metocean / diwasp Goto Github PK

I recently observed a mismatch in output, where two machines that ran the same code utilizing this public domain package, had a significant difference in output (but no warning or error message).

The reason was one of the machines had the Signal Processing Toolbox installed, which was listed as a dependency on running the matlab.codetools.requiredfilesandproducts function.

We had observed that the machine with the Signal Processing Toolbox installed and listed as a dependency, gave an incorrect output. On the other hand, the machine with the correct output had only base MATLAB as a dependency.

Is there a script available which regroups the functions and performs the directional analysis?

would anybody here be keen to see a version of this in python?

Hi,
you did not provide a vertical displacement transfer function, but I presume it should be like elevation (elev) multiplied by depth attenuation term Kz:

function trm=dspz(ffreqs,dirs,wns,z,depth)
Kz=sinh(z*wns)./sinh(depth*wns);
%include a maximum cuttoff for the vertical displacement response function
Kz(find(Kz<0.1))=0.1;
trm=Kz*ones(size(dirs));

Then, when derivating once, assuming the wave is expressed as exp(i*(wns*cos(dirs)*x+wns*sin(dirs)*y-ffreqs*t)), we get a transfer function multiplied by -i*ffreqs as we can see in the velz function.

Then, if we derivate a second time versus time t to get vertical acceleration accz, I expect another multiplication by -i*ffreqs, thus i^2*ffreqs.^2, thus -ffreqs.^2. I see the opposite sign in accz, which seems suspect to me... Unless I misunderstood some conventions?

That's the same for accx and accy since commit ae31ec2 shouldn't they be same as dspx and dspy multiplied by -ffreqs.^2?

Is there a simple example for new users? I see the function 'testspec.m', but it needs to set the input parameters. For beginners, it is better to give a very simple example.

In the last part of "makewavedata.m", the transfer function trf(:,:,j) is used to derive different datatypes in the following line of code:

eta(i,j)=sum(sum(eamp.*trf(:,:,j).*cos(layout(1,j)*wns*cos(dirs)+layout(2,j)*wns*sin(dirs)-freqmat*t(i)+randphase)));

However, the trf here will introduce imaginary part in the simulated data. pr #1 gave a solution to this.