Example run without -m

The command line is:

$ max_diff_nc.sh 111/histrac.nc 113/histrac.nc

In this example, max_diff_nc is invoked without the options -s, -m or -i. Here is the output:

standard output comment 
Found  25  variable(s) in group /
in the first file.
The program found 25 variables in the root group of the first file. In this example, there are no subgroups. 23 of the variables are identical to the variables with same name in the second file. As the option -s was not used, those 23 identical variables do not appear below. Only two variables for which differences were found appear. 
---------------------------------------------
Variable "masse":

 Relative difference for non-zero values:

Maximum difference:  2.7E-03
 Occurring at:
data_old(14 7 4 1) =    6.97899061E+15
data_new(14 7 4 1) =    6.95995316E+15
masse is a 4-dimensional array. The location of the maximum is given by the set of 4 indices: 14 7 4 1. These are the Fortran indices:
  • the list of dimensions is reversed from the list given by ncdump -h;
  • the indices start at 1 in each dimension
(The location of the maximum is repeated for the two arrays but, of course, it is by definition the same location.) You can see that 6.97899061E+15 / 6.95995316E+15 - 1 ~= 2.7E-03		 
 Absolute difference:

Maximum difference:  2.3E+13
 Occurring at:
data_old(14 7 5 1) =    8.52667026E+15
data_new(14 7 5 1) =    8.50399659E+15
There is no zero value in data_old or data_new so the program only prints the maximum absolute difference for the whole arrays. 
||data_new - data_old||_infinite
/ ||data_old||_infinite =  2.6E-03
 This is the infinite norm (that is, maximum of absolue value) of the difference divided by the infinite norm of data_old. Note that you had the infinite norm of the difference just above (2.3E+13 in this example) so this also gives you the infinite norm of data_old. 
-------------------------------------------
Variable "d_tr_cv_RN":

 Relative difference for non-zero values:

Maximum difference: 1.6E+07
 Occurring at:
data_old(3 16 1 1) =  1.6982631E-24
data_new(3 16 1 1) =  2.7377417E-17


 Absolute difference when there is a zero:

Maximum difference: 5.9E+03
 Occurring at:
data_old(19 15 1 1) =  -5863.641
data_new(19 15 1 1) =  0.

 Absolute difference:

Maximum difference: 6.6E+03
 Occurring at:
data_old(11 12 1 1) =  -8759.474
data_new(11 12 1 1) =  -2166.5264
There are zero values at some locations in data_old or data_new so the program prints two results: for locations where there is a zero value and for the whole arrays.