How do I append data to a MAT file from C++?

17 views (last 30 days)
From a C++ program, I would like to continuously append data to a MAT file while the program runs. The program doesn't have to append to an existing file, but after creating the file at the beginning the program needs to append save data to the MAT file over long periods of time. mxCreateNumericMatrix appears to require the size of the variable to be known with no options to append. The program cannot keep all the data in memory and then write it upon completion.
Thanks!
  1 Comment
shuangshuang kong
shuangshuang kong on 15 Apr 2018
have you solve this problem yet? can you tell me how to do?because i have also met such problem,and i dont know how to figure it out.

Sign in to comment.

Accepted Answer

James Tursa
James Tursa on 30 Jun 2015
Why does it have to be a mat file? The simplest solution is to simply write the values in binary as you go, and then when you are done close the file. Then in MATLAB use fopen and fread to read in the data (in pieces if necessary because of memory constraints).
  2 Comments
James Tursa
James Tursa on 1 Jul 2015
MATLAB compresses mat files ... maybe that is what you mean by "optimized loading".
Is it your desire to write a MATLAB variable to a mat file incrementally without specifying the size, and then at the end specify the size of the variable? And do this for several variables in the same mat file?
If the variable can't be held in memory at once, is it your plan to load only pieces of it later?

Sign in to comment.

More Answers (2)

Adam Hug
Adam Hug on 30 Jun 2015
I think your best bet will be to use MATLAB Engine. This is an API that lets you run MATLAB commands using C/Fortran programs. See the following doc link for more information and examples:
  1 Comment
Sean de Wolski
Sean de Wolski on 1 Jul 2015
One could also use MATLAB Compiler based executables or libraries for this.

Sign in to comment.


Muthu Annamalai
Muthu Annamalai on 1 Jul 2015
Edited: Muthu Annamalai on 1 Jul 2015
The two alternatives are nicely laid out by Adam, and James.
  1. Adam's recommendation is to use the Mex API.
  2. James's recommendation is to use the load(), csvload() like commands from MATLAB
I think following James's approach maybe quicker, and all you need in the short run.

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by