cusna 0.1.1

Bug fixes

• cusna_fran() now honours RSiena’s simstats0c contract: fra and sc are periods-by-parameters matrices (one row per period). In 0.1.0 fra was the period-summed vector and sc held the last period only, so siena07() stopped at its first call (in colSums(zz$fra)). The closure also checks that the parameter vector passed by siena07() matches effect_names and conditional, draws its seeds from R’s random number stream (so sienaAlgorithmCreate(seed = ) makes a run reproducible, and the global seed is no longer reset on every call), and counts its calls in attr(fran, "counter")$i.
• mom_estimate() can now return the simulated networks: with return_nets = TRUE the fit keeps the phase-3 end networks in fit$sims (one chains-by-n-by-n integer array per period, the layout cusna_gof_distribution() takes) and, for co-evolution, the end behavior in fit$sims_behavior. In 0.1.0 the simulator was always called without its end networks.

Packaging and documentation

• New dataset s50 (three waves of friendship among 50 girls, with alcohol use; the RSiena teaching panel), used by the examples and vignettes.
• All examples run: no \dontrun{} remains, and the SAOM examples use s50. The siena07() vignette runs its example when RSiena is installed.
• Under R CMD check --as-cran (_R_CHECK_LIMIT_CORES_) the OpenMP engine uses at most two threads.
• New stergm_simulate(): draws STERGM transitions on the constrained spaces of stergm_cmle().
• The description, configure messages and cusna_has_cuda() documentation now state that the R package builds the CPU backend of the engine; the CUDA backend of the same C interface is built separately.

cusna 0.1.0

First release: a self-contained native engine (C ABI over CUDA kernels and C++ host logic) for the simulation and estimation of network models, callable from R without a Python runtime. The CRAN build is CPU-only from source; the GPU path compiles when a CUDA toolkit is detected at configure time.

• SAOM (the RSiena model family). saom_data(); effect constructors (cusna_effect(), cusna_beh_effect(), cusna_rate_effect(), cusna_interaction()); the full Method-of-Moments estimator mom_estimate() / mom_control() returning a cusna_fit object with summary(), coef(), vcov(), and as.data.frame() methods; behavior co-evolution, composition change, conditional and unconditional estimation; multi-network co-evolution (saom_multinet_data(), mom_estimate_multinet()); and a siena07() simulation backend (cusna_fran()). Data preparation, effect preprocessing, and moment targets are validated bit-for-bit against the reference implementation; estimates agree within simulation standard errors.
• ERGM. Sufficient statistics (ergm_stats()), a TNT sampler (ergm_simulate()), pseudo-likelihood (ergm_mple()), and MCMC maximum likelihood (ergm_mcmle()), matching ergm::ergm() on benchmark models.
• Temporal ERGM. tergm_mple() (pooled MPLE with block bootstrap, matching btergm), temporal simulation (tergm_simulate()), and the separable stergm_cmle() (formation/persistence, matching tergm CMLE).
• ALAAM. alaam_mple() (exactly reproducing the corresponding glm), alaam_mcmle(), and a Gibbs simulator (alaam_simulate()).
• Low-level host statistics (cusna_network_stats(), cusna_behavior_stats(), cusna_gof_distribution()) reproduce RSiena targets on public datasets to machine precision.