Filter Design Toolbox

eps

Return the quantized relative accuracy for quantized filters, quantizers, and quantized FFTs

Syntax

• eps(q)
  eps(hq)
  eps(f)
  

Description

eps(q) returns the quantization level of quantizer object q, or the distance from 1.0 to the next largest floating-point number when q is a floating-point quantizer object.

eps(hq) returns the quantization level of quantized filter hq, or the distance from 1.0 to the next largest floating-point number when hq is a floating-point quantized filter.

eps(f) returns the quantization level of quantized FFT f, or the distance from 1.0 to the next largest floating-point number when f is a floating-point quantized FFT.

Examples

• q = quantizer('float',[6 3]); 
  eps(q) 
  

returns the value 0.25. The following code

• hq = qfilt;
  f = qfft;
  eps(hq)
  eps(f)
  

returns the values

• coefficient   3.051757813e-005
         input   3.051757813e-005
        output   3.051757813e-005
  multiplicand   3.051757813e-005
       product   9.313225746e-010
           sum   9.313225746e-010
  

for the quantizers in the quantized filter, and

•  coefficient   3.051757813e-005
         input   3.051757813e-005
        output   3.051757813e-005
  multiplicand   3.051757813e-005
       product   9.313225746e-010
           sum   9.313225746e-010
  

for the quantizers in the quantized FFT.

Algorithm

For fixed-point or floating-point numbers,  where e is the relative accuracy of the quantizer and f is the fraction length of the quantizer.

See Also
eps, exponentbias, exponentlength, exponentmax, exponentmin

disp

errmean