Mastodawn

Hi,
a few Latex questions to ppl who submit papers to math journals:
1. do you convert your latex file to the style of the journal at the submission stage, or at the production stage (when the paper is accepted)?
(by "conversion", I mean using the sty file and/or the tex template provided by the journal, etc.
2. Do you usually find it easy/difficult to do this conversion?
3. Have you ever given up submitting to a particular journal because this conversion seemed too cumbersome?

Disclaimers:

a. I'm interested in your answers because I am about to start my term as a JE for JRSSB. But I'm curious to hear your view more generally about this process for any math journal.

b. feel free to DM me if you don't want to answer publicly.

c. For the first point, my view is that authors should be expected/required to apply the style only at the production stage. There is no point wasting the time of authors with this (especially if the acceptance rate is low).

Show thread

Dennis Prangle Nov 15

@nchopin 1) Usually in an early draft, to get an accurate idea of the page count during drafting and to avoid any surprises like a 2 column format 2) Usually easy 3) I think I have given up in the past when I just couldn't get compilation to work with the style file

Dennis Prangle Jan 29, 2025

Show thread

Terence Tao Jan 29, 2025

The last comparable shock to my field was the 2008 global financial crisis. Due to that crisis, university budgets and endowments were suddenly upended. My colleagues and I received salary cuts; but more importantly, many open positions evaporated, and an entire generation of promising mathematicians was at risk of having their careers cut short. However, in the US a stimulus package was passed which, among other things, allowed the NSF to greatly expand its postdoctoral program on an emergency basis, and also maintain or temporarily expand other sources of NSF funding. This allowed hiring for that season to proceed at something close to normal levels, and in the end there was no major disruption to the expectations for what the career track of an academic mathematician in the US would be. To continue the mathematical formalism at the start of this post, the "mean field" of the mean field game stayed within the "effective regime" in which it could be understood by simpler models.

A similar stabilization of expectations for the future would be quite welcome and helpful at this point. But the current administration has thus far not demonstrated much ability to solve mean field games effectively. (7/7)

Dennis Prangle Jan 23, 2025

Can anyone recommend an undergraduate friendly introduction to particle filters? Ideally something concentrating on the algorithm details and the intuition rather than the theory - I know of lots of good resources for that!

Show thread

Dennis Prangle Nov 29, 2024

@christianp I'm a Bristol mathematician now, but sadly have a medical appointment that day. I hope the talk goes well!

Show thread

Dennis Prangle Sep 17, 2024

@nchopin Another nice alternative to scholar is https://search.crossref.org/

Crossref Metadata Search

Dennis Prangle Aug 19, 2024

Emtiyaz Khan Aug 16, 2024

Call for papers for AISTATS 2025 is out now
http://aistats.org/aistats2025/call-for-papers.html

You can also (self-)nominate for reviewer/AC position
https://forms.gle/saediFoTznTQ7heC6

We are (trying to) switch to OpenReview so submission details will be available a bit later.

One major change is that the author list at the abstract submission deadline will be considered final, and no changes in authorship will be allowed. This is to avoid new unforeseen COI after the bidding.

Call for Papers

Dennis Prangle Jul 1, 2024

New paper "Flexible tails for normalizing flows" https://arxiv.org/abs/2406.16971 with Tennessee Hickling, exploring links between generative models and extreme value theory. Come and ask me about it at the #isba2024 wednesday poster session (or any other time during the conference!)

Flexible Tails for Normalizing Flows

Normalizing flows are a flexible class of probability distributions, expressed as transformations of a simple base distribution. A limitation of standard normalizing flows is representing distributions with heavy tails, which arise in applications to both density estimation and variational inference. A popular current solution to this problem is to use a heavy tailed base distribution. Examples include the tail adaptive flow (TAF) methods of Laszkiewicz et al (2022). We argue this can lead to poor performance due to the difficulty of optimising neural networks, such as normalizing flows, under heavy tailed input. This problem is demonstrated in our paper. We propose an alternative: use a Gaussian base distribution and a final transformation layer which can produce heavy tails. We call this approach tail transform flow (TTF). Experimental results show this approach outperforms current methods, especially when the target distribution has large dimension or tail weight.

arXiv.org

Dennis Prangle May 2, 2024

Very exciting preprint, this could be a big advance in making SBI practical https://arxiv.org/abs/2404.09636

All-in-one simulation-based inference

Amortized Bayesian inference trains neural networks to solve stochastic inference problems using model simulations, thereby making it possible to rapidly perform Bayesian inference for any newly observed data. However, current simulation-based amortized inference methods are simulation-hungry and inflexible: They require the specification of a fixed parametric prior, simulator, and inference tasks ahead of time. Here, we present a new amortized inference method -- the Simformer -- which overcomes these limitations. By training a probabilistic diffusion model with transformer architectures, the Simformer outperforms current state-of-the-art amortized inference approaches on benchmark tasks and is substantially more flexible: It can be applied to models with function-valued parameters, it can handle inference scenarios with missing or unstructured data, and it can sample arbitrary conditionals of the joint distribution of parameters and data, including both posterior and likelihood. We showcase the performance and flexibility of the Simformer on simulators from ecology, epidemiology, and neuroscience, and demonstrate that it opens up new possibilities and application domains for amortized Bayesian inference on simulation-based models.

arXiv.org

Dennis Prangle Mar 28, 2024

Rest of World Mar 28, 2024

Protests have become opportunities for authorities to net thousands of faces via CCTV, van-mounted cameras, and police mobile devices, which can then be added to facial recognition databases. A Russian civil society group believes Moscow police used the technology to track down people who attended opposition leader Alexei Navalny’s funeral.

Read: How governments are using facial recognition technology to crack down on protesters
https://restofworld.org/2024/facial-recognition-government-protest-surveillance?utm_source=mast

How governments are using facial recognition to crack down on protesters

Mass protests used to offer a degree of safety in numbers. Facial recognition technology changes the equation.

Rest of World