search

inlabru-org / inlabru

inlabru
https://inlabru-org.github.io/inlabru/
76 stars 21 forks source link

gaussian model fails in 'backend-clean' #61

Closed stanbiryukov closed 5 years ago

stanbiryukov commented 5 years ago

Hi all,

I'm using the latest pull of the backend-clean branch and receive the following error when running a gaussian family model. Fits for other appropriate families proceed fine.

cmpT <- layer ~ -1 + loc_field(map = coordinates, model = spde) fitT <- bru(cmpT, hsN, family='gaussian')

iinla() iteration 1 [ max: 1 ].

Unused entry [INLA.Data1!initial0]=[4]
Unused entry [INLA.Data1!fixed0]=[0]
Unused entry [INLA.Data1!hyperid0]=[65001]
Unused entry [INLA.Data1!prior0]=[loggamma]
Unused entry [INLA.Data1!parameters0]=[1 5e-05]
Unused entry [INLA.Data1!to.theta0]=[function (x) <<NEWLINE>>log(x)]
Unused entry [INLA.Data1!from.theta0]=[function (x) <<NEWLINE>>exp(x)]
Unused entry [INLA.Data1!initial1]=[72.08731]
Unused entry [INLA.Data1!fixed1]=[1]
Unused entry [INLA.Data1!hyperid1]=[65002]
Unused entry [INLA.Data1!prior1]=[none]
Unused entry [INLA.Data1!to.theta1]=[function (x) <<NEWLINE>>log(x)]
Unused entry [INLA.Data1!from.theta1]=[function (x) <<NEWLINE>>exp(x)]

inla_build: [/private/var/folders/ct/r8t91xwj0mb9q5_ytp_c3z340000gp/T/RtmpYwAELK/file21a143a4cf6/Model.ini] contain[13] unused entries. PLEASE CHECK Error in (function (data, model, stackmaker, n = 10, result = NULL, family, : INLA returned message: The inla-program exited with an error. Unless you interupted it yourself, please rerun with verbose=TRUE and check the output carefully. If this does not help, please contact the developers at help@r-inla.org. In addition: Warning message: In inla.inlaprogram.has.crashed() : The inla-program exited with an error. Unless you interupted it yourself, please rerun with verbose=TRUE and check the output carefully. If this does not help, please contact the developers at help@r-inla.org.

stanbiryukov commented 5 years ago

Doing some more debugging I realized this was an issue with the testing branch of INLA I was using and not inlabru.
Gaussian fits work after re-installing the stable version via install.packages("INLA", repos=c(getOption("repos"), INLA="https://inla.r-inla-download.org/R/stable"), dep=TRUE)