help with single precision floating point arithmetic

Hello Can you help me with the single precision floating point arithmetic in matlab I need to generate twiddle factors and obtain the output in the binary format( which is the standard IEEE 754 single precision 32-bit floating point format).Is there any function in MATLAB that will directly convert any value such as sine(pi/4)into the single precision format?

Thanks!

Answer 1

[Walter Roberson]

dec2bin(typecast(single(pi),'uint32'),32)

Answer 2

[Titus Edelhofer]

Hi,

guess what: the function is called 'single' ;-).

Titus

Answer 3

[Daniel Shub]

You do need to be careful with MATLAB and converting to single. Potentially not obvious is that

isequal(single(pi*pi), single(pi)*single(pi))

Also, I believe that in some cases single precision numbers are converted to double precision during calcualtions

Answer 4

[Aparna]

@titus , i tried using the function but it doesn't produce the output in the 32 bit precision floating point format.

p.s. i need the output in binary

thanks for helping me with this.

Answer 5

[Titus Edelhofer]

Hi Aparna,

I understand:

num2hex(single(pi))

Is that what you are looking for?

Titus

Answer 6

[Aparna]

hey! no i do not want the answer in Hexadecimal format i need it in binary 32 bit format. Could you help me with that?

[Titus Edelhofer]

binary 32 bit format? So 32 zeros and ones?

dec2bin(hex2dec(num2hex(single(pi))))

and add zeros on the left for having indeed 32 bits ...

Titus

[Raiyyan Masumdar]

Hi Titus,

I was actually looking for conversion of a matrix of real numbers into single precision binary. Any idea?

[James Tursa]

So, did you try the above code? It works for matrices also.

[Raiyyan Masumdar]

yes it works, thank you !!!

Answer 7

[Aparna]

hey actually it works i guess! thank you daniel and titus

[Aparna]

hey thank you so much it worked:)

Answer 8

[Tony Scarcia]

This will convert a single precision float to 32 bit binary using IEEE 754 format where;

• s is the sign bit (bit 31)
• e is the exponent (bits 30 to 23)
• m is the mantissa (bits 22 to 0)

% a is the float number
% b is the 32 bit binary converted number
b = padarray(hexToBinaryVector(num2hex(single(a)))',32-length(hexToBinaryVector(num2hex(single(a)))'),'pre')';
sign = b(1)
exp = b(2:9)
mantissa = b(10:32)