FMINCON requires the following inputs to be of data type double: 'X0'.

I am getting the subject mentioned error in matlab code... i spent consecutive 3 days to recitify and solve but not succesfful uptil now... It is requested to help... Many thanks inadvance.... Followis is the code that i use....

options = optimoptions('fmincon','Display','iter','Algorithm','sqp');
problem.options = options;
problem.solver = 'fmincon';

[NumRSUs,MuRSU,MaxNumtk,E,DPrc,pc,Muc,Mui,Muk,B,pi,pm,W,Wfog,t,T,cdk,j,i,TasksperVehicle,NTotal,Numv,Numtk,TijPrime,loop1,Fprime,PhikPrime,D,F,Ravailk,Phik,theta1,matrixforAlgo3] = FCVECInitializeParameters08November2021V5();
% StoreInitialization = struct(NumRSUs,MuRSU,MaxNumtk,E,DPrc,pc,Muc,Mui,Muk,B,pi,pm,W,Wfog,t,T,cdk,j,i,TasksperVehicle,NTotal,Numv,Numtk,TijPrime,loop1,Fprime,PhikPrime,D,F,Ravailk,Phik,theta1,matrixforAlgo3);
global StoreInitialization;
StoreInitialization = struct('NumRSUs_struct',NumRSUs,'MuRSU_struct',MuRSU,'MaxNumtk_struct',MaxNumtk,'E_struct',E,'DPrc_struct',DPrc,'pc_struct',pc,'Muc_struct',Muc,'Mui_struct',Mui,'Muk_struct',Muk,'B_struct',B,'pi_struct',pi,'pm_struct',pm,'W_struct',W,'Wfog_struct',Wfog,'t_struct',t,'T_struct',T,'cdk_struct',cdk,'j_struct',j,'i_struct',i,'TasksperVehicle_struct',TasksperVehicle,'NTotal_struct',NTotal,'Numv_struct',Numv,'Numtk_struct',Numtk,'TijPrime_struct',TijPrime,'loop1_struct',loop1,'Fprime_struct',Fprime,'PhikPrime_struct',PhikPrime,'D_struct',D,'F_struct',F,'Ravailk_struct',Ravailk,'Phik_struct',Phik,'theta1_struct',theta1,'matrixforAlgo3_struct',matrixforAlgo3);
% NumRSUs,MuRSU,MaxNumtk,E,DPrc,pc,Muc,Mui,Muk,B,pi,pm,W,Wfog,t,T,cdk,j,i,TasksperVehicle,NTotal,Numv,Numtk,TijPrime,loop1,Fprime,PhikPrime,D,F,Ravailk,Phik,theta1,matrixforAlgo3;
D_Cloud = zeros(1,Numtk);
F_Cloud = zeros(1,Numtk);
theta1 = Algorithm1_24December2021V8(Numv,Numtk,Ravailk,i,j,F,theta1);
N_Total = Numv;
%                             non-orthogonal multiple access (NOMA) technique, the communication resource is divided into resource
L = 10*N_Total;               % blocks (RBs), expressed as B1 = {1,2,.....L} for uplink and
M = 10*N_Total;               % B2 = {1,2,.....,M} for downlink.
[Tctotal,Ttotal,tTijtoRij,ttransij,texeij,tkij] = FCVEC4AllOutputs05January2022V3(D_Cloud,DPrc,pc,Muc,Mui,Muk,B,pi,pm,W,Wfog,t,T,cdk,j,i,TasksperVehicle,NTotal,Numv,Numtk,TijPrime,loop1,Fprime,PhikPrime,D,F,Ravailk,Phik,theta1,matrixforAlgo3);
for iloopagn12 = 1:1:Numtk           
    if (theta1(1,iloopagn12) == 1)                 % Only those tasks that are to be offloaded to cloud-server are segregated....
        D_Cloud(1,iloopagn12) = D(1,iloopagn12); 
    end
end
[ElapsedTime,Ttotal,matrixforAlgo3,MatTkAllocVehsLM,summation2] = Algorithm3_04January2022V6(NumRSUs,MuRSU,L,M,DPrc,pc,Muc,Mui,Muk,B,pi,pm,W,Wfog,t,T,cdk,j,i,TasksperVehicle,NTotal,Numv,Numtk,TijPrime,loop1,Fprime,PhikPrime,D,F,Ravailk,Phik,theta1,matrixforAlgo3,Ttotal,tTijtoRij,ttransij,texeij,Tctotal);
tic
theta = optimvar('theta',Numtk','Type','integer','LowerBound',0,'UpperBound',1); 
ttransij1 = optimvar('ttransij1',Numtk,'Type','continuous','LowerBound',min(ttransij),'UpperBound',max(ttransij)); 
texeij1 = optimvar('texeij1',Numtk,'Type','continuous','LowerBound',min(texeij),'UpperBound',max(texeij)); 
rng(0) % For reproducibility
% OptimizationConstraints
constr = optimconstr(Numtk,1);
X0.theta = theta1;
X0.ttransij1 = ttransij;
X0.texeij1 = texeij;
for (loop1 = 1:1:Numtk)
    if(matrixforAlgo3(loop1,4) == 0)
        D_Cloud(1,loop1) = D(1,loop1);
        F_Cloud(1,loop1) = F(1,loop1);
    end
end
fun = @(X0,StoreInitialization,ttransij1,texeij1,theta,D_Cloud,F_Cloud,DPrc,pc,Muc,Mui,Muk,B,pi,pm,W,Wfog,t,T,cdk,j,i,TasksperVehicle,NTotal,Numv,Numtk,TijPrime,loop1,Fprime,PhikPrime,D,F,Ravailk,Phik,theta1,matrixforAlgo3)FCVEC06January2022V2(X0,StoreInitialization,ttransij1,texeij1,theta,D_Cloud,F_Cloud,DPrc,pc,Muc,Mui,Muk,B,pi,pm,W,Wfog,t,T,cdk,j,i,TasksperVehicle,NTotal,Numv,Numtk,TijPrime,loop1,Fprime,PhikPrime,D,F,Ravailk,Phik,theta1,matrixforAlgo3);

nonlcon1 = @(Numtk,constr,ttransij1,texeij1,cdk)nlcons(Numtk,constr,ttransij1,texeij1,cdk);
problem.nonlcon = nonlcon1(Numtk,constr,ttransij1,texeij1,cdk);                  % Non-linear constraint....
problem.objective = fun(X0,StoreInitialization,ttransij1,texeij1,theta,D_Cloud,F_Cloud,DPrc,pc,Muc,Mui,Muk,B,pi,pm,W,Wfog,t,T,cdk,j,i,TasksperVehicle,NTotal,Numv,Numtk,TijPrime,loop1,Fprime,PhikPrime,D,F,Ravailk,Phik,theta1,matrixforAlgo3);                                                        % Objective function.....
problem.x0 = X0;
x = fmincon(problem);
% x = fmincon(fun,x0,A,b,Aeq,beq,lb,ub,nonlcon)

Answer 1

[Alan Weiss]

Without testing anything, I see a problem in this line of code:

fun = @(X0,StoreInitialization,ttransij1,texeij1,theta,D_Cloud,F_Cloud,DPrc,pc,Muc,Mui,Muk,B,pi,pm,W,Wfog,t,T,cdk,j,i,TasksperVehicle,NTotal,Numv,Numtk,TijPrime,loop1,Fprime,PhikPrime,D,F,Ravailk,Phik,theta1,matrixforAlgo3)FCVEC06January2022V2(X0,StoreInitialization,ttransij1,texeij1,theta,D_Cloud,F_Cloud,DPrc,pc,Muc,Mui,Muk,B,pi,pm,W,Wfog,t,T,cdk,j,i,TasksperVehicle,NTotal,Numv,Numtk,TijPrime,loop1,Fprime,PhikPrime,D,F,Ravailk,Phik,theta1,matrixforAlgo3);
...
problem.objective = fun(X0,StoreInitialization,ttransij1,texeij1,theta,D_Cloud,F_Cloud,DPrc,pc,Muc,Mui,Muk,B,pi,pm,W,Wfog,t,T,cdk,j,i,TasksperVehicle,NTotal,Numv,Numtk,TijPrime,loop1,Fprime,PhikPrime,D,F,Ravailk,Phik,theta1,matrixforAlgo3); 

For fmincon your objective function must be a function of one variable only. You can have

fun = @(x)otherfunction(x,stuff,otherstuff)

This is a function of x alone. Here, x can be an array or vector of any size, but it is just one argument, not multiple arguments.

If you want to use a bunch of named arguments in your code, then I suggest that you take a serious look at the Problem-Based Optimization Workflow.

Alan Weiss

MATLAB mathematical toolbox documentation

[muhammad ilyas khattak khattak]

@Alan Weiss☺☺...Thanks, Alan once again...I exactly did by encapsulating the parameters into a structure and then calling "fmincon"...Below are the final fmincon commands/statments.... Also in the attachment in pdf format (Convertable/can be copied to .m format) , is the full code that contains all the helping functions/main functions in case it is needed to run.....If I get successful execution/obtain minimization of what I am doing, a great part of my work will be solved... Therefore, I am trying to contact you again and again..... I am literally lacking words to say thanks... Thanks once again in advance, even if you spare a little time, it is much for me........

fun = @(x)FCVEC06January2022V2(x,StoreInitialization);
ttransij1 = StoreInitialization.ttransij_struct;
texeij1 = StoreInitialization.texeij_struct;
cdk = StoreInitialization.cdk_struct;
ncon = @(x)nlcons(x,Numtk,constr,ttransij1,texeij1,cdk);
x = fmincon(fun, X0, A, b, Aeq, beq, lb, ub, ncon, options);

[Alan Weiss]

You have the following lines in your code:

X0.theta = theta1;
X0.ttransij1 = ttransij;
X0.texeij1 = texeij;

These lines make X0 a structure, not a double. You need X0 to represent the starting value of x, where x is a vector or possibly an array, not a structure.

(By the way, this is the right thing to do in the problem-based approach, but since you do not have optimization variables in your problem and you are calling fmincon, you are not using the problem-based approach.)

I do not know whether you have theta, for example, as a variable that you want to incorporate as part of x, meaning a variable that you want to change to reach an optimum. If so, then you would probably be better served by using the problem-based approach.

But if you don't want to change so much of your code, then instead you need to figure out exacty which indices in X correspond to theta1, which correspond to ttransij, and which correspond to texeij. For example, if theta1 is a 20-by-40 array, ttransij is a vector of length 30, and texeij is a 50-by-5 array, then you could have:

• theta1 is X(1:800)
• ttransij is X(801:830)
• texeij is X(831:1080)

No structures, just indices corresponding to the optimization quantities.

Good luck,

Alan Weiss

MATLAB mathematical toolbox documentation

[muhammad ilyas khattak khattak]

@Alan Weiss...Thanks alan once again... I am lacking words to say thanks as your guidance has streamlined my code on a possible path to get near to a solution and to an error-free attempt..... At the moment, I am getting one another error after encapsulating all my parameters in one single "x" vector... For sake of simplicity, I am attaching my edited/updated code (thought long code, infact very long code)...and the error that I recive is ........

K>> x = fmincon(fun(x), X0, A, b, Aeq, beq, lb, ub, ncon(x,Numtk,constr,ttransij1,texeij1,cdk), options);

Error using AllinOne14January2022V1>nlcons (line 2818)

Conversion to OptimizationConstraint from logical is not possible.

Error in

AllinOne14January2022V1>@(x,Numtk,constr,ttransij1,texeij1,cdk)nlcons(x,Numtk,constr,ttransij1,texeij1,cdk)

(line 976)

ncon = @(x,Numtk,constr,ttransij1,texeij1,cdk)nlcons(x,Numtk,constr,ttransij1,texeij1,cdk);

I am pretty hopful that your possible guidance again will put me on a good/right path/direction one step closer to an error-free attempt... Thanks once again in advance

[Alan Weiss]

Instead of trying to read your very long code, I will simply comment on the issues that I see in the ncon argument.

You write

ncon = @(x,Numtk,constr,ttransij1,texeij1,cdk)nlcons(x,Numtk,constr,ttransij1,texeij1,cdk);

This is not legal syntax. The ncon argument, exactly as the fun argument, must be a function of one variable only, something like

ncon = @(x)nlcons(x,Numtk,constr,ttransij1,texeij1,cdk);

The other issue I see is that you appear to be trying to use both the problem-based approach (the first line of code is p = optimproblem('ObjectiveSense', 'minimize'); ) and the solver-based approach (your solution code is x = fmincon(fun(x), X0, A, b, Aeq, beq, lb, ub, ncon(x,Numtk,constr,ttransij1,texeij1,cdk), options); ). You cannot do that. If you are goiing to call fmincon, then don't make an optimization problem. You have to decide on one approach and stick with it. Apparently you are trying to use optimization variables in your nonlinear constraint. Don't do that.

Also, your code for nlcons is not in the correct format for the solver-based approach:

constr(loop1,1) = ttransij1(loop1,1) + texeij1(loop1,1) <= cdk;

This is a problem-based inequality. For the solver-based approach, you need to write your nonlinear inequality constraints according to the instructions in Nonlinear Constraints.

However, if what you are trying to do is follow the problem-based approach, then follow the directions in Problem-Based Optimization Workflow. Follow all of them, inclusing not calling fmincon, but instead calling solve. But at this point it is probably difficult to change over, so I suggest you simply stick with the solver-based approach.

Good luck,

Alan Weiss

MATLAB mathematical toolbox documentation

[muhammad ilyas khattak khattak]

@Alan Weiss Thank you alan.. Its very fruitful and also definitely I need guidance in the configuration/management portion as you pointed out instead of scrolling through all my code... Scrolling through all code is my job as I am dealing with it... Thanks once again... As you helped and guided me to restrict ncon to only one varialbe , probably "x", that I did now... And also I restricted the entire code to be "solver-based"... Now I get an error that does not recognize the input variable "x" in the objective function...For sake of simplicity, I have saved the errors that is the "try-1" in attached pdf file format..Also, I shared other tryes by name try-2 , try-3 that I attempted...😊😊😊😊🤝🙏🏻

[Alan Weiss]

You need your objective and nonlinear constraint functions to be functions of one variable only. Something like this:

fun = @(x)FCVEC06January2022V2(x,PUT ALL OTHER FUNCTION ARGUMENTS HERE);
ncon = @(x)nlcons(x,PUT ALL OTHER FUNCTION ARGUMENTS HERE);
[sol,fval] = fmincon(fun,x0,A,b,Aeq,bq,lb,ub,ncon,options)

Alan Weiss

MATLAB mathematical toolbox documentation

[muhammad ilyas khattak khattak]

@Alan Weiss 🙏🏻😊🤝😊 IThank you Alan ... I used single variable that is "x" for both objective and nonlinear constraint functions....Now Ithat i am getting "Too many output arguments." error......Many thanks in advance once again... The error that I recieve is:

K>> x = fmincon(fun, X0, A, b, Aeq, beq, lb, ub, ncon, options);

Error using AllinOne14January2022V1>FCVEC06January2022V2

Too many output arguments.

Error in AllinOne14January2022V1>@(x)FCVEC06January2022V2 (line 950)

fun = @(x)FCVEC06January2022V2;

Error in fmincon (line 571)

[initVals.f,initVals.g] = feval(funfcn{3},X,varargin{:});

Caused by:

Failure in initial objective function evaluation. FMINCON cannot continue.

[Alan Weiss]

Your options specify that the objective function returns a gradient.

options = optimoptions(@fmincon,'Algorithm','interior-point',...
 'SpecifyObjectiveGradient',true,'SpecifyConstraintGradient',true,...
 'PlotFcn',{@optimplotx,@optimplotfval,@optimplotfirstorderopt});

However, your objective function does not seem to return a gradient:

function Ttotal = FCVEC06January2022V2(x)
% If this were to return a gradient the definition would be
% function [Ttotal,grad] = FCVEC06January2022V2(x)
end

I suggest that you remove these two name-value arguments and try again:

'SpecifyObjectiveGradient',true,'SpecifyConstraintGradient',true,...

Alan Weiss

MATLAB mathematical toolbox documentation

[muhammad ilyas khattak khattak]

@Alan Weiss Right Alan.... Thanks I ammended options as you guided....and executed the "fmincon" with "objective" and "nlcon" functions....As a result, I recieved the error that I used to recive when "options" with two (02) "gradient" sub-options as true..... The value of "x" is passed to the "fun" that is the objective function but when starts using the value of "x" inside the objetive function , it gives the error "Not enough input arguments."..... I tried with some altered configurations of "fun" objective function and the error remains the same in all the four (04) attemtps....

Try 1:

I received suggestions on mathworks online from Alan Wies to try change the ‘options’ by removing the two gradient options elements….

So I changed the options to ::

options = optimoptions(@fmincon,'Algorithm','interior-point',...
    'PlotFcn',{@optimplotx,@optimplotfval,@optimplotfirstorderopt});

Try 2:

options = optimoptions('fmincon');

Try 3:

options = optimoptions(‘fmincon’,'Algorithm','interior-point');

While keeping objective, non linear constraints and fmincon same in all the above as follows….

x = fmincon(fun, X0, A, b, Aeq, beq, lb, ub, ncon, options);
ncon = @nlcons;
fun = @(x)FCVEC06January2022V2;

So, now the new error am getting is:

Not enough input arguments.

Error in AllinOne14January2022V1>FCVEC06January2022V2 (line 1515)

Numtk = x(1,size(x,2)-1);

Error in AllinOne14January2022V1>@(x)FCVEC06January2022V2 (line 950)

fun = @(x)FCVEC06January2022V2;

Error in fmincon (line 562)

initVals.f = feval(funfcn{3},X,varargin{:});

Caused by:

Failure in initial objective function evaluation. FMINCON cannot continue.

While keeping objective, non linear constraints and fmincon as follows….

x = fmincon(fun, X0, A, b, Aeq, beq, lb, ub, ncon, options);
ncon = @nlcons;
fun = @(x)FCVEC06January2022V2;

So, now the above mentioned same error (as in try-1) am getting is:

Not enough input arguments.

Error in AllinOne14January2022V1>FCVEC06January2022V2 (line 1515)

Numtk = x(1,size(x,2)-1);

Error in AllinOne14January2022V1>@(x)FCVEC06January2022V2 (line 950)

fun = @(x)FCVEC06January2022V2;

Error in fmincon (line 562)

initVals.f = feval(funfcn{3},X,varargin{:});

Caused by:

Failure in initial objective function evaluation. FMINCON cannot continue.

While keeping objective, non linear constraints and fmincon as follows….

x = fmincon(fun, X0, A, b, Aeq, beq, lb, ub, ncon, options);
ncon = @nlcons;
fun = @(x)FCVEC06January2022V2;

% Try 4: 
% Configuration:
options = optimoptions('fmincon','Algorithm','interior-point');
x = fmincon(fun(x), X0, A, b, Aeq, beq, lb, ub, ncon, options);

% Error received:

Not enough input arguments.

Error in AllinOne14January2022V1>FCVEC06January2022V2 (line 1516)

Numtk = x(1,size(x,2)-1);

Error in AllinOne14January2022V1>@(x)FCVEC06January2022V2 (line 950)

fun = @(x)FCVEC06January2022V2;

Answer 2

[Walter Roberson]

fun = @(x)FCVEC06January2022V2;

That says that fun is to become an anonymous function that accepts zero or one input arguments, and when invoked, ignores the input argument and calls FCVC06January2022V2 with no parameters.

Refer back to where Alan said

fun = @(x)FCVEC06January2022V2(x,PUT ALL OTHER FUNCTION ARGUMENTS HERE);
ncon = @(x)nlcons(x,PUT ALL OTHER FUNCTION ARGUMENTS HERE);

If you do not happen to have any other function arguments, then you would still need to (x) part,

fun = @(x)FCEVEC06January2022V2(x)
ncon = @(x)nlcons(x)

but in your earlier versions you did need additional arguments so we suspect that you still need some.

[muhammad ilyas khattak khattak]

@Walter Roberson @Alan Weiss 🙏🏻🤝😊😊😊Highly respected Walter and also highly respected professional, Alan, I am lacking words to thank you for responding to my queries whenever I post …. Walter! Sir, I configured handles of the “non-linear constraint” function, “nlcons” and objective function, “FCVEC06January2022V2” as you guided in the above post… Which is give in the following text…

% Configuration:

ncon = @(x)nlcons(x);
fun = @(x)FCVEC06January2022V2(x);
x = fmincon(fun, X0, A, b, Aeq, beq, lb, ub, ncon, options);

However, I am getting now an error which is “index in position 2 is invalid”… I have double-checked the line 1667 which contains assigning values to variable “TijPrime” from vector variable “x”…… I think I should mention here that variable “x” is a single vector variable which I am using to pass values to objective function handled by “fun” and non-linear constraint function handled by using “nlcon”…..I am attaching the completing code for convenience which need not to be scrolled/read entirely…However, I am attaching the completing code for your convenience so that every part of code is present with you when you need to look/scrutinize a particular portion from the entire code..I do not expect that errors will be emitted at single attempt …However, I expect that I may get one step closer to an error-free attempt… Many thanks once again for precious time that mathworks staff particularly you people are giving to my problem….🙏🏻🤝😊😊😊

K>> x = fmincon(fun, X0, A, b, Aeq, beq, lb, ub, ncon, options);

Index in position 2 is invalid. Array indices must be positive integers or logical values.

Error in AllinOne14January2022V1>FCVEC06January2022V2 (line 1667)

TijPrime = x(1,Numtk+size(ttransij,2) + size(texeij,2)+ size(D_Cloud,2)+size(F_Cloud,2)+size(MuRSU,2)+...

Error in AllinOne14January2022V1>@(x)FCVEC06January2022V2(x) (line 948)

fun = @(x)FCVEC06January2022V2(x);

Error in finDiffEvalAndChkErr

Error in finitedifferences

Error in computeFinDiffGradAndJac

Error in barrier

Error in fmincon (line 834)

[X,FVAL,EXITFLAG,OUTPUT,LAMBDA,GRAD,HESSIAN] = barrier(funfcn,X,A,B,Aeq,Beq,l,u,confcn,options.HessFcn, ...

[Walter Roberson]

It looks to me as if insted of using parameterization to store constant information, that the constant information has been added on to the end of the vector x that is passed into the function.

If lb and ub had been used to pin that portion of the vector to be constant then there might be some hope that would work. But ub and lb are empty here, so fmincon() is going to vary all of the contents of the vector. Including the element that holds the number of Tk (NumTk).

[muhammad ilyas khattak khattak]

@Walter Roberson@Alan Weiss .. Thanks once again for being with me for my several queries all the time 🙏🙏🙏…..I defined lower and upper bounds (lb and ub) exactly of the same length as that of “x”….And now the error / response is changed.. I am now getting the error/response “Exiting due to infeasibility: 166 lower bounds exceed the corresponding upper bounds”…. I am getting the response, while I am using the following configuration/settings of “fmincon”…..

If I get through this phase, I am pretty confidant, that an error-free run will be very soon….. Many thanks once again for answering my querries … 🙏🙏🙏.

Try 1:

x = fmincon(fun, X0, A, b, Aeq, beq, lb, ub, ncon, options);

[Walter Roberson]

you will have to show us how you set lb and ub

Answer 3

[muhammad ilyas khattak khattak]

@Walter Roberson 🤝✌️😊☺️😊👍 Attached MATLAB code contains the assigned lb and ub….Tha is the matlab code when opened in MATLAB editor then from line-52 till line-2406 contains the assigned values to lb and ub…..Only, if you scroll down from line-52 till line-2406, you will find how and what values are assigned to lb and ub….. Here I think I should share with you that the length of both lb and ub is eactly equal to that of “x”….

I doubt this online editor is not allowing the length portion of my code to be appended here that contain the assigned values to lb and ub…..So, I had attach the entire updated code in matlab file "AllinOne14January2022V1."

Also, please ask freely that you need to know about this code, I will answer soon and precisely as possible..

I am getting very hopeful this time as I think I am getting nearer with help of response from you and other staff of the mathworks…. 🙏🙏🙏

Many Thanks once again….

[Walter Roberson]

Near line 2099 you have

lb(1,Numtk+size(StoreInitialization.ttransij_struct,2)+ size(StoreInitialization.texeij_struct,2)+...   % StoreInitialization.Tctotal_struct
    size(StoreInitialization.D_Cloud_struct,2)+size(StoreInitialization.F_Cloud_struct,2)+...
    size(StoreInitialization.MuRSU_struct,2)+size(StoreInitialization.MaxNumtk_struct,2)+...
    size(StoreInitialization.E_struct,2)+size(StoreInitialization.DPrc_struct,2)+...
    size(StoreInitialization.pc_struct,2)+size(StoreInitialization.Muc_struct,2)+...
    size(StoreInitialization.Mui_struct,2)+size(StoreInitialization.Muk_struct,2)+ size(StoreInitialization.B_struct,2)+...
    size(StoreInitialization.B_struct,2)+size(StoreInitialization.pm_struct,2)+size(StoreInitialization.W_struct,2)+...
    size(StoreInitialization.Wfog_struct,2)++size(StoreInitialization.t_struct,2)+size(StoreInitialization.T_struct,2)+...
    size(StoreInitialization.cdk_struct,2)+size(StoreInitialization.j_struct,2)+size(StoreInitialization.i_struct,2)+...
    size(StoreInitialization.TasksperVehicle_struct,2)+size(StoreInitialization.NTotal_struct,2)+size(StoreInitialization.Numv_struct,2)+...
    size(StoreInitialization.Numtk_struct,2)+size(StoreInitialization.TijPrime_struct,2)+...
    size(StoreInitialization.loop1_struct,2)+size(StoreInitialization.Fprime_struct,2)+...
    size(StoreInitialization.PhikPrime_struct,2)+size(StoreInitialization.D_struct,2)+size(StoreInitialization.F_struct,2)+...
    size(StoreInitialization.Ravailk_struct,2)+size(StoreInitialization.Phik_struct,2)+size(StoreInitialization.theta1_struct,2)+...
    (4*size(StoreInitialization.matrixforAlgo3_struct(:,1),1))+size(StoreInitialization.tTijtoRij_struct,2)+...
    size(StoreInitialization.Ttotal_struct,2)+1:...
    Numtk+size(StoreInitialization.ttransij_struct,2)+ size(StoreInitialization.texeij_struct,2)+...
    size(StoreInitialization.D_Cloud_struct,2)+size(StoreInitialization.F_Cloud_struct,2)+...
    size(StoreInitialization.MuRSU_struct,2)+size(StoreInitialization.MaxNumtk_struct,2)+...
    size(StoreInitialization.E_struct,2)+size(StoreInitialization.DPrc_struct,2)+...
    size(StoreInitialization.pc_struct,2)+size(StoreInitialization.Muc_struct,2)+...
    size(StoreInitialization.Mui_struct,2)+size(StoreInitialization.Muk_struct,2)+...
    size(StoreInitialization.B_struct,2)+size(StoreInitialization.pi_struct,2)+...
    size(StoreInitialization.pm_struct,2)+size(StoreInitialization.W_struct,2)+...
    size(StoreInitialization.Wfog_struct,2)+size(StoreInitialization.t_struct,2)+...
    size(StoreInitialization.T_struct,2)+size(StoreInitialization.cdk_struct,2)+...
    size(StoreInitialization.j_struct,2)+size(StoreInitialization.i_struct,2)+...
    size(StoreInitialization.TasksperVehicle_struct,2)+size(StoreInitialization.NTotal_struct,2)+...
    size(StoreInitialization.Numv_struct,2)+size(StoreInitialization.Numtk_struct,2)+...
    size(StoreInitialization.TijPrime_struct,2)+size(StoreInitialization.loop1_struct,2)+...
    size(StoreInitialization.Fprime_struct,2)+size(StoreInitialization.PhikPrime_struct,2)+...
    size(StoreInitialization.D_struct,2)+size(StoreInitialization.F_struct,2)+...
    size(StoreInitialization.Ravailk_struct,2)+size(StoreInitialization.Phik_struct,2)+...
    size(StoreInitialization.theta1_struct,2)+(4*size(StoreInitialization.matrixforAlgo3_struct(:,1),1))+...
    size(StoreInitialization.tTijtoRij_struct,2)+size(StoreInitialization.Ttotal_struct,2)+...
    size(StoreInitialization.Tctotal_struct,2)) = repelem(min(StoreInitialization.Tctotal_struct),1,Numtk);
ub(1,Numtk+size(StoreInitialization.ttransij_struct,2)+ size(StoreInitialization.texeij_struct,2)+... 
    size(StoreInitialization.D_Cloud_struct,2)+size(StoreInitialization.F_Cloud_struct,2)+...
    size(StoreInitialization.MuRSU_struct,2)+size(StoreInitialization.MaxNumtk_struct,2)+...
    size(StoreInitialization.E_struct,2)+size(StoreInitialization.DPrc_struct,2)+...
    size(StoreInitialization.pc_struct,2)+size(StoreInitialization.Muc_struct,2)+...
    size(StoreInitialization.Mui_struct,2)+size(StoreInitialization.Muk_struct,2)+ size(StoreInitialization.B_struct,2)+...
    size(StoreInitialization.B_struct,2)+size(StoreInitialization.pm_struct,2)+size(StoreInitialization.W_struct,2)+...
    size(StoreInitialization.Wfog_struct,2)++size(StoreInitialization.t_struct,2)+size(StoreInitialization.T_struct,2)+...
    size(StoreInitialization.cdk_struct,2)+size(StoreInitialization.j_struct,2)+size(StoreInitialization.i_struct,2)+...
    size(StoreInitialization.TasksperVehicle_struct,2)+size(StoreInitialization.NTotal_struct,2)+size(StoreInitialization.Numv_struct,2)+...
    size(StoreInitialization.Numtk_struct,2)+size(StoreInitialization.TijPrime_struct,2)+...
    size(StoreInitialization.loop1_struct,2)+size(StoreInitialization.Fprime_struct,2)+...
    size(StoreInitialization.PhikPrime_struct,2)+size(StoreInitialization.D_struct,2)+size(StoreInitialization.F_struct,2)+...
    size(StoreInitialization.Ravailk_struct,2)+size(StoreInitialization.Phik_struct,2)+size(StoreInitialization.theta1_struct,2)+...
    (4*size(StoreInitialization.matrixforAlgo3_struct(:,1),1))+size(StoreInitialization.tTijtoRij_struct,2)+...
    size(StoreInitialization.Ttotal_struct,2)+size(StoreInitialization.Tctotal_struct,2)+1:...
    Numtk+size(StoreInitialization.ttransij_struct,2)+ size(StoreInitialization.texeij_struct,2)+...
    size(StoreInitialization.D_Cloud_struct,2)+size(StoreInitialization.F_Cloud_struct,2)+...
    size(StoreInitialization.MuRSU_struct,2)+size(StoreInitialization.MaxNumtk_struct,2)+...
    size(StoreInitialization.E_struct,2)+size(StoreInitialization.DPrc_struct,2)+...
    size(StoreInitialization.pc_struct,2)+size(StoreInitialization.Muc_struct,2)+...
    size(StoreInitialization.Mui_struct,2)+size(StoreInitialization.Muk_struct,2)+...
    size(StoreInitialization.B_struct,2)+size(StoreInitialization.pi_struct,2)+...
    size(StoreInitialization.pm_struct,2)+size(StoreInitialization.W_struct,2)+...
    size(StoreInitialization.Wfog_struct,2)+size(StoreInitialization.t_struct,2)+...
    size(StoreInitialization.T_struct,2)+size(StoreInitialization.cdk_struct,2)+...
    size(StoreInitialization.j_struct,2)+size(StoreInitialization.i_struct,2)+...
    size(StoreInitialization.TasksperVehicle_struct,2)+size(StoreInitialization.NTotal_struct,2)+...
    size(StoreInitialization.Numv_struct,2)+size(StoreInitialization.Numtk_struct,2)+...
    size(StoreInitialization.TijPrime_struct,2)+size(StoreInitialization.loop1_struct,2)+...
    size(StoreInitialization.Fprime_struct,2)+size(StoreInitialization.PhikPrime_struct,2)+...
    size(StoreInitialization.D_struct,2)+size(StoreInitialization.F_struct,2)+...
    size(StoreInitialization.Ravailk_struct,2)+size(StoreInitialization.Phik_struct,2)+...
    size(StoreInitialization.theta1_struct,2)+(4*size(StoreInitialization.matrixforAlgo3_struct(:,1),1))+...
    size(StoreInitialization.tTijtoRij_struct,2)+size(StoreInitialization.Ttotal_struct,2)+...
    size(StoreInitialization.Tctotal_struct,2)+size(StoreInitialization.texeij_struct,2))...
    = repelem(max(StoreInitialization.Tctotal_struct),1,Numtk); % Tctotal_struct

However the section of lb that you write into is not the same as the section of ub that you write into, so the part of ub corresponding to that stretch stays 0 while the lb becomes 0.2

I would recommend that you rewrite that code quite a bit, so that you keep a cummulative counter of where you are writing into, along the lines of

len = numel(tc);
x(idx:idx+len-1) = tc;
lb(idx:idx+len-1) = tc_lb;
ub(idxL:idx+len-1) = tc_ub;
idx = idx + len;

len = numel(Mui);
x(idx:idx+len-1) = Mui;
lb(idx:idx+len-1) = Mui_lb;
ub(idx:idx+len-1) = Mui_ub;

and so on. No copying and re-copying of all those size() things and possibly getting them wrong as you copy: keep a running total and use simple indexing as you write into the arrays.

It would probably be a good idea to do something similar as you pull values out of the array.

[muhammad ilyas khattak khattak]

@Walter Roberson 🙏🏻☺️🙂Thank you Walter for pin-pointing the exact line 2099 that contains lower bound “lb” becoming larger in magnitude than its corresponding upper bound “up”…

Also, I got some more errors that I somehow rectified/corrected….

Also, you are very right to use simpler indexing however, in parallel, I will try to use simple-indexing…. However, currently, I received this error.. I am pretty confidant that sooner I will get an error-free run of my code….

The error that I am revieving now is:

Error in fmincon (line 834)

[X,FVAL,EXITFLAG,OUTPUT,LAMBDA,GRAD,HESSIAN] = barrier(funfcn,X,A,B,Aeq,Beq,l,u,confcn,options.HessFcn, ...

For convenience, let me share with you the updated matlab file, “AllinOne14January2022V1”...as I said, i like to append here only the required portion of the entire code but due to the length limitation of this online editor, its not easy, therefore, I need to upload the entire matlab file... Thanks once again..🏻☺️🙂

[Alan Weiss]

Please report your error as everything in red, not just a line or two. I, for one, cannot understand what is going on from the one line you report.

Alan Weiss

MATLAB mathematical toolbox documentation

[Walter Roberson]

In your nonlinear constraint function you have

ceq = +Inf;

which is not a constraint that can ever be satisfied. fmincon() says that infinite constraint values and infinite values for the function itself are "undefined" values and refuses to proceed.

[muhammad ilyas khattak khattak]

@Walter Roberson@Alan Weiss☺🙏🏻️🙂 Thanks Walter….I changed the “+Inf” to another suitable value.. And also as Alan suggested to share the red colored error…. I am sharing the snapshot of red colored error in jpg format …. However, I also append the actual matlab file so that its convenient to read/use for you people to some extent..…

[Walter Roberson]

Line 3167

Numtk = x(1,Numtk+size(ttransij,2) + size(texeij,2)+ size(D_Cloud,2)+size(F_Cloud,2)+size(MuRSU,2)+...
    size(MaxNumtk,2)+ size(E,2)+ size(DPrc,2)+size(pc,2)+size(Muc,2)+size(Mui,2)+size(Muk,2)+...
    size(B,2)+size(pi)+size(pm,2)+ size(W,2)+size(Wfog,2)+size(t,2)+...
    size(T,2)+size(cdk,2)+size(j,2)+ size(i,2)+ size(TasksperVehicle,2)+size(NTotal,2)+size(Numv,2)+1:...
    Numtk+size(ttransij,2)+ size(texeij,2)+ size(D_Cloud,2)+size(F_Cloud,2)+...
    size(MuRSU,2)+size(MaxNumtk,2)+size(E,2)+size(DPrc,2)+size(pc,2)+size(Muc,2)+...
    size(Mui,2)+size(Muk,2)+size(B,2)+size(pi)+size(pm,2)+size(W,2)+size(Wfog,2)+...
    size(t,2)+size(T,2)+size(cdk,2)+size(j,2)+size(i,2)+size(TasksperVehicle,2)+...
    size(NTotal,2)+size(Numv,2)+1);

You do not take floor() or ceil() or round() the value that you retrieve.

You are accessing x(1,1071) at that point. lb(1071) is 1 and ub(1071) is +inf . So MATLAB is going to choose values from anywhere in that range, not confined to integers. So although it might point to a 66 now, next run it might point to 66.000001349 and since you use it as part of your indexing, you are going to be using non-integer subscripts.

You cannot use integer constraints for fmincon. If you take floor() or fix() or round(), it will explore "near" the first integer and decide that the function is flat in that parameter and will not search properly over that parameter

[muhammad ilyas khattak khattak]

@Walter Roberson 🙏🏻🤝😊Thanks walter, With your help, I very carefully used the floor command and did my best to avoid use of “floor” to get maximum possible precision by “fmincon”… With this help, I got great improvement uptill now in a way that now all the errors related to indexing seems almost vanished…

However, now I receive “Solver stopped prematurely. fmincon stopped because it exceeded the function evaluation limit, options.MaxFunctionEvaluations = 1.000000e+05.”…

What I did was to alter/change the “MaxFunctionEvaluations” several times and observed/executed several runs..

Can you suggest what other configurations can I do to get closer to error-free run of the code….

It seems that I am pretty near the error-free run of the code…. And use/tabulate my results in excel….

I am appending both the updated MATLAB file and snapshot in jpg format of the “error” that I receive… 🙏🏻🤝😊 Thanks once again…

[Alan Weiss]

You have a lot of variables, so you need to have a very high function evaluation limit. Try setting MaxFunctionEvaluations to 1e6 or even more. And you might then have to raise your iteration limit, too.

FYI, fmincon is not issuing an error, it is just an exit message telling you why the solver stopped.

I'm glad that fmincon is now proceeding.

Alan Weiss

MATLAB mathematical toolbox documentation

[muhammad ilyas khattak khattak]

@Walter Roberson@Alan WeissThank you Alan, With your suggestions of re-configuring the limit, I obtained further progress… However, one another error that I am receiving is “Index in position 2 exceeds array bounds (must not exceed 4763).”….

Sir, I checked the indices at line 3762..this is the line pointed in generated error….However, the indices are correct.. Despite the indices being correct, I am getting this error “Index in position 2 exceeds array bounds (must not exceed 4763…

Maybe, I am missing some calculations in indices that need to be highlighted…

Once, I get through this error, I will then start tallying obtained results to obtain the required theme of my on-going observations….

Your help/time is needed in pointing out the exact thing that I am missing…🙏🏻😊☺🥲🙂

I am appended snapshot of the red-colored error that I am recieving and also updated matlab code... for possible convenience....

Thanks once again for your time and energy….🙏🏻😊☺🥲🙂

[Walter Roberson]

Numtk starts at 262, and you have various variables that are Numtk long in some dimension.

But you failed to lock in Numtk using ub and lb to ensure that it can never change. So it does change. It becomes slightly less -- something whose floor is 261. Then because it is used as the length of other variables, those other variables become shorter.

So eventually instead of Numv being pulled out of element 1070 of x, it gets pulled out of element 1066 -- because 4 variables in x before it each got 1 element shorter because floor(Numtk) reduced to 261 instead of 262. And what is at x(1070) ? Well, it happens to be a value that is almost 100000000 .

Now: line 3672

Ravailk = x(1,floor(Numtk+size(ttransij,2) + size(texeij,2)+ size(D_Cloud,2)+size(F_Cloud,2)+size(MuRSU,2)+...  % "Ravail" keeps number of OBCU-mounted
    size(MaxNumtk,2)+ size(E,2)+ size(DPrc,2)+size(pc,2)+size(Muc,2)+size(Mui,2)+size(Muk,2)+...          % fog-vehicles with redundant/available
    size(B,2)+size(pi)+size(pm,2)+ size(W,2)+size(Wfog,2)+size(t,2)+...                                   % computation-resources.....
    size(T,2)+size(cdk,2)+size(j,2)+ size(i,2)+ size(TasksperVehicle,2)+size(NTotal,2)+...
    size(Numv,2)+ size(Numtk,2)+size(TijPrime,2)+size(loop1,2)+...
    size(Fprime,2)+size(PhikPrime,2)+size(D,2)+size(F,2)+1):...
    floor(Numtk+size(ttransij,2)+ size(texeij,2)+ size(D_Cloud,2)+size(F_Cloud,2)+...
    size(MuRSU,2)+size(MaxNumtk,2)+size(E,2)+size(DPrc,2)+size(pc,2)+size(Muc,2)+...
    size(Mui,2)+size(Muk,2)+size(B,2)+size(pi)+size(pm,2)+size(W,2)+size(Wfog,2)+...
    size(t,2)+size(T,2)+size(cdk,2)+size(j,2)+size(i,2)+size(TasksperVehicle,2)+...
    size(NTotal,2)+size(Numv,2)+size(Numtk,2)+size(TijPrime,2)+size(loop1,2)+...
    size(Fprime,2)+size(PhikPrime,2)+size(D,2)+size(F,2)+Numv))

Notice the Numv at the end: it is being used to construct a range of indices into x. But it is now nearly 10^8 so you are constructing an index range up to close to 10^8 . For an array that has less than 5000 elements.

[Walter Roberson]

Note: I refuse to debug any more of this code. You need to completely redesign it.

Any value that is fixed (unchanging) during optimization should be passed in by parameterization http://www.mathworks.com/help/matlab/math/parameterizing-functions.html . This should include the sizes of all of the elements.

If you need to pack a lot of variable data (that truly changes) into a single vector, then write a function to unpack it -- a function that uses the static size information that was passed in by way of parameterization.

If you want to have entities that randomly interact (so part of your vector is random indices of which ones are interacting) then make sure you declare those index parameters as integer parameters and make sure you use lb and ub for them that lock them in to valid ranges according to the size of the data.

I do not recommend randomly changing the number of items you are using as you progress. However, it can be productive to use a fixed number of items together with binary "indicator" variables that switch elements on or off. Binary indicator variables should be declared as integer with range 0 to 1... and your unpacking function can logical() them as it unpacks them from x.

I repeat: I refuse to debug your current structure any further. No, you are not "just one or two errors away from being finished": your code is jam-packed with similar errors that you just haven't gotten to yet. The design you are using is too fragile to continue with.