Levels of Modelling Computation – Distinct but Dependent

Posting during a week’s business trip to China, with the “Great Firewall” constraints on social media and Google services – including no access to any of my email (!) or any TV media. A fascinating visit to Wuhan capital of Wuchang, the home of the original 1911 Chinese revolution and more, but I digress.

One of the web-pages cached on my phone was the one extracted below from HuffPo’s UK version Tech pages. Fusion Reactor ‘Breakthrough’ Could Finally Hold The Key To Giving Us Clean, Abundant Power (16 February 2016 by Thomas Tamblyn Technology Editor, The Huffington Post UK)

Apart from the usual media hype of the key or final piece in the science jigsaw – mediated by “could” in the title and “may” in the opening sentence …

Researchers may finally have overcome one of the biggest obstacles to making nuclear fusion live up to its dream of giving us clean, virtually limitless energy.

… the view of science as a completable jigsaw is misguided anyway, even as a mathematical model – ask Godel, but again I digress.

While we fundamentally understand the [fusion principle] it has been extremely difficult to predict how the hot gas will behave […] various forms of turbulence have taken place within the gas causing the heat to drop and in turn breaking the fusion process […] researchers [at MIT] believe they’ve finally worked out how to predict this turbulence.

[That “finally” again – aaagh – there is no jigsaw to complete! No finite number of parts with a single correct arrangement. But I still digress.]

China has become the latest country to join the race in creating the first sustainable fusion reactor.

Interesting, see earlier post on Chinese nuclear power success, but incidentally not related to why I’m in China.

Using some of the world’s most powerful supercomputers the researchers were able to create an almost exact model of the plasma inside the reactor and discovered something completely unexpected.

The turbulence was actually being caused on two very distinct levels, both at the ion level and at the far far smaller electron level. It had until now been presumed that the ion turbulence would simply overpower the smaller turbulence but these models now show that they’re intrinsically linked, and impossible to separate.

The “almost exact” will turn out to be significant, but for now, whilst this is a computer model, physics is at base fundamentally computational. And whilst the scale of computation here is impressive (more details – read the story) it is an established fact that repeated computation throws up unpredictable results (at higher “evolved” levels than the starting point) even if the computation is simply and repeatedly algorithmic. (See Hofstadter, see Dennett, distinctly and together.)

The distinct levels (gestalts) ARE distinct – have their own identity, properties and behaviours – and whilst related, connected, even dependent, the lower does NOT cause the higher in any reductively, predictable way. Knowing the lower tells you nothing (very little) about the higher. Small differences in either the starting point or in the algorithmic relationship will lead to entirely different outcomes, and neither can you predict how small is too small to ever circumvent the problem. (In the specific example of the two-level plasma turbulence, the turbulence is already metaphorical I suspect, compared to “normal” fluid flow, but turbulence is notoriously chaotic, unpredictable and empirically stochastic – which spookily IS related to why I’m in China.)

Science ain’t that scientific. Not all, not even all good science is that repeatable.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.