[R-sig-Geo] [FORGED] spatstat - user-supplied list of point patterns in envelope() error

[Rolf Turner]

Dear Joe,

Further to my previous response to your post:  I have now been in touch 
with Adrian Baddeley and he has confirmed that the problem is as I 
conjectured:   The patterns in "pp_list" need to be compatible with 
"xx".  The error message was a bit misleading.  (Adrian has now adjusted 
the error message so as to make it more clear what the problem actually is.)

However there was also a genuine bug (that was revealed by your 
question; thank you!!!) in the software, so even if pp_list had been a 
list of patterns of the appropriate type, you would still have got an 
error.  (A spurious error, caused by the bug; different message of course.)

The bug has been been fixed in the development version of spatstat 
(1.60-1.022).  You will need to install that from github:

library(remotes)
install_github('spatstat/spatstat')

or wait for the next "official release" of spatstat on CRAN.  Such 
releases happen about every three months; the last one was on
23/June/2019.

cheers,

Rolf

> On 12/08/19 8:51 AM, Joe Lewis wrote:
> 
>> Dear list,
>>
>> I'm trying to use a user-supplied list of point patterns in envelope()
>> rather than test against the CSR. Page 400 of Spatial Point Patterns:
>> Methodology and Applications in R states that "the argument simulate
>> can be a list of point patterns". However, I get the following error
>> when I try  to supply a list of ppp:
>>
>>   ekls <- envelope.lpp(Y =  xx, fun = linearK, nsim=5, simulate = 
>> pp_list)
>> Error in envelopeEngine(X = X, fun = fun, simul = simrecipe, nsim = 
>> nsim,  :
>>    ‘simulate’ should be an expression, or a list of point patterns
>>
>> for reference,
>>
>> xx
>> Point pattern on linear network
>> 680 points
>> Linear network with 22 vertices and 21 lines
>> Enclosing window: rectangle = [284086.69, 309740] x [709900, 726547.7] 
>> metres
>>
>> class(xx)
>> [1] "lpp" "ppx"
>>
>> pp_list
>> List of point patterns
>>
>> Component 1:
>> Planar point pattern: 513 points
>> window: rectangle = [284086.69, 309740] x [709900, 726547.7] metres
>>
>> Component 2:
>> Planar point pattern: 422 points
>> window: rectangle = [284086.69, 309740] x [709900, 726547.7] metres
>>
>> Component 3:
>> Planar point pattern: 495 points
>> window: rectangle = [284086.69, 309740] x [709900, 726547.7] metres
>>
>> Component 4:
>> Planar point pattern: 557 points
>> window: rectangle = [284086.69, 309740] x [709900, 726547.7] metres
>>
>> Component 5:
>> Planar point pattern: 576 points
>> window: rectangle = [284086.69, 309740] x [709900, 726547.7] metres
>>
>> class(pp_list)
>> [1] "ppplist" "solist"  "anylist" "listof"  "list"
>>
>> Any ideas why the error is occurring? Thanks.
> 
> It's hard to say without having a *reproducible* example.  Since we 
> don't have access to "xx" or to "pp.list" we cannot experiment to see 
> what's going on.
> 
> One problem that leaps out at me --- although it doesn't seem that this 
> should trigger the error message that you received --- is that the 
> entries of pp_list appear to be *planar point patterns* (of class "ppp") 
> whereas "xx" is a pattern on a linear network (of class "lpp"). 
> Consequently there is a fundamental incompatibility here.
> 
> However I don't see why you would get the error message that you did.  I 
> am CC-ing this email to Adrian Baddeley who has more insight than I, and 
> may be able to point you in the right direction.  Adrian is kind of 
> overwhelmed with work at the moment, so it may be a while till you hear 
> from him.
> 
> If you provide a reproducible example I *may* be able to help.