magrid configuration brainstorming

Discuss how polywell fusion works; share theoretical questions and answers.

Moderators: tonybarry, MSimon

Randy
Posts: 82
Joined: Mon Sep 20, 2010 5:40 am
Location: Texas

Post by Randy »

mattman,

If you are unable to find employment in a fusion energy related field may I suggest trying "Space Exploration Technologies Corporation - SpaceX".
http://www.spacex.com/
The owner 'Elon Musk' created pay-pal and also owns a lot of Tesla Motors too. SpaceX is hiring now because of the space station cargo (Dragon spacecraft) contract with NASA. I think you would like working for a company like this. If I were younger I'd definitely drop them a resume'.

~Randy

prestonbarrows
Posts: 78
Joined: Sat Aug 03, 2013 4:41 pm

Re:

Post by prestonbarrows »

Randy wrote:... You simply cannot plot all the information at once – it’s too mind-boggling. You’re better off learning how to plot only certain subsets of the 3D magnetic field lines in question. In 2D, magnetic field lines (B-lines) form closed loops. In 3D, magnetic field lines (B-lines) generally – never end!...
Can't say that I have looked at your code at all but speaking in general for magnetic fields, this is not true.

From Gauss's Law Div(B) = 0
In practice, this means that if you draw any sort of box in space, the magnetic flux into and out of the box must be zero. As a consequence of this, ALL magnetic field lines are closed loops. An open magnetic field line would require a magnetic monopole which as far as we know do not exist.

Unfortunately, computer models will always be discrete approximations of continuous fields. This can lead to floating point errors or round off errors etc.This could be what is causing your field lines to never end. Small errors change a closed loop into a drifting spiral.

http://www.particleincell.com/blog/2011 ... tegration/ is an interesting link relating to similar issues in PIC codes.

Randy
Posts: 82
Joined: Mon Sep 20, 2010 5:40 am
Location: Texas

Re: Re:

Post by Randy »

prestonbarrows wrote:
Randy wrote:... You simply cannot plot all the information at once – it’s too mind-boggling. You’re better off learning how to plot only certain subsets of the 3D magnetic field lines in question. In 2D, magnetic field lines (B-lines) form closed loops. In 3D, magnetic field lines (B-lines) generally – never end!...
Can't say that I have looked at your code at all but speaking in general for magnetic fields, this is not true.

From Gauss's Law Div(B) = 0
In practice, this means that if you draw any sort of box in space, the magnetic flux into and out of the box must be zero. As a consequence of this, ALL magnetic field lines are closed loops. An open magnetic field line would require a magnetic monopole which as far as we know do not exist.

Unfortunately, computer models will always be discrete approximations of continuous fields. This can lead to floating point errors or round off errors etc.This could be what is causing your field lines to never end. Small errors change a closed loop into a drifting spiral.

http://www.particleincell.com/blog/2011 ... tegration/ is an interesting link relating to similar issues in PIC codes.
You're exactly right Preston. Magnetic field lines ALWAYS form closed loops due to Gauss's Law. Where you stated:

"Unfortunately, computer models will always be discrete approximations of continuous fields. This can lead to floating point errors or round off errors etc.This could be what is causing your field lines to never end. Small errors change a closed loop into a drifting spiral."

This computational error is what I meant when I said that the 3d B-loops never end. It's not as simple as giving the step routine a starting point and stepping until you get back to your starting point; because, you'll never get there (due to these small errors). I haven't found a good way to solve the problem of when to quit adding new line segments to a loop yet. I tried defining a starting plane perpendicular (normal) to the starting B-field vector and adding segments to both sides of it until I crossed the plane again, but sometimes (due to these small errors) the loop line starts curving away from the plane and never crosses it again. So I still have do it by hand using trial and error.

I was also working on an automatic step length routine a while back but never finished it. i.e., Something that would automatically take smaller steps in volumes of large B-field direction change and larger steps in volumes of small B-field direction change. This involved setting up an octree class for efficient look-up on large simulations. As of now I only have the octree class partially completed and haven't performed any experiments with implementation yet.

~Randy

Post Reply