Unable to perform assignment because value of type 'optim.pro​blemdef.Op​timization​Expression​' is not convertible to 'double'.

Hello!

I'm working on an optimization problem where I need to manage the availability of multiple cores for task execution. My current approach involves finding the earliest available core and updating its availability after assigning a task. However, I'm encountering an issue when trying to update the hapsCoreAvailability array with the new start time and execution time.

the error is

Unable to perform assignment because value of type 'optim.problemdef.OptimizationExpression' is not convertible to
'double'.
Error in Solve_Linear_Problem (line 355)
hapsCoreAvailability(coreIdx) = T_start + execTime;
Caused by:
Error using double
Conversion to double from optim.problemdef.OptimizationExpression is not possible.

here is the code

 % Initialize the queue time matrix
Queue_time = optimexpr(N, numNodes, num_vehicles);

% Define variables

T_start = optimvar('T_start', 1, 'Type', 'continuous', 'LowerBound', 0);
% Constraints
Queue_constraints1 = [];
Queue_constraints2 = [];
% Track availability of each HAPS core
hapsCoreAvailability = zeros(1, hapsCapacity); 
nodeQueues = cell(numNodes, 1); % Queue times for each node
% Process tasks by generation time
for taskIdx = 1:num_vehicles
    for subtaskIdx = 1:N
        for nodeIdx = 1:numNodes
                
        % Extract generation time, execution time, and uplink time
        taskGenTime = generation_times_matrix(subtaskIdx, nodeIdx, taskIdx);
        execTime = Execution_time(subtaskIdx, nodeIdx, taskIdx);
        uplinkTime = uplink_time(subtaskIdx, nodeIdx, taskIdx);
        
        % Calculate task arrival time considering uplink time
        taskArrivalTime = taskGenTime + uplinkTime;
        
        if nodeIdx == 4 % If assigned to HAPS
            % Find the earliest available core
            % [earliestCoreTime, coreIdx] = min(hapsCoreAvailability);
            coreIdx = -1;
            earliestCoreTime = inf;

            % Iterate through each core's availability time
            for i = 1:length(hapsCoreAvailability)
                % Check if the current core's availability is earlier than the current earliest time
                if hapsCoreAvailability(i) < earliestCoreTime
                    % Update earliest core time and index
                    earliestCoreTime = hapsCoreAvailability(i);
                    coreIdx = i;
                end
            end
            
            % Start time is the maximum of arrival time and core availability
             
            % T_start should be greater than or equal to both T_arrival and T_core
              Queue_constraints1 = [Queue_constraints1, T_start >= taskArrivalTime];
              Queue_constraints1 = [Queue_constraints1, T_start >= earliestCoreTime];

          
            Queue_time(subtaskIdx, nodeIdx, taskIdx) = T_start - taskArrivalTime;

            
            % Update the core availability time
             hapsCoreAvailability(coreIdx) = T_start + execTime;  %% here is the error 
        else % For UAVs
            % Queue tasks based on previous completion
            if isempty(nodeQueues{nodeIdx})
                startTime = taskArrivalTime;
            else
               T_queue  =  nodeQueues{nodeIdx}(end);
                 

                    % T_start should be greater than or equal to both T_arrival and T_queue
                      Queue_constraints2 = [Queue_constraints2, T_start >= taskArrivalTime];
                      Queue_constraints2 = [Queue_constraints2, T_start >= T_queue];

                    
            end
            % Calculate the queue time for the current subtask
            Queue_time(subtaskIdx, nodeIdx, taskIdx) = T_start - taskArrivalTime;
            % Update departure time for the current subtask
            departureTime = T_start + execTime;
            nodeQueues{nodeIdx} = [nodeQueues{nodeIdx}, departureTime];
        end
    end
end
end
 
prob.Constraints.queue_time_constraints = Queue_constraints1;
prob.Constraints.queue_time_constraints = Queue_constraints2;

How can I correctly update the hapsCoreAvailability array with the new start time and execution time in the context of my optimization problem? Is there an alternative way to manage and update core availability when dealing with OptimizationExpression objects?

Any advice or suggestions would be greatly appreciated!

Thank you!

Answer 1

[Torsten]

If you use optimvars on the right-hand side of the assignment

hapsCoreAvailability(coreIdx) = T_start + execTime;

(T_start is an optimvar !), hapsCoreAvailability must be an optimexpr, not a double array:

hapsCoreAvailability = zeros(1, hapsCapacity);

Consider using functions as fcn2optimexpr. Inputs can be optimvars or optimexpr, but in the function, they can be used as if they were doubles. Outputs from the function can again be doubles, and constraints using the outputs can be defined using the outputs in the calling program. So in the function itself, all operations are performed using doubles.

[Maria]

Thank you for your response.

I tried to replace the following part with function and i used as fcn2optimexpr, but i get this error :

Unable to use a value of type optim.problemdef.OptimizationNumeric as an index.

Error in Solve_Linear_Problem (line 343)

hapsCoreAvailability(coreIdx) = T_start + execTime;

            coreIdx = -1;
            earliestCoreTime = inf;
            % Iterate through each core's availability time
            for i = 1:length(hapsCoreAvailability)
                % Check if the current core's availability is earlier than the current earliest time
                if hapsCoreAvailability(i) < earliestCoreTime
                    % Update earliest core time and index
                    earliestCoreTime = hapsCoreAvailability(i);
                    coreIdx = i;
                end
            end
            
      [earliestCoreTime, coreIdx] = fcn2optimexpr(@findEarliestCore,hapsCoreAvailability);      

[Torsten]

Not enough information to give an answer. I do not even see the line where MATLAB complains in the piece of code you included.

If you try to use an optimvar as an array index, I doubt that this is allowed:

x = [1 2 3 4 5 6 7 8];
index = optimvar('index','Type','integer','LowerBound',1,'UpperBound',8);
obj = x(index);

Unable to use a value of type optim.problemdef.OptimizationVariable as an index.

prob = optimproblem('Objective',obj);
solve(prob)