[Eeglablist] saving .set files as singles

Arnaud Delorme arno at salk.edu
Mon Nov 7 14:18:42 PST 2005

Elliot Cohen wrote:

> Hi,
> In linux, when saving datasets, it seems that eeglab automatically 
> converts the data array to singles (from matlab's default double).  
> Can anyone comment on why this is done?  Can this behaviour be turned 
> off, without changing the pop_saveset code?

This is done to save space on disk. The reason is that for EEG 
datafiles, having double precision (64 bits) is not necessary (for most 
EEG recording systems, signal above bit 17 are usually considered 
noise), so the signal may be saved as single precision (32 bits). In 
Matlab 7, all computation is done in single precision, but in Matlab 5 
or 6, EEGLAB is required to use double precision for the computation. 
For these Matlab versions, once the EEGLAB file is loaded, the data 
array is simply converted to double

EEG.data = double(EEG.data);



More information about the eeglablist mailing list