fastSaveUpdate - Efficiently update variables in MAT-file

Faster alternative to built-in 'save -append' command, which becomes increasingly slow as the number of variables and size of MAT-file grows

fastSaveUpdate - Efficiently update existing variables in MAT-file

This function provides a faster alternative to MATLAB's built-in

'save -append' command, which can become increasingly slow as the

number of variables and size of a MAT-file grows. The slowdown occurs

because 'save -append' must scan the entire MAT-file to determine

whether each variable exists, and may need to rewrite large portions

of the file if a variable already exists.

Instead of relying on '-append', this function first loads the

variables in the MAT-file that are NOT being updated, then saves

them along with the updated variables. This avoids repeated lookup

and rewriting overhead.

Usage:

fastSaveUpdate(matfile, variablesToSave, progressMSG)

Inputs:

matfile - String specifying the path to the .mat file.

variablesToSave - (Optional) Cell array of variable names (as strings) to

update or add in the MAT-file. If omitted or empty,

all variables in the current workspace will be used.

progressMSG - (Optional) verbose for progress messages [0 for no, 1 for yes]

Example:

x = rand(100);

y = sin(1:100);

z = cos(1:100);

save('myData.mat', 'x', 'y', 'z'); % Initial save

fastSaveUpdate('myData.mat', {'y','z'}, 1); % Update variables y and z efficiently

Notes:

- *** MATLAB's built-in 'save -append' functionality is faster when

appending new variables that do not already exist in the MAT-file.

The current function, fastSaveUpdate, is faster only for updating the

existing variables.

- If a variable listed in 'variablesToSave' already exists in the MAT-file,

it will be overwritten.

- All other existing variables in the file are preserved.

- This function may use more RAM memory than 'save -append', as it temporarily

loads existing variables into RAM memory before re-saving.

- For large files with many variables, this approach is often significantly

faster than using 'save -append'.

See also: save, load, who

-------------------------------------------------------------------------

Author: Pritesh Parmar, PhD

Changelog:

v1.0 - (04/15/2025) Initial version implementing faster variable update via selective loading and saving

v1.1 - (04/16/2025) Minor changes; added performance analysis comparing the current function

versus MATLAB's built-in 'save -append'

-------------------------------------------------------------------------