Using 'if' to conduct specific matrix cell operations

I'm trying to use the 'if' function on specific matrix cells to create a new matrix then find the product of that matrix, which will ultimately be used to minimuze a risk function. So far I have Points 'P' and the associated failure 'Pfailgrid', both 1x200 matrices, but regardless of my changes in P (which are binary), y = 0. Any tips would be appreciated, I'm sure I haven't clarified well enough so please let me know if I can do so. Thank you.

function y = DistObj(P)
Pfailgrid = readmatrix('PfailColumn.xlsx'); %1x200
P = readmatrix('Points.xlsx'); %1x200
if P(:,1) <= 1
    P2(:,1) = 1-Pfailgrid(:,1);
else 
    P2(:,1) = 1;
end
y = prod(P2); % makes 1x200 = Ppass
end

Answer 1

[Matt J]

P2=ones(size(P));
subset=(P<1);
P2(subset)=1-Pfailgrid(subset);

[Daniel McDonald]

Matt, I think this will work, my plan is to run this in an fmincon function to solve for P, which is now P2, would that work?

[Matt J]

It might be preferable to minimize the equivalent cost function sum(log(P2)), rather than prod(P2), to avoid overflow/underflow issues.

[Daniel McDonald]

That didn't seem to work. Here's what I've got now:

function y = DistObj(P)
    Pfailgrid = readmatrix('PfailColumn.xlsx'); %1x200
    P = readmatrix('Points.xlsx'); %1x200
    P2=ones(size(P));
    subset=(P<1);
    P2(subset)=1-Pfailgrid(subset);
    y = sum(log(P2)); % makes scalar = Ppass
end

It ends up with -Inf when P = [0;1;1;1;1;0;0;0;...;0] and Pfailgrid=[1;0.09;0.09;0.09;1;1;1;1...;1] which should be resulting in a scalar value between 0 and 1.

[Matt J]

The input P being passed as input to DistObj isn't being used. It is being overwritten in the line,

P = readmatrix('Points.xlsx'); %1x200

[Daniel McDonald]

Matt, thanks a ton so far, this is honestly extremely helpful. I may have a confusion about how to fundamentally make a function (not good) but I'm hoping I've corrected that. Since I'm really trying to send P2 to DistObj, I've added that in instead. I'm not sure that works since I've got P2(subset) as well though and I may still be overwriting it. I also tried DistObj(~) which didn't work either. Do I need to make up some other variable (like x?)?

function y = DistObj(P2)
    Pfailgrid = readmatrix('PfailColumn.xlsx'); %1x200
    P = readmatrix('Points.xlsx'); %1x200
    P2 = ones(size(P));
        subset = (P<1);
    P2(subset)=1-Pfailgrid(subset);
    y = sum(log(P2)); % makes scalar = Ppass
end

This all still returns -Inf so I don't think it's working yet. Thanks for all your help.

[Matt J]

No, the output y still doesn't depend on the input P2 at all. You overwrite the input P2 in this line

P2 = ones(size(P));

and the original values of P2 never affect anything.

[Daniel McDonald]

Matt, I think this fixed it. I definitely had some fundamental errors about making this function and, in a quest for an answer, forgot what I was trying to solve for! By changing y = DistObj(x) and setting P = x(1) I believe that resolves the issue. The code below is what I plan to use in the optimization:

% Objective function - Solving for P
function y = DistObj(x)
    Pfailgrid = readmatrix('PfailColumn.xlsx'); %1x200
    P = x(1); %readmatrix('Points.xlsx'); %1x200
    P2 = ones(size(P));
        subset = (P<1);
    P2(subset)=1-Pfailgrid(subset);
    y = sum(log(P2)); % makes scalar = Ppass
%    y = prod(P2);
end

%Import excel matrices
Dist = readmatrix('Distance.xlsx');
Pfailgrid = readmatrix('Pfail10_20.xlsx');

f = @(x)DistObj(x);

%Call x0
x0 = readmatrix('Points.xlsx');

%upper and lower bounds
lb = readmatrix('lb.xlsx');
ub = readmatrix('ub.xlsx');
nonlcon = @DistNonlCons;

%Run fmincon to find surface points with minimum risk
PointsNew = fmincon(f,x0,[],[],[],[],[],[],[]);
PointsNew;

The second fmincon portion is still a work in progress but hopefully should be smoother! Thanks for your help and please let me know if you've got any additional comments!