Five questions for 2023

UPDATE (Feb 2023): Already 3 questions answered!

We are coming out, slowly, from an age of confusion.

We are not naive dreamers when we support Open Science.

Here are five personal questions for the next year.

1. Can we steal from Open Science? This is the question which has to get an answer for the ANR BIGBEN project. UPDATE: until now the Agence Nationale de la Recherche answered me that the purpose of Open Science is to be appropriated by the community… huh?

UPDATE: Yes, the french Agence Nationale de la Recherche supports the steal from Open Science. See the response

“Now the french Agence Nationale de la Recherche answers clearly to my questions concerning the awarded ANR BIGBEN project, with pearls like:

“Indeed, the fact that you have made your work and results freely available should have been precisely with the aim of making them accessible to the community so that it can appropriate them and then use them to move forward.”

[from correspondence with the ANR prior to their official (sketchy) answer]”

and the flawed “avis”, along with my public rebuttal.

I shall put this wonderful quote on all public presentations 🙂

2. Will the private sector engage into supporting Open Science? For them OS is the archenemy (how to make money if my work is not protected by IP) and the main source of ideas. When they say that “ideas are cheap”, they really mean “my work is to scale an idea until it becomes successful, not to create new ideas”. Respect for scaling, but new ideas are the secret for future success. Until now the private sector is bent to destroy all the sources of new ideas.

3. Who will make the software infrastructure to share, collect and study microbiome data in real time, globally? Molecular Reality is the nanopore sensor I dreamed about, along with a dna printer, as the I/O for the molecular computers. At some point somebody will put something like this on phones. More importantly, those who will win the competition of apps and software infrastructure will face not only extreme wealth, but also Open Science questions.

UPDATE: All best wishes to Molecular Reality. They dream even bigger: to put an AGI Maxwell Demon in charge over each nanopore.

4. AI chatbots will replace search engines? Recall that all those chatbots are trained on public data. While it makes perfect commercial sense to give to the masses mediocre answers for any question, it will also increase the general dumbness and conformity.

UPDATE: Yes. See for example phind.com .

So, the last question is:

5. Will Sci-Hub, LibGen or other emerging hackers produce a science friendly alternative? Science grows and thrives always on the fringe. Where people are OK with conformist answers, researchers can’t help but looking to find a crack to open the shell and go in unknown territory.

Matematica foris

Randomness is a theory of the rest of the world. Fortuna is a function which converts the outside (foris) into a number.

[source] Enough with old stories. What about pseudorandom number generators (PRNG)?

Randomness is everywhere. There is a clear advantage of random algorithms vs deterministic algorithms. We can turn a random algorithm into a deterministic one by using a PRNG as the source of randomness.

There is another related thread, namely decentralized computing. Indeed, we take as granted that we can model decentralized computing via asynchronous automata. But an async automaton simulates decentralized computing if we accept the hypothesis that from the point of view of one user, actor, etc, the rest of the world which participates to the computation behaves randomly.

If we put together the two ideas:

• that we can turn a random algorithm into a deterministic one via a PRNG,
• that we can model the rest of the world as random,

then we arrive to the following: a PRNG is a model of the rest of the world.

Usually we make a model of something of interest, a phenomenon, process, by a simplification and a formalization of the said phenomenon or process. Now, for some models we also need to place them in the world, therefore a PRNG seems like an interesting choice to do that.

Of course, a PRNG is used as a source of randomness but also the output or state of the algorithm has an influence on the rest of the world, therefore it should be used as a salt for the PRNG.

All in all it may happen that most of the computation spent in the simulation of the model is for the PRNG.

Problems with the ANR Bigben project

Due to unethical behavior of de Saxce, the principal investigator of the project Bigben, recently funded by the french Agence Nationale de la Recherche (ANR), I asked the ANR for a reevaluation of the project and a public response.

My work on hamiltonian inclusions, aka SBEN, is central and the main novelty of this project. After winning the ANR competition, the principal investigator misrepresented my work and engaged in unethical behavior. I keep the correspondence which proves this, for the interested colleagues, although I would rather hope that ANR takes the steps to self regulate in this matter.

I shall update with the ANR response or reaction, if any.

UPDATE: ANR kindly answered, see this post, but not exactly to my questions, so I replied.

As concerns the scientific part, a detailed explanation will be available. I am sad that a beautiful principle of dissipation as minimal disclosed information is dumbed down to an old idea. The Brezis-Ekeland-Nayroles (BEN) principle in quasistatic plasticity is just a particular example of my general theory (and the only new contribution of de Saxce) and sadly, not the feasible way to exploit the hamiltonian inclusions, except in the most trivial situations.

To transform the hamiltonian inclusions into symplectic BEN then into generalized bipotentials(BIG) BEN is only a game where by slight name changes de Saxce tries to appropriate my ideas. There is no scientific content in these name changes or particular examples.

Even the names are misleading, for example there are no generalized bipotentials, they are the same ones with respect to the symplectic duality (my work, not de Saxce’s). The point is not about bipotentials!

One needs to show how this principle can be used for simulations and for this there exist other, new ways.

Here you can see slides from 2014 and all the actors of the present project.

For my work on this subject see:

[1] M. Buliga, Hamiltonian inclusions with convex dissipation with a view towards applications, Mathematics and its Applications 1, 2 (2009), 228-251, arXiv:0810.1419

[2] M. Buliga, G. de Saxce, A symplectic Brezis-Ekeland-Nayroles principle, Mathematics and Mechanics of Solids 22, 6, (2017), arXiv:1408.3102

[3] M. Buliga, A stochastic version and a Liouville theorem for hamiltonian inclusions with convex dissipation (2018), arXiv:1807.10480

[4] M. Buliga, On the information content of the difference from hamiltonian evolution (2019), arXiv:1902.04598