fsolve with variable undeclared - converting to web app

I am attempting to convert a function that used to solve with vpasolve into a function that will solve on the matlab web app.

The original equation to solve was the following:

 app.Q_Intrinsic=pi*app.ng*app.Ring_Length*sqrt(a)/app.lambda0/(1-a)
 

This works fine on a desktop compiled version of the app due to symbolic toolbox meaning I can define the variable a as syms a, before I run vpa solve. However, without syms toolbox, I am unsure how to format the following lossCalc function to allow me to pass in the LossFunction into fsolve.

function lossCalc(app)

    a = fsolve(@(a) lossFunc,1);%field loss
    alpha = -2*log(a)/app.Ring_Length;
    app.Alpha = alpha/100*4.34; %cm

end

function fval = lossFunc(app,a)

    app.Q_Intrinsic=pi*app.ng*app.Ring_Length*sqrt(a)/app.lambda0/(1-a);
end

Answer 1

[Varun]

Hi George,

Looks like you want to solve the “Loss Function” equation using “fsolve”. You have shared a code snippet for the same which seems to be incomplete and some syntactical errors.

Please find the updated code which is complete and flawless:

function lossCalc(app)

% correct way to pass anonymous function which calls lossFunc.

a = fsolve(@(x) lossFunc(x,app), 0.5);

alpha = -2*log(a)/app.Ring_Length;[AK1] 

app.Alpha = alpha/100*4.34; % cm

end


function fval = lossFunc(x, app)

app.Q_Intrinsic = pi*app.ng*app.Ring_Length*sqrt(x)/app.lambda0/(1-x);

% lossFunc should also return the value of app.Q_Intrinsic

fval = app.Q_Intrinsic;

end

In this modified version:

1. The anonymous function “@(x) lossFunc(x,app)” is passed as the first argument to “fsolve”. This anonymous function takes ‘x’ as the input and calls the “lossFunc” function passing input arguments ‘x’ and “app”(additional argument).
2. The “lossFunc” function takes two input arguments, ‘x’ and “app”. It calculates the intrinsic Q factor (app.Q_Intrinsic) based on the given parameters of app and input argument ‘x’.
3. The calculated app.Q_Intrinsic value is returned as the output of lossFunc.

You can also adjust the initial guess value (0.5 in this case) passed to “fsolve” as needed. This value should be close to the expected solution for faster convergence.

Please refer the following page to learn more on using “fsolve”:

https://www.mathworks.com/help/optim/ug/fsolve.html?searchHighlight=fsolve&s_tid=srchtitle_support_results_1_fsolve

Hope this helps.