FIR
- Finite Impulse Response filter

Moving averages, summation
(AmiBroker 5.40)

The output is convolution of input aray with coefficents table (impulse response table).

The function performs automatic normalization of coefficient table if necessary (if its sum is not 1)

INPUTS:

• array - input array to be smoothed
• coefficients - array of FIR coefficients
• size - number of coefficients ( filter_order + 1 )

It is functional (but 2+ times faster) equivalent of following AFL:

function FIR_AFL( input, coeff, size )
{
 result = 0;
 sumcoeff = 0;
 for( i = 0; i < Min( size, BarCount ); i++ )
 {
   sumcoeff += coeff[ i ];
   result += coeff[ i ] * Ref( input, - size + i + 1 );
 }

 return result/sumcoeff;
}

// Another example: ALMA which is FIR Filter with
// shifted Gaussian coefficents can be implemented as follows:

function ALMA_AFL( input, range, Offset, sigma )
{
 local m, im, s, Coeff;

 m = floor( Offset * (range - 1) );
 s = range / sigma;

 for( i = 0; i < Min( range, BarCount ); i++ )
 {
   im = i - m;
   Coeff[ i ] = exp( - ( im * im )/ ( 2 * s * s ) );
 }

 return FIR( input, Coeff, range );
}