ctmm vs ctmmweb variogram output

[Kerry Kilshaw]

Hi Chris

I hope this email finds you well? I have a very quick question. I'm running the same telemetry data using ctmm and then also with the batch script in ctmmweb which is much faster but the ctmmweb seems to be selecting different best models to ctmm for example OU anisotropic instead of OUF anisotropic and these do not fit the model as well, is this an odd thing to be occurring? Just curious as to why this might be happening?

Thanks

Kerry

[Christen Fleming]

Hi Kerry,

Are you running ctmmweb locally so that the ctmm versions are the same? If so, they should be calling the same functions under the hood and should be giving the same results.

If your likelihood is multimodal, which usually only happens with data quality issues, then you can get different local modes on different runs, but you can check the log-likelihood to see which is the better solution.

Best,

Chris