Can I convert a named vector or list with random-walk perturbations to rw.sd function? about pomp HOT 4 CLOSED

How about the following, where x is your named vector?

x <- do.call(rw.sd,as.list(x))

x is now in a format acceptable to mif2.

How have you been treating initial value parameters?

from pomp.

Thanks for your quick response. Your suggestion works fine.
The initial value parameters are pooled in a named vector which seems to work in mif2.
Thanks a lot for your help!

from pomp.

Hmm, I worry about what "seems to work" might actually mean. You might want to check that those initial value parameters are really being treated as such, i.e., that they are only being perturbed at the appropriate times.

BTW, it is possible to pass the rw.sd as a matrix, which may be better for your purposes. This should be an npar x ntimes matrix with rownames, where the (i,j)-th entry represents the s.d. of the (normal) perturbation to be applied to parameter i before advancing the state from time t[j-1] to time t[j].

from pomp.

In my opinion, x <- do.call(rw.sd,as.list(x)) to define the value of rw.sd may be unsuitable because this setting can not work on the initial parameter value which only need the random walk at the first time-point(ivp).
In my situation, a function to do the MIF2 with the paramnames in pomp varied by the analysis purposes was created. I follow your step to create a matrix which is npar x ntimes with rownames, but how can i pass it to the value of the rw.sd augment?

from pomp.