More than one million derivatives computed in 2019 – continues to enjoy great popularity

Last year more than 1.4 million derivations were computed by almost 50,000 users. In most cases we don’t know the purpose of the derivatives, but sometimes users contact us with suggestions for additional features or just to say thanks. Such feedback, which we are always happy to receive, shows the wide range of applications of matrix derivatives. is used worldwide in almost all areas of science and technology. One application that we are especially happy about is in the field of probabilistic programming languages, a topic that also interests us very much.  Last year we organized a workshop on probabilistic programming in Hanover. One of the best-known probabilistic programming languages is Stan. The main developer of STAN, Bob Carpenter, has written a post about on the widely read blog of statistician Andrew Gelman. Bob Carpenter uses to compute Hessian matrices for Laplace approximations and for third-order derivatives used in Hamiltonian-Monte-Carlo methods. Details can be found in the mentioned blog entry.