- abi
- glr
- linprog via convex conjugate
- add link to Krylov for projected gradient descent
- add example of MDA (e.g. with KL divergence)
- use https://github.com/triscale-innov/GFlops.jl to count number of floating point ops and plot that to show how things scale
- group posts by "stories" (maybe change the part xxx and just add a number or whatever) (in tags)
- QN methods and (L)BFGS
- Kalman filter
- Nesterov (and related) accelerations
- LP and dual LP (and relation with convex conjugate)
- ABI, MAP
- generate correlation matrix:
L = LowerTriangular(randn(5, 5))
D = Diagonal(1 ./ vec(sqrt.(sum(abs2, L, dims=2))))
C = D * L * L' * D
could be interesting to see whether that covers the space, maybe look at spectrum