Skip to content

RandomFields: Randomness in realisations of sequential simulation?

2 messages · Andre Jung, Edzer Pebesma

Original
Andre Jung
# 
Dear all,

I've some general question regarding the implementation of
randomness in RandomFields and sequential simulation.

(1) Is there a difference between running
    krige(nsim = 5, ....)
     and running
    krige(nsim = 1, ....)
     for 5 times?

(2) Does repeating of the simulation with the same random number seed
     (how to keep the seed?) lead to the same realisations? And: how
     random is the path?

(3) Can I fix the parameters (which ones?) to reproduce the same
     realisations a later time?


Another question is: How can I generate an 'e-type' of, let's say 10, 
realisations, and do some post-simulation analysis like defining 
probability to be above or below a certain threshold?


Many thanks in advance!

best,
Andre
14 days later
Edzer Pebesma
#
Andre Jung wrote:
My answers only apply to krige in package gstat; I believe this is what
you refer to with "sequential simulation".
in the second case, 5 different random paths are followed.
?set.seed
?set.seed
?apply ; I believe the asdar book has examples (see www.asdar-book.org)