News

Short-courses on spatial modeling using R-INLA

posted 4 Nov 2019, 10:01 by Paula Moraga

Paula Moraga from the University of Bath, UK, will deliver training on spatial modeling and visualization using R-INLA next year:

ENAR 2020 Spring Meeting
Nashville, USA
March 22-25, 2020

International Statistical Ecology Conference
Sydney, Australia
June 22-26, 2020

The course materials are based on the book "Geospatial Health Data: Modeling and Visualization with R-INLA and Shiny" by Paula Moraga (2019, Chapman & Hall/CRC Biostatistics Series).


Testing-version 19.09.15

posted 15 Sep 2019, 11:03 by Havard Rue   [ updated 15 Sep 2019, 11:03 ]

There are some changes in this version that is good to point out.
  • I have upgraded the Mac for which the 'Mac-stuff' is being built, and it is now done on MacOSX Mojave. Also, the gcc/g++ compiler is upgraded. If there are any issues related to this, please let us know by email to help@r-inla.org or to the discussion list.
  • To ease the installation of alternative Linux builds, I added a tool that does this (almost) automatically, so please run the function  INLA:::inla.binary.install()   and chose the Linux binary you want to have installed. Let us know if you found this useful and we can export this function in a future release.
H

Problems with openblas

posted 10 Sep 2019, 21:24 by Havard Rue   [ updated 10 Sep 2019, 22:01 ]

There is a problem with the library OpenBLAS  version > 0.3.5, (at least 0.3.6) which for some Linux systems (like Fedora 30), gives 
an error like

List of hyperparameters: 
theta[0] = [Link sn alpha]
theta[1] = [Log precision for group]
Optimise using DEFAULT METHOD


file: smtp-taucs.c  hgid: 750f3d92b580  date: Tue Sep 10 11:23:39 2019 +0300
Function: GMRFLib_factorise_sparse_matrix_TAUCS(), Line: 876, Thread: 0
Fail to factorize Q. I will try to fix it...


file: smtp-taucs.c  hgid: 750f3d92b580  date: Tue Sep 10 11:23:39 2019 +0300
Function: GMRFLib_factorise_sparse_matrix_TAUCS(), Line: 876, Thread: 0
Fail to factorize Q. I will try to fix it...


file: smtp-taucs.c  hgid: 750f3d92b580  date: Tue Sep 10 11:23:39 2019 +0300
Function: GMRFLib_factorise_sparse_matrix_TAUCS(), Line: 876, Thread: 0
Fail to factorize Q. I will try to fix it...


file: smtp-taucs.c  hgid: 750f3d92b580  date: Tue Sep 10 11:23:39 2019 +0300
Function: GMRFLib_factorise_sparse_matrix_TAUCS(), Line: 876, Thread: 0
Fail to factorize Q. I will try to fix it...



GMRFLib version 3.0-0-snapshot, has recived error no [12]
Reason    : The Newton-Raphson optimizer did not converge
Message   : Condition `lambda < 1.0 / lambda_lim' is not TRUE
Function  : GMRFLib_init_GMRF_approximation_store__intern
File      : approx-inference.c
Line      : 2947
RCSId     : file: approx-inference.c  hgid: 750f3d92b580  date: Tue Sep 10 11:23:39 2019 +0300

Error in inla.inlaprogram.has.crashed() : 





There are two ways around this.

1. Force to load the openblas library that is in the INLA package
instead of the one on the system, by doing

        cd ~/R/x86_64-redhat-linux-gnu-library/3.6/INLA/bin/linux/64bit
        cp libopenblas.so.0 first

   where you have to change the path depending on where the INLA package is installed.

2. Bypass the use of OpenBLAS and use the Intel MKL instead

                library(INLA)
                inla.setOption(mkl=TRUE)



H




Postdoc position at KAUST 2020-

posted 5 Sep 2019, 00:57 by Havard Rue   [ updated 5 Sep 2019, 00:58 ]

There is an opening for a Postdoc position in my research group at KAUST for up to 3 years.
Seriously good conditions: excellent research environment, no teaching, free housing, free health insurance,  salary in USD, no local tax, etc.

Contact me if you're interested and we can discuss details. 



Code for the report "Improving Bayesian Local Spatial Models in Large Data Sets"

posted 22 Jul 2019, 05:10 by Havard Rue

Code for the above arxiv-report by A.Lenzi et at, is made available here


Statsref article about INLA

posted 7 Jul 2019, 04:15 by Havard Rue

Sara and Andrea have submitted a contribution (as far as I know) to Wiley Statsref about INLA. Worth reading. 

H

Congratulations!!!

posted 26 Jun 2019, 12:57 by Havard Rue   [ updated 26 Jun 2019, 13:04 ]

Centers for Disease Control and Prevention (Atlanta, USA) awarded the paper


by Diba Khan, Lauren M. Rossen, Holly Hedegaard and Margaret Warner, the Charles Shepard Science award, in category Data Methods and Study Design. This is the biggest scientific achievement award at CDC.





Missing links

posted 10 Jun 2019, 13:04 by Havard Rue

There may be some missing links when the web server was moved from ntnu.no to r-inla-download.org, but it translates rather easily, like

https://www.math.ntnu.no/inla/r-inla.org/papers/Simpson_2016_LGCP.zip

is moved to

https://inla.r-inla-download.org/r-inla.org/papers/Simpson_2016_LGCP.zip



Workshop in Glasgow

posted 21 May 2019, 09:42 by Elias Krainski

There will be a workshop on spatiotemporal modelling using R-INLA from July 3 to July 5 at the Institute of Biodiversity, Animal Health, and Comparative Medicine (IBAHCM) at the University of Glasgow. This workshop will focus on the application of spatiotemporal modelling techniques for ecological and public health data. For details, please contact Dr. Laurie Backer by laurie.baker@glasgow.ac.uk.

A change in the Weibull-likelihood (from version 19.05.17)

posted 17 May 2019, 12:27 by Havard Rue

From (testing) version 19.05.17 there is a change in the internal representation of one of the hyperparameters in the Weibull family likelihoods. The alpha parameter needed internal scaling to avoid some optimisation problems that were experienced. The default prior is for this reason slightly changed, and new user-defined prior has to be given to obtaining the same results as earlier. 

Sorry about that.
H

1-10 of 153