splitting up a graph into 2 parts

I'm trying to divide my triangular wave f1 into 2 parts, left and right of the upper point

x = -10:0.1:10;

lower_1 = 0:.005:1;

upper_1 = 0:.005:1;

y2 = 0:.005:1;

f1 = trapmf(x,[-2 0 0 2]);

[a,ix]=max(f1);

   for k = 1:ix
       lower_1(k,:) = [x(k), f1(k)];
   end

   for k1 = ix:201
      upper_1(k1,:) = [x(k1), f1(k1)];
   end

this seems correct to me, but it's giving me a dimension error in the second for loop.

any ideas why?

[Purushottama Rao]

what is ix in your code?

[soloby]

ix is the iteration of my maximum value which is 1

note: [a,ix]=max(f1)

Answer 1

[Guillaume]

I would think you get the error in the first loop. You're trying to put two values, the vector [x(k) f1(k)] into a scalar lower_1(k) (note that the , : in your expression has no effect since lower_1 is a column vector. lower_1(k, :) is the same as lower_1(k, 1) is the same as lower_1(k)).

Whatever you're trying to do, you don't need a loop anyway. If you're trying to create an nx2 matrix of x and f1 split at the max:

x = -10:0.1:10;
lower_1 = zeros(numel(x), 2);
upper_1 = zeros(numel(x), 2);
f1 = trapmf(x,[-2 0 0 2]);
[~, firstmax] = max(f1);
lower_1(1:firstmax, :) = [x(1:firstmax), f1(1:firstmax)];
upper_1(firstmax:end, :) = [x(firstmax:end, :), f1(firstmax:end, :)];

Note that the above assumes that trapmf returns a column vector (I don't have the fuzzy logic toolbox so can't check).

Also note, that nowhere have I hardcoded the size of the vectors (201). I use numel or end, so if it ever changes, there's nothing to change in the code.

[soloby]

I actually get through the first loop just fine but get stuck on the second, but would love to do it without using a loop.

you're right using your code, f1 would be 1x201 but lower_1 and upper_1 are both 201x2, gives me a dimension error. would using something like an inverse be the solution to this?

[Guillaume]

if f1 is a row vector (1x201), then you just need to transpose it:

f1 = trapmf(x, [-2 0 0 2]).'; %note the .' to transpose

[soloby]

x = -10:0.1:10;

lower_1 = zeros(numel(x), 2);

upper_1 = zeros(numel(x), 2);

f1 = trapmf(x,[-2 0 0 2]).';

[~, firstmax] = max(f1);

lower_1(1:firstmax, :) = [x(1:firstmax), f1(1:firstmax)];

upper_1(firstmax:end, :) = [x(firstmax:end, :), f1(firstmax:end, :)];

still giving me an error

"Error using horzcat
Dimensions of matrices being concatenated are not consistent.

Error in (line 6)  
lower_1(1:firstmax, :) = [x(1:firstmax), f1(1:firstmax)];"

did you mean to change f1 to its transverse in the variable statement or the coding?

[Guillaume]

Please, use the code formatting button ( {}Code) rather than putting spaces between lines of code.

Sorry, for some reason I thought that x was a column vector but it's a row vector. Depending on what you want, x first column, f second column or x first row, f second row, you can either transpose x and leave the rest as is, or not transpose f, swap the dimensions of lower and upper and concatenate x and f verticually:

%|x| and |f| as column vectors:
x = (-10:0.1:10)';
%... rest of the code as above

%|x| and |f| as row vectors
x = -10:0.1:10;
lower = zeros(2, numel(x));
upper = zeros(2, numel(x));
f = trapmf(x, [-2 0 0 2]);
[~, firstmax] = max(f1);
lower(1:firstmax) = [x(1:firstmax); f(1:firstmax)]; %note the ; instead of , for concatenation.