EMC2 news

[ladajo]

It is First Principles based, and supported by coders and physicists (International even...)

I don't think anything I have seen here over the years can compete. The processing power requirements alone exceed anything realistic folks here have access to for any kind of a timely output on a run. This is not like cobbling together video cards or game consoles, while powerful for low end researchers, it does not compete with actual supercomputing time.
Imagine if you had access to something like this:
Lots of Horsepower

[happyjack27]

It is First Principles based, and supported by coders and physicists (International even...)

I don't think anything I have seen here over the years can compete. The processing power requirements alone exceed anything realistic folks here have access to for any kind of a timely output on a run. This is not like cobbling together video cards or game consoles, while powerful for low end researchers, it does not compete with actual supercomputing time.
Imagine if you had access to something like this:
Lots of Horsepower

I asked about the software and you responded by talking about the hardware that ran the software.

That tells me nothing at all about the software.

It's like i asked how good a pizza tasted and you told me they have these ovens that can cook a lot of pizzas.

[happyjack27]

It is First Principles based, and supported by coders and The processing power requirements alone exceed anything realistic folks here have access to for any kind of a timely output on a run. "

Ok so you're telling me the algorithm is horribly inefficient.

[paperburn1]

t is First Principles based, and supported by coders and physicists (International even...)
edge for plasma sims. and proprietary.


I think he answered as much as he was able

[paperburn1]

are they simulating above 6.6 meters size?

https://www.nextbigfuture.com/2017/05/l ... plans.html

[ladajo]

No, not yet. Scaling is next.

I can share that it runs on 1000+ cells when it runs.
I would not say it is horribly inefficient, I would say it is high fidelity and very defendable.

[mvanwink5]

PB, I had missed the May2017 LM NextBigFuture article. Thanks.

Ladajo, re:

No, not yet. Scaling is next.

so, the Polywell size will be larger than expected (greater than 6.6 m size)?

[Betruger]

Reads as a direct reply - that they have yet to scale what they currently have to 6.6m and/or larger.

[ladajo]

Correct. Scaling is next.
The sim has furthered understanding of what was seen in the lab.

[mvanwink5]

So, the model is working, up and running. Nice to hear. Given talk of scaling, I presume that the model shows plasma is still contained with a potential well formed. How deep of a well would be the question?

Timetable on scaling results? Year's end? I'm hoping to hear something from Tri Alpha Energy on their C2-W testing (They have given it a name now, 'Norman').

Something should break through, maybe LM will feel the pressure to say something with GF talking site location for prototype.

[happyjack27]

No, not yet. Scaling is next.

I can share that it runs on 1000+ cells when it runs.
I would not say it is horribly inefficient, I would say it is high fidelity and very defendable.

Wait, so then it doesn't use floating point numbers?

Hard to imagine a scenario where scaling isn't first.

[ladajo]

You can't do scaling until you have the first principles model figured out. There are some physics going on that were not understood, or to a degree even known. It is a very complex system.

[happyjack27]

You can't do scaling until you have the first principles model figured out. There are some physics going on that were not understood, or to a degree even known. It is a very complex system.

You can't have the reprentation of first principles figured out until you have the data structures planned out. And you can't do that until you've solved scalability.

Sounds like what you're saying is it's not scalable and they're going to have to start over from scratch.

They haven't developed simulation software, they just ran a few in silico experiments to refine a model. So now they have a refined model.

Sounds like they've still got a lot of work to do.

[ladajo]

Nope.
I don't think you are tracking on what was done.
There was a need to establish a first principles based model which recreated and explained the observed physics from the lab. With a proven model, scaling parameters can be manipulated to test scaling functions of the various parameters.Without the proven base model, the results are meaningless. This is a complex system with competing non-linear functions creating a chaotic equlibrium.you don't get to presuppose the outcomes of changed factors without knowing the base functions.

[ladajo]

To be very clear, in vivo preceded in silico. Now in silico will be used to predict the next in vivo.Based on this, the next in vivo will be based.
Lab to sim to scaled up sim to scaled up lab.