[SOLVED] Gfortran's strange behavior

AlexBB · 11-28-2014, 03:49 PM

Hi there,

My environment is Oracle VM (Ubuntu)==>GFortran.

I have a problem with my GFortran. I have an input file (anglesV1.dat) with two columns of numbers:

Code:

  0.7439803D-01  0.1107149D+01         
  0.7321435D-01  0.1099019D+01         
  0.7203546D-01  0.1090621D+01         
  0.7086159D-01  0.1081944D+01        
  0.6969303D-01  0.1072974D+01        
  0.6853005D-01  0.1063698D+01         
  0.6737297D-01  0.1054101D+01        
  0.6622212D-01  0.1044169D+01

Then I run a subroutine that reads them and prints them on the screen:

Code:

     subroutine determinant                 
           real*8 :: phi,theta
           integer*4 :: counter
           open (unit=1,file='anglesV1.dat', status='old')
           counter = 0
10         FORMAT (2F12.5)
           do while (counter .lt. 20)
             counter = counter + 1
             read (1,10) theta,phi
             write (6,10) theta,phi
           enddo
           close(unit=1)
         end subroutine determinant

This is what I get in the terminal:

Code:

     0.74398   -10.11071
     0.73214   -10.10990
     0.72035   -10.10906
     0.70862   -10.10819
     0.69693   -10.10729
     0.68530   -10.10636
     0.67373   -10.10541
     0.66222   -10.10441

I don't understand the result. In the first column all exponents are -1. It means that the numbers (mantissas) must be multiplied by 10 to the power -1, meaning it is 1/10, correct? So when I use F12.6 format the numbers in the first column should come out like this:

Code:

     0.074398   
     0.073214 etc.,

In the second column the exponents are +1 meaning that the mantissas should be multiplied by 10.0, that is:

Code:

                 1.107149    
                 1.099019

Why do I have negative sign in the output for the second column?

Please help, thanks.

suicidaleggroll · 11-28-2014, 04:29 PM

This

Code:

  0.7439803D-01  0.1107149D+01         
  0.7321435D-01  0.1099019D+01         
  0.7203546D-01  0.1090621D+01         
  0.7086159D-01  0.1081944D+01        
  0.6969303D-01  0.1072974D+01        
  0.6853005D-01  0.1063698D+01         
  0.6737297D-01  0.1054101D+01        
  0.6622212D-01  0.1044169D+01

Is not 2F12.5, it's 2D15.7, so when you tell it to parse it as 2F12.5 you'll get the wrong result.
Reading this line

Code:

  0.7439803D-01  0.1107149D+01

as 2F12.5, will see this

Code:

  0.7439803D

for the first number, and this

Code:

-01  0.11071

for the second number.

Either remove the format code from the read, or change it to match the format of the data you're reading.

AlexBB · 11-28-2014, 05:35 PM

I am in AWE with you, suicidaleggroll. I will try it in a minute. Many thanks.

AlexBB · 11-28-2014, 05:36 PM

It's awesome! It worked. Another lesson for me. Thanks again.