[R-sig-Geo] alltypes & envelopes

[Adrian.Baddeley at csiro.au]

Thus spake Jan Quets <Jan.Quets at ua.ac.be>:

> with "envelope" it is possible to generate a summary statistic (e.g. G-function) 
> of a given unitype point pattern, together with a monte-carlo-envelope of a 
> null-model.

Again this is a question about the package 'spatstat'.

> with "envelope" it is also possible to save the monte-carlo-generated patterns 
> of a null-model by setting the argument "savepatterns=TRUE"

> with "alltypes" it is possible to generate a summary statistic (e.g. n x n G-functions) > of a given n-multitype point pattern, together with monte-carlo-envelopes 
> of a CSR-model.

> But is it also possible with "alltypes" to generate monte-carlo-envelopes of 
> a non-CSR-null-model, for instance by setting simulate as an argument in "alltypes" 
> equal to an expression which generates a n-multitype null model pattern, 
> "nsim" times, like is the case with "envelope"?

Yes. 
According to help(alltypes), any additional arguments ("...") are passed to envelope(). 
 e.g.

    data(amacrine)
    z <- alltypes(amacrine, Gcross, envelope=TRUE, simulate=expression(rlabel(amacrine)))
    plot(z)

Since 'simulate' is not the name of any formal argument of 'alltypes', the argument 'simulate' is passed through to envelope(). That means each call to envelope() is executed with the argument simulate=expression(rlabel(amacrine)). 

Note that 'simulate' is an expression that is evaluated if we want to generate ONE simulated realisation. It will be evaluated nsim times to get nsim random patterns.

> Is it possible (as it is with "envelope") in "alltypes" to set the argument 
> "savepatterns=TRUE", and hereby saving the generated "nsim" n-multitype
>  point patterns of the null-model, which can be used to generating another
>  envelope in an other summary statistic (e.g. J-function)?

Yes.
See above.

   data(amacrine)
   z <- alltypes(amacrine, Gcross, envelope=TRUE, 
              simulate=expression(rlabel(amacrine)),
              savepatterns=TRUE)

To extract the simulated patterns for the plot panel in (say) the top left position,

    z11 <- z[1,1]
    s11 <- attr(z11, "simpatterns")

====
Prof Adrian Baddeley (UWA/CSIRO)
CSIRO Mathematics, Informatics & Statistics
Leeuwin Centre, 65 Brockway Road, Floreat WA 6014, Australia
Tel: 08 9333 6177 | Fax: 08 9333 6121 | Mob: 0410 447 821