Tuomo Valkonen

@tuomov
7 Followers
19 Following
26 Posts

I use mathematics to build algorithms, high-performance software, and R&D programmes—especially in optimisation, inverse problems, and scientific computing.

I am currently open to selected consulting and short-term collaborations, including technical writing, mentoring, and visiting positions.

I am particularly interested in work with scientific, social, or ecological impact—especially around livable, car-free cities.

Please see my website for details.

⎯⎯⎯⎯⎯⎯⎯⎯⎯

Posts delete after 30 days.

Websitehttps://tuomov.iki.fi/

I need to move my team to #Matrix. Somehow. The public servers are not very reliable.

But let us not forget that Element has also been significantly enshittified from what it was. Labs is video rooms and other junk now. #LaTeX math rendering option is gone. Undocumented in some random config file.

#enshittification

#Zoom just keeps on getting more and more #enshittified. Now in 7.x the chat sidebar takes half the screen, and the whitespace around messages the rest.

In other words, the standard disease of graphical designers with ten 27" screens cramming their disgusting junk 🤮 down 13" laptop users' throat.

⬇️

#enshittification

Research should be treated more like open source software: continuously improved, corrected, extended, and scrutinized. But both #peerreview ers and #opensource developers could benefit from less gatekeeping and more constructive engagement with users and contributors.

The academic certification system is broken because it insists on conflating itself with publication and dissemination—roles that it no longer serves. We need to move to structures that embrace living documents that evolve over time and are reviewed, revised, and extended organically through real use—not static PDFs frozen forever after a handful of perfunctory referee reports.

⬇️

Once journals start thinking of themselves as products competing for attention, incentives shift toward short, fashionable, easily marketable papers. Extensive, self-contained works become liabilities rather than assets.

⬇️

Scientific journals act as though they were tabloid magazines. But that is not their function. Their purpose is to evaluate novelty, assess correctness, and place their stamp on the work. Nothing more. Editorial opinions belong elsewhere. Who reads journals?!? Today, scientific publication happens on #arXiv. Discovery happens through search engines, social networks, citations, talks, and personal recommendations.

⬇️

It is becoming nearly impossible to get journals to stamp extensive research. Everything has to be splittable into attention-deficit TikTok-sized bites.

And this paper is actually too short. There’s still much to do.

⬇️

RE: https://mastodon.social/@tuomov/116768577960200379

Let’s see how many days it takes for Set-Valued and Variational Analysis to send this 56-page behemoth back as a desk reject boomerang 🪃.

They have a 40-page limit.

No point in even trying luxury journals with 25 page limits, or worse.

(At least machine learning conferences allow massive appendices, but proofs don’t belong in appendices! They are the work!)

⬇️

… derive corresponding optimality conditions, and formulate nonsmooth adjoint inclusions based on both the Fréchet and limiting coderivatives. Founded on these results, we propose a single-loop algorithm that accommodates a wide range of inner and adjoint steps, including those of primal-dual methods. We prove its convergence. Numerical experiments on total variation regularised inverse problems demonstrate the practicality of the approach.

Our #article “Single-loop approaches to nonsmooth bilevel optimisation”, which will form the final part of the #PhD thesis of Ensio Suonperä, is now available.

https://arxiv.org/abs/2606.19143

Abstract: We study #bilevel #optimization problems in which the inner problem is represented as a set-valued, parametric constraint. We develop relevant optimistic and pessimistic calculus rules, …
⬇️