Diamond¶
Diamond utilizes iterative, quasi-Newton 2nd-order solvers to estimate certain generalized linear models (glms) with known covariance structure. A common use is fitting mixed effects models, with their covariance already being known by another means (e.g. after fitting in R using lme4). These 2nd-order iterative solvers are considerably faster than a full-blown solution, assuming that the covariances are known. Currently, Diamond does not solve for the covariance structure. This must be input a-priori. In addition, only logistic and ordinal logistic response variables are currently implemented. See the Readme for more details, this blog post for the math behind Diamond, and this blog post for more on mixed-effects models and the specifics of Diamond.
Contents: