For fi objects a and b, there is a difference between

a = b

and

a(:) = b

In the first case, a = b replaces a with b while a assumes the value, numerictype object and fimath object associated with b. In the second case, a(:) = b assigns the value of b into a while keeping the numerictype object of a. You can use this to cast a value with one numerictype object into another numerictype object.

For example, cast a 16-bit number into an 8-bit number.

a = fi(0, 1, 8, 7)

a = 
     0

          DataTypeMode: Fixed-point: binary point scaling
            Signedness: Signed
            WordLength: 8
        FractionLength: 7

b = fi(pi/4, 1, 16, 15)

b = 
    0.7854

          DataTypeMode: Fixed-point: binary point scaling
            Signedness: Signed
            WordLength: 16
        FractionLength: 15

a(:) = b

a = 
    0.7891

          DataTypeMode: Fixed-point: binary point scaling
            Signedness: Signed
            WordLength: 8
        FractionLength: 7

This example defines a variable acc to emulate a 40-bit accumulator of a DSP. The products and sums in this example are assigned into the accumulator using the syntax acc(1)=... Assigning values into the accumulator is like storing a value in a register. To begin, turn the logging mode on and define the variables. In this example, n is the number of points in the input data x and output data y, and t represents time. The remaining variables are all defined as fi objects. The input data x is a high-frequency sinusoid added to a low-frequency sinusoid.

fipref('LoggingMode', 'on');
n = 100;
t = (0:n-1)/n;
x = fi(sin(2*pi*t) + 0.2*cos(2*pi*50*t));
b = fi([.5 .5]);
y = fi(zeros(size(x)), numerictype(x));
acc = fi(0.0, true, 40, 30);

The following loop takes a running average of the input x using the coefficients in b . Notice that acc is assigned into acc(1)=... versus using acc=..., which would overwrite and change the data type of acc .

for k = 2:n
    acc(1) = b(1)*x(k);
    acc(1) = acc + b(2)*x(k-1);
    y(k) = acc;
end

By averaging every other sample, the loop shown above passes the low-frequency sinusoid through and attenuates the high-frequency sinusoid.

plot(t,x,'x-',t,y,'o-')
legend('input data x','output data y')

The log report shows the minimum and maximum logged values and ranges of the variables used. Because acc is assigned into, rather than overwritten, these logs reflect the accumulated minimum and maximum values.

logreport(x, y, b, acc)

                     minlog         maxlog     lowerbound     upperbound     noverflows    nunderflows
           x      -1.200012       1.197998             -2       1.999939              0              0
           y     -0.9990234      0.9990234             -2       1.999939              0              0
           b            0.5            0.5             -1      0.9999695              0              0
         acc     -0.9990234      0.9989929           -512            512              0              0

Display acc to verify that its data type did not change.

acc

acc = 
   -0.0941

          DataTypeMode: Fixed-point: binary point scaling
            Signedness: Signed
            WordLength: 40
        FractionLength: 30

Reset the fipref object to restore its default values.

reset(fipref)