Wrong print negative number with fprintf in hex

Why can't print function fprintf negative values in hex?

For example:

rng(73);
N = 20;
r = (randn(N, 1) + 1j.*randn(N, 1)) * 10000;
for i = 1:N
    fprintf('Real[%d] = %04X; ', i, int16(real(r(i))));
    fprintf('Imag[%d] = %04X; \n', i, int16(imag(r(i))));
end

And result:

Real[1] = 0D6C; Imag[1] = 1DF2; 
Real[2] = 053A; Imag[2] = -1.007700e+04; 
Real[3] = 00E1; Imag[3] = -1.237000e+04; 
Real[4] = 0D85; Imag[4] = 0435; 
Real[5] = -1.700000e+02; Imag[5] = -1.814200e+04; 
Real[6] = -4.073000e+03; Imag[6] = -5.949000e+03; 
Real[7] = -1.211600e+04; Imag[7] = -1.676000e+04; 
Real[8] = 0875; Imag[8] = -5.985000e+03; 
Real[9] = 3021; Imag[9] = 19A0; 
Real[10] = -1.096100e+04; Imag[10] = 378C; 
Real[11] = -3.266000e+03; Imag[11] = 20AB; 
Real[12] = 0B82; Imag[12] = 167C; 
Real[13] = -1.231200e+04; Imag[13] = -1.007300e+04; 
Real[14] = 270D; Imag[14] = 091E; 
Real[15] = -1.406600e+04; Imag[15] = 13FC; 
Real[16] = 056C; Imag[16] = 0E25; 
Real[17] = -6.861000e+03; Imag[17] = -1.780700e+04; 
Real[18] = 1EE4; Imag[18] = -6.337000e+03; 
Real[19] = 1EF1; Imag[19] = 01D4; 
Real[20] = -1.694700e+04; Imag[20] = -8.530000e+02; 

May be some cast need before?

Answer 1

[Walter Roberson]

rng(73);
N = 20;
r = (randn(N, 1) + 1j.*randn(N, 1)) * 10000;
for i = 1:N
    fprintf('Real[%d] = %04X; ', i, typecast(int16(real(r(i))), 'uint16')); fprintf('Imag[%d] = %04X; \n', i, typecast(int16(imag(r(i))), 'uint16'));
end

Real[1] = 0D6C; Imag[1] = 1DF2; 
Real[2] = 053A; Imag[2] = D8A3; 
Real[3] = 00E1; Imag[3] = CFAE; 
Real[4] = 0D85; Imag[4] = 0435; 
Real[5] = FF56; Imag[5] = B922; 
Real[6] = F017; Imag[6] = E8C3; 
Real[7] = D0AC; Imag[7] = BE88; 
Real[8] = 0875; Imag[8] = E89F; 
Real[9] = 3021; Imag[9] = 19A0; 
Real[10] = D52F; Imag[10] = 378C; 
Real[11] = F33E; Imag[11] = 20AB; 
Real[12] = 0B82; Imag[12] = 167C; 
Real[13] = CFE8; Imag[13] = D8A7; 
Real[14] = 270D; Imag[14] = 091E; 
Real[15] = C90E; Imag[15] = 13FC; 
Real[16] = 056C; Imag[16] = 0E25; 
Real[17] = E533; Imag[17] = BA71; 
Real[18] = 1EE4; Imag[18] = E73F; 
Real[19] = 1EF1; Imag[19] = 01D4; 
Real[20] = BDCD; Imag[20] = FCAB; 

[vadim onuchin]

yes, in newer version it's working. but in matlab r2019b not working.

It's bug in matlab r2019b.

[Walter Roberson]

I installed and tested R2019b. Given the above code, it produces exactly the same output as abover.

It is not a bug that %x does not handle negative values; %x only handles positive values. Using typecast() as I show here works

Answer 2

[Swastik]

Hi Vadim,

I understand that you want to print negative hexadecimal number, also I assume the 2’s complement of the negative number’s absolute value is to be printed in hex. So, -1 would print as 0xFFFF according to this.

I found that fprintf in MATLAB does not print negative numbers with the hexadecimal format(%x), this is mentioned in the documentation of fprintf which reads:

“If you specify a conversion that does not fit the data, such as a text conversion for a numeric value, MATLAB overrides the specified conversion, and uses %e.”

Here is the link for the same:

https://www.mathworks.com/help/releases/R2019b/matlab/ref/fprintf.html#btf8xsy-1_sep_shared-formatSpec

In this case, MATLAB expects an unsigned integer to the format specifier “%x”, however does not find one and defaults to “%e”.

I would like to suggest the following workaround to print the hexadecimal representation of a negative number,

If a is a signed integer, then we can apply the following operation to make sure it prints in hexadecimal every time.

a = -1 

a = -1

fprintf("%04x", a + (a<0)*2^32) 

ffffffff

Hope this helps.

[vadim onuchin]

I want to print int16 numbers. Your example working for double numbers.