NEWS

---

lassoinf 0.1.0 (2026-10-20)

• First CRAN release.
• LassoInference: selective intervals and p-values for the (randomized) bounded LASSO, with dense matrices or matrix-free operators (dense_operator(), xtvx_operator()).
• glmnet_inference(): inference from a fitted glmnet object at a given lambda, without randomization or with carving (selection_rows).
• glmnet_problem() and glmnet_problem_from_fit(): the LASSO problem glmnet solved, with the Hessian at a one-step relaxed fit by default (information = "relaxed").
• Inference for user-specified estimands (contrast_inference(), custom_estimand_inference(), estimand_summary()), inactive variables (inactive_summary(), SelectionCoordinates) and screening (ScreenedSelection).