NEWS

---

dream 2.1.2

Minor Changes

• Added three new functions for the examination of relational event model fits. First, gof_rem() computes the proportion of correctly predicted dyadic events for a dream_rem relational event model fit. Secondly, we added the plot() and residuals() S3 methods for the dream_rem object class. Please see the function documentation pages for more information on these new functions. Happy dreaming!

dream 2.1.1 (2026-05-29)

Major Changes

• The package has undergone a new API for the analysis of relational event sequences aimed at reducing redundant user-inputs and increasing useability. Importantly, the package now contains two S3 object classes: dream_sequence and dream_rem, where dream_sequence is the object class created after using create_res() in the dream package to generate the post-processing relational event sequence (i.e., (sampled) realized events and sampled control/null events) or after using dream_sequence(), a new function to create the S3 object based upon a user-generated processing sequence. Moreover, in prior versions of the package, users had to input 6 arguments into the functions to compute various sufficient network statistics (e.g., four-cycles, repetition), these 6 arguments are now pulled from the dream_sequence data object. Additionally, the functions that compute the sufficient network statistics add the vector of computed statistics to the user-inputted dream_sequence object as an additional element in the statistics list. The estimate_rem() function now relies upon the user-created dream_sequence objects and extracts the relational event model covariates from the data object. Finally, dream_rem S3 objects now contain a new set of functions to extract relevant model information: vcov(), logLik(), coef(), and predict().
• create_riskset_dynamic() and create_riskset_constant() have been replaced
  by the new function create_res(), which internally is the same as the aforementioned functions and generate risksets based upon the riskset argument. We encourage users to read the help page for the create_res() to see the updated set of full riskset possibilities.
• As mentioned above, the dreamstats_ series of functions have been updated to reduce the number of arguments supplied by the user. Previously, the following arguments needed to be specified: time, sender, receiver, eventID, observed, and sampled. However, the new updated functions internally create these based upon the data argument which is an S3 dream_sequence object.
• The newly created S3 dream_sequence object, which is created using the create_res() function, also has a function named dream_sequence() that can create a dream_sequence object based upon the user-inputted values. In addition, the as.data.frame() has been supplied to extract the post-processing relational event sequence and the computed statistics. Users may find this function useful if they would like to estimate the relational event model by a different modeling function such as the clogit() function in the survival package or a user-created function.
• The new package version also includes a new suite of functions that expands the set of exogenous statistics for the estimation of relational event models. These are dreamstats_actor(), dreamstats_actorfe(), dreamstats_dyadic(), dreamstats_dyadfe(), and dreamstats_event(). These functions, respectively, allow users to (1) add actor-level time-varying and time-invariant exogenous covariates, (2) add actor-level (sender and receiver) fixed effects, (3) add dyadic-level time-varying and time-invariant exogenous covariates, (4) add dyadic-level fixed effects, and (3) add event-level exogenous covariates.
• Lastly, the estimate_rem() function now allows for the estimation of interval timing relational event models. In terms of estimation routines, the c++ functions to compute the model log-likelihoods, hessians, and gradients are now computed using RcppArmadillo with the arma set of functions.
• All functions that were deprecated in dream 1.1.1 were removed, in addition to the past function arguments for the dreamstats_ functions.

dream 1.1.1 (2026-03-25)

Major Changes

• All remstats_-related functions were renamed to dreamstats_ to provide a more consistently named API for the dream package. As a result, all remstats_ original functions are now deprecated starting on version 1.1.1, but remain available in this update.
• create_riskset() was renamed to create_riskset_constant().
• All functions that were deprecated in dream 1.0.0 were removed.

dream 1.0.1 (2026-02-07)

Major Changes

• Implemented a new function to create time-dynamic risk sets for one- and two-mode relational event sequences. The new function is named create_riskset_dynamic().

Minor Changes

• Updated the remstats_-related C++ files to delete past network edges (i.e., past event weights) that are smaller than the past event cut off value (i.e., dyadic_weight) when this feature is selected by the user. That is, when the dyadic_weight argument is non-zero.

dream 1.0.0 (2026-01-20)

Major Changes

• Added a NEWS.md file to track changes to the package.
• Two key changes occured in the new version of dream package. First, the majority of functions now depend upon c++ routines via the Rccp package to improve computational runtime. For instance, the new function to compute the repetition network statistic, remstats_repetition(), is around 90 times faster than the previous function, computeRepetition(), in dream version 0.0.1. Secondly, the dream package has an updated API to allow better categorization for the functions. Please run ?dream for more details on the new API. Based upon this transition, the original functions are now deprecated starting on version 1.0.0, but remain available in this update.
• The following functions were deprecated for remstats_triads(): computeISP(), computeITP(), computeOSP(), computeTriads, and computeOTP().
• The following functions were deprecated for remstats_degree(): computeSenderOutdegree(), computeReceiverOutdegree(), computeSenderIndegree(), and computeReceiverIndegree().
• estimateREM() was deprecated for estimate_rem_logit().
• simulateREseq() was deprecated for simulate_rem_seq().
• computeFourCycles() was deprecated for remstats_fourcycles().
• computeRemDyadCut() was deprecated for remstats_dyadcut().
• computePersistence() was deprecated for remstats_persistence().
• computePersistence() was deprecated for remstats_persistence().
• processOMEventSeq() and processTMEventSeq() were deprecated for createriskset().
• computePrefAttach() was deprecated for remstats_prefattachment().
• computeRepetition() was deprecated for remstats_repetition().
• computeRepetition() was deprecated for remstats_repetition().
• remExpWeights() was deprecated and will not be replaced in the current version as the updated functions do not require it.
• remExpWeights() was deprecated and will not be replaced in the current version as the updated functions do not require it.
• computeTMDegree() was deprecated for netstats_tm_degreecent().
• computeTMEgoDis() was deprecated for netstats_tm_egodistance().
• computeBCConstraint() was deprecated for netstats_tm_constraint().
• computeBCES() was deprecated for netstats_tm_effective().
• computeBCRedund() was deprecated for netstats_tm_redundancy().
• computeBurtsConstraint() was deprecated for netstats_om_constraint().
• computeBurtsES() was deprecated for netstats_om_effective().
• computeHomFourCycles() was deprecated for netstats_tm_homfourcycles().
• computeLealBrokerage() was deprecated for netstats_om_pib().
• computeNPaths() was deprecated for netstats_om_nwalks().
• computeTMDens() was deprecated for netstats_tm_density().