Having generated text/thought about a subject, they don't gain understanding, only more context that needs to be always carefully recreated when following up.

You can't blindly trust them, but there absolutely is a place for them in situations where being explicit about the reasoning is the point.

I use LLMs to generate mathematical derivations and proof prototypes, so explicitly demonstrating reasoning is a central part of what I use them for.

I don't want to trivialise the problems involved, their failure modes are weird and very different. The boundary of what they can/t do is jagged.

The reasoning of LLMs can be examined in exactly the same way as that of a human colleague: by asking for it. For both humans and LLMs the answer you get will only have a very indirect relationship with the process that generated the answer.

Solar and batteries are cheap enough that most people can get most of their electricity from them, and save money. This equation gets better and better over time as their costs decline.

All details in a new blog post: nworbmot.org/blog/solar-b...

„“

Maybe the time has finally come to link up PyPSA and PowerDynamics, like we promised in proposals 8ish years ago ;)

PowerDynamics 4.0 Release Workshop

October 16th 13:00-15:00 CEST - Online

Sign up here to find out how to build and simulate sophisticated dynamical models of power grids with Julia:

www.listserv.dfn.de/sympa/info/p...

PowerDynamics specifically contains features to initialize a valid initial state from a power flow solution, and an initial model library for power grid components.

We have an easy* to use prototype of the IEEE39 bus system. Get in touch if you want to use this!

*ish

- Heterogeneous node and edge dynamics
- Modular equation-based modeling
- Symbolic simplification
- Parallel on CPU and GPU (experimental)
- Automatic Differentiation compatible (enables NeuralODE on nodes, parameter fitting, sensitivity analysis, optimization of dynamic properties, etc...)

We have released new, heavily reworked versions of NetworkDynamics.jl and PowerDynamics.jl for simulating dynamical systems on networks, and power grids specifically.

juliadynamics.github.io/NetworkDynam...
juliaenergy.github.io/PowerDynamic...

Congratulations! Incidentally we just released PowerDynamics.jl 4.0, the latest iteration of the dynamics work we started when PyPSA was still young. A more or less complete rewrite and the first version that is actually suitable for Engineering level dynamics studies.

PyPSA v1.0 is officially here - 10 years and 2 days since the first git commit. This milestone release brings major new features, completely new documentation, and a fresh landing page - congratulations to all the PyPSA team! 🎉

🔗 New Documentation: docs.pypsa.org
🔗 New Landing Page: pypsa.org

Habermas worked extensively on the conditions under which open discourse can establish truths (including moral truths).

Discourse Ethics, communicative rationality, and hierarchyless discourse, are key terms. From this analysis the outcome of the internet is maybe not so surprising.

This month, CTCS (IIT Madras) & PIK present a webinar:
📢: Dynamics of modern power grids - A complex systems perspective
🎙️: Dr. Frank Hellman @various.complexities.eu, @pik-potsdam.bsky.social, Germany
📅: September 29 |⏰19:30 IST | 16:00 CEST,
🔗: us06web.zoom.us/webinar/regi...
#ComplexSystems #phd

How individual systems and their connections react to each other to ensure stability of the whole. We live in a complex world with a plethora of connections and interactions. Many important processes, such as power grid dynamics, ecological interactions, opinion formation, or brain plasticity, can ...

Our new article in Chaos. One of my favorite articles so far. Connecting different disciplines and building on recent breakthroughs in ancient fields (Linear Algebra!).

link.growkudos.com/1ebwrn9qkn4

doi.org/10.1063/5.02...

As the hydrogen bubble deflates, what are the alternatives? We present a "minimal methanol economy": using methanol as a gap-filler for the few sectors electrification can't reach.

New working paper together with Philipp Glaum, @fneum.bsky.social, @millinger.bsky.social:

arxiv.org/abs/2505.09277

Unsplash image of the Earth, mostly the nightside with a tracery of city lights on every continent.

OK, this is wild.

In September 2023, geophysicists across the world started monitoring a very odd signal coming from the ground under them.

It was picked up in the Arctic. And Antarctica. It was detected everywhere, every 90 seconds, as regular as a metronome, for *nine days*.

What the HELL?

1/