"end-1" as parameter

I have a question on MATLAB Onramp Chapter 5.1 Indexing into Arrays, Task 3:

Under what "real" scenario would I use the "end-1" command to find a value?

In the given example of MATLAB Onramp, the task is to find the value in the 6th row and 3rd column. I understand that I can use the keyword "end" to either find a row or column index to reference of the last element, which is quick and straightforward in a dataset, but why would I use the "end-1" as parameter? Doesn't it make it too complicated to find an "unknown" value?

Any further explanation on this is much appreciated.

[John D'Errico]

It would help if you provide a link. Without that or a copy of the code/question, it is difficult to know what you are talking about.

[Steven Lord]

I believe that's part of MATLAB Academy.

Answer 1

[Guillaume]

Well, end-1 is not an unknown location. It's the element before the last. There are plenty of scenarios where you'd want the nth element before the end, so end-n is very useful. e.g.:

• Implementing a LIFO (Last In First Out) stack. When you want to pop the last element from the stack:

function [stack, element] = pop(stack)
   element = stack(end);      %get last element
   stack = stack(1:end-1);  %return the stack without the last element
end

• When you want to remove the border around an image:

function croppedimage = cropborder(sourceimage, bordersize)
   croppedimage = sourceimage(bordersize+1:end-bordersize, bordersize+1:end-bordersize, :);
end

• split a vector into two:

function [firsthalf, secondhalf] = splitmiddle(vector)
    firsthalf = vector(1:ceil(end/2));
    secondhalf = vector(ceil(end/2)+1:end);
end

• etc.

Answer 2

[Steven Lord]

Let's say I wanted to add each element of a vector to the next element. I could do this with a for loop, or I could use a vectorized approach. But I can't add the last element of the vector with the next element -- there isn't a next element! So I add every element of the vector EXCEPT the last (which I retrieve using the index vector 1:end-1) to the one just after that (which I retrieve using the index vector (1:end-1)+1 or 2:end.)

x = 1:10;
y = x(1:end-1) + x(2:end);

We introduced in release R2016a a function named movsum that you could use instead to compute y, but the indexing approach works back probably to Cleve's original Fortran version of MATLAB.

Answer 3

[Chen Lin]

Hi all, thanks for the quick responses.

And John was right, so is Steven's assumption. The original question came from MATLAB Onramp Chapter 5.1, Task 3 (https://matlabacademy.mathworks.com/R2016a/portal.html?course=gettingstarted#chapter=5&lesson=1&section=1).

It was given: _>> load datafile >> data

data =

    3.0000    0.5300    4.0753       NaN
   18.0000    1.7800    6.6678    2.1328
   19.0000    0.8600    1.5177    3.6852
   20.0000    1.6000    3.6375    8.5389
   21.0000    3.0000    4.7243   10.1570
   23.0000    6.1100    9.0698    2.8739
   38.0000    2.5400    5.3002    4.4508_

Task 3: _Info: Note that you can use arithmetic with the keyword end. For example:

>> x = A(end-1,end-2)_

Try creating a scalar variable p that contains the value in the second to last (end-1) row and 3rd column of data.

I didn't understand the meaning of "the second to last" (end-1) row first, but I figured out now. It was easy.

And thanks to Guillaume's practical examples. Perfect gents. Much appreciated!