Error using reshape: Size arguments must be real integers. Except all values are positive whole numbers?

*** I have attached the variable in question as a .mat file. But I am not sure if anyone will be able to view it. Please comment if not and I will try and share an excel version of the information. ****

I have some data inside a variable. It is dynamically allocated so sometimes the values are different. But I have created a way to make sure that I only create a matrix that can be reshaped. Here is the code I have:

num_steps = raw_data.num_steps;

% Reshape value matrix and time/source vector
mat_size = size(raw_data.variable_mat);

Unable to resolve the name raw_data.variable_mat.

% If the variable matrix has dimensions that will cause
% probems for reshaping, just remove data points to make it
% work.
reshape_val = floor(mat_size(2) / num_steps);
if reshape_val ~= mat_size(2) / num_steps
    raw_data.variable_mat = raw_data.variable_mat(:,1:reshape_val);
end

mat_size = size(raw_data.variable_mat); % Recreate mat_size for new variable_mat

raw_data.variable_mat = reshape(raw_data.variable_mat, mat_size(1), mat_size(2) / num_steps, num_steps);

But when I run this code, I receive the following error:

Error using reshape
Size arguments must be real integers.

This does not make sense to me as, in this case, you would technically have:

reshape(raw_data.variable_mat,2,10453,3)

where you should receive a 2x10453x3 matrix given that all values for size arguments are real integers (positive whole numbers).

Can anybody help me see what I am doing incorrectly here?

Answer 1

[Jan]

Change:

raw_data.variable_mat = raw_data.variable_mat(:,1:reshape_val);

to

raw_data.variable_mat = raw_data.variable_mat(:,1:(reshape_val * num_steps));

In the original version you crop too much and mat_size(2) / num_steps need not be an integer.

[Steven Manz]

Such an easy fix. Cannot believe I missed that. Thank you so much!

Answer 2

[Voss]

I think the main problem is that you are removing an incorrect number of elements (if I understand what you are trying to do). Have a look at the code below. I removed some semi-colons to show the variables' values, but other than that I only changed how the truncation of raw_data.variable_mat was done:

load('raw_data');

num_steps = raw_data.num_steps

num_steps = 3

% Reshape value matrix and time/source vector
mat_size = size(raw_data.variable_mat)

mat_size = 1×2

2 31358

% If the variable matrix has dimensions that will cause
% probems for reshaping, just remove data points to make it
% work.
reshape_val = floor(mat_size(2) / num_steps)

reshape_val = 10452

if reshape_val ~= mat_size(2) / num_steps
%     raw_data.variable_mat = raw_data.variable_mat(:,1:reshape_val);
    raw_data.variable_mat = raw_data.variable_mat(:,1:(reshape_val*num_steps));
end

mat_size = size(raw_data.variable_mat) % Recreate mat_size for new variable_mat

mat_size = 1×2

2 31356

raw_data.variable_mat = reshape(raw_data.variable_mat, mat_size(1), mat_size(2) / num_steps, num_steps);

size(raw_data.variable_mat)

ans = 1×3

2 10452 3

Now raw_data.variable_mat is the right size (maybe), but you need to also be sure that the elements are in the order you want after reshaping.

[Steven Manz]

Yeah, the issue is that I totally forgot to add that I need to multiply the the value by the number of steps. Probably time to stop for the day as I can tell my brain is fried. Thank you for the help!