Subscript indices must either be real positive integers or logicals.

Hi,

I see I am really bad at indexing so if someone could tell me some general rules (but less general than in online tutorials) I would be grateful.

*Subscript indices must either be real positive integers or logicals.

Error in PL2ex1 (line 4) meandata = mean(nonoise.data_out);*

File to read was uploaded.

%PL2ex1

nonoise = load('verg1');
meandata = mean(nonoise.data_out);
trials = repmat(meandata, 1000, 1);

%add noise with normal distribution (randn) and repeat
sigma = randn(1);
for i = 1:1000
    trials(i) = trials(i) + randn(1)*randn(size(trials(i)));
end

%compare plots and mean of trials
x = linspace(0, 10); % define axis x

figure
subplot(3,3,1)
n =(randi(1000));
plot(x, trials(n))
title(sprintf('Trial %d', n))

subplot(3,3,2)
n =(randi(1000));
plot(x, trials(n))
title(sprintf('Trial %d', n))

subplot(3,3,3)
n =(randi(1000));
plot(x, trials(n))
title(sprintf('Trial %d', n))

subplot(3,3,4)
n =(randi(1000));
plot(x, trials(n))
title(sprintf('Trial %d', n))

subplot(3,3,5)
n =(randi(1000));
plot(x, trials(n))
title(sprintf('Trial %d', n))

subplot(3,3,6)
n =(randi(1000));
plot(x, trials(n))
title(sprintf('Trial %d', n))

subplot(3,3,7)
n =(randi(1000));
plot(x, trials(n))
title(sprintf('Trial %d', n))

subplot(3,3,8)
n =(randi(1000));
plot(x, trials(n))
title(sprintf('Trial %d', n))

%mean
subplot(3,3,9)
means = zeros(1,400);
for i = 1:400
    means(1,i) = mean(trials(:,i));
end
plot(x, means)
title('Means')

Answer 1

[Star Strider]

In the Command Window, type:

whos mean

The output should be nothing (just the >> cursor).

If you get something like this:

    Name      Size            Bytes  Class     Attributes

    mean      1x1                 8  double

that means you have assigned a variable to ‘mean’, ‘overshadowing’ the mean function.

Solution: Rename the variable.

[Katarzyna Wieciorek]

I had problem like this before posting a question (I came up with such idea on my own), I edited and saved file, clear all but it seems MATLAB needed something more because after typing whos mean (with no result) and trying to run again there was no more error. I am sorry for bothering you, I thought it was something more serious than 'quit and run again' because I tried many times before.

It works but not well, so if I can ask, why my plots (excluding mean) are blank? They are blank whatever I set as x or when x is removed from plot, axes are adjusted but I see no points.

[Image Analyst]

Please attach verg1.mat so we can try your code, both to fix the plotting problem and to see if there is a mean inside it. Generally it's safer to assign load() a structure than to just let it "poof" variables into existence.

s = load('verg1.mat');  % Read variables into fields of a structure.
myMean = s.mean;

Of course there should not even be a variable called mean in your mat file, but unfortunately it seems that is the most likely cause of the error.

[Star Strider]

What size is ‘trials’ when you actually go to plot it?

Note that:

x = linspace(0, 10); % define axis x

produces a 100-element vector from 0 to |10}, but in every plot:

n =(randi(1000));

is a single random integer between 1 and 1000.

[Katarzyna Wieciorek]

Image Analyst: It is attached from the beginning. I predicted such need :)

Star Strider: What I wanted to do was:

trials = repmat(meandata, 1000, 1); 
- 1000 copies of meandata with noise added later

n =(randi(1000));
plot(x, trials(n)) 
- plot random trial

Where is my mistake?

[Star Strider]

Put simply:

n =(randi(1000));

is a single random integer between 1 and 1000. It is not a vector.

[Katarzyna Wieciorek]

I wanted to display one of copies (copy number random n) from array 'trials', why it doesn't work?

[Star Strider]

Since ‘trials’ is a (1000x400) matrix, you have to specify that you want a row and not a single element.

You will also need to define ‘x’ as a 400-element vector.

See if this does what you want:

x = linspace(0, 10, 400); % define axis x

then:

plot(x, trials(n,:))

[Katarzyna Wieciorek]

Now I have:

Index exceeds matrix dimensions.

Error in PL2ex1 (line 20) title(sprintf('Trial %d', n))

I don't understand why, I didn't change anything concerning n :(

[Star Strider]

I don’t know either. I assume ‘n’ is an integer between 1 and 1000. You might see if changing ‘%d’ to ‘%.0f’:

title(sprintf('Trial %.0f', n))

changes anything.

I can’t see how that line would throw that error.

[Katarzyna Wieciorek]

No errors now, but the same graph in every subplot -.- It is so confusing.

[Star Strider]

See if this slight modification of your code — that does all your first 8 plots in the for loop — improves it:

nonoise = load('verg1');
meandata = mean(nonoise.data_out);
trials = repmat(meandata, 1000, 1);
x = linspace(0, 10, 400); % define axis x

trials = trials + randn(size(trials))*0.5;

for k1 = 1:8
    subplot(3,3,k1)
    n = randi(1000);
    plot(x, trials(n,:))
    title(sprintf('Trial %d', n))
end

%mean
subplot(3,3,9)
means = mean(trials);
plot(x, means)
title('Means')

Experiment with different constants (actually, the standard deviation, here 0.5) to get the randomness you want. you will see that the plots are now all different. I believe the noise you were adding was too small to be visible on your plots.

Answer 2

[Roger Stafford]

I would guess that somewhere in your system you have defined a variable named 'mean', in which case matlab misinterprets the calls to the function 'mean' as referring to that variable. That is the only way I can account for the error message. You should never use the names of matlab functions as variables for just this reason.