dsem 2.0.0

• Changing gmrf_parameterization = "gmrf_project", given V = t(G)*G, to invert Vinv = invertSparseMatrix(V) where Vinv is then dense and then casting Vinv2 = asSparseMatrix( Vinv ), rather than a sparseLDLT for solve( V, I-P ), because the latter seems numerically unstable when P has a high condition number (e.g., the moose-wolf vignette in tinyVAST)
• Add option to specify a path based fixed at another variable
• Renamed gmrf_parameterization = "conditional_krig" as gmrf_parameterization = "mvn_project" and confirmed it in simple case
• Renamed gmrf_parameterization = "separable" as full and projection as project
• Fixed bug where predict for type="link" was pulling x_tj rather than z_tj and therefore was missing the initial conditions, mean value, and only worked for gmrf_parameterization = "separable"
• Add stabilize_Q option to dsem_control, adding a diagonal component to t(Gamma)*Gamma to ensure it’s PD
• Added gmrf_parameterization = "gmrf_project" (which maintains sparsity while allowing latent variables with zero exogenous variance, or manifest variables with measurement error and zero exogenous variance), and which automatically switches to separable if no variables have zero variance
• Added an integrated test confirming that gmrf_project, mvn_project and separable (with a small extra non-zero variance inflation) are all identical in a logistic regression that involves loops (i.e., zero-variance for intermediate latent variable)
• Switching dsem_control to use gmrf_parameterization = "gmrf_project" as default
• Turned off integrated-tests for dsemRTMB (which threw an error with updates to RTMB, and is not being used anyway)

dsem 1.7.0

• Add option to turn off useless NA/NaN function evaluation function evaluations from nlminb (enabled by default, but overriden in dsem_control)
• Use Matrix::bandSparse to simplify logic in make_matrices
• Add experimental function partition_variance
• Extend total_effect to compute result for either pulse or press experiment
• Adding exploratory option gmrf_parameterization = "conditional_krig" that excludes family = "fixed" variables from the GMRF density calculation while still conditioning upon their value(s)

dsem 1.6.0

• Export and document make_matrices
• Remove ggm from Imports (because it Imports graph which is not on CRAN) and instead define a locally copy of relevant functions. Also adding Giovanni M. Marchetti (the maintainer of ggm as contributor in DESCRIPTION)
• Update test_dsep(.) to allow options for imputing missing data prior to test, and using default impute_data = "by_test" based on explorations to date
• Add example for total_effect
• Adding “model structura” vignette and re-organizing pkgdown page

dsem 1.5.0

• Add option to simulate data with or without missingness pattern of original data
• Change test_dsep(.) to have option to impute missing data, and changing that to be the default behavior

dsem 1.5.0

• Fix bug when adding missing variances (PR 36)
• Fix bug in detecting high ratio of variances (issue 35)
• Added test_dsep(.) to calculate a p-value for the probability that a a data set arose from the specified model (highly experimental)
• Added total_effect(.) to calculate total effects
• Fixed bug where dsem crashed when quiet = FALSE and running without data e.g., as qualitative network model
• Added vignette showing how to approximate diffusive movement using DSEM with paths among adjacent strata

dsem 1.4.1

• Edits to allow DSEM to run when all path coefficients are fixed (contributed by Maurice Goodman)

dsem 1.4.0

• Adding option for lower and upper bounds
• Adding stepwise_selection for automated stepwise model selection
• Adding plot option to use ggraph as alternative to previous igraph option
• Adding convert_equations to extend sem::specifyEquations and simplify specification for large models
• Adding argument prior_negloglike as interface to specify Bayesian priors and/or likelihood penalties in dsem
• Adding dsemRTMB using RTMB as interchangeable alternative to dsem using TMB

dsem 1.3.0

• Adding option to specify constant marginal variance, as alternative to existing calculation which results in a constant conditional variance but a changing marginal variance along the causal graph

dsem 1.2.1

• removing checkDepPackageVersion(dep_pkg="Matrix", this_pkg="TMB") from .onLoad() as requested by K. Kristensen

dsem 1.2.0

• Adding option to specify covariance via argument covs
• Adding options to specify gmrf_parameterization=“projection”
• Adding vigette outlining how to fit dynamic factor analysis
• Fix bug arising when tsdata had two or more columns sharing a single variable name
• Adding make_dfa helper function
• Updating bering_sea dataset to include extra year of cold-pool, and changing vignette to match the published specification and results
• Updating the lag indexing for the Klein-I model in the vignette to use positive values for lags, and updating saved MCMC results to match that corrected specification

dsem 1.1.0

• Adding option to specify covariance in Gamma

dsem 1.0.2

• Eliminate eval usage
• Add codecov Action and badge
• Change default behavior so that all variables in tsdata have a standard deviation by default

dsem 1.0.1

• Fix bug in simulate.dsem to keep up with changing interface in dsem
• Update CITATION to indicate accepted paper
• Remove fit_tmb to eliminate cryptic warning messages and simplify code

dsem 1.0.0

• Initial public release