SecondDerivative2012.pdf
web.mit.edu · 5,012 words · saved by 1 readers
N/A
Numerical Computation of Second Derivatives 1 with Applications to Optimization Problems Philip Caplan – pcaplan@mit.edu Abstract Newton’s method is applied to the minimization of a computationally expensive objective function. Various methods for computing the exact Hessian are examined, notably adjoint-based methods and the hyper-dual method. The hyper-dual number method still…
related reading
- Hessian matrix - Wikipediaen.wikipedia.org
- Scalable Gradients for Stochastic Differential Equationsarxiv.org
- Differentiable Programming from Scratchthenumb.at
- Deriving Muonjeremybernste.in
- 1806.07366arxiv.org
- Mathematical optimization - Wikipediaen.wikipedia.org
- 2404.17625arxiv.org
- Muon: An optimizer for hidden layers in neural networks | Keller Jordan blogkellerjordan.github.io
- Why Momentum Really Worksdistill.pub
- Calculus on Computational Graphs: Backpropagation -- colah's blogcolah.github.io
- matrixcookbook.pdfmath.uwaterloo.ca
- Ji-Ha's Blogjiha-kim.github.io