MATLAB Answers

Add structs in a saved array of structures

4 views (last 30 days)
Miquel
Miquel on 27 Jan 2020
Commented: Miquel on 28 Jan 2020
Hello,
I have a function (fun1) which generates several results (r1,r2,r3...). I organise them in a struct.
results=struct('results1',{r1},'results2',{r2},'results3',{r3},...);
This struct is the output of the funtion.
I have to run this same function for several variables and I want to keep all the results in the same array of structs. Therefore could I do:
results(1)=fun1(variable1,otherinputs);
results(2)=fun1(variable2,otherinputs);
results(3)=fun1(variable3,otherinputs);
...
However, the function deals with huge sets of data so it can take on the order of 45 minutes to run, and I have to run it more than 20 times. Is there any way to save the output struct so that each time that I run the function it adds the output to the previous ones? I mean to do something similar to what I do in the workspace but directly in a .mat file.
Thanks!

  0 Comments

Sign in to comment.

Accepted Answer

Mohammad Sami
Mohammad Sami on 28 Jan 2020
You can use the matfile function to write directly to a mat file
a = matfile('newfile.mat','Writable',true);
n = 10;
for i = 1:n
% your code
a.mystruct(i,1) = struct('abc',123,'def',456);
end
clear a;

  6 Comments

Show 3 older comments
Miquel
Miquel on 28 Jan 2020
BTW, is it advisable to start the loop in the end for allocation issues (as Jan always remarks in the forum)? Or is this not the case with structs?
n = numberofvariables;
for i = n:-1:1
a.results(i,1)=fun1(variable(i),otherinputs(i));
end
Stephen Cobeldick
Stephen Cobeldick on 28 Jan 2020
Unless the structure itself** is huge, iterating in reverse is unlikely to make much difference. But you can experiment and tell us the results.
** not the data it contains!
Miquel
Miquel on 28 Jan 2020
Ok, thanks. No, I generate 2 stucts that are only 7 fields and around 20... rows? So I guess it should make almost no difference. And I will be using the computer most of the time while Matlab is running. Thanks both!

Sign in to comment.

More Answers (0)

Sign in to answer this question.

Products


Release

R2018b