reshape() error in the function created by matlabFunction()

3 ビュー (過去 30 日間)
Valeri Aronov
Valeri Aronov 2021 年 8 月 21 日
回答済み: Valeri Aronov 2021 年 9 月 3 日
matlabFunction() has generated for me AmplAndDers() to be used in fminunc(). The very first objective function evaluation fails:
Error using reshape
Number of elements must not change. Use [] as one of the size inputs to automatically calculate the appropriate size for that dimension.
Error in AmplAndDers (line 93)
Hess = reshape([-t2.*t3.*t4. ... .*(3.0./4.0)],[4,4]);
The signature of the function is:
function [Ampl,GradA,Hess] = AmplAndDers(C1_0,C2_0,R1_0,R2_0,w,x1,x2,x3,x4)
I call it with a w being an array. Two other outputs are returned correctly. I do not understand what does the above suggestion mean:
Use [] as one of the size inputs to automatically calculate the appropriate size for that dimension.
If I manually change the above reshape() call to end with:
...*(3.0./4.0)],[4,4,length(w)]);
the program runs OK.
How can avoid the manual intervention?
  4 件のコメント
Valeri Aronov
Valeri Aronov 2021 年 8 月 24 日
編集済み: Valeri Aronov 2021 年 8 月 24 日
This is the script that creates the function in question:
syms s
syms C1 C2 R1 R2 w real
syms C1_0 C2_0 R1_0 R2_0 real
syms T(s, C1, C2, R1, R2)
T(s, C1, C2, R1, R2) = 1/(C1*C2*R1*R2*s^2 + (C2*R1 + C2*R2)*s + 1);
A = abs(T(1j*w, C1, C2, R1, R2));
A = rewrite(A, 'sqrt')
syms x1 x2 x3 x4 real
x=[x1,x2,x3,x4];
A = subs(A, [C1, C2, R1, R2], [C1_0, C2_0, R1_0, R2_0] .*x);
Ampl = A;
GradA = gradient(A, x);
Hess = hessian(A, x);
x0 = [0.1e-6, 1.5e-9, 16000, 11000];
f = logspace(1,5,101);
A_Init = double(abs(T(1j*f, x0(1), x0(2), x0(3), x0(4))));
% Covert symbolic functions to the function
matlabFunction(Ampl, GradA, Hess, 'File', 'AmplAndDers');
This how I tried to follow MATLAB hint and failed:
% matlabFunction(Ampl, GradA, Hess, 'File', 'AmplAndDers', 'Vars', {C1_0,C2_0,R1_0,R2_0,w[],x1,x2,x3,x4});
Thanks
Valeri Aronov
Valeri Aronov 2021 年 8 月 24 日
Oops. The call is here:
function [x, fval] = OptimizeFilter(A0, f, x0)
options = optimoptions('fminunc','Algorithm','trust-region','SpecifyObjectiveGradient',true, HessianFcn','objective');
[x,fval,exitflag,output] = fminunc(@Target, [1,1,1,1], options)
function y = Target(x)
[aCurrent, GradW, HessW] = AmplAndDers(x0(1), x0(2), x0(3), x0(4), f, x(1), x(2), x(3), x(4));
...
end
end

サインインしてコメントする。

採用された回答

Valeri Aronov
Valeri Aronov 2021 年 9 月 3 日
The error is known to MATLAB and will be fixed in the future

その他の回答 (1 件)

KSSV
KSSV 2021 年 8 月 21 日
Read the documentation of reshape and understand it. The error is clear, you are trying to create extra elements than present in the array while reshaping.
Example:
A = rand(1,25) ;
B = reshae(A,5,5) ; % 5*5 = 25, no error as same elements present
C = reshape(A,6,5) ; % 6*5 = 30, error as A as 30 elements only.
  1 件のコメント
Valeri Aronov
Valeri Aronov 2021 年 8 月 21 日
編集済み: Valeri Aronov 2021 年 8 月 22 日
The body of AmplAndDers() function and the reshape() call code in it is generated by matlabFunction(). I can't see how reading the documentation of reshape() can possibly help me. And, yes, the error is clear indeed.

サインインしてコメントする。

カテゴリ

Help Center および File ExchangeMatrices and Arrays についてさらに検索

製品


リリース

R2021a

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by