Calling the size of an input

I call different models in a function. I want to call the size of the inputs (number of parameters the model takes) from the function I'm calling them. E.g.,

function y = model1(t,x,Pm,Other_Inputs)
  Some stuff = Pm(1);
  Some other stuff = Pm(2);
  ...
  More stuff = Pm(n);
  %some stuff it does using the inputs.
  end

Now I want to call it with something like

function x = caller(Model)
 %For example
  P = size(Pm,1)
  z = randi(P,1)); 
  [~,yx2] = ode15s(@(t,x) model1...
  (t,x,z,Other_Inputs)t,InitalValue); 
  end

Is this possible? I know some basics about varargin and nargin, but I'm not sure if they are what I should use here. Thanks for any help.

[Jan]

I do not understand the question. What is "P" in the caller? Where is the problem? Which step does not work?

[Philip Berg]

Ahh... Sorry let me try to clarify. I have some different models that take different amount of parameters, P. In other words, P is a row vector. Now, from my main function I will call some model that is given as a input, but for now I also have to specify how many parameters the model takes as an input. I want this to be automated, so when I call a model, the function figures out the size of P instead of me giving it as an input. The problem is that I don't know how to get the size of P in the caller function. What I want to know is n from P(n) in the model. Do I make more sense?

Answer 1

[Star Strider]

‘I also have to specify how many parameters the model takes as an input.’

If ‘Pm’ is a vector and all the parameters are in the same order, the models themselves can decide.

For example:

function out1 = model1(t,x,Pm,Other_Inputs)
N = 3;
P1 = Pm(1:N);
. . .
end

function out2 = model2(t,x,Pm,Other_Inputs)
N = 5;
P2 = Pm(1:N);
. . .
end

This assumes I understand what you want to do.

[Philip Berg]

Yes, you are sort of getting me, the issue is still that these models are called from another function. This function wants to know the length of Pm as it is defined in the models. That is the value of n from Pm(n). So I want to randomize n numbers in the function that calls the model. Am I making any sense at all? Edit: It wants to know the value you have for N.

[Star Strider]

Perhaps name them by the number of inputs?

function out5 = model5(t,x,Pm,Other_Inputs)
N = 5;
. . .
end

I normally discourage the use of eval, feval or their friends, but using them with functions that include the parameter length as part of the name may be the only way to do what you want. See the documentation on eval and feval to see where I’m going with this. Use the sprintf function to generate the function names:

N = 5;
fname = sprintf('model%d', N);
result = feval(fname, t,x,Pm,Other_Inputs);

You could even put that into a loop and save ‘result’ as a cell array:

result{k1} = feval(fname, t,x,Pm,Other_Inputs);

You will obviously have to experiment with this, since I still only have a vague idea of what you want to do.

[Philip Berg]

Yes, I know eval etc slows down the script. I just want to randomize n numbers. Where n is the number of parameters the model being called by the main function. But I would like that number to be given from how many parameters the model takes. I'm not sure how to make this clearer. The only way I figured it out is to put an output vector of zeroes that corresponds to the number of parameters and then do size on that vector. Edit: Oh and thank you for the help---of course!

[Star Strider]

My pleasure!

Answer 2

[Steven Lord]

So you want to turn some unknown number of parameters into individual inputs of another function? In this case I would use a cell array and turn it into a comma-separated list when you call the other function. Put the following code in a file named helper.m:

function helper(vectorOfParameters)
cellOfParameters = num2cell(vectorOfParameters);
showParameters(cellOfParameters{:});

function showParameters(varargin)
for whichParameter = 1:nargin
  fprintf('Parameter %d is %g.\n', whichParameter, varargin{whichParameter});
end

Try calling the main function like this:

helper(1:5)
helper([99 4334 237 1 55 999 -17 6])

Inside helper, the num2cell function accepts the vector of parameters and creates a cell array of the same length with each element of the vector stored in the corresponding cell of the cell array. When I call showParameters, I use this syntax: cellOfParameters{:} which is just like:

showParameters(cellOfParameters{1}, cellOfParameters{2}, cellOfParameters{3}, ...)

except I don't have to write out all 1, 2, 3, 20, etc. parameters in the function call. Then inside showParameters I iterate over the number of inputs MATLAB passed into the function (which the nargin function returns) and use the loop variable to index into the cell array varargin.

[Philip Berg]

Well I see that I'm not very good at explaining my problem, sorry still a matlab novice... "So you want to turn some unknown number of parameters into individual inputs of another function?" Well, the number of parameters is know inside the models. Its the function that calls the model that wants to know how many they are. So when you call a model that has 5 parameters instead of 4---you just specify the name of model as input, and then in accordance with the number of parameters the model has. The main function (the one calling the model) will then randomizes that many. So say Model1 takes 3 parameters, and model2 take 5, then MainFunction(Model1) should give 3 random numbers while MainFunction(Model2) would give 5 random numbers.