Maximization of a Function

Hi All,

I am trying to find the maximization of the following function with bus frequency as a variable.

operator_profit(bus_frequency) =((flow_bus * distance * bus_fare - bus_operation_cost * bus_frequency) + subsidy);
max_profit(bus_frequency) = -@(operator_profit)
max_profit = @(bus_frequency) -((bus_flow*distance*bus_fare - bus_operation_cost * bus_frequency)) + subsidy;
x0 = [0 0];
[xmin, gmin] = fminsearch(max_profit(bus_frequency),x0)

When I run it, it shows following error;

Error: File: Bus_model.m Line: 40 Column: 48
Invalid expression. Check for missing or extra characters.

I don't know how to fix it. Kind guidance and help will be appreciated.

Answer 1

[Stephen23]

The syntax

.. = -@(..)

is not valid for multiple reasons:

1. the output of @ is a function handle. Arithmetic operations are not defined for function handles (only for numeric/char/logical arrays).
2. that syntax specifies one input variable operator_profit, but then inside the function nothing is defined. This is definitely not a valid syntax: a function has to do something. Read more here:

https://www.mathworks.com/help/matlab/matlab_prog/anonymous-functions.html

It seems like your goal is to simply take the negative of a function, in order to maximise the function. The correct way to negate the output of a function handle is to call it, which you can also do inside an anonymous function:

baz = @(x) -foo(x);

However I doubt that much of the rest of your code does anything useful. In particular:

1. you appear to have defined operator_profit as an array, whose existence is of little relevance to the anonymous function because you redefined it as an input argument.
2. ditto for bus_frequency.
3. max_profit is (implicitly?) defined and then ignored and redefined.
4. fminsearch requires a function handle as its first input, but instead of providing it with a function handle you call the function max_profit(bus_frequency) and provide fminsearch with whatever output that has (most likely an array of some kind).
5. x0 is defined with two elements, whereas most likely you should be providing a scalar for that (intended) function to minimize.
6. ... probably more bugs, I gave up at that point.

As far as I can tell, your confusion arises from your approach of thinking that you need to include bus_frequency everywhere. You don't. Most likely you should be doing something like this:

max_profit = @(bus_freq) bus_flow*distance*bus_fare - bus_operation_cost*bus_freq + subsidy;
[xmax, gmax] = fminsearch(@(bf)-max_profit(bf),0)

Answer 2

[Bjorn Gustavsson]

When I create a sequence of functions like that I always have to have the input-variables explicitly visible on the right-hand side. So this line looks suspicious:

max_profit (bus_frequency) =-@ (operator_profit)

I would have thought that you at least have to make the input-variable to operator_profit visible:

max_profit (bus_frequency) =-@ operator_profit(bus_frequency);

Also the error-message mentions Bus_model but that is something not anywhere else in your message...

HTH

[Nazam Ali]

Hi @Bjorn

Thank you for the answer. I have clearly defined input variable "bus_frequency" earlier. Don't know why it's not working. I understand your comment. But error is still persistant.

[Bjorn Gustavsson]

Actually even the assignment of max_profit I suggested is wrong. The version I suggested assigns the function-handle (RHS) to some elements determined by the current value of bus_frequency in the array max_profit. What you most likely want is closer to this:

max_profit = @(bus_frequency) -operator_profit(bus_frequency);

There we at least create a function-handle variable max_profit that has an input variable bus_frequency and is the negative of the function operator_profit. That function you could now minimize with fminsearch:

[xmin, gmin] = fminsearch (max_profit, x0)

Or you could do this without the intermediate steps:

operator_profit = @(bus_frequency)((flow_bus * distance * bus_fare --bus_operation_cost * bus_frequency) + subsidy);
[xmin, gmin] = fminsearch (@(b_f) -operator_profit(b_f), x0);

First we have to create a function-handle that depends on the variable that we want to optimize the function over, then, since we want to maximize operator_profit we need to minimize the negative of that function, therefore we create an anonymous function that is just that in the call to fminsearch. I have no idea what the values are for your other parameters in the operator_profit expression, flow_bus, distance, bus_fare, bus_operation_cost, and subsidy have to exist when you define operator_profit - either as variables or functions without inputs.

HTH