Issue #143 , filed late last year, raised the question of whether, or when, we would get to Eigen 5.0.*. As detailed in the nearly 100 messages in that issue, it took a while.
As of September 2026, and thanks to the stan team, we can build against a very large number of packages---but a rather large number remains problematic. Per a reverse dependency run (summarised here ), we now have '472 successes, 69 failures, 24 skipped'. This issue will aim to regroup in sub-issues the required work on different packages. As of right now, one PR has been filed, and more will follow.
Help is definitely welcome.
If you can work on one of the packages, please do. In general the CRAN version 0.3.4.0.2 passes for the CRAN dependencies, and the candidate version 0.4.9.9-2 at r-universe (and here at GitHub) represents the release candidate with Eigen 5.0.1 which breaks some packages. The key will to adjust the affected packages to that they still build under the CRAN and the r-universe version permitting a smooth upgrade of the RcppEigen package.
Individual packages:
admixr2: Reveals a latent bug in softmax(), PR submitted Address a subtle bug revealed by Eigen 5.0.1 LeidenPharmacology/admixr2#158
AsyPeer uses Eigen::all, PR at Replace Eigen::all with Eigen::indexing::all MathieuLambotte/AsyPeer#1
bWGR uses Eigen::all and Eigen::MappedSparseMatrix, PR at Adjust code to Eigen 5.0.0 alenxav/bWGR#3
CDatanet uses Eigen::all, PR at Replace Eigen::all with Eigen::indexing::all ahoundetoungan/CDatanet#2
clustTMB uses unqualified array<>, PR by @andrjohns already at Fix compatibility with Eigen 5 Andrea-Havron/clustTMB#35
Colossus needs a new functor_traits, PR at For Eigen 5.0.* add functor_traits ericgiunta/Colossus#41 , merged 2026-09-28
crctStepdown needs a new function_traits, PR at For Eigen 5.0.* add functor_traits samuel-watson/crctStepdown#6 , merged 2026-09-30
enrichit needs Map<SparseMatrix<>>, PR at For Eigen 5.0.* switch to Map<SparseMatrix> YuLab-SMU/enrichit#2
FactorHet needs Map<SparseMatrix<>>, PR at For Eigen 5.0.* switch to Map<SparseMatrix> mgoplerud/FactorHet#2
fastglm needs Map<SparseMatrix<>>, PR at For Eigen 5.0.* switch to Map<SparseMatrix> jaredhuling/fastglm#18
fastQR uses Eigen::all, no repo we could find, patch emailed
fdaPDE needs Map<SparseMatrix<>>, no repo we could find, patch emailed
forestBalance needs Map<SparseMatrix<>> For Eigen 5.0.* switch to Map<SparseMatrix> jaredhuling/forestBalance#1
galamm needs Map<SparseMatrix<>> and (vendored) autodiff change, PR at For Eigen 5.0.* switch to Map<SparseMatrix>, update autodiff ropensci/galamm#260
gedi2 needs Map<SparseMatrix<>>, PR at For Eigen 5.0.* switch to Map<SparseMatrix> csglab/gedi2#28 ; merged 2026-10-02
gKRLS needs Map<SparseMatrix<>>, PR at For Eigen 5.0.* switch to Map<SparseMatrix> mgoplerud/gKRLS#5
glmmrOptim needs a new function_traits, PR at For Eigen 5.0.* add functor_traits samuel-watson/glmmrOptim#16 , merged 2026-09-30
gRbase needs Map<SparseMatrix<>> plus one change, patch emailed; (gRain and gRim build with updated gRbase)
GTEs can remove MappedSparseMatrix, PR at For Eigen 5.0.* remove MappedSparseMatrix yzhou1999/GTEs#3
hts amplified existing PR Use Map<SparseMatrix<.>> over MappedSparseMatrix<.> earowang/hts#71
lime needs Map<SparseMatrix<>>, PR at For Eigen 5.0.* use Eigen::Map<Eigen::SparseMatrix>> tidymodels/lime#205
mcmcsae needs Map<SparseMatrix<>>, no repo we could find, patch emailed
MetricGraph needs Map<SparseMatrixM<>, PR at For Eigen 5.0.* switch to Map<SparseMatrix> davidbolin/MetricGraph#25
mmrm GH repo already covers this, release needed, issue filed New CRAN release to aid with Eigen 5.0.1 transition openpharma/mmrm#575
nebula needs Map<SparseMatrix<>>, PR at For Eigen 5.0.* switch to Map<SparseMatrix> lhe17/nebula#77
ngme2 needs a CHOLMOD tweak, PR at Updated CHOLMOD support for Eigen 5.0.1 davidbolin/ngme2#11 , merged 2026-09-30
oem needs Map<SparseMatrix<>>, PR at For Eigen 5.0.* switch to Map<SparseMatrix> jaredhuling/oem#26
OptimalBinningWoE GH repo version no longer relies on RcppEigen, CRAN upload needed, issue filed New CRAN release to aid with Eigen 5.0.1 transition evandeilton/OptimalBinningWoE#19 ; new version 1.14.0 on CRAN 2026-09-30
pipq(-r) needs a NOEXCEPT update already in (C++) pipq, issue filed Updated piqp for Eigen 5.0.1 PREDICT-EPFL/piqp-r#6
ppforest2 uses Eigen::all, PR at Replace Eigen::all with Eigen::indexing::all andres-vidal/ppforest2-core#2 , merged 2026-09-30, new version on CRAN 2026-10-01
QuantilePeer uses Eigen::all, PR at Replace Eigen::all with Eigen::indexing::all ahoundetoungan/QuantilePeer#1
robCompositions need cassert header, PR at Add cassert header needed when building with Eigen 5.0.1 matthias-da/robCompositions#29
sanic needs Map<SparseMatrix<>>, PR at For Eigen 5.0.* switch to Map<SparseMatrix> nk027/sanic#6
serrsBayes needs Map<SparseMatrix<>>, PR at For Eigen 5.0.* switch to Map<SparseMatrix> mooresm/serrsBayes#4 , merged 2026-10-02
spaMM needs Map<SparseMatrix<>>, no public repo we could find, patch emailed
SpatialGEV needs pow() adjustment, PR at For Eigen 5.0 add a missing element-wise pow() meixichen/SpatialGEV#26
spGARCH needs Map<SparseMatrix<>>, no public repo we could find, patch emailed
stelfie uses unqualified array<>, PR by @andrjohns already at Fix compatibility with next RcppEigen cmjt/stelfi#8
tsgarch needs pow() adjustment, PR at For Eigen 5.0 add a missing element-wise pow() tsmodels/tsgarch#9
tsissm uses Eigen::all, PR at For Eigen 5.0.1 switch to Eigen::indexing::all tsmodels/tsissm#2
TwoStepSDFM uses Eigen::all and Eigen::last, PR at Replace Eigen::{all,last} with Eigen::indexing::{all,last} SiSanchopancho/TwoStepSDFM#2
ungroup needs Map<SparseMatrix<>>, PR at For Eigen 5.0.* switch to Map<SparseMatrix> mpascariu/ungroup#18
WpProj needs Map<SparseMatrix<>>, PR at For Eigen 5.0.1, use Map<SparseMatrix> ericdunipace/WpProj#8
More to follow, see full list
NB: There is a second set of package failing with segmentation violation because StanHeaders / rstan needs an update. That is not covered by this ticket.
Issue #143, filed late last year, raised the question of whether, or when, we would get to Eigen 5.0.*. As detailed in the nearly 100 messages in that issue, it took a while.
As of September 2026, and thanks to the stan team, we can build against a very large number of packages---but a rather large number remains problematic. Per a reverse dependency run (summarised here), we now have '472 successes, 69 failures, 24 skipped'. This issue will aim to regroup in sub-issues the required work on different packages. As of right now, one PR has been filed, and more will follow.
Help is definitely welcome.
If you can work on one of the packages, please do. In general the CRAN version 0.3.4.0.2 passes for the CRAN dependencies, and the candidate version 0.4.9.9-2 at r-universe (and here at GitHub) represents the release candidate with Eigen 5.0.1 which breaks some packages. The key will to adjust the affected packages to that they still build under the CRAN and the r-universe version permitting a smooth upgrade of the RcppEigen package.
Individual packages:
array<>, PR by @andrjohns already at Fix compatibility with Eigen 5 Andrea-Havron/clustTMB#35functor_traits, PR at For Eigen 5.0.* add functor_traits ericgiunta/Colossus#41, merged 2026-09-28function_traits, PR at For Eigen 5.0.* add functor_traits samuel-watson/crctStepdown#6, merged 2026-09-30function_traits, PR at For Eigen 5.0.* add functor_traits samuel-watson/glmmrOptim#16, merged 2026-09-30cassertheader, PR at Add cassert header needed when building with Eigen 5.0.1 matthias-da/robCompositions#29pow()adjustment, PR at For Eigen 5.0 add a missing element-wise pow() meixichen/SpatialGEV#26array<>, PR by @andrjohns already at Fix compatibility with next RcppEigen cmjt/stelfi#8pow()adjustment, PR at For Eigen 5.0 add a missing element-wise pow() tsmodels/tsgarch#9More to follow, see full list
NB: There is a second set of package failing with segmentation violation because StanHeaders / rstan needs an update. That is not covered by this ticket.