drmike wrote:But the idea of "quick and dirty" doesn't apply without a lot of assumptions. I think it will actually be easier to do a brute force calculation than to pursue an analytical solution. But I think over the long run both need to be done along with experiments for a full picture to emerge.
OK, so I don't have the physics background to be able to do the analytical solutions, although I've been reading what you've posted.
But numerical brute-force simulation is probably doable, for me and any other interested (but not PhD Physics) readers. I like the idea of pursuing both!
The major advantage of doing both is that they each independently check the other. It should be possible to force the numerical crunching to match the analytical assumptions, and it should get close to the same answer. Sometimes qualitatively close is all you need - if it doesn't behave similarly, there's a bug some where!
And you have to have experiments to double check it all too.
But it takes time and effort - so the more people trying to do something, the better!
Alchemist wrote:Don't know if any of you have seen this yet, but it looks like it could provide the processing power to track a bunch of ions and electrons at once:
This is my first post, although I've been lurking for months.
I live in Albuquerque, just 45 minutes from Santa Fe. I've been tempted to take a day off, locate the office there, and beg for a chance to look around. However I'm guessing that this behavior would not endear me to Dr. Nebel Et Al. I'm not sure about the classification status either.
My M.S. is in CS, and my undergrad was EE and Math. It happens that a good portion of my career, including up to now, has to do with data visualization and algorithm development. I emailed Tom Ligon to see if he thought my skills might be useful to the folks in Santa Fe. He told me he thought they were in the head-down-best-not-to-interrupt mode and referred me here (a location which I had already found on my own as well).
I'm betting that at some point I'll end up working with reactor simulation code, but it seems to me the most immediate thing I could do would be to help with visualization of the results of other people's simulations. This especially in view of drmike's comments about having to debug his visualization code. I think I could fairly easily develop a visualization application that would make it easier (and fun!) to do things like see the fields and particle motions, spin the viewpoint and zoom in and out freely to get a better 3D understanding, put cut planes through the 3D data to see the fields and particle motion in user-chosen planes, use interactive coloring to enhance features of the data, etc.
So my question is this: does anybody (drmike, indrek, scareduck, and MSimon in particular) have simulation result data files lying around describing (1) magnetic fields from the magrid, (2) electric fields from an assumed electron distribution within the reactor, (3) motion over time of a particular individual electron or multiple individual electrons, and/or (4) motion over time of positive ions?
I'm not partial here--I'm just looking for data to build a visualization application around. If somebody is just doing brute force and is calculating the individual positions and mutual electric field effect of hundreds of thousands of electrons, rather than assuming a distribution generating a field and seeing how an individual electron moves, I'd be interested in that too.
I'd need format information as well--how the data files are arranged, and what the data actually is (e.g. the data file is the magnetic field in space with space divided into cubic shaped cells. The first line gives the number of cells in each dimension, and then each line after that is the x, y, and z value of the magnetic field for a particular cell. OR the data file is the motion of an electron over time. The first line gives information about the number of timesteps and the length of each timestep, each succeeding line is the xyz position of the electron at each succeeding timestep, or whatever...)
I've seen some of the stuff posted here, and I even followed a link to some data files. I might be able to do some interesting stuff rooting around, but I thought the direct request approach might be more efficient.
My enthusiasm is high, but my time is limited (wife, three kids, job...). I'm about to go on a vacation for a good part of June and was hoping I could see some data and stew over it while I'm traveling.
It's been a while since I looked at what I've posted on my web page. The data I create is all binary with no descriptors, but that is easy enough to add. I just have the code posted and some of my own crude visualization, so it would be pretty easy to upload or mail you raw data output from the code. Most files are in the 1 to 10 k byte range, so pretty small - but they are all 8 byte double float binary data.
I'll be really happy to help you play! I'm thinking about hardware at the moment, but it won't take long to get the simulation codes running to crank more data, let alone send data I've already got. You can send me mail directly and then post results here to everyone.
I finally got my computer to a state where I can do useful 3D modeling. By adding a graphics card that has a linux driver I got 3D acceleration to work with Ubuntu. Last night I got 3D arrows to work (except on a perfect axis - gotta fix that bug!) so converting my previous data sets into 3D viewable files is the next step.
I chose blender as the view engine, but *.obj files since almost every 3D modeler can import that file type. For file space, I may have to zip them.
But I think it will be advantageous if anybody can view the data sets at every possible angle. It will more accurately transfer information about what the B and E fields look like.
One advantage of blender is that it runs on linux, mac and windows.
Hopefully I'll have something useful in a week or so.
I'm honored that you are thinking of me. However, I'm a general systems/control kind of guy. I don't do much modeling code (except of the very simplest kind).
My thing is taking the physics and turning it into a power producer. Safely and economically.
Engineering is the art of making what you want from what you can get at a profit.
I finally got a 3D plot working. My first example is the "tombo coil", which is 12 wires in zig-zag configuration described elsewhere in the forum by tombo. The data is in *.obj format which can be read by blender and many other 3D rendering programs. The files that generated all the data along with the data are located at http://www.eskimo.com/~eresrch/Fusion/tombo_coil/
The code that creates the magnetic fields is this file
The code that creates the 3D image vector data is this file.
And the 3D output data is gziped into http://www.eskimo.com/~eresrch/Fusion/t ... eld.obj.gz
If you save the *.gz, then gunzip it you can "import" into blender as a "wavefront *.obj" file. It is then easy to rotate the image or translate it in 3D so all angles can be viewed.
The color of the arrows is set up to mark B field magnitude with red as "hot" and blue as "cold". It's too much fun just to play with the camera and lighting, so I'd better get back to the standard MaGrid B and E fields next!
I hope this helps with visualization. It will be interesting to try to make 3D movies of plasma motion.....
Check to see if AutoCAD can import *.obj files first! I picked that format because it is almost universal - it is in the public domain. My bet is you don't have to relearn anything,
you can just "import" a foreign format and autocad will convert it to its own correct format.
Once you can do that, let me know what you think of the "arrows". I'm thinking I need to reduce the size as the magnitude of the field shrinks below a certain point, and use color to tell where the field varies a lot. The light blue zones are almost zero field, yet the arrows are the same size as where the field is strong.
I'll see if I can get E and B fields in the same plot for the MaGrid fields I've already computed. That should be interesting!
I got the electric and magnetic fields for a 1/48 MaGrid done. You can find them here with the code that generated it here. I used the output from spfieldgen.c modified to use MAXDIM of 25.
I played with blender zooming in an around the data. It's pretty cool, but I'm not so sure I really understand it all. There are places where the E and B fields cross, and places where they point in similar directions. Exactly how that plays out with the plasma is another story.
It's been a lot of fun doing theory, but now I want to build some hardware. I'm going to start with welding power supplies - high currents and low voltage. If nothing else I can run some magnets with that, and maybe build a vacuum chamber with a little bit of welding practice!
I found some AutoCAD plugins for Wavefront .obj files. So anybody should be able to view this data.
Hello Dr. Mike,
Thank you for the sim results. I have Blender running OK on my Mac and I can visualise the field vectors. I am still learning Blender, so I have yet to work out sections ...
Would it be possible to you to publish two sections through your dataset? Both along the polar axis. One running through an equatorial coil centre, and one running through the point where two equatorial coils touch.
For the sake of further conversation, we might call these two planes Meridian Zero (through the equatorial coil centre) and Meridian 45 (through the equatorial coil touchpoint).
The plots along each of these planes would be:-
*B Field (magnetic) ... X and Y as distance from the polar axis, transparent to red colour as field strength. This is a 2D plot. Overlaid on this is
an arrow display indicating the B field vector on a given grid throughout the section. The section is viewed at an isometric angle (i.e. about 30 - 45 degrees to the plane). The arrowlengths are proportional to field strength.
*E Field (electric) ... X and Y as distance from the polar axis, blue colour as electron density, red colour as ion density (if these are determined individually) or red to blue colour as E field strength if not.