Stan 1.1.1 - Changelog

What's new in Stan 1.1.1:

February 11th, 2013

Bug Fixes:
· fixed bug in comparison operators, which swapped operator< with operator with operator>= semantics
· auto-initialize all variables to prevent segfaults
· atan2 gradient propagation fixed
· fixed off-by-one in NUTS treedepth bound so NUTS goes at most to specified tree depth rather than specified depth 1
· various compiler compatibility and minor consistency issues
· fixed bug in metaprogram preventing lower/upper bound constraints on matrices
· fixed print error for number of kept samples
· fixed floating point literal precision issue in code generation
· fixed bug in bernoulli_log for boundary chance of success theta=0 or theta=1
· many doc patches (mostly due to user comments -- thanks!)
· replace boost sign() to avoid compiler conflicts
· trapping mismatched dimension assignments in arrays, matrices, and vectors

Enhancements:
user's guide chapters w. sample models:
· gaussian processes
· measurement error and meta-analysis
· clustering (soft k-means, LDA, naive Bayes)
· ARCH, GARCH model section in regression chapter

sample models:
· hidden Markov models (HMMs)
· non-negative matrix factorization (NNMF)

speed improvements to multivariate models and matrix solvers:
· mdivide_left, mdivide_left_tri_low, mdivide_right,
· mdivide_right_tri_low
· determinant, log determinant
· inverse

· much more extensive probability tests
· unstacked vari for multivariate auto-diff unfolding
· faster multiply self transpose / columns_dot_self
· cleaned up error messages for size mismatches in accessors
· simplified vector view expression template parameterization
· cleaned up many --pedantic compiler warnings

New Functions:
· log absolute determinant, with optimized gradients
· model timing and n_eff output in CSV for all test models
· faster Phi_approx computing an approximate cumulative unit normal density
· added dims() function to extract dimensions of arrays of scalars, vectors, and matrices
· added size() function to extract the number of elements in an array

ongoing vectorizations and reparameterizations of probability functions:
· binomial_logit

probability functions:
· multivariate normal, precision parameterizations