Matrix Optimization, optimizing a 336 by 2 matrix, x

Hello,

I'm working on , with certain constraints as specified in the code. Unfortunately I'm encountering a bunch of warnings that i'm just not able to get rid of despite multiple code edits, mentioned below. Any advise would be help full.

CODE

%global variables
global Tch Pch Mch Ad C1Tin Ocd mw1 mw2 mtw SpP ncd OEn NEn C1w C2w Tow EnvAirP PolyCompEff IsenExpo PolyCompExpo CompRatio MaxCP Cp_air HxEff WaterIn WaterOut Cp_water IFChP IChT R PolyTurbEff PolyTurbExpo Pexp Vch C1P C2P C1Tout C2Tin C2Tout HXout;

EnvAirP = 101;
PolyCompEff = 0.85;
IsenExpo = 1.4;
PolyCompExpo = 1.503;
CompRatio = 8;
MaxCP = 18.6;
Cp_air = 1;
HxEff = 0.8;
WaterIn = 293.15;
WaterOut = 368.15;
Cp_water = 4.184;
IFChP = 2000;
IChT = 400;
R = 0.287;
PolyTurbEff = 0.8;
PolyTurbExpo = 1.297;
Pexp = 700;
Vch = 10;
Pch = 0;


%Matrix input variables

%C1Tin - Compressor 1 Inlet air temp
%Amd - Company air mass flow Demand
%Ocd - Overall Old Company Power Demand
%Ad - Air Density


x0 = [0,0]; 

lb = [0,0];

ub = [];

x = patternsearch(@fun,x0,[],[],[],[],lb,ub,@nlinconst);


C1P = EnvAirP*CompRatio;
C2P = C1P*CompRatio;
C1Tout(:,1) = C1Tin(:,1)*(CompRatio^(PolyCompExpo/(PolyCompExpo - 1)));
C2Tin(:,1) = WaterIn + HxEff*(WaterIn - C1Tout(:,1));
C2Tout(:,1) = C2Tin(:,1)*CompRatio^(PolyCompExpo/(PolyCompExpo - 1));
HXout(:,1) = WaterIn + HxEff*(WaterIn - C2Tout(:,1));

mw1(:,1) = x(:,1).*Cp_air.*(C1Tout(:,1) - C2Tin(:,1))./(Cp_air*(WaterOut - WaterIn));
mw2(:,1) = x(:,1).*Cp_air.*(C2Tout(:,1) - HXout(:,1))./(Cp_air*(WaterOut - WaterIn));
mtw(:,1) = mw1(:,1) + mw2(:,1);
SpP(:,1) = (PolyCompEff^(-1))*Cp_air.*C1Tin(:,1).*(CompRatio^(PolyCompExpo/(PolyCompExpo - 1)));
ncd(:,1) = C1w(:,1) + C2w(:,1) - Tow(:,1) + Ocd(:,1) - SpP(:,1).*x(:,2)./Ad(:,1).*2118.9;
OEn = sum(Ocd(:,1),'all')*0.5;
NEn = sum(ncd(:,1),'all')*0.5;

for i = 1:336
        
    Tch(i+1,3) = Tch(i,3) + (Mch(i,1)^-1).*(1-IsenExpo^-1).*(x(i,1).*HXout(i,1) - x(i,2).*Tch(i+1,1)).*1800;
    
end





function [c,ceq] = nlinconst(x)  


c(:,1) = OEn - NEn;
c(:,2) = 700 - Pch(:,1);
c(:,3) = x(:,1) - 0.2;
c(:,4) = x(:,2) - 0.2;
c(:,5) = Pch(:,1) - 6464;
c(:,6) = C1w(:,1) - 18.6;
c(:,7) = C2w(:,1) - 18.6;
c(:,8) = Mch(:,1) - 350;




ceq(1,1) = Tch(1,1) - IChT;
ceq(1,2) = Pch(1,1) - IFChP;
ceq(336,2) = Pch(336,1) - IFChP;
ceq(:,3) = C1w(:,1) - (PolyCompEff^-1).*x(:,1).*Cp_air.*C2Tin.*(CompRatio^(PolyCompExpo/(PolyCompExpo - 1)) - 1);
ceq(:,4) = Pch(:,1).*Vch./(R.*Tch(:,1)) - Mch(:,1);


ceq(:,6) = PolyTurbEff.*x(:,2).*Cp_air.*Tch(:,1).*(1-(Pch(:,1)./Pexp).^((PolyTurbExpo - 1)./PolyTurbExpo));
ceq(:,7) = EnvAirP.*CompRatio - C1P ;
ceq(:,8) = C1P(:,1).*CompRatio - C2P;
ceq(:,9) = C1Tin(:,1).*CompRatio^(PolyCompExpo./(PolyCompExpo - 1)) - C1Tout(:,1) ;
ceq(:,10) = WaterIn + HxEff.*(WaterIn - C1Tout(:,1)) - C2Tin(:,1);
ceq(:,11) = C2Tin(:,1).*CompRatio^(PolyCompExpo./(PolyCompExpo - 1)) - C2Tout(:,1) ;
ceq(:,12) = WaterIn + HxEff.*(WaterIn - C2Tout(:,1)) - HXout(:,1);

ceq(:,13) = x(:,1).*Cp_air.*( C1Tout(:,1) - C2Tin(:,1))./(Cp_water.*(WaterOut - WaterIn)) - mw1(:,1) ;
ceq(:,14) = x(:,1).*Cp_air.*(C2Tout(:,14) - HXout(:,1))./(Cp_air.*(WaterOut - WaterIn)) - mw2(:,1) ;
ceq(:,15) = mw1(:,1) + mw2(:,1) - mtw(:,1) ;
ceq(:,16) = (Mch(:,1) + (x(:,1) - x(:,2) ).*1800).*R.*x(:,1)./Vch - Pch(:,1);



end



function y = fun(x)

ncd(:,1) = C1w(:,1) + C2w(:,1) - Tow(:,1) + Ocd(:,1) - SpP(:,1).*x(:,2)./Ad(:,1).*2118.9;

y = max( ncd(:,1));

end

Errors and Warnings

Warning: The value of local variables may have been changed to match the globals. Future versions of MATLAB will require that you declare a variable to be

global before you use that variable.

> In test_4 (line 3)

Warning: The value of local variables may have been changed to match the globals. Future versions of MATLAB will require that you declare a variable to be

global before you use that variable.

> In test_4 (line 3)

Warning: The value of local variables may have been changed to match the globals. Future versions of MATLAB will require that you declare a variable to be

global before you use that variable.

> In test_4 (line 3)

Warning: The value of local variables may have been changed to match the globals. Future versions of MATLAB will require that you declare a variable to be

global before you use that variable.

> In test_4 (line 3)

Unrecognized function or variable 'EnvAirP'.

Error in test_4>fun (line 91)

C1P = EnvAirP*CompRatio;

Error in funevaluate (line 54)

f = feval(FUN,reshapeinput(Xin,X),varargin{:});

Error in poptimfcnchk (line 23)

[y,count,cache] = funevaluate(FUN,Xin,X,'init',cache,[],[],objFcnArg{:});

Error in patternsearch (line 262)

[Iterate,OUTPUT.funccount] = poptimfcnchk(FUN,nonlcon,X0,Iterate, ...

Error in test_4 (line 39)

x = patternsearch(@fun,x0,[],[],[],[],lb,ub,@nlinconst);

Caused by:

Failure in initial user-supplied objective function evaluation. PATTERNSEARCH cannot continue.

[Sargondjani]

You only assign values to variables like C1P after calling patternsearch, so basically C1P has no vlue when you start optimization. You need to put the block below patternsearch above it.

Furthermore, I would recommend against using global variables. I usually put all my parameter in a structure, with each variable being a field:

par.C1P = EnvAirP*CompRatio;

etc. and then pass "par' as an argument in your functions.

[Ebin Daniel]

Thank you so much, both your suggetions were helpful. I'm still new to structure so I just bundled all the variables in the X variable and passed it into the non-lin constraint function. I'm continuing to debug and had to update my code, and now am dealing with another set of errors, again any help would be appreciated.

Code

EnvAirP = 101;
PolyCompEff = 0.85;
IsenExpo = 1.4;
PolyCompExpo = 1.503;
CompRatio = 8;
MaxCP = 18.6;
Cp_air = 1;
HxEff = 0.8;
WaterIn = 293.15;
WaterOut = 368.15;
Cp_water = 4.184;
IFChP = 2000;
IChT = 400;
R = 0.287;
PolyTurbEff = 0.8;
PolyTurbExpo = 1.297;
Pexp = 700;
Vch = 10;



%Matrix input variables

%C1Tin - Compressor 1 Inlet air temp
%Amd - Company air mass flow Demand
%Ocd - Overall Old Company Power Demand
%Ad - Air Density


x = ones(336,2);
Pch = ones(336,1);
Mch = ones(336,1);
Tch = ones(336,1);



C1P = EnvAirP*CompRatio;
C2P = C1P*CompRatio;
C1Tout(:,1) = C1Tin(:,1)*(CompRatio^(PolyCompExpo/(PolyCompExpo - 1)));
C2Tin(:,1) = WaterIn + HxEff*(WaterIn - C1Tout(:,1));
C2Tout(:,1) = C2Tin(:,1)*CompRatio^(PolyCompExpo/(PolyCompExpo - 1));
HXout(:,1) = WaterIn + HxEff*(WaterIn - C2Tout(:,1));

C1w(:,1) = (PolyCompEff^-1).*x(:,1).*Cp_air.*C1Tin.*(CompRatio^(PolyCompExpo/(PolyCompExpo - 1)) - 1);
C2w(:,1) = (PolyCompEff^-1).*x(:,1).*Cp_air.*C2Tin.*(CompRatio^(PolyCompExpo/(PolyCompExpo - 1)) - 1);
Pch(:,1) = (Mch(:,1) + (x(:,1) - x(:,2) ).*1800).*R.*x(:,1)./Vch;
Mch(:,1) = Pch(:,1).*Vch./(R.*Tch(:,1));

Tow(:,1) = PolyTurbEff.*x(:,2).*Cp_air.*Tch(:,1).*(1-(Pch(:,1)./Pexp).^((PolyTurbExpo - 1)./PolyTurbExpo));

mw1(:,1) = x(:,1).*Cp_air.*(C1Tout(:,1) - C2Tin(:,1))./(Cp_air*(WaterOut - WaterIn));
mw2(:,1) = x(:,1).*Cp_air.*(C2Tout(:,1) - HXout(:,1))./(Cp_air*(WaterOut - WaterIn));
mtw(:,1) = mw1(:,1) + mw2(:,1);
SpP(:,1) = (PolyCompEff^(-1))*Cp_air.*C1Tin(:,1).*(CompRatio^(PolyCompExpo/(PolyCompExpo - 1)));
ncd(:,1) = C1w(:,1) + C2w(:,1) - Tow(:,1) + Ocd(:,1) - SpP(:,1).*x(:,2)./Ad(:,1).*2118.9;
OEn = sum(Ocd(:,1),'all')*0.5;
NEn = sum(ncd(:,1),'all')*0.5;

for i = 1:335
        
    Tch(i+1,1) = Tch(i,1) + (Mch(i,1)^-1).*(1-IsenExpo^-1).*(x(i,1).*HXout(i,1) - x(i,2).*Tch(i+1,1)).*1800;
    
end


x0 = [0,0]; 

lb = [0,0];

ub = [];

x = patternsearch(@fun,x0,[],[],[],[],lb,ub,@nlinconst);

x(:,3) = C1w(:,1);
x(:,4) = C2w(:,1);
x(:,5) = Mch(:,1);
x(:,6) = Tch(:,1);
x(:,7) = IChT;
x(:,8) = Pch(:,1);
x(:,9) = IFChP;



function [c,ceq] = nlinconst(x)  


c(:,1) = 700 - x(:,8);
c(:,2) = x(:,1) - 0.2;
c(:,3) = x(:,2) - 0.2;
c(:,4) = x(:,8) - 6464;
c(:,5) = x(:,3) - 18.6;
c(:,6) = x(:,4) - 18.6;
c(:,7) = x(:,5) - 350;
ceq(1,1) = x(1,6) - x(:,7);
ceq(1,2) = x(1,8) - x(:,9);
ceq(336,2) = x(336,8) - x(:,9);


end



function y = fun(ncd)


y = max( ncd(:,1));

end

Errors

Index in position 2 exceeds array bounds (must not exceed 2).

Error in test_4>nlinconst (line 90)

c(:,1) = 700 - x(:,8);

Error in poptimfcnchk (line 64)

[cineq,ceq] = feval(nonlcon,reshapeinput(Xin,X),conFcnArg{:});

Error in patternsearch (line 262)

[Iterate,OUTPUT.funccount] = poptimfcnchk(FUN,nonlcon,X0,Iterate, ...

Error in test_4 (line 75)

x = patternsearch(@fun,x0,[],[],[],[],lb,ub,@nlinconst);

Caused by:

Failure in initial user-supplied nonlinear constraint function evaluation. PATTERNSEARCH cannot continue.

[Walter Roberson]

x0 = [0,0]; 

That is a vector of length 2.

x = patternsearch(@fun,x0,[],[],[],[],lb,ub,@nlinconst);

and that vector of length 2 is being used as the initial location for searching, so that tells patternsearch() that there are two variables being optimized over.

ceq(336,2) = x(336,8) - x(:,9);

That line tells us that your nonlinear constraints think that you have at least 336*8 = 2688 variables rather than just 2 variables.

When you pass in a row vector of length 2 as your x0, then that row vector will not be automatically copied as often as needed to act as initial conditions for the optimization. You need to provide an x0 that has exactly as many elements as you are optimizing over.

c(:,1) = 700 - x(:,8);

Your code is expecting a 2D array with at least 8 columns. However:

https://www.mathworks.com/help/gads/patternsearch.html#buxdit7-x0

Initial point, specified as a real vector. patternsearch uses the number of elements in x0 to determine the number of variables that fun accepts.

So what your objective function will receive and what your nonlinear constraint function will receive, will be a vector, not an array.

ceq(1,1) = x(1,6) - x(:,7);

If you did have an array of x (perhaps you reshape() into an array as the first step in the routine, which is absolutely a valid thing to do), then x(:,7) would be a vector of length 336, and subtracting that from x(1,6) would give a vector of length 336... which you are trying to store inside the scalar location ceq(1,1)

ceq(1,2) = x(1,8) - x(:,9);

Assuming you changed that to

ceq(:,2) = x(1,8) - x(:,9);

then because those are equality constraints, that forces all members of x(:,9) to be exactly equal to x(1,8) . But that is a linear equality constraint that you should express through the Aeq, beq matrix rather than through the nonlinear constraint function.

c(:,1) = 700 - x(:,8);
c(:,2) = x(:,1) - 0.2;

Those are linear inequalities that should be expressed through the A, b matrix rather than through the nonlinear constraints function.

Linear constraints are much more efficient than nonlinear constraints.

[Ebin Daniel]

Thank you Walter, i'll be working on the x0 parameter to reflect the needed number of variables.

As for the ceq constraints i believe i made an error before hand,

I meant to have the following as my code

ceq(1,2) = 2000 - x(1,8);

ceq(336,2) = 2000 - x(336,8);

I wanted to make sure that the first and last values of x(:,8) column were to remain 2000

[Walter Roberson]

To force 2000 use lb / ub. As a general rule, lb/ub is the most efficient, then linear inequality, then linear equality, and nonlinear inequality is much less efficient, with nonlinear equality being expensive. For efficiency, you should code any given constraint for fastest execution.

[Ebin Daniel]

I want each cell potential value within the x(;,8) column matrix to have an ub of 6464 and an lb of 700, which i tried to set up as

c(:,2) = 700 - Pch(:,1);

c(:,5) = Pch(:,1) - 6464;

In the code above.

I just need the first and last cell values (i.e. x(1,8) and x(336,8)) of the x(:,8) matrix to be 2000, i.e. (x(1,8) = x(336,8) = 2000)

[Walter Roberson]

I recommend against doing that through the nonlinear constraints.

x0 = zeros(336, 8);
lb = zeros(336,8);
ub = inf(336,8);
lb(:,8) = 700;
lb(1,8) = 2000;
lb(336,8) = 2000;
ub(:,8) = 6464;
ub(1,8) = 2000;
ub(336,8) = 2000;

A = []; b = [];
Aeq = []; beq = [];
x = patternsearch(@fun, x0(:), A, b, Aeq, beq, lb(:), ub(:), @nlinconst);

and leave out those linear constraints such as 700-Pch(:,10 from your nonlinear constraint function.

Answer 1

[Ebin Daniel]

I essentially want to optimize 2 column matrices x(:,1) and x(:,2) and have constraints that are dependant on function valus as stated below, I have made a few changes to the code and want to see if my syntax matches the logic i am trying to achieve.

C1Tin is a 336 by 1 matrix

%Matrix input variables that will be imported

%C1Tin - Compressor 1 Inlet air temp


x0 = zeros(336,2); 

lb = [0,0];

ub = [0.2, 0.2];

x = patternsearch(@fun,x0,[],[],[],[],lb,ub,@nlinconst);



function [c,ceq] = nlinconst(C1Tin, x0)

[c,ceq] = nlin(@const, x0);

 function [c,ceq] = const(x)  
    
PolyCompEff = 0.85;
IsenExpo = 1.4;
PolyCompExpo = 1.503;
CompRatio = 8;
Cp_air = 1;
HxEff = 0.8;
WaterIn = 293.15;
Pch(1,1) = 2000;
Tch(1,1) = 400;
R = 0.287;
Vch = 10;

C1Tout(:,1) = C1Tin(:,1)*(CompRatio^(PolyCompExpo/(PolyCompExpo - 1)));
C2Tin(:,1) = WaterIn + HxEff*(WaterIn - C1Tout(:,1));
C2Tout(:,1) = C2Tin(:,1)*CompRatio^(PolyCompExpo/(PolyCompExpo - 1));
HXout(:,1) = WaterIn + HxEff*(WaterIn - C2Tout(:,1));

C1w(:,1) = (PolyCompEff^-1).*x(:,1).*Cp_air.*C1Tin.*(CompRatio^(PolyCompExpo/(PolyCompExpo - 1)) - 1);
C2w(:,1) = (PolyCompEff^-1).*x(:,1).*Cp_air.*C2Tin.*(CompRatio^(PolyCompExpo/(PolyCompExpo - 1)) - 1);

Mch(1,1) = IFChP*Vch/(R*Tch(1,1));

for i = 1:335
    Tch(i+1,1) = Tch(i,1) + (Mch(i,1)^-1).*(1-IsenExpo^-1).*(x(i,1).*HXout(i,1) - x(i,2).*Tch(i,1)).*1800;
    Pch(i+1,1) = (Mch(i,1) + (x(i,1) - x(i,2) ).*1800).*R.*x(i,1)./Vch;
    Mch(i+1,1) = Pch(i,1).*Vch./(R.*Tch(i,1));
end

  

c(:,1) = 700 - Pch(:,1);
c(:,2) = Pch(:,1) - 6464;
c(:,3) = C1w(:,1) - 18.6;
c(:,4) = C2w(:,1) - 18.6;
c(:,6) = Mch(:,1) - 350;


ceq(1,1) = 400 - Tch(1,1);
ceq(1,2) = 2000 - Pch(1,1);
ceq(336,2) = 2000 - Pch(336,1);


end

end


function y = fun(ncd)


y = max( ncd(:,1));

end

The following is the errors I have recieved

Not enough input arguments.

Error in test_4>nlinconst (line 80)

[c,ceq] = nlin(@const, x0);

Error in poptimfcnchk (line 64)

[cineq,ceq] = feval(nonlcon,reshapeinput(Xin,X),conFcnArg{:});

Error in patternsearch (line 262)

[Iterate,OUTPUT.funccount] = poptimfcnchk(FUN,nonlcon,X0,Iterate, ...

Error in test_4 (line 74)

x = patternsearch(@fun,x0,[],[],[],[],lb,ub,@nlinconst);

Caused by:

Failure in initial user-supplied nonlinear constraint function evaluation. PATTERNSEARCH cannot continue.

[Walter Roberson]

x = patternsearch(@fun, x0, [], [],[],[], lb, ub, @(T1cin) nlinconst(T1Cin, x0));

However, your nlconst function ignores T1Cin, and I am having difficulty coming up with a reason why you would want to pass in the initial values x0 to your nonlinear constraints.

[Ebin Daniel]

You're right, I wouldn't have to. I'll get rid of the x0.

Thanks

[Ebin Daniel]

I tried a nested function so i could pass multiple variables to the constraint function easily with patternsearch within the parent function and I beleiev I'm closer to the solution and have just a few hitches.

nlinconst(C1Tin, Ocd, Ad)


function nlinconst(C1Ti, Od ,A)


PolyCompEff = 0.85;
IsenExpo = 1.4;
PolyCompExpo = 1.503;
CompRatio = 8;
Cp_air = 1;
HxEff = 0.8;
WaterIn = 293.15;
IFChP = 2000;
R = 0.287;
PolyTurbEff = 0.8;
PolyTurbExpo = 1.297;
Pexp = 700;
Vch = 10;
Tch(1,1) = 400;
Pch(1,1) = 2000;


x0 = zeros(336,2); 

lb = [0,0];

ub = [0.2, 0.2];

x = patternsearch(@fun,x0,[],[],[],[],lb,ub,@const);



 function [c,ceq] = const(x)  
    

C1Tout(:,1) = C1Ti(:,1)*(CompRatio^(PolyCompExpo/(PolyCompExpo - 1)));
C2Tin(:,1) = WaterIn + HxEff*(WaterIn - C1Tout(:,1));
C2Tout(:,1) = C2Tin(:,1)*CompRatio^(PolyCompExpo/(PolyCompExpo - 1));
HXout(:,1) = WaterIn + HxEff*(WaterIn - C2Tout(:,1));

C1w(:,1) = (PolyCompEff^-1).*x(:,1).*Cp_air.*C1Ti.*(CompRatio^(PolyCompExpo/(PolyCompExpo - 1)) - 1);
C2w(:,1) = (PolyCompEff^-1).*x(:,1).*Cp_air.*C2Tin.*(CompRatio^(PolyCompExpo/(PolyCompExpo - 1)) - 1);

Mch(1,1) = IFChP*Vch/(R*Tch(1,1));

for i = 1:335
    Tch(i+1,1) = Tch(i,1) + (Mch(i,1)^-1).*(1-IsenExpo^-1).*(x(i,1).*HXout(i,1) - x(i,2).*Tch(i,1)).*1800;
    Pch(i+1,1) = (Mch(i,1) + (x(i,1) - x(i,2) ).*1800).*R.*x(i,1)./Vch;
    Mch(i+1,1) = Pch(i,1).*Vch./(R.*Tch(i,1));
end

  

c(:,1) = 700 - Pch(:,1);
c(:,2) = Pch(:,1) - 6464;
c(:,3) = C1w(:,1) - 18.6;
c(:,4) = C2w(:,1) - 18.6;
c(:,6) = Mch(:,1) - 350;


ceq(1,1) = 400 - Tch(1,1);
ceq(1,2) = 2000 - Pch(1,1);
ceq(336,2) = 2000 - Pch(336,1);


 end




function y = fun(x)

C1Tout(:,1) = C1Ti(:,1)*(CompRatio^(PolyCompExpo/(PolyCompExpo - 1)));
C2Tin(:,1) = WaterIn + HxEff*(WaterIn - C1Tout(:,1));



C1w(:,1) = (PolyCompEff^-1).*x(:,1).*Cp_air.*C1Ti.*(CompRatio^(PolyCompExpo/(PolyCompExpo - 1)) - 1);
C2w(:,1) = (PolyCompEff^-1).*x(:,1).*Cp_air.*C2Tin.*(CompRatio^(PolyCompExpo/(PolyCompExpo - 1)) - 1);


Tow(:,1) = PolyTurbEff.*x(:,2).*Cp_air.*Tch(:,1).*(1-(Pch(:,1)./Pexp).^((PolyTurbExpo - 1)./PolyTurbExpo));
SpP(:,1) = (PolyCompEff^(-1))*Cp_air.*C1Ti(:,1).*(CompRatio^(PolyCompExpo/(PolyCompExpo - 1)));
ncd(:,1) = C1w(:,1) + C2w(:,1) - Tow(:,1) + Od(:,1) - SpP(:,1).*x(:,2)./A(:,1).*2118.9;
    
    
y = max( ncd(:,1));

end

end

Errors

Error using poptimfcnchk (line 118)

Constraint function must return real value.

Error in patternsearch (line 262)

[Iterate,OUTPUT.funccount] = poptimfcnchk(FUN,nonlcon,X0,Iterate, ...

Error in test_4>nlinconst (line 97)

x = patternsearch(@fun,x0,[],[],[],[],lb,ub,@const);

Error in test_4 (line 68)

nlinconst(C1Tin, Ocd, Ad)

[Walter Roberson]

Put in a conditional breakpoint on the end statement of the const function. Make the condition

~all(isfinite(c)) || ~all(isfinite(ceq)) || any(imag(c) ~= 0) || any(imag(ceq) ~= 0)

[Ebin Daniel]

would i have to use a seperate function like the 'dbstop'? Also I'm unsure if I'm calling the nlinconst function right.

[Walter Roberson]

No, you would not have to use a seperate function. Inside your const function, your last line of code is

ceq(336,2) = 2000 - Pch(336,1);

and the "end" statement that is after that is considered to be part of the same function. In the editor, right-click on the circle that is between the line number and the text area and select Set Conditional Breakpoint . In the box that comes up, put in

~all(isfinite(c)) || ~all(isfinite(ceq)) || any(imag(c) ~= 0) || any(imag(ceq) ~= 0)

and then click OK. Then run your code. If you still get the error without the breakpoint firing, then my analysis of your code is not correct.

You are calling the nonlinear constraint function correctly.