How to implement class based unit tests for functions that return random results

I have to write class based unittests for a function which returns random values. I have ideas on how to do that. For example see the testing methods described in this answer of this stackoverflow question.

I want to know how can we implement this for my function using Matlab's class based (matlab.unittest.TestCase) unit test methods.

Lets call my function which returns the random value as "myRandRetFunc" and it looks like following. How can I implement unit testing for it for things described here or for some other ideas you have in mind.

function rndRetVal = myRandRetFunc(inp1, inp2, inp3)
    arg1 = aComplexFunc(inp1, inp2)
    arg2 = anotherComplexFunc(inp2, inp3)
    rndRetVal = rand(arg1, arg2)
end

Answer 1

[Steven Lord]

The four phases of a four phase test are Setup, Exercise, Verify, and Teardown. In this case, you want to Setup the random number generator to a known fixed state before the code that relies upon it is Exercised. Use the rng function to do this.

[Akshay Jajoo]

How should I be implementing this? I tried the following three approaches and the last two are not working. In the first one I am getting the following warning for passing the argument 'testCase': "Argument 'testCase' is unused. Should this method be static?"

methods(TestClassSetup)
        function setRandNumber(testCase)
            rng(06);
        end
end

OR

methods(TestClassSetup)
        function setRandNumber(testCase)
            testCase.rng(06);
        end
 end

OR

methods(Static)
        function setRandNumber()
            rng(06);
        end
end

[Steven Lord]

Since you know you're not going to use the input, tell MATLAB that explicitly.

methods(TestClassSetup)
        function setRandNumber(~)
            rng(06);
        end
end

[Akshay Jajoo]

This works. Thanks.