Warning: "in 'table_lookup': extrapolating at ..." about pomp

[kingaa]

pomp incorporates time-varying covariates via a lookup table. Whenever your model needs the value of a covariate at a particular time, it uses the lookup table provided (via the covar and tcovar options of pomp), linearly interpolating if necessary. It can happen that the required computation depends on values of the covariates outside the time interval covered by the lookup table. When this happens, pomp does linear extrapolation, with a warning.

This can be a problem if the linear extrapolation is not accurate. To avoid this, be sure to give values for the covariates over a time interval that covers the full range of relevant times, i.e., t0 to the time of the last observation.

Note that the ODE integration algorithms underlying trajectory can sometimes evaluate the covariates beyond the last observation time. To avoid warnings (and the potential inaccuracy they signal) in this case, it may be necessary to supply values for the covariates somewhat beyond the final observation time.

Thanks for bringing this issue up. It should have long since been in the FAQ, and now it will be. Please help me polish up the FAQ by helping me clarify the above.

[tomessilva]

pomp incorporates time-varying covariates via a lookup table. Whenever your model needs the value of a covariate at a particular time, it uses the lookup table provided (via the covar and tcovar options of pomp), linearly interpolating if necessary. It can happen that the required computation depends on values of the covariates outside the time interval covered by the lookup table. When this happens, pomp does linear extrapolation, with a warning.

Yes, this part was clear from the documentation and the warning messages themselves.

This can be a problem if the linear extrapolation is not accurate. To avoid this, be sure to give values for the covariates over a time interval that covers the full range of relevant times, i.e., t0 to the time of the last observation.

I think I did things correctly here (covariates are defined, with no missing values, from the t0 [t=0] to the time of the last observation [t=127]). I do have missing values in the observables, but I think those are being dealt with correctly (and those occur, obviously, within the period from t0 to the time of the final observation, so they shouldn't be the source of these warnings).

Note that the ODE integration algorithms underlying trajectory can sometimes evaluate the covariates beyond the last observation time. To avoid warnings (and the potential inaccuracy they signal) in this case, it may be necessary to supply values for the covariates somewhat beyond the final observation time.

Your explanation seems like the most plausible one and, in a practical sense, the suggestion you provide is probably the easiest way to deal with the issue. In my case, the covariates are more or less locally constant, so I duplicated the t=127 row of covariates and set it to t=150 (so the extrapolation is constant) and... yes, the warnings went away.

Thanks for your help and clarifications.

[kingaa]

This is now FAQ 4.5.