I have Octave 3.2-4 on Ubuntu 11.10. This is the officially supported version for Ubuntu. Where it seems that GNU Octave's save
function does not support binary data saving in single precision float format in MAT-files. For instance:
octave:119> p_t = single(rand(1,5))
p_t =
0.038507 0.117961 0.894508 0.788297 0.651727
octave:120> class(p_t)
ans = single
octave:121> save -v6 P_T.mat p_t
octave:122> clear all
octave:123> load -v6 P_T.mat p_t
octave:124> p_t
p_t =
0.038507 0.117961 0.894508 0.788297 0.651727
octave:125> class(p_t)
ans = double
I have tried searching for a known issue of Octave, but could not find anything... Anyone else noticed this problem?