• Work on the forthcoming sTiles integration
  • Minor code revision and optimization
  • Code cleanup
  • Revise installation scripts
  • Add PIT to gcpo calculations and output
  • Fixed in issue with feature ‘lp.scale’
  • Minor modifications
  • Code optimization and code improvements
  • Computing initial values are now default enabled
  • Adding SIMDe code for critical parts
  • Linux: added clone_targets builds
  • Added support for aarch64 Linux
  • Code optimization
  • New version of family ‘lavm’
  • Improve startup message
  • Added vignette about ‘Alocal’
  • Code cleanup
  • Added warning for irregular locations used in models ar/ar1/seasonal
  • New likelihood ‘lavm’ (disabled other circular ones)
  • Updated documentation
  • Fix memory leak when factorisation fails
  • Code cleanup
  • Fixed an issue with user-defined integration points
  • Code cleanup
  • Redefinition of circular likelihood ‘vm’, added ‘nvm’
  • Added new likelihoods ‘1poisson’ and ‘1poissonS’
  • Some changes in dynamic library loading order for binaries
  • Some code optimisation
  • Added new external-package to be built INLAtools
  • Fixed an issue in inla.cgeneric.q
  • Fixed a regression in group cv with given groups
  • Added new likelihood ‘gammasv’ (experimental)
  • Fixed a regression with decreasing link-functions
  • Fixed and issue with ‘inla.group.cv’ and passing the argument ‘type.cv’ through
  • Added functions for the range prior in PRW2
  • Added (experimental) latent model PRW2
  • Code cleanup, improvement and optimisation
  • Fixed an regression for family=“lognormal” (appeared on Mac M)
  • Code cleanup, improvement and optimisation
  • Code improvement and optimisation
  • Fixed a regression for 25.09.03
  • Made opt.num.threads default TRUE
  • Improved sTiles integration
  • Code optimisation and cleanup
  • More work on the sTiles integration
  • Improved some documentation
  • Code cleanup and improvement
  • Improved openmp code
  • Code cleanup and improvement
  • Improved data-storage
  • sTiles integration work
  • Fixed a regression in KLD calculations (in output summaries)
  • fmesher deprecation info update
  • Added argument type.cv for group-cv
  • Fixed a regression in family coxph
  • Added adaptive num.threads control at critical places (default off)
  • Added cloglike: user-likelihood code in C (experimental)
  • Improvement for ‘gcpo’ calculation
  • Code optimisation for some output calculations
  • Last (planned) build before summer
  • Improved gcpo performance (TAUCS)
  • More work on the sTiles integration (gcpo)
  • Code improvement and optimization
  • New link function added for circular likelihoods: tan.pi
  • Code development for NUMA aware cache merged (default off)
  • General code improvement
  • New overall CPU-time summary
  • Revise bfgs parameters into two steps corresponding to part I and II in the optimization
  • Implement remap_cache for many rhs solves for TAUCS
  • Code improvement
  • Improved hyperpar optimization, more stable
  • Fixed a regression with lincomb (thx to AZ)
  • Suggested package HKprocess is removed
  • Build with R-4.5
  • Build to make it work with R-4.5 on linux
  • Fixed an namespace error in the barrier models
  • Refer inla.mesh.components to fmesher, from fmesher 0.3.0.9005
  • Add new external package: graphpcor
  • Added info about new releases
  • Updated family ‘mv’
  • Updated code for pc-priors for ‘vm’
  • Code development work
  • New likelihoods (experimantal) ‘vm’ and ‘dgompertz’
  • Minor naming-changes in (experimental) gev-link function
  • Various code improvements and cleanup
  • Fix for variable-parsing in family ‘coxph’. Updated doc
  • Better memory management / first touch
  • Code optimisation
  • Fixed a minor issue with many-model-runs
  • New license: MIT
  • New experimental family: obeta
  • Fixed an issue with family=coxph
  • Code maintenance
  • Make ‘lp_scale’ work on the lin.predictor except the offset
  • More code optimization for larger models
  • Fixed an issue in estimation of the Hessian
  • Fixed an issue in the ‘gev’ link
  • Code improvements targeted for larger models
  • Code improvements
  • Code improvement for survival-families
  • GEV link now accept negative tail parameter
  • Revised family=binomialmix
  • Added cloglog link in family=occupancy
  • Code cleanup and improvement
  • Dependency of ‘sp’ is removed
  • Default malloc.lib is now ‘mi’
  • Code improvement
  • Fixed a bug when using family ‘egp’
  • Update vignette ‘SPDE2mdatasurv’
  • Code cleanup and improvement
  • Changes in family binomialmix
  • Code optimization for Mac Intel/ARM
  • Code cleanup and optimization
  • Some minor changes in family=egp and binomialmix
  • Fix a constant in barrier model (see diff for details)
  • Code cleanup
  • New family: ‘egp’ (experimental)
  • Minor fixes and improvements
  • Code improvement and optimisation
  • Fixed a regression for Intel Mac run-script
  • Link with armpl-library for Mac-ARM64
  • New experimental likelihood ‘binomialmix’
  • Various code improvements
  • Increased number of parameters for model ‘intslope’
  • Various code improvements
  • Fixed a regression from Version 24.06.09
  • Code optimization and development
  • Code improvement
  • Fixed regression using inla.group.cv
  • Better warning for utf8 characters in home-directory
  • More work on control.mode
  • Fixed an issue with A.local
  • Documentation fixes
  • More robust code for `survival models
  • Remove inla.mode=“twostage”
  • Fixed regression in control.mode
  • Support for inla.mdata-objects in inla.stack (experimental)
  • New likelihood exppower (experimental)
  • Fixed a regressin in model iidkd
  • Fixed missprints in group cv vignette
  • Updated plot
  • Added new experimental likelihood
  • Change: do not load INLA-package by default using rgeneric or rprior
  • New (experimental) likelihood: occupancy
  • Optimized family betabinomial
  • Minor code work
  • Increased max.dim for model iidkd from 20 to 24
  • Optimized family betabinomial
  • Fixed (I hope) PARDISO issue for Mac (Intel and M…) due to OpenMP issues
  • New and smaller default value for tolerance.step
  • Fixed a regression in cgeneric.h in cache-macros
  • Revise plot(..., plot.opt.trace=TRUE) plot
  • Rewrote ‘read.graph’ to avoid recursion
  • First build with R-4.4
  • New likelihood mgamma
  • Fixes in inla.knmodels
  • Fixed a regression in VB correction for the mean
  • Updated vignette (barrier global)
  • Minor code changes
  • New likelihood ‘npoisson’
  • Robustify VB mean correction
  • Various updates
  • New likelihood stochvolln
  • Updated vignette barrier_global
  • More code-cleanup
  • Limit dimension of the low-rank VB correction when replications and/or group is very high
  • Code optimization and cleanup
  • Code optimization
  • Fixed issue in rprior
  • More development for gcpo
  • Update vignette about CODA
  • gcpo development (new features) and code improvement
  • likelihood fl changes
  • MatrixModels moved to Imports
  • Code optimization and improvement
  • Updated PARDISO library to version 8.2
  • Various code improvements
  • Fixed an regression since cf5714584c6ea59911526fb3952193bfb9f550c6 for the marginal of one hyperparameter only
  • Improved function inla.spy
  • Minor code edits
  • Code optimization
  • Remove experimental status of many likelihoods and model components
  • Fixed an issue with scopy
  • New vignette added (barrier model on the world’s oceans)
  • Small fix for ‘malloc.lib’ for Mac’s
  • Code cleanup and improvements
  • New option ‘malloc.lib’ in ‘inla.setOption’ (experimental)
  • Fixed a bug: prior means was not accounted for, doing the VB mean correction
  • Added new (experimental) likelihood: tpoisson
  • Some code cleanup
  • Fixed a bug: prior means for fixed effects was not included for compact mode
  • New experimental likelihoods: rcpoisson and fl
  • New arguments to control.vb$f.enable.limit to limit maximum dimension of vb-corrections (which is multiplied with replicates and group)
  • Code cleanup
  • Minor bug fix
  • Improved control of rounding errors for numerical derivatives
  • New experimental link functions, gev and cgec
  • Code cleanup
  • Remove library jemalloc in builds
  • Cleanup code for sparse dot-products
  • Intel MKL no longer supports MacOS and the Accelerate framework is now used instead in the build
  • Some code optimization
  • Optimization of sparse dot-products
  • Code cleanup
  • Fixe issue with ‘rmodemodex’
  • Some code improvement and minor fixes
  • Fixed false warning for ‘scopy’
  • Fixed issue for ‘safe’-option
  • New version of scopy (experimental)
  • Added support for defining priors in R (experimental, see rprior)
  • Some code improvements
  • Updated vignette about group-cv
  • Fixed issue (hopefully) with saturated DIC
  • New option, ‘internal.opt’ in ‘inla.setOption’
  • Fixed regression error: ‘resultmiscmiscnfunc’
  • Made link ‘cloglog’ and ‘ccloglog’ more numerical stable
  • Link with PARDISO8-library (not yet Windows), get license at ‘https://panua.ch
  • Switch to static MKL libraries for Mac Intel
  • New (experimental) likelihood ggaussian and ggaussianS
  • Fix for readlink on older MacOS
  • Development build
  • Development build
  • Fix for sparse-matrix in data-argument
  • Refactor internal functions
  • Cleanup after a merge done by error
  • Improve storage when selection is used
  • Various minor development
  • Development bug-fix
  • Fixed memory leaks
  • Bux-fix to work around regression bug in package ‘excursions’
  • Internal development
  • Fixed issue with inla.call="submit"
  • Increased precision for copy
  • New family stdgaussian (where the precision is fixed to be 1)
  • New family nzPoisson (for Poisson without zero’s)
  • Internal code-cleanup and some minor fixes
  • A new approach to recover from failure (option safe=TRUE)
  • Option control.inla=list(b.strategy="keep") is a new default.
  • Various code improvement
  • Bug-fix release (see commit log)
  • New vignette added
  • Some improvement in the optimiser
  • Fix option control.compute=list(q=TRUE) for the default mode
  • Depend on the new fmesher package. By default, the new methods are used silently instead of the old fmesher standalone binary, and are meant to produce the same meshes etc as before. During a transition period, one can switch between the two code bases, as well as turn on informative deprecation warnings that point to which fmesher R function replaces the existing ones. See https://inlabru-org.github.io/fmesher/articles/inla_conversion.html for more details.
  • Fixed regression with argument cdf=.
  • Various code improvement and minor fixes.
  • Some code rewrite
  • Fixed an issue if joint-priors (control.expert) for R-4.3.
  • Fix vignettes
  • Some improvement in TAUCS linear solve(s)
  • Fixed an issue with Gaussian likelihood and non-identity link
  • Fixed issue with priors with no parameters and inla.rerun
  • Fixed an issue with computing moments from a marginal
  • Update deprecation information (sp/rgdal)
  • Remove CYGWIN code
  • Various namespace and inla/inlabru connection updates
  • Documentation updates to protect the user’s filespace
  • Various code-cleanup work
  • Tolerance was set too low
  • Robustify deviance residuals
  • Improved version of inla.rerun
  • Issues reported in 23.06.12 fixed
  • Documentation and dependency updates (roxygen transition is now complete)
  • Removed AMDC from default reordering (TAUCS only)
  • Internal code reorganisation
  • Optimisation of cpo/pit/po/dic
  • Optimisation of spde2-models
  • Optimisation of poisson/binomial/nbinomial (with default link and no offset)
  • Various optimisation work
  • New toolchain for building on Windows (MinGW)
  • Removed openblas-related code (and specific builds)
  • Removed inla.setOption-options: mkl, blas.num.threads, vecLib, vecLibPath and CYGWIN
  • Added new external package: fbesag
  • First build with R-4.3 and MacOS 12.6
  • Code improvement and optimisation
  • Updated some documentation
  • Reduced memory usage for data-rich models
  • Added new option control.compute=list(save.memory=FALSE|TRUE) for more aggressive savings
  • Change how equal correlations are defined in gcpo
  • Code optimization
  • Reduce memory consumption
  • New diagnostic: result$misc$warnings
  • Added documentation for copy
  • New scopy model added (experimental)
  • Added option hessian.correct.skewness.only in control.inla
  • Minor code changes
  • Fixed inconsistent treatment of offset() and offset= in the output, compared with mode='classic'
  • Small fix in inla.dryrun
  • Change to dynamic libs for Mac
  • Update config-output with vb-variance corrections
  • Added INLAspacetime functions in barrier.R
  • Added inla.dryrun (experimental)
  • Fixed an issue with cgeneric
  • Dot-product optimization for MKL
  • Improved Qinv for TAUCS
  • Added global-constraint(s) to the expert option
  • Fixed an issue with group-cv vigette
  • Fixed issue with model="slm" (thx to RB and VGB)
  • New vigette about CV and group-CV
  • New option to turn off online-optimisations
  • New option to measure gain in dot-product optimisations
  • Various code-cleanup and optimisation
  • Improved code for family="nbinom" and "bell"
  • New vignette (SPDE2mdatasurv)
  • Small change in how saturated deviance is computed
  • Updated family=bell
  • Fixed an issue with selection of nodes for vb.correct
  • Added family=bell (experimental)
  • Workaround for PARDISO issue with diagonal matrix
  • Various code-cleanup
  • Add linear predictor summaries to config
  • Vignette about rgeneric is updated
  • Updated doc about inla.posterior.sample
  • Fixed issue with missing covariance in A-matrix
  • Fix for control.mode=list(..., fixed=TRUE)
  • Various code-cleanup and improvement
  • Remove rgeos dependency
  • Remove rgdal dependency
  • Add new link-function: ccloglog
  • New argument to inla.group.cv
  • safe-mode improved in the inla()-call
  • Various code-cleanups
  • Fixed an issue with ptweedie
  • Code cleanup
  • Sync C++ code from the fmesher package
  • Added log-likelihood output to configs
  • Fixed a regression for model="fgn"
  • Improved code for model="binomial" and logit-link
  • Increased kmax from 10 to 20 for model="iidkd"
  • Fixed an issue with lincombs and PARDISO library
  • Minor changes to integrate cgeneric-code better
  • New stable build
  • Minor changes to integrate cgeneric-code better
  • Some minor code fixed
  • Development build to test cgeneric update
  • Fixed an issue with cache and nested parallelism (mostly cgeneric-related)
  • Added friends-list to gcpo
  • Expanded support for non-unit-radius spheres in fmesher
  • Experimental support for external-packages
  • Experimental support for new 0inflated models
  • Various code cleanup
  • Fix class inheritance checks to use inherits(), to allow e.g. inlabru output to work with inla.cpo, etc.
  • Fixed issue with DIC in experimental-mode
  • rename experimental-mode into compact-mode (experimental-mode will still work)
  • Make compact-mode the default.
  • bug-fix build
  • Added new vignette
  • Added default values of safe, verbose’ and keep to options
  • Added new check for valid arguments (default off for the moment)
  • control.compute=list(residuals=TRUE) provide expected deviance residuals, but it is experimental for the moment
  • Optimisation of gcpo and inner-products
  • Port for upcoming Matrix 1.5-2 change in kronecker()
  • Various minor code improvements
  • Bug fix
  • Internal changes and bug fix.
  • Bug fix
  • Added new experimental cure-model for parametric survival (documentation will come later).
  • The weibullcure-model is removed as its covered by the new cure-model.
  • Optimization: Models with many constraints
  • Optimization: Models with ‘betabinomial’ likelihood
  • Fixed regression in initial value for model="copy"
  • Code cleanup
  • Rebuild package to due broken ‘22.10.05’ version
  • Code cleanup and optimization of dot-products
  • Fixed an issue with ‘control.fixed=list(remove.names=…)’
  • MKL code for nested indexed inner products
  • Computation of initial values for experimental mode
  • Various more minor code improvements
  • Updated version of fmesher, the program which do the mesh-stuff
  • Various code-improvement
  • Some internal (work-in-progress) updates
  • Various code-improvement and cleanup
  • Code-optimization of spde2-models
  • New likelihood model cennbinomial2 (experimental)
  • New likelihood model gaussianjw (experimental)
  • New function inla.group.cv (experimental)
  • Added support for Predictor and APredictor in lincomb for experimental mode
  • Code improvement for spde2 models
  • Add support for ELLIPSOID radius get/set for epsg:4326
  • Improved gcpo with singular covariance matrices
  • Various minor tweaks
  • Make argument f.enable.limit work the replications and groups
  • Fixed an issue with vb variance correction (experimental code)
  • Added new vignette about inla.posterior.sample.eval
  • Improve the openmp code for gcpo
  • Avoid deprecated class coercions in inla.as.dgTMatrix()
  • Fixed an issue with rgeneric
  • New vignette added
  • Various code improvements
  • Fixed regression from Version_22.07.15
  • Code-improvement and minor fixes
  • Fixed regression from Version_22.07.15
  • Code-improvement and minor fixes
  • Code-improvement/fix
  • Code-improvement
  • Changed stopping criteria for VB corrections
  • Code-improvement
  • More work on the new experimental feature (internal)
  • Code-improvement
  • Adding new experimental feature (internal)
  • Added more to gcpo-output
  • Code-improvement
  • Bug-fix release
  • The mode in the output summary is back.
  • Family gev is disabled, using bgev instead
  • Code improvement, cleanup and bug-fixes
  • Changes in work-in-progress features
  • Minor changes to move to R-4.2
  • Code improvement, cleanup and bug-fixes
  • More work on the gcpo-feature
  • Removed threadprivate from code (almost), which was a huge rewrite.
  • Made linesearch in the optimization more robust and stable
  • Improved stupid-search
  • Internal changes in the smart-optimization
  • (This is likely the last build with R-4.1 before we switch to R-4.2)
  • Added theta-correction for gcpo computations
  • Code improvement and cleanup
  • Improve gcpo-handling of singular cases. group.size=-1 now gives the CPO.
  • Bug-fix release
  • Bug-fix and code improvement release
  • Bug-fix and code improvement release
  • Bug-fix release
  • Bug-fix release
  • Bug-fix release
  • Bug-fix release
  • Improved the code for output which is now way faster for data-rich models
  • Internal changes in how integrals are computed and how marginals densities are represented. This which will make the results tiny tiny different from earlier versions.
  • Code-improvement for data-rich models
  • Minor bug-fixes
  • Fixed regression issue with strategy="laplace" which has been broken from version 22.02.16 (ok in 22.01.25).
  • The value of mode in summary output is not longer computed for some densities (like for mixtures and non-linear transformation), as its pretty expensive to do so accurately. It is then shown as NA. This might be fixed later.
  • Code cleanup and a bug-fix
  • Dense matrices are now store column-wise in cgeneric
  • New example added to cgeneric, see vignette
  • Changed default integration stratey for experimental mode
  • Option safe=TRUE is now default

  • Code cleanup

  • Code optimization for TACUS

  • Nested parallelism is now enabled on Windows

  • Issue with gcpo and safe=T fixed
  • Fixed issue with A-matrix in gcpo
  • Better TAUCS performance: improved storage scheme and new code tp allow for linear solves with many rhs.

  • gcpo-improvement. The default is now group.size=1, and it acts like cpo.

  • Fixed an issue for propagation an error from the cpo-calculations

  • Various smaller changes

  • Improved the TAUCS-interface performance.
  • Various code improvements.
  • New options for the group-cpo feature.

  • Fixed CRAN-check issues.

  • VB correction for experimental mode do now include nodes from all model components, not only the short ones and fixed effects.

  • New group-cpo feature, very experimental at the moment (internal use only).

  • Parts of the VB code has been improved for simplicity and speed.

  • inla.call="remote" now use zstd for parallel compression, meaning that zstd must be available on both sides.

  • PARDISO license help and startup msg has been disabled.

  • Minor fixes, code improvements and cleanup

  • New feature inla.cgeneric.q

  • New and updated PARDISO library

  • Code improvements and cleanup

  • Fixed regression bug in knmodels
  • Code improvements and cleanup
  • Fixed bug in the variable expansion in some arguments that could give different results than before
  • Added new variant for nbinomial likelihood
  • Added graph$cc$mean to help defining intercepts for disconnected regions.

  • Can now define the license key for PARDISO directly as inla.setOption(pardiso.license="<KEY>")

  • Print total RAM in Gb in the inla-program output, for easier debugging of these problems.

  • Using inla.call="remote" will now return error code that can be controlled by try, so that r=try(inla(..., inla.call="remote")) should work as intended.

  • Add a refine step for parallel.linesearch, so if this is enabled, then it will do a restart at the end with this option disabled.

  • Minor changes when option safe is enabled.

  • Fixed regression error with plot(result, plot.prior=TRUE)

  • Fixed a regression error from version 21.12.21 which expanded some variables incorrect.

  • Option safe in function inla

  • Minor bug fixes

  • Improved handling of NAN and INF
  • A bug-fix for matern2d model
  • Various code improvements
  • Minor bug fixes for cgeneric for Windows
  • A cgeneric interface like rgeneric but with C-code. This is for internal testing only.

  • Various code improvements and minor fixes

  • Allow for probit cdf in likelihood model pom

  • Various code improvements

  • Small fixes
  • added new link powerprobit. Not completely done yet.
  • Remove jemalloc-option that creates warning message for Mac
  • Minor internal changes
  • Add jemalloc options for Linux/Mac
  • Minor internal changes for robustness
  • Upgraded compilers to gcc-11 on Windows
  • Option for parallel linesearch has been renamed into parallel.linesearch in control.inla.

  • This is fix for a build-error in the Windows binary for version 21.11.14

  • Various minor internal code changes and improvements

  • Parallel linesearch implemented, see option bfgs.version in control.inla. This option is highly experimental at the moment and work in progress, and should not be used.

  • Change default settings for family="tweedie"

  • Implemented an adaptive parallel/serial version of Qx

  • Improved perforance for parallel linear solve with PARDISO

  • Use stable Legendre polynomial evaluation for spherical covariances

  • Add support for Fedora Linux 35

  • Fixed dic.sat when all response are NA

  • Various internal code changes and improvements

  • New experimental feature lp.scale. No good documentation yet, but that will come soon.

  • Robustify inla.iidkd.sample function for numerical singular matrices

  • Corrected inla.stack.join handling of factor variables that in some cases converted variables to/from factors unintentionally.

  • inla.mode="experimental" got its fitted values back.

  • Internal code improvement
  • Better special number

  • New option to inla.iidkd.sample

  • New latent model iidkd which use a different parameterisation than iid3d and similar ones.
  • Fixed crash when using int.strategy="user" with inla.mode="experimental"
  • Minor fixes
  • Fixed crash when using linear combinations with inla.mode="experimental"
  • Minor fixes
  • Added option scale to family="xbinomial"
  • Various bug-fixes
  • If control.predictor=list(compute=TRUE) then the marginal densities for the linear predictor and fitted values are computed, but the new default is that only the summary statistics are returned not the marginal densities. This is because the storage requirement for the marginals can be substantial. The marginal densities can be returned setting control.compute=list(return.marginals.predictor=TRUE)

  • The argument control.results are removed as it was not really in use.

  • The new default value for b.strategy is b.strategy="skip". See ?control.inla for details

  • New option inla.timeout, similar to fmesher.timeout; see ?inla.setOption

  • CPO-marginals are returned in when control.compute=list(config=TRUE)

  • There is a change how a (numerical) singular constraint is treated, hopefully this new way is more robust.

  • New option inla.mode, which define how to arrange the internal model formulation. One of "classic", "twostage" and "experimental". The default is "classic", which is unchanged behaviour compared for earlier versions. The other two are highly experimental for the moment. See ?inla

  • Massive code clean-up and some minor fixes.

  • Added support for Mac M1 (native build with R-4.1)

  • New twostage options (highly experimental)

  • Massive code overhaul

  • Various minor fixes

  • Some minor bug-fixes and code improvements
  • Mac only: Added optional path to vecLib BLAS and LAPACK libraries.

  • Adding argument .special to inla.surv-object.

  • Some code improvement

  • Mac only: Link with vecLib BLAS and LAPACK libraries by default. Turn off with inla.setOption(vecLib=FALSE).

  • Adding an experimental improvement for strategy=simplified.laplace (not enabled by default)

  • Some code improvement

  • General code improvement
  • New family stochvolsn

  • New family cenpoisson2

  • Improve robustness for survival models

  • General code improvement

  • Changes in inla.jp; see help-page for details.
  • New family gompertz

  • inla.binary.install() now do md5-checksum check

  • Code optimization for barrier models

  • Internal code changes for family=tweedie
  • Internal code changes
  • Fixed some missprints in doc
  • Trying to get vigettes to work again
  • Added family logperiodogram back in

  • Added new family agaussian

  • Improved initial values for x

  • Improved computations for saturated likelihood

  • Various minor code cleanup

  • Default strategy is now simplified.laplace for smaller models (< 5000 nodes) and adaptive for larger ones

  • Minor bug-fixes

  • Improve parallel rerodering for PARDISO
  • Improve coxph for large data
  • Improve internal CPU-timing output
  • Use METIS5 with PARDISO
  • Family fmri

  • Speed improvements for family tweedie

  • New family tweedie

  • Optimization work, code improvement and maintenance

  • Added a new experimental feature for running PARDISO in parallel

  • Various minor fixes and improvements

  • Bug-fix release only
  • Added new section to the vignette about rgeneric

  • Option optimize in inla.rgeneric.define

  • Cleanup in some output format.

  • Various internal changes.

  • Code improvement and optimization for rgeneric.
  • Some OpenMP improvements
  • New family gammajw

  • Fixed an OpenMP deadlock case, using rgeneric

  • PARDISO version 7 is included for Mac and Linux.

  • Improved paralellism and nested paralellism

  • Speedup improvements, especially for models with many constraints.

  • Improved default plot(result). Argument cex=.. will now work

  • New shorthand feature in inla.posterior.sample.eval for extracing model components from samples

  • Added fmesher.timeout option, see ?inla.setOption

  • Using a non-zero seed in inla.posterior.sample will now force serial computations.

  • A lot of internal code cleanup and improvements

  • New likelihood poisson.special1

  • Minor code cleanup

  • Minor changes in code/doc to adapt to R-4.0

  • Bug fixes and some optimisation work
  • Likelihood model sn is redone.

  • Various minor changes.

  • Model iid added to possible models for the baseline hazard in the Cox-ph model

  • Likelihood models sn and sn2 have been replaced with a new version sn. The parameterisation is different and now done correct.

  • The likelihood models sn and sn2 are now disabled. They need a rewrite to be done right.

  • For Linux, then build so-libraries are loaded before the system ones. This behaviour can be reverted by setting the environment variable INLA_NATIVE_LD_LIBRARY_PATH.

  • Dimension for CCD integration have been increased from 38 to 52.

  • Default prior for the intercept and the skewness in the skew-normal link have changed.

  • Various minor changes

  • A small bug-fix with argument -t
  • Some improvements in the install scripts
  • More work on the nested parallelism stuff
  • Argument num.threads="A" now means num.threads="A:1"
  • Some changes for nested paralellism
  • There is a minor change in inla.qsample and inla.posterior.sample. If argument seed!=0 then serial mode is now forced. Earlier, an error would be raised if parallel mode was also requested.

  • Argument remove.names in control.fixed

  • Bug-fixes for nested parallism

  • Some optimization improvement

  • Some minor fixes in the install

  • Linux builds now link with PARDISO version 7 (beta)

  • MKL is now default for both Mac and Linux. If this cause any issues, revert back by setting inla.setOption(mkl=FALSE).

  • As of today, the TAUCS library does not play well with MKL on Mac for some reason, and this is silent accounted for. It is possible to bypass the internal check, and if that becomes in issue, just define mkl=FALSE as above.

  • Updated some install scripts

  • Various minor fixes
  • MKL binaries are now linked with libiomp5
  • Changed to the gcc-10 compiler suite for Ubuntu 1804 and 2004.
  • Fixed an issue if the optimiser does not move and the use of directions
  • Minor fixes
  • Code is prepared for nested parallism, which most gain obtained when using the PARDISO library

  • Argument num.threads is now in the format A:B, where A threads for the outer layer and B threads for the inner layer. And this applies for several functions.

  • The speed have improved, mostly when using the PARDISO library. Hopefully, nothing is broken.

  • likelihood beta now allow for cencoring near 0 and 1, see inla.doc("beta", section="likelihood")

  • Default value for control.inla$h is now 0.005, which replace the previous value of 0.01. We might change this later, but we have enough experience to know that 0.01 is slightly to large.

  • control.inla$optimise.strategy="smart" is now default. By construction, this should be both faster and more robust. Using "plain" reverts back the old behaviour.

  • control.inla$use.directions=TRUE is now default. This option estimate numerical gradients/Hessian in directions with more change than for coordinate-wise directions.

  • Add new experimental feature control.inla=list(use.directions=TRUE/FALSE)
  • New experimental optimise strategy control.inla=list(optmise.strategy="smart" that is hopefully faster and as safe as the default one. Maybe even more safe and robust.

  • More work on the nested paralellism

  • More work on nested parallelism and some code changes
  • Code changes to reduce memory usage
  • Linux/Mac binaries are now linked with jemalloc
  • Improved nested parallelism and some code changes
  • More work on nested parallelism
  • Option control.inla$lincomb.derived.only is now disabled.

  • Testing nested parallelism openmp.nested with num.threads="A,B". Work in progress

  • plot(result) will now produce a plot of the CPO/PIT for each likelihood family (if available), instead of a joint plot as earlier.

  • New vignette about jmarginal

  • Added new likelihood models zeroinflatedcenpoisson0 and zeroinflatedcenpoisson1

  • Link-model sn is updated, as well as the PC-prior for the skewness therein, and the added intercept model.

  • Revised the PIT calculations for family cenpoisson

  • Code rewrite to (try to) prevent Inf for DIC calculations

  • Minor fixed in inla.binary.install

  • Build-script changes and misc fixes
  • Packages mpoly and symmoments are added to the Suggests-list.

  • Added info about inla.prune() to the startup message

  • More features for jmarginal added

  • Build-scripts fixes

  • Fix for a rare fmesher issue

  • Improved the code for the DIC calculations to make them more stable

  • Some improvment in the PC-prior for the SN-link

  • Fixed inla.link.sn to vectorise over argument a as sn-package do not do that properly itself.
  • Package built with R-4.0
  • Vignette added for family=bGEV
  • Some internal changes due to the migration to git
  • Added Qprior.diag to the output when config=TRUE. The off-diagonals of this matrix are the same as Q in the same configuration, so only the diagonal of Qprior is stored.

  • Added some internal experimental code

  • PARDISO interface: internal check added

  • Fixed an wrong assert with family=bgev

  • Improve some code in the PARDISO interface
  • Improved the computation of the third derivative in the log likelihood.
  • Improved WKT support for PROJ6/GDAL3
  • Support for PROJ6/RGDAL3 for handling CRS information for spatial objects.
  • For the intslope-model: made all gamma’s default fixed to 1, so its similar in style the copy-feature.

  • Added argument constr it inla.rjmarginal

  • Added argument ask to function inla.prune

  • Argument cyclic=TRUE in f() should not set constr=FALSE when default is constr=TRUE

  • Change the scale.model=TRUE code for RW1/RW2 so the scaling for the continous case is the same as for the discrete case when the locations are eqvidistant.

  • Disable link sslogit

  • Fixed an issue with model besag2
  • Fixed an issue with plot(r,plot.prior=TRUE) for some priors
  • Remove the experimental status of inla.posterior.sample.eval

  • Added function inla.prune which will remove binaries not supported by the running OS, to reduce the size of the package.

  • Added method summary and print to class inla.jmarginal

  • Add check for NA/NaN/Inf in mesh creation input ocations

  • Make sure that skewness is not to high in inla.posterior.sample

  • Added new argument tag to inla.coxph

  • inla.rjmarginal.eval, to evaluate samples from a join approximations

  • Names of samples are now “sample:1”, “sample:2”, and should be coherent over all functions. Similar, their contents, its like “x:1”, “x:2”, etc.

  • Fixed a bug setting prior for the log baseline hazard in inla.coxph

  • Small fix so that result$mode$x is written out in the case where nhyper=0 and num_threads>1
  • Minor internal changes.
  • Added link loga. Not yet documented.

  • First try on a new feature to more easily approximate the joint marginal for a subset of the latent field. This is a new option selection and corresponding inla.rjmarginal() to sample from it.

  • Added check that model="linear" is not used with replicate or group, which is not intention.

  • MCMC mode is now disabled

  • Skewness correction is now back as default, in inla.posterior.sample()

  • Added family xbinomial that allow non-integer response.

  • Likelihood model bgev add (not yet complete), and was renamed from the experimental likelihood model gev2.

  • If inla.call="remote" is set, then INLA:::inla.call.builtin() is used if inla.qinv() and/or inla.qsolve() are used while constructing the model.

  • Updated file jointdataCD4.rds in exampledata/
  • Fixed a bug in the PIT calculations for the zeroinflated, type 0, of poisson, binomial and nbinomial.
  • Added option b.strategy in control.inla to control what to do with the linear term when the cmin option is in effect

  • Added in-interval observed event in inla.surv

  • Added dplyr as suggested package as dplyr::bind_rows can replace INLA::inla.rbind.data.frames

  • Added argument E, or log(offset), to likelihood gammacount, so its equal to family poisson for alpha=1.

  • Minor changes

  • Added a check that discrete observations are indeed integers, like for Poisson, Binomial, etc

  • The function inla.binary.install is now exported.

  • Added new likelihood family, xpoisson, which allows continous response: see the documentation for details (and note the error-check now done for discrete observations)

  • Added new likelihood dgp (discrete generalized Pareto)

  • Code clean-up (contpoisson and qcontpoisson)

  • Made inla.pardiso.check() a bit more informative if there is an error.

  • Improved documentation of inla.posterior.sample and inla.coxph
  • Fixed an issue with NA data in the family gev2
  • Updated some documentation about the pc.gevtail prior.
  • Reverted inla.posterior.sample back to the old version, the new experimental version is available as INLA:::inla.posterior.sample.new
  • Error in Epil data-set, y[31] should be 23 not 21.
  • Updated the vignette about the multinomial distribution

  • New experimental windows binary built with x86_64-w64-mingw32-gcc, version 7.3, and linked with the pardiso library. Its stored in bin/windows/experimental

  • Updated inla.qreordering and updated leuk-demo.R example file (and the corresponding zip-file).

  • Cache values of qgamma to speedup Gamma quantile regression
  • Added a scaling constant for the precision parameter in the qkumar likelihood (to avoid instabilities). See updated documentation for details.

  • inla.posterior.sample now correct for possible skewness by default: see ?inla.posterior.sample for details.

  • Likelihoodmodel betabinomialna
  • Default prior for the tail parameter in likelihood model gp, have changed to pc.gevtail, and the name change from shape to tail. It is now required to define a interval for the tail parameter, similar to pc.gevtail.
  • Code-improvement for the loggamma-function
  • barrier.R updated (minor fix and code edits)
  • Disable some debug output
  • Fixed a bug in the nmixnb likelihood.
  • Preserve names in inla.posterior.sample.eval if present.
  • More work on the skew-normal link model
  • INLA:::inla.binary.install() is a new interactive tool to install alternative Linux builds.
  • Added skew-normal link-model sn for binary data, with its PC-prior
  • Added robit link model.

  • Improved the stability of the saturated deviance calculations

  • Fixed INLA:::inla.is.list.of.lists to cover the case where the arguments are a list of named lists

  • New (experimental) likelihood: gev2

  • Fixed, again, an issue with (parallel) PARDISO and many linear combinations.

  • Minor code changes in doc.R

  • Removed must-be-enabled warnings in some surival models, from Oct 25 2017

  • Added PC-prior for the Weibull likelihood models. The prior is derived for variant = 1, which is the good parameterisation.

  • Added missing to.theta and from.theta functions in likelihoods sn and sn2

  • Fix some documentation in marginal.R (refering to the obsolete function inla.marginal.transform)

  • Fixed an issue with (parallel) PARDISO and many linear combinations.

  • Set StagedInstall:no to work around installation problems for MacOS and R-3.6
  • The internal parameterisation of the alpha-parameter for the Weibull likelihood familes, has been redefined/scaled, to fix some optimisation issues. This means that the default prior has changed (a little) and user-defined priors has to change to account for this new internal parameterisation (sorry about that). See the documentation for details.
  • Option short.summary will use a version of summary with less output, maybe more suitable for Markdown documents.
  • Added exampledata directory for various example datasets

  • Code cleanup and improved some input-error checking.

  • Fixed an error in the cache-system for model="rgeneric" and model="dmatern". Most notably with option openmp.strategy="pardiso.parallel".
  • Removed the weight correction for the computation of the cpo for int.design="user.expert"
  • Option int.strategy="user.expert", see the vignette about user-defined integration points.
  • Merge also cpo and po results in inla.merge()
  • Fixed an issue with AR-model and group
  • Small fix in model dmatern
  • Redirect error output of some warning messages in the remote-feture section from MacOSX to Linux.
  • Faster return when mu is zero for rgeneric
  • Changed from PARDISO.PARALLEL to PARDISO.SERIAL in inla.qsample
  • Optimize the nhrs for inla.qsolve for PARDISO
  • Several fgn-models are now fine
  • Fixed CPU timing with the PARDISO library
  • Do not need to optimize reordering when PARDISO is used.
  • Fixed input-test using inla.qsample with selection-argument.
  • Added back family = "normal" which is now translated to family = "gaussian" internally.
  • More work and fixes in inla.merge
  • Simplied print.inla output

  • New method merge and function inla.merge, for merging inla-objects

  • Store control.family after processing, in the result$.args argument, not just the calling value.

  • New parameter for Gaussian likelihood: Fixed offset in the variance.

  • Updated envir definition in the rgeneric documentation and examples.

  • Removed testing code for likelihood model testbinomial1

  • Added new likelihood gamma.surv

  • Cleaned up the use of temporary dir and files

  • General code clean-up

  • Increased maximum number of covariates in likelihood models nmix and nmixnb from 10 to 15
  • Added a new test-script
  • New models, loggamma and mloggamma in mix.

  • Minor changes in some build scripts.

  • New option mkl in inla.setOption() to chose MKL-buildt binaries.
  • Linux binaries now buildt with Ubuntu1804.
  • MKL-versions are included for MacOSX, and Linux (both dynamic and static).
  • New latent model intslope
  • Improved control.mix interface and code
  • Likelihood model nbinomial2
  • New function inla.priors.used
  • Export class inla
  • New latent model: dmatern

  • Improved the numerics for computing the scaling of the RW1 and RW2 models.

  • New option control.inla=list(tolerance.step=), to control the RMS of the step-size for the inner optimization.

  • Changed, slightly, the initial values for the exponent in the Weibull likelihood models, to a value close to zero instead of zero.

  • New vignette about how to deal with multinomial data.

  • Added option verbose to inla.qsample() and inla.posterior.sample()

  • Performance improvement using the PARDISO library
  • Argument selection in inla.posterior.sample and inla.qsample.
  • Added new function inla.pardiso.check()

  • Added COPYRIGHTS file

  • Separated the quantile link for the binomial response, into individual (model="quantile") and population (model="pquantile")

  • Added new strategy control.inla=list(strategy="adaptive") which use the simplified.laplace approximations for fixed effects and low-dimensional model components, and the gaussian approximation otherwise. The argument adaptive.max in control.inla determines what is low-dimensional in this context (default 10).

  • Removed some code not used anymore

  • NEWS page created (see news(package="INLA"))
  • Added vignette about the conditional logit model (thanks to Stefani Muff)
  • Fixed missprints in the documentation for model ar1c (Thanks to Virgilio Gomez Rubio)
  • Fixed documentation about argument blas.num.threads in inla()
  • Package built with R-3.5, both stable and testing
  • Package built for R-3.4, both stable and testing.
  • Likelihood model pom (proportional odds model)