The hardware and bandwidth for this mirror is donated by METANET, the Webhosting and Full Service-Cloud Provider.
If you wish to report a bug, or if you are interested in having us mirror your free-software or open-source project, please feel free to contact us at mirror[@]metanet.ch.
dmrfit(data, structure = ), no longer
fail and return NULL (“Optimization failed”) for some network
structures. The trust-region step is now computed on the free parameters
only (the thresholds and the interactions present in the structure).
Unconstrained fits are unchanged.dmrfit_bayes(): the default initial step size of the
adaptive FisherMALA sampler is now sigma2 = 1.0 (was 0.1),
the value used in Arena and Marsman (2026). The step size adapts during
the burn-in, so the target distribution is unchanged, but the draws for
a given seed differ from version 0.1.0.dmrfit() and
dmrfit_bayes() are now more stable across seeds: when few
posterior draws lie near zero, the density at zero is estimated from a
normal approximation of the draws, instead of a kernel estimate from
those few draws or a floor. The estimator used for each interaction is
in savage_dickey$density_at_zero, and
summary() reports how many Bayes factors use the normal
approximation.These binaries (installable software) and packages are in development.
They may not be fully stable and should be used with caution. We make no claims about them.