Confusion between real-world and floating-point values

Hello there,

I'm converting floating to real-world data values with the code snippet below.

I notice that after scaling, floating point values are no longer fractional, implying that data loss has occurred.

Thank you in advance

fi(linspace(-5,5,10),true,32,28)

ans = 
   -5.0000   -3.8889   -2.7778   -1.6667   -0.5556    0.5556    1.6667    2.7778    3.8889    5.0000

          DataTypeMode: Fixed-point: binary point scaling
            Signedness: Signed
            WordLength: 32
        FractionLength: 28

storedInteger(fi(linspace(-5,5,10),true,32,28))

ans = 1×10

-1342177280 -1043915662 -745654044 -447392427 -149130809 149130809 447392427 745654044 1043915662 1342177280

storedInteger(fi(linspace(-5,5,10),true,32,28)) / 2 ^28

ans = 1×10

-5 -4 -3 -2 -1 1 2 3 4 5

I'm expecting fractional length, but I'm wondering where it's gone. Thanks !!

Answer 1

[Walter Roberson]

You forgot to take into account that an integer data type operated on with a double precision number, returns the integer data type

fi(linspace(-5,5,10),true,32,28)

ans = 
   -5.0000   -3.8889   -2.7778   -1.6667   -0.5556    0.5556    1.6667    2.7778    3.8889    5.0000

          DataTypeMode: Fixed-point: binary point scaling
            Signedness: Signed
            WordLength: 32
        FractionLength: 28

SI = storedInteger(fi(linspace(-5,5,10),true,32,28))

SI = 1×10

-1342177280 -1043915662 -745654044 -447392427 -149130809 149130809 447392427 745654044 1043915662 1342177280

class(SI)

ans = 'int32'

SI / 2^28

ans = 1×10

-5 -4 -3 -2 -1 1 2 3 4 5

double(SI) / 2^28

ans = 1×10

-5.0000 -3.8889 -2.7778 -1.6667 -0.5556 0.5556 1.6667 2.7778 3.8889 5.0000

[Life is Wonderful]

Ahaah ,Thanks makes perfect sense.