Higher rank tensor: how to export and read it as a file?

Hello,

I have a rank-3 tensor, say simply

tensor=zeros(dA,dB,dC);

where dA, dB and dC are the respective dimensions. I like to export it as a file and later on import it again to continue some calculation in another Matlab script. How do I do this? As far as I see, dlmwrite, cswwrite and writematrix only work for matrices and give me some error saying "transpose not defined...".

Many thanks!

Answer 1

[Divyam]

Hi @Philipp,

If you have access to the tensor toolbox, you can simply use the export_data function to export the tensor data and save it to a file. The following command should work in your case.

export_data(tensor, "tensor_data.tensor")

For more information on exporting and importing tensor data using the tensor toolbox, you can refer to this article: Importing and Exporting Tensor Data (tensortoolbox.com)

[Philipp]

Unfortunately I do not have access to the tensor toolbox as I am using Octave... :(

Anyway, I found out that I can use the reshape command as follows:

tensor_new = reshape(tensor,[dA*dB,dC])

then export tensor2 and after importing it again convert back via

tensor = reshape(tensornew,[dA,dB,dC]))

That seems to work well. :) Hopefully, I am not overseeing something...

[Divyam]

@Philipp that should work well, there shouldn't be any issues with converting a tensor to an array. I dont think that you are overseeing something just keep the dimensions in check.