Input combination of string and number to a function which only accepts numbers?

I am using EEGLAB on Matlab and would like to cut sections of the EEG between blocks. In EEGLAB the function is pop_select which accepts numbers separated by semicolon, e.g. [0 100; 120 150; 160 170], to select e.g. the period from 0 to 100 ms, etc.

I would like to create this input as an array, input_array, which is fed directly to the function, e.g. pop_select( EEG, 'notime', input_array), since I don't know beforehand which time periods I want to select.

However, the input must be separated by semicolon, which means that input_array contains strings and cannot be converted to a number array. Is there a way to overcome this?

[Rik]

How exactly does that mean input_array contains strings?

I only see a 3x2 numeric array.

[0 100; 120 150; 160 170]

ans = 3×2

0 100 120 150 160 170

[Martin Randau]

When constructing the array I need to insert the semicolon as a string. The function pop_select will not accept the input without semicolon separating the time intervals. So ";" is a string, so is "[" and "]" which I have to put at the end and start of the input, respectively.

Answer 1

[Cris LaPierre]

MATLAB uses semicolons to construct arrays. A semicolon represents the end of a row. Anything following it is placed on the next row.

You can learn more about this in Ch 4 of MATLAB Onramp.

[Martin Randau]

The problem is I need to pass an entire expression consisting of both strings, "[", "]" and ";" as well as numbers. I cannot pass e.g. [0 100] [125 150] - the numbers in brackets must be separated by a semicolon.

[Cris LaPierre]

Go through Ch 4. You do not appear to yet understand MATLAB syntax.

[Martin Randau]

Thank you, I will do that, but let me explain the problem a bit better with examples from my script.

EEG.event.type contains event codes formatted as string. '88' and '99' marks the start and end of a block in the EEG session. I start by saving these events in a string array and their time points are converted to seconds as per the EEG sampling rate.

boundaryStart = find(strcmp({EEG.event.type}, '88'));

boundaryEnd = find(strcmp({EEG.event.type}, '99'));

boundaryStart_sec = [EEG.event(boundaryStart).latency]*1/EEG.srate;

boundaryEnd_sec = [EEG.event(boundaryEnd).latency]*1/EEG.srate;

For example:

boundaryStart_sec =

Columns 1 through 10

14.004 74.221 197.12 292.9 388.25 481.98 577.12 675.46 775.76 873.22

Column 11

968.02

and

boundaryEnd_sec =

Columns 1 through 10

40.743 161.16 283.66 380.89 475.19 569.82 665.91 761.84 865.15 960.26

Column 11

1056.1

I want to remove the EEG data between blocks and can do this "by hand" with pop_select:

EEG = pop_select( EEG, 'notime', ...,

[EEG.xmin boundaryStart_sec(2); ...,

boundaryEnd_sec(2) boundaryStart_sec(3); ...,

boundaryEnd_sec(3) boundaryStart_sec(4); ...,

boundaryEnd_sec(4) boundaryStart_sec(5); ...,

boundaryEnd_sec(5) boundaryStart_sec(6); ...,

boundaryEnd_sec(6) boundaryStart_sec(7); ...,

boundaryEnd_sec(7) boundaryStart_sec(8); ...,

boundaryEnd_sec(8) boundaryStart_sec(9); ...,

boundaryEnd_sec(9) boundaryStart_sec(10); ...,

boundaryEnd_sec(10) boundaryStart_sec(11); ...,

boundaryEnd_sec(11) EEG.xmax]);

end

Now I want this input to be automatic since the number of blocks is different for each dataset. pop_select only accepts numbers in brackets separated by ;.

I do this:

str(1) = strcat("[", num2str([EEG.xmin boundaryStart_sec(2)]), "; ");

for k = 2:length(boundaryStart_sec)-1

str(k) = strcat(num2str([boundaryEnd_sec(k) boundaryStart_sec(k+1)]), "; ")

k=k+1;

end

input_str = strjoin(str)

string = str2double(input_str)

But string is NaN since input_str also contains [, ] and ;.

I take full responsibility for the stupidity of this solution, which does not work.

[Cris LaPierre]

You are thinking about this too literally. Don't create a string. Create a numeric array, which is covered in Ch 4 of MATLAB Onramp.

Test the following in your code to convince yourself it works.

input_array = [ EEG.xmin boundaryStart_sec(2);
                boundaryEnd_sec(2) boundaryStart_sec(3);
                boundaryEnd_sec(3) boundaryStart_sec(4);
                boundaryEnd_sec(4) boundaryStart_sec(5);
                boundaryEnd_sec(5) boundaryStart_sec(6);
                boundaryEnd_sec(6) boundaryStart_sec(7);
                boundaryEnd_sec(7) boundaryStart_sec(8);
                boundaryEnd_sec(8) boundaryStart_sec(9);
                boundaryEnd_sec(9) boundaryStart_sec(10);
                boundaryEnd_sec(10) boundaryStart_sec(11);
                boundaryEnd_sec(11) EEG.xmax];
EEG = pop_select( EEG, 'notime', input_array)

And to prove that semicolons are just for creating a new row, you don't even need them if you actually use a linefeed when building your array.

input_array = [ EEG.xmin boundaryStart_sec(2)
                boundaryEnd_sec(2) boundaryStart_sec(3)
                boundaryEnd_sec(3) boundaryStart_sec(4)
                boundaryEnd_sec(4) boundaryStart_sec(5)
                boundaryEnd_sec(5) boundaryStart_sec(6)
                boundaryEnd_sec(6) boundaryStart_sec(7)
                boundaryEnd_sec(7) boundaryStart_sec(8)
                boundaryEnd_sec(8) boundaryStart_sec(9)
                boundaryEnd_sec(9) boundaryStart_sec(10)
                boundaryEnd_sec(10) boundaryStart_sec(11)
                boundaryEnd_sec(11) EEG.xmax];
EEG = pop_select( EEG, 'notime', input_array)

Since boundaryEnd_sec and boundaryStart_sec are already vectors, you could do it even simpler doing this:

input_array = [EEG.xmin boundaryEnd_sec(2:end); boundaryStart_sec(2:end) EEG.xmax]';
EEG = pop_select( EEG, 'notime', input_array)

[Stephen23]

"...pop_select only accepts numbers in brackets separated by ;."

Then why are you converting everything to character?

[Steven Lord]

I believe your pattern is:

% Sample data
boundaryEnd_sec = 1:11

boundaryEnd_sec = 1×11

1 2 3 4 5 6 7 8 9 10 11

boundaryStart_sec = 0.5:10.5

boundaryStart_sec = 1×11

0.5000 1.5000 2.5000 3.5000 4.5000 5.5000 6.5000 7.5000 8.5000 9.5000 10.5000

EEG = struct('xmin', 0, 'xmax', 12);

endings = [EEG.xmin; reshape(boundaryEnd_sec(2:end), [], 1)];
starts = [reshape(boundaryStart_sec(2:end), [], 1); EEG.xmax];
M = [endings, starts]

M = 11×2

0 1.5000 2.0000 2.5000 3.0000 3.5000 4.0000 4.5000 5.0000 5.5000 6.0000 6.5000 7.0000 7.5000 8.0000 8.5000 9.0000 9.5000 10.0000 10.5000

This assumes boundaryEnd_sec and boundaryStart_sec each have at least 2 elements. No string processing necessary. The fact that you're eliminating the first element of both boundaryEnd_sec and boundaryStart_sec feels a little odd; usually I'd expect you'd skip the first element of one and the last element of the other.

[Martin Randau]

I thank you all very much for the assistance. I've learnt something new today.

This code did the trick:

endings = [EEG.xmin; reshape(boundaryEnd_sec(2:end), [], 1)];

starts = [reshape(boundaryStart_sec(2:end), [], 1); EEG.xmax];

deleteData = [endings, starts];

EEG = pop_select( EEG, 'notime', deleteData);

I eliminate the first two items in start and end because the firls block is a training block not included in the analysis. :)

[Cris LaPierre]

Incidentally, that produces exactly the same result as this code I shared earlier:

input_array = [EEG.xmin boundaryEnd_sec(2:end); boundaryStart_sec(2:end) EEG.xmax]';
EEG = pop_select( EEG, 'notime', input_array)

[Martin Randau]

Yes, sorry, I forgot to add that both solutions work! Thanks again! :)