,"-function",rosenbrock);
[ opt , f , index ] = optimbase_function ( opt , [0.0 0.0] , 1 );
expectedf = 1
disp(f)
opt = optimbase_destroy(opt);```

### Example : Passing extra parameters

In the following example, we consider a function which has two additional parameters `a` and `b`. In this case, we can configure the "-function" option as a list, where the first element is the function and the two extra arguments are located at the end of the list.

```function [f, index]=rosenbrock2(x, index, a, b)
f = a*(x(2)-x(1)^2)^2 + (b-x(1))^2;
endfunction

opt = optimbase_new();
opt = optimbase_configure(opt,"-numberofvariables",2);
nbvar = optimbase_cget(opt,"-numberofvariables");
a = 100;
b = 1;
opt = optimbase_configure(opt,"-function",list(rosenbrock2,a,b));
[ opt , f , index ] = optimbase_function ( opt , [0.0 0.0] , 1 );
expectedf = 1
disp(f)
opt = optimbase_destroy(opt);```