Subtracting elements with constraints

Hi all,

I have an array A and I need to subtract the elements of each column using the cumsum function:

B=bsxfun(@minus, A(1,:), cumsum(A(2:end,:)));

However I have some constraints that should not violated (min=1.3, max=11.7) Is there any way to do the calculations within these limits?

thanks

[Walter Roberson]

What should happen if it goes outside that range? And since the items might be positive or negative, it could potentially go back into the range again.

[Guillaume]

What does "constraints that should not be violated" actually mean. What should happen to avoid violating these constraints?

[Nikolas Spiliopoulos]

If the constraints are violated it just put the maximum or minimum value respectively! If it's within the range it does the calculations.

Answer 1

[Stephen23]

Your question is not clear where you want these constraints applied.

Option one: constrain output values:

 B = bsxfun(@minus, A(1,:), cumsum(A(2:end,:)));
 B = max(1.3, min(11.7, B));

Options two: constrain cumsum values:

 tmp = max(1.3, min(11.7, cumsum(A(2:end,:))));
 B = bsxfun(@minus, A(1,:), tmp);

[Guillaume]

And note that if you're using R2016b, the bsxfun is not needed

BorTMP = A(1, :) - cumsum(A(2:end, :));  %in R2016b

[Walter Roberson]

Another possibility: that at each point the sum needs to be in that range, and if adding the next element would move it out of the range then clamp it at the edge of the range, but allow the next element to move it away from the edge of the range. For example, if the lower limit was 1, then for

A = 6 2 2 2 2 -3

cummulative

C =  4 2 1 1 4
         ^ ^

the clamping has been applied at the ^ points, but no history of how far outside the range it would go, so the -(-3) moves it 3 away from the bottom of the range

I am not sure at the moment that a vectorized version of this approach is possible.

[Nikolas Spiliopoulos]

thanks a lot, I want to constrain the output values!