Max Hird

[Max Hird @max_hird]

@NoJusticeMTG Story somewhat debunked by Dan Davies: https://t.co/3jaFRCUKiJ

[Stat.CO Papers @StatCOupdates]

Max Hird, Samuel Livingstone. [statCO]. High-dimensional Adaptive MCMC with Reduced Computational Complexity. https://t.co/KmjotMtxeY

[Stat.CO Papers @StatCOupdates]

Max Hird, Florian Maire, Jeffrey Negrea. [statCO]. A Non-asymptotic Analysis for Learning and Applying a Preconditioner in MCMC. https://t.co/MTLGXGbyCB

[Max Hird @max_hird]

@konstmish See https://t.co/5rfNYj3dsG

[Max Hird @max_hird]

@konstmish In the optimisation context you are correct, but if we take a sample X~π(-f(x))dx, the expectation of ∇f(X)∇f(X)^T is the same as the expectation of ∇^2f(X). This fact has actually been exploited in preconditioning for sampling

[Max Hird @max_hird]

@GatsbyUCL @KevinHanHuang1 Very, very well deserved

[Max Hird @max_hird]

@konstmish That's right! See also Section 3.5 here: https://t.co/zxcZTfsetg for an analogous story in sampling, where the common misconception is that we should be preconditioning with the diagonals of the covariance.

[Max Hird @max_hird]

@readalanread account*

[Max Hird @max_hird]

@readalanread I've been enjoying your recent film reviews Alan, have you ever considered getting an acconut on Letterboxd?

[Max Hird @max_hird]

@readalanread @BAFTA The finale was certainly embarrassing. In my view the film portrayed trauma (of architect + patron + ...) but did not give it nearly enough content. All the viewer knew was the fact of trauma, but beyond that things were basically unexplored, which was unsatisfactory.

[Michael Choi @michaelchchoi]

How to use assignment problems, Sylvester's equations, permutations and projections to understand and speedup Markov chain mixing? Check out this work!

[arXiv math.PR Probability @mathPRb]

Michael C. H. Choi, Max Hird, Youjia Wang: Improving the convergence of Markov chains via permutations and projections https://t.co/cO9nrta4eb https://t.co/TKxuqBhZYA

[Max Hird @max_hird]

@sp_monte_carlo Relatedly: https://t.co/MRspyJpXZo

[Max Hird @max_hird]

@FJavierRubio1 @sp_monte_carlo Were they pronouncing it in a greek way or a none-greek way?

[Max Hird @max_hird]

@sp_monte_carlo I think you're fighting a losing battle with this one Sam

[Max Hird @max_hird]

@sp_monte_carlo I usually look at a picture of mathematicians whose names I haven't heard before. Finding Frigyes Riesz's picture on his wiki was like a jumpscare.

[Max Hird @max_hird]

@Almost_Sure Ta very much