armacmp

armacmp copied to clipboard

During the compilation step a lof of problems later in the C++ compilation can already be detected. It would be nice to give the user very helpful error messages in...

dirkschumacher

Single value vs. vector sampling

1

There are two sets of underlying samplers in _Rcpp_. Samplers returning a `NumericVector` are from `Rcpp::` and those that return a `double` are from `R::`. E.g. ``` rnorm(1, 2, 3)...

coatless

Sample from a wishart distribution ```r rWishart(n, df, Sigma) ``` ```cpp arma::rwishart(S, df) ```

coatless

Add `apply`?

1

`apply(X, 1, fun)` => `X.each_row(fun)` `apply(X, 2, fun)` => `X.each_col(fun)` But that modifies X in place 🤔

dirkschumacher

If input are integers, then return is `icolvec`, if input are doubles then `colvec`

dirkschumacher

Vectorized versions of * `dpois` * `dnbinom` * `dnorm` * `dbinom` * `dbeta` * `plogis` We could use `Rcpp` wrappers and then convert to `arma::vec`...

dirkschumacher

Something like this ```R arma_optim( data = list(design_matrix = type_matrix(), response = type_colvec()), evaluate_with_gradient = function(beta, g) { diff

dirkschumacher

dirkschumacher