HDF5 transpose data?

I have a HDF5 file called "output.h5" and it have a dataset called "density" which is a matrix with 21 rows and 2 cols.

I read this HDF5 dataset, but I got the transposed version of my density matrix (2 rows, 21 cols). Why? How can I solve it?

Here is my HDF5 dump of this file:

h5dump output.h5
HDF5 "output.h5" {
GROUP "/" {
   GROUP "my" {
      GROUP "path" {
         GROUP "to" {
            DATASET "density" {
               DATATYPE  H5T_IEEE_F64LE
               DATASPACE  SIMPLE { ( 21, 2 ) / ( 21, 2 ) }
               DATA {
               (0,0): 0, 2,
               (1,0): 1, 2,
               (2,0): 2, 4,
               (3,0): 3, 8,
               (4,0): 4, 14,
               (5,0): 5, 22,
               (6,0): 6, 32,
               (7,0): 7, 44,
               (8,0): 8, 58,
               (9,0): 9, 74,
               (10,0): 10, 92,
               (11,0): 11, 112,
               (12,0): 12, 134,
               (13,0): 13, 158,
               (14,0): 14, 184,
               (15,0): 15, 212,
               (16,0): 16, 242,
               (17,0): 17, 274,
               (18,0): 18, 308,
               (19,0): 19, 344,
               (20,0): 20, 382
               }
            }
         }
      }
   }
}
}

Thanks

Answer 1

[Anudeep Kumar]

Hey Nycholas,

I believe this maybe due to the difference in the way different languages store data. C uses row-major and MATLAB uses column-major to store data. So when you read a dataset from an HDF5 file in MATLAB, the dimensions may appear transposed.

There is a detailed discussion on the same in the following MATLAB Answer link :

https://www.mathworks.com/matlabcentral/answers/308303-why-does-matlab-transpose-hdf5-data

You can use the transpose operator to obtain data in your intended shape as follows

data = h5read('output.h5', '/my/path/to/density')';