Converting decimal value to binary

I have to convert decimal value to binary ,in hich my decimal value is in negative and positive

my values are 

a=[ 0.0803
    0.0726
    0.0469
   -0.0038
   -0.0421]

please help how to convert to binary

Answer 1

[Harshit]

http://www.mathworks.in/help/matlab/ref/dec2bin.html. See this. It will do 2's complement for negative numbers

[Pat]

But that pages says d must be a nonnegative integer smaller than 2^52.

another thing do you have and idea how to convert the above values into 14 bit

[Harshit]

ok sorry. i didn't notice. for negative integers a

if true
  % a=-x;
b=dec2bin(2^n+a);
end

for fractions convert first them into positive say -16.56 = -17 + 0.42. Get the binary of negative using the scheme above and get the binary of 0.42 bu multiplying by 2 at each instant and gathering the integer part. You will do it ofcourse on the basis of precision you required. n denotes the number of integer bit for you.

[Pat]

ok harshit thanks can you please tell how to convert values in a to signed 14 bit,those are ouputs of wavelet coefficients of a soundwave

Answer 2

[Walter Roberson]

There are three competing standards for the representation of negative integers.

There are many methods for the representation of fractions. The ones used to represent floating point double precision numbers use some schemes that required work to deal with unusual cases.

We cannot tell you how to do the conversion until you tell us which representation you wish to use.

[Pat]

Walter can you please tell what is

which representation you wish to use.

[Walter Roberson]

Is the representation one's complement, two's complement, or separate-sign? How many bits of exponent? How many bits of mantissa? Is the exponent separate-sign or two's complement or biased? What is the power that the exponent is to be raised to? Is the first bit of the mantissa to be explicit, or is it to be "hidden"?

For example, show us the exact bit breakdown you wish to use for -1E+42

[Pat]

Walter any of the one which will be useful for sampling the signal into 14 bits

[Walter Roberson]

Use the following standard but delete the last two bits of the mantissa:

http://en.wikipedia.org/wiki/Half-precision_floating-point_format#IEEE_754_half-precision_binary_floating-point_format:_binary16