EMC2 news

Point out news stories, on the net or in mainstream media, related to polywell fusion.

Moderators: tonybarry, MSimon

Diogenes
Posts: 6967
Joined: Mon Jun 15, 2009 3:33 pm

Re: EMC2 news

Post by Diogenes »

happyjack27 wrote:
ladajo wrote: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.

Yes, we get it. You think you are a super duper brilliant software guy, and obviously professional nuclear physicists don't know what they are doing and are not fit to shine your intellectual shoes.


For those of you who think the "News" section ought to remain free of snark, I wish to say I forbore several times before now.
‘What all the wise men promised has not happened, and what all the damned fools said would happen has come to pass.’
— Lord Melbourne —

ScottL
Posts: 1122
Joined: Thu Jun 02, 2011 11:26 pm

Re: EMC2 news

Post by ScottL »

Diogenes wrote:Yes, we get it. You think you are a super duper brilliant software guy, and obviously professional nuclear physicists don't know what they are doing and are not fit to shine your intellectual shoes.


For those of you who think the "News" section ought to remain free of snark, I wish to say I forbore several times before now.
This is a silly argument. Physicists are not coders and coders are not physicists (with exceptions of course). If you believe that physicists write code at the same level as a properly trained software engineer, you're going to have a bad day. While I completely disagree with HappyJacks approach, he is not necessarily wrong. The problem is that there isn't enough information due to Ladajo being unable to divulge it at least yet. We're operating on Ladajo's brief bits of information which is likely just repeated from what he hears (or sees). That being said, we don't know Ladajo's qualifications to assess coding efficiencies, data model efficiences, etc. It is entirely possible that the group doing this work actually contains 1 or more software engineers who are absolutely brilliant, but we have no way of knowing currently.

Now for the anecdotal. I'm a software developer (Comp. Sci. and EE degree). I have worked predominantly in the academic world, although for the last 6 years, I have been in an IT head department of of a major university with about 400 staff in our specific department. During my previous time within the academic departments, I noted countless times (if I'm being honest, every single time) that professors and grad students are some of the worst coders and modellers I have ever seen. Not because their end result is wrong, but because of the inefficient means they go about the work. It is like watching a cave man trying to figure out fire at times. That is not to say it can't happen, but every experience I have ever had with supposed top of their field scientists with regard to attempted coding has been terrible. There is a reason computer science is its own field.

mvanwink5
Posts: 2143
Joined: Wed Jul 01, 2009 5:07 am
Location: N.C. Mountains

Re: EMC2 news

Post by mvanwink5 »

Since a specialized supercomputer is being used, I was supposing that the software code was tailored to it, but the physics was defined by Park and who he is working with. The physics behind the modeling is the key, that is the problem in the past. For instance the tokamak group found that electron physics could be separated from the ions and that was a huge modeling breakthrough. That may be just a part of the EMC2 modeling story.

Whatever.
Counting the days to commercial fusion. It is not that long now.

Diogenes
Posts: 6967
Joined: Mon Jun 15, 2009 3:33 pm

Re: EMC2 news

Post by Diogenes »

ScottL wrote:

Not because their end result is wrong, but because of the inefficient means they go about the work. It is like watching a cave man trying to figure out fire at times. That is not to say it can't happen, but every experience I have ever had with supposed top of their field scientists with regard to attempted coding has been terrible. There is a reason computer science is its own field.


I am well aware that many people with computer science degrees are highly concerned about code efficiency. My point here is that while it may be desirable, it isn't essential for this purpose that the code be efficient. What is essential is that it gets the physics right, and if you accomplish that bit, how long it takes to run a simulation is really irrelevant to the problem, isn't it?


Efficient Code is good, but not a necessity. Code that gets the physics accurate is the necessity.


Computer horsepower is not nearly the problem it once was.
‘What all the wise men promised has not happened, and what all the damned fools said would happen has come to pass.’
— Lord Melbourne —

krenshala
Posts: 914
Joined: Wed Jul 16, 2008 4:20 pm
Location: Austin, TX, NorAm, Sol III

Re: EMC2 news

Post by krenshala »

Diogenes wrote:I am well aware that many people with computer science degrees are highly concerned about code efficiency. My point here is that while it may be desirable, it isn't essential for this purpose that the code be efficient. What is essential is that it gets the physics right, and if you accomplish that bit, how long it takes to run a simulation is really irrelevant to the problem, isn't it?


Efficient Code is good, but not a necessity. Code that gets the physics accurate is the necessity.
The issue is not "inefficient code is bad", but instead is "inefficient coders can do things wrong with the code, and the results from that aren't obviously incorrect, leading to really bizarre bugs -- some of which aren't even noticeable unless you already know what the correct (physics) answer is before you run it through the computer, and sometimes its not obvious even then."

ScottL
Posts: 1122
Joined: Thu Jun 02, 2011 11:26 pm

Re: EMC2 news

Post by ScottL »

krenshala wrote:
Diogenes wrote:I am well aware that many people with computer science degrees are highly concerned about code efficiency. My point here is that while it may be desirable, it isn't essential for this purpose that the code be efficient. What is essential is that it gets the physics right, and if you accomplish that bit, how long it takes to run a simulation is really irrelevant to the problem, isn't it?


Efficient Code is good, but not a necessity. Code that gets the physics accurate is the necessity.
The issue is not "inefficient code is bad", but instead is "inefficient coders can do things wrong with the code, and the results from that aren't obviously incorrect, leading to really bizarre bugs -- some of which aren't even noticeable unless you already know what the correct (physics) answer is before you run it through the computer, and sometimes its not obvious even then."
This a hundred fold. Krenshala hit the proverbial nail on the head. Bad coders make simple mistakes. A great and exceedingly common example is choosing the wrong data type. You choose the wrong one and you could be introducing round errors that could have a domino effect across your entire model. As I mentioned before, I'm not saying they don't have some top notch coders, but we honestly don't know. When it comes to work like this though, I can tell you rounding errors should be a legitimate concern.

As for Diogenes response, inefficiency can be hugely important. Most people assume that the difference between one approach the other is a few minutes, but that is not necessarily the case at all. It can be the difference of running in an hour and running for half a year, or worse. Granted, it sounds like their runs are in a timely manner, but as mentioned above, data typing off the top of my head is just one of many problems that could be introduced. There is a lot more to coding than the outside person seems to be aware of which is ultimately disappointing in this day and age. A good coder doing this work will be able to read the equation(s), understand them, and know which and how to apply any needed algorithms, transformations, etc. that the physicist potentially has no clue about.

paperburn1
Posts: 2484
Joined: Fri Jun 19, 2009 5:53 am
Location: Third rock from the sun.

Re: EMC2 news

Post by paperburn1 »

There are two types of coders in the sim world
1. Those that can extrapolate from incomplete data
I am not a nuclear physicist, but play one on the internet.

hanelyp
Posts: 2261
Joined: Fri Oct 26, 2007 8:50 pm

Re: EMC2 news

Post by hanelyp »

There seems to be some confusion between 2 different meanings of scaling. In the software world scaling refers to how well code deals with a larger system or dataset. In physics scaling refers to how a system changes behavior at different length scales. A simulation that deals well with physical scaling may scale horribly as a program. Or a simulation may use methods to scale well as a program but do poorly at modeling physical scaling.
The daylight is uncomfortably bright for eyes so long in the dark.

ladajo
Posts: 6258
Joined: Thu Sep 17, 2009 11:18 pm
Location: North East Coast

Re: EMC2 news

Post by ladajo »

Yes.
What matters at hand is physics scsling.
And, again, this project is being worked by coders and physicists who are top level people.
There is an argument to be made that the code is the most advamced 3d plasma code there is right now. It is cutting edge. And there is interest by other parties to make use of ot.
Relax guys, the team knows what it is doing.
The development of atomic power, though it could confer unimaginable blessings on mankind, is something that is dreaded by the owners of coal mines and oil wells. (Hazlitt)
What I want to do is to look up C. . . . I call him the Forgotten Man. (Sumner)

happyjack27
Posts: 1439
Joined: Wed Jul 14, 2010 5:27 pm

Re: EMC2 news

Post by happyjack27 »

Diogenes wrote:
happyjack27 wrote:
ladajo wrote: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.

Yes, we get it. You think you are a super duper brilliant software guy, and obviously professional nuclear physicists don't know what they are doing and are not fit to shine your intellectual shoes.


For those of you who think the "News" section ought to remain free of snark, I wish to say I forbore several times before now.
If you think that this is about ego, you've missed the point by a mile!

I got up to spead on the comments and I have to second that - yes, the nail was hit on the head!

happyjack27
Posts: 1439
Joined: Wed Jul 14, 2010 5:27 pm

Re: EMC2 news

Post by happyjack27 »

The proverbial elephant in the room is the electromagnetic fields from the moving particles.

ladajo
Posts: 6258
Joined: Thu Sep 17, 2009 11:18 pm
Location: North East Coast

Re: EMC2 news

Post by ladajo »

Elephant yes, ignored no.
They matter and are now well understood.
Complex equilibriums.
The development of atomic power, though it could confer unimaginable blessings on mankind, is something that is dreaded by the owners of coal mines and oil wells. (Hazlitt)
What I want to do is to look up C. . . . I call him the Forgotten Man. (Sumner)

hanelyp
Posts: 2261
Joined: Fri Oct 26, 2007 8:50 pm

Re: EMC2 news

Post by hanelyp »

Moving charged particles create a magnetic field, which influences the path of the particles, which if your configuration isn't dynamically stable can lead to a very complex mess. All the good plasma codes at least try to account for this. Some might, in the interest of efficiency, use shortcuts that fail with an unstable plasma. The polywell appears to be dynamically stable within the parameters of interest.
The daylight is uncomfortably bright for eyes so long in the dark.

mvanwink5
Posts: 2143
Joined: Wed Jul 01, 2009 5:07 am
Location: N.C. Mountains

Re: EMC2 news

Post by mvanwink5 »

One way to find out if the calculations / simulations are good enough...
Counting the days to commercial fusion. It is not that long now.

Betruger
Posts: 2321
Joined: Tue May 06, 2008 11:54 am

Re: EMC2 news

Post by Betruger »

Either way this is such a better state than the embargo's complete blackout from a few years ago.
You can do anything you want with laws except make Americans obey them. | What I want to do is to look up S. . . . I call him the Schadenfreudean Man.

Post Reply