How can I align pos/neg numbers with fprintf?

I'm trying to recreate the a specific file format (K file), that looks as follows

     219  8.0199807e-001 -2.2447408e-002  4.5720000e-002
     220  8.0199807e-001 -2.2447408e-002  5.0800000e-002
     221  8.0199807e-001  2.3252592e-002  2.2204460e-016

My code now can assemble the matrix correctly, but I'm having trouble aligning positive value numbers with negative ones. For instance, the code above shows up as:

          219 1.8290000e+00 -2.7420000e-02 1.0160000e-03 
    220 1.8290000e+00 -2.7420000e-02 5.0800000e-03 
    221 1.8290000e+00 9.1400002e-03 -3.5560000e-02

The value in the 3rd row and column is left adjusted which is a huge hassle later on. I can't simply add a "+".

What can I do to properly align them?

The format code I have now for the fprintf command to write them is:

    formatspec = ('\t %4i %9.7e %9.7e %9.7e \n');

Any help is greatly appreciated!

Answer 1

[Star Strider]

Probably the easiest way is to precede each width specifier with a space.

Example:

x  = [pi; -pi]
s1 = sprintf('% .4f\n', x)

Another option is to force the sign to be printed:

x  = [pi; -pi]
s2 = sprintf('%+.4f\n', x)

producing respectively:

s1 =

   3.1416
  -3.1416

s2 =

+3.1416
-3.1416

Note that ‘sprintf’ and ‘fprintf’ are essentially the same, except for the nature of the output. I chose ‘sprintf’ to illustrate the concept.

Answer 2

[Image Analyst]

9.7 is not big enough to hold the variable so it automatically expands the field, and because one has - and one doesn't, it expands them by different amounts. Try %15.7e or something like that.

Answer 3

[dpb]

>> xx=[219  8.0199807e-001 -2.2447408e-002  4.5720000e-002
   220  8.0199807e-001 -2.2447408e-002  5.0800000e-002
   221  8.0199807e-001  2.3252592e-002  2.2204460e-016];
>> fmt=['\t %4i' repmat('%12.3e',1,2) '%12.3e\n'];
>> fprintf(fmt,xx.')
    219   8.020e-01  -2.245e-02   4.572e-02
    220   8.020e-01  -2.245e-02   5.080e-02
    221   8.020e-01   2.325e-02   2.220e-16
>>

Should work. I shortened the fields but the alignment should work fine if you count spaces correctly. NB there are no extra blanks in the formatting string.

Unless the file format demands the tab I'd forego it and use a fixed width field of the proper width for the integer field, too. Tabs are dependent upon the displaying device for interpretation so can often cause problems altho only the single one at the beginning of the line is less of an issue than trying to use them embedded within a line.

Let's see, for your 7 decimal places means you need--

     7
plus 4 for the exponent field
plus 1 for the decimal point
plus 1 for the sign field
plus 1 (at least) for the space between fields

That's a field width of 14 and you asked for only 9. Hence Matlab did what you asked; used the minimum possible to fill the field resulting in the shifting of position when the positive sign wasn't needed.

AFAIK, there's no standard way in C to write the three-digit exponent field; if you must have that you'll have to write it specifically.