I want to know how to writ code to net1 = train(net,input,z) but z is double(complex)??

net1 = train(net, input, z);

Error using nntraining.setup>setupPerWorker

Targets T is complex.

Error in nntraining.setup (line 77)

[net,data,tr,err] = setupPerWorker(net,trainFcn,X,Xi,Ai,T,EW,enableConfigure);

Error in network/train (line 342)

[net,data,tr,err] = nntraining.setup(net,net.trainFcn,X,Xi,Ai,T,EW,enableConfigure,isComposite);

[KSSV]

It looks your target is complex number. NN do not support complex values. Instead you may take target as two values real and imaginary seperately.

[Darshini]

If I can't change the Z function, how should I code it?

z= ((exp(sin(x))).*(cos(y)))+((sin(y)).*log(abs(tan(x.^2))+(x.^3)));

Answer 1

[Chunru]

net1 = train(net,input,[real(z) imag(z)])   

[Darshini]

it show

Error using image

Complex values are not supported. Specify the color data as numeric or logical values.

[Chunru]

imag (not image)

[real(z) imag(z)] or [real(z); imag(z)] (or similar form depending on the orientation of your data)

[Darshini]

now it show

Error using network/train

Inputs and targets have different numbers of samples.

[Darshini]

now it show

Error using network/train

Inputs and targets have different numbers of samples.

again

If I can't change the Z function, how should I code it?

z= ((exp(sin(x))).*(cos(y)))+((sin(y)).*log(abs(tan(x.^2))+(x.^3)));

[Chunru]

Attach your code to demonstrate the problem. Otherwise it's difficult to guess what is wrong.

Answer 2

[Walter Roberson]

z= ((exp(sin(x))).*(cos(y)))+((sin(y)).*log(abs(tan(x.^2))+(x.^3)));

z appears to be your target for a fitting function. You tell us that z is complex-valued.

Create a 2 x N array xy with x in the first row and y in the second row.

Create a 2 x N array zri with real(z) in the first row and imag(z) in the second row.

Now you can train xy as the features and zri as the targets.

You might possibly have to transpose xy tne N x 2 and zri to N x 2 for the particular training function you use.