Until 1956, parity conservation was believed to be one of the fundamental geometric conservation laws (along with conservation of energy and conservation of momentum). However, in 1956 a careful critical review of the existing experimental data by theoretical physicists Tsung-Dao Lee and Chen Ning Yang revealed that while parity conservation had been verified in decays by the strong or electromagnetic interactions, it was untested in the weak interaction. They proposed several possible direct experimental tests. The first test based on beta decay of Cobalt-60 nuclei was carried out in 1956 by a group led by Chien-Shiung Wu, and demonstrated conclusively that weak interactions violate the P symmetry or, as the analogy goes, some reactions did not occur as often as their mirror image.
In 1964, James Cronin, Val Fitch and coworkers provided clear evidence (which was first announced at the 12th ICHEP conference in Dubna) that CP-symmetry could be broken. This work won them the 1980 Nobel Prize. This discovery showed that weak interactions violate not only the charge-conjugation symmetry C between particles and antiparticles and the P or parity, but also their combination. The discovery shocked particle physics and opened the door to questions still at the core of particle physics and of cosmology today.
...Just a reminder that CoE and CoM were once partners with another "absolute" law. Do I expect either CoE or CoM to fall to microwaves bouncing about in a copper cone? Not really. But I can dream. You get a Nobel Prize, and you get a Nobel Prize, and you...
I took a quick look through the MEEP source code and didn't see any assembly optimizations. By identifying bottlenecks and hand writing SIMD assembly routines, you can often improve performance on the order of several hundred percent (it's my day job). If you folks have a set of representative input data, I'll try to take a look with a profiler in search of low-hanging optimization fruit.
Quote from: VAXHeadroom on 07/02/2015 01:18 amQuote from: Rodal on 07/02/2015 12:57 amQuote from: aero on 07/02/2015 12:36 amThe higher density views are uploaded.https://drive.google.com/folderview?id=0B1XizxEfB23tfkVzeXVub2NpTm5fanZTTTdrLXNiT3VHaV9FYnB6TVpDUmJsWjRQbEUwdE0&usp=sharingThese Meep runs were made at resolution = 250 which is 2.5 times higher than previously uploaded views. These are the 14 final time slices of a 32 cycle run. 14 images for each view, separated by 0.1 cycle of the drive frequency 2.45 GHz. The Gaussian noise bandwidth of 2.45 GHz * .025 was used which seems reasonable for a magnetron. The 58 mm dipole antenna was located parallel to and 1/4 wavelength from the small end plate, excited with the Ez field component.Included are two models of the 10.2 inch NSF-1701 cavity, one using copper and the other using Perfect metal.I expect questions.Looking forward to seeing somebody make movies from these higher density runs !On it.First cut at a POVRay 3D animation. Let me know what anybody would like to see changed (slower/less/no rotation, less transparency, more loops etc). Personally loving this EY-EZ videoEX-HX videoHY-HZ video
Quote from: Rodal on 07/02/2015 12:57 amQuote from: aero on 07/02/2015 12:36 amThe higher density views are uploaded.https://drive.google.com/folderview?id=0B1XizxEfB23tfkVzeXVub2NpTm5fanZTTTdrLXNiT3VHaV9FYnB6TVpDUmJsWjRQbEUwdE0&usp=sharingThese Meep runs were made at resolution = 250 which is 2.5 times higher than previously uploaded views. These are the 14 final time slices of a 32 cycle run. 14 images for each view, separated by 0.1 cycle of the drive frequency 2.45 GHz. The Gaussian noise bandwidth of 2.45 GHz * .025 was used which seems reasonable for a magnetron. The 58 mm dipole antenna was located parallel to and 1/4 wavelength from the small end plate, excited with the Ez field component.Included are two models of the 10.2 inch NSF-1701 cavity, one using copper and the other using Perfect metal.I expect questions.Looking forward to seeing somebody make movies from these higher density runs !On it.
Quote from: aero on 07/02/2015 12:36 amThe higher density views are uploaded.https://drive.google.com/folderview?id=0B1XizxEfB23tfkVzeXVub2NpTm5fanZTTTdrLXNiT3VHaV9FYnB6TVpDUmJsWjRQbEUwdE0&usp=sharingThese Meep runs were made at resolution = 250 which is 2.5 times higher than previously uploaded views. These are the 14 final time slices of a 32 cycle run. 14 images for each view, separated by 0.1 cycle of the drive frequency 2.45 GHz. The Gaussian noise bandwidth of 2.45 GHz * .025 was used which seems reasonable for a magnetron. The 58 mm dipole antenna was located parallel to and 1/4 wavelength from the small end plate, excited with the Ez field component.Included are two models of the 10.2 inch NSF-1701 cavity, one using copper and the other using Perfect metal.I expect questions.Looking forward to seeing somebody make movies from these higher density runs !
The higher density views are uploaded.https://drive.google.com/folderview?id=0B1XizxEfB23tfkVzeXVub2NpTm5fanZTTTdrLXNiT3VHaV9FYnB6TVpDUmJsWjRQbEUwdE0&usp=sharingThese Meep runs were made at resolution = 250 which is 2.5 times higher than previously uploaded views. These are the 14 final time slices of a 32 cycle run. 14 images for each view, separated by 0.1 cycle of the drive frequency 2.45 GHz. The Gaussian noise bandwidth of 2.45 GHz * .025 was used which seems reasonable for a magnetron. The 58 mm dipole antenna was located parallel to and 1/4 wavelength from the small end plate, excited with the Ez field component.Included are two models of the 10.2 inch NSF-1701 cavity, one using copper and the other using Perfect metal.I expect questions.
(define-param srctype Ex) ;direction of the source current
@Dr. Rodal,Yesterday while reading this document http://meepunits.wikia.com/wiki/Meep_unit_transformation_WikiI found this comment in the code for example 2 Quote(define-param srctype Ex) ;direction of the source currentIt had never dawned on me that the x on Ex was the direction of the current. I have been alligning my Ez source antenna in the y direction thinking that in the real world, it couldn't matter. But maybe in the numerical model world? I changed direction of the antenna in the NSF-1701 copper model to the z direction, still using Ez as the source current and made a resonance run in Meep.The antenna direction does not change the resonant frequency but it increased the quality factor Q, by 40 times.I have since ran and uploaded the full set of .csv files, naming them zCopper .csv to differentiate them from the previously named Copper .csv files. They appear to have much larger numbers than before. You may wish to look at them.
Quote from: kitsuac on 07/02/2015 05:10 pmI took a quick look through the MEEP source code and didn't see any assembly optimizations. By identifying bottlenecks and hand writing SIMD assembly routines, you can often improve performance on the order of several hundred percent (it's my day job). If you folks have a set of representative input data, I'll try to take a look with a profiler in search of low-hanging optimization fruit.Attached find a control file that is representative. It seems that the Harminv module is slower than the normal continuous wave generation. To run the code, first change the extension from .txt to .ctl. See my system's run command at the end of the file. Three dimensional runs are by far the slowest so they could use some help.Luck, and thanks. aero
Before anybody gets excited about this intriguing result, this maybe a coincidence. We need to check other time steps to see whether this flow is consistent or whether it reverses like a sinusoid function in time.Well, if it averages out to zero, at least we have a vibration to start the thrust.
Skepticism, like chastity, should not be relinquished too readily.
Quote from: aero on 07/03/2015 01:12 amBefore anybody gets excited about this intriguing result, this maybe a coincidence. We need to check other time steps to see whether this flow is consistent or whether it reverses like a sinusoid function in time.Well, if it averages out to zero, at least we have a vibration to start the thrust. Those Poynting vector results are for the cases "Copper" (the original csv files here: https://drive.google.com/folderview?id=0B1XizxEfB23tfmkxNm1Ha1YxR1NZU2ZjUUpBUVVGV0M4QUVxaGYySEVFam5jVzdRYy0tSWs&usp=sharing from your message http://forum.nasaspaceflight.com/index.php?topic=37642.msg1398791#msg1398791).Can you also give me csv files for a couple IMMEDIATELY prior time steps ? Or would you need to re-run the case to do that? (If that's the case, I would RE_RUN It exactly the same, as we have a lot of time invested in this case, and that's better than starting with another model)
Quote from: Rodal on 07/03/2015 01:32 amQuote from: aero on 07/03/2015 01:12 amBefore anybody gets excited about this intriguing result, this maybe a coincidence. We need to check other time steps to see whether this flow is consistent or whether it reverses like a sinusoid function in time.Well, if it averages out to zero, at least we have a vibration to start the thrust. Those Poynting vector results are for the cases "Copper" (the original csv files here: https://drive.google.com/folderview?id=0B1XizxEfB23tfmkxNm1Ha1YxR1NZU2ZjUUpBUVVGV0M4QUVxaGYySEVFam5jVzdRYy0tSWs&usp=sharing from your message http://forum.nasaspaceflight.com/index.php?topic=37642.msg1398791#msg1398791).Can you also give me csv files for a couple IMMEDIATELY prior time steps ? Or would you need to re-run the case to do that? (If that's the case, I would RE_RUN It exactly the same, as we have a lot of time invested in this case, and that's better than starting with another model)No, I don't need to re-run the case. I have time steps 0 thru 13 in the h5 files. But lets be sure what you want to see before I make and upload the files. (I could upload all 14 in perhaps a day's time.) But you would be looking at time slices at the base only. Another option would be to look at a single time slice but look at slices of the cavity from large to small end, or left to right or front to back. There are a lot of those. More that I could generate or you could look at but we could pick and choose judiciously. 248 slices perpendicular to the x axis, and 265 slices parallel to it, in both y and z directions. Of course some of those x slices are outside of the cavity (look at a .csv file to find where the slice should start and end.) Same with y and z which cut "off-center" slices parallel to the x axis.And do look at the zCopper .csv files. It is exactly the same model except for the antenna being rotated by 90 degrees. I have the same data set for both copper cases. And the signal is much stronger in the zCopper data set.
Quote from: aero on 07/03/2015 01:58 amQuote from: Rodal on 07/03/2015 01:32 amQuote from: aero on 07/03/2015 01:12 amBefore anybody gets excited about this intriguing result, this maybe a coincidence. We need to check other time steps to see whether this flow is consistent or whether it reverses like a sinusoid function in time.Well, if it averages out to zero, at least we have a vibration to start the thrust. Those Poynting vector results are for the cases "Copper" (the original csv files here: https://drive.google.com/folderview?id=0B1XizxEfB23tfmkxNm1Ha1YxR1NZU2ZjUUpBUVVGV0M4QUVxaGYySEVFam5jVzdRYy0tSWs&usp=sharing from your message http://forum.nasaspaceflight.com/index.php?topic=37642.msg1398791#msg1398791).Can you also give me csv files for a couple IMMEDIATELY prior time steps ? Or would you need to re-run the case to do that? (If that's the case, I would RE_RUN It exactly the same, as we have a lot of time invested in this case, and that's better than starting with another model)No, I don't need to re-run the case. I have time steps 0 thru 13 in the h5 files. But lets be sure what you want to see before I make and upload the files. (I could upload all 14 in perhaps a day's time.) But you would be looking at time slices at the base only. Another option would be to look at a single time slice but look at slices of the cavity from large to small end, or left to right or front to back. There are a lot of those. More that I could generate or you could look at but we could pick and choose judiciously. 248 slices perpendicular to the x axis, and 265 slices parallel to it, in both y and z directions. Of course some of those x slices are outside of the cavity (look at a .csv file to find where the slice should start and end.) Same with y and z which cut "off-center" slices parallel to the x axis.And do look at the zCopper .csv files. It is exactly the same model except for the antenna being rotated by 90 degrees. I have the same data set for both copper cases. And the signal is much stronger in the zCopper data set.If you want to save time, we could start by looking ONLY at the plane that has "Y" as normal (the x z plane).Please post csv files forExy, Eyy, Ezy, Hxy, Hyy, Hzy, (I need all 6 to compute the Poynting vector)for 3 time steps immediately preceding the last time step for which you already have the csvYes, I will look at the zCopper files when i have a chance. Thanks
Quote from: Rodal on 07/03/2015 02:04 amQuote from: aero on 07/03/2015 01:58 amQuote from: Rodal on 07/03/2015 01:32 amQuote from: aero on 07/03/2015 01:12 amBefore anybody gets excited about this intriguing result, this maybe a coincidence. We need to check other time steps to see whether this flow is consistent or whether it reverses like a sinusoid function in time.Well, if it averages out to zero, at least we have a vibration to start the thrust. Those Poynting vector results are for the cases "Copper" (the original csv files here: https://drive.google.com/folderview?id=0B1XizxEfB23tfmkxNm1Ha1YxR1NZU2ZjUUpBUVVGV0M4QUVxaGYySEVFam5jVzdRYy0tSWs&usp=sharing from your message http://forum.nasaspaceflight.com/index.php?topic=37642.msg1398791#msg1398791).Can you also give me csv files for a couple IMMEDIATELY prior time steps ? Or would you need to re-run the case to do that? (If that's the case, I would RE_RUN It exactly the same, as we have a lot of time invested in this case, and that's better than starting with another model)No, I don't need to re-run the case. I have time steps 0 thru 13 in the h5 files. But lets be sure what you want to see before I make and upload the files. (I could upload all 14 in perhaps a day's time.) But you would be looking at time slices at the base only. Another option would be to look at a single time slice but look at slices of the cavity from large to small end, or left to right or front to back. There are a lot of those. More that I could generate or you could look at but we could pick and choose judiciously. 248 slices perpendicular to the x axis, and 265 slices parallel to it, in both y and z directions. Of course some of those x slices are outside of the cavity (look at a .csv file to find where the slice should start and end.) Same with y and z which cut "off-center" slices parallel to the x axis.And do look at the zCopper .csv files. It is exactly the same model except for the antenna being rotated by 90 degrees. I have the same data set for both copper cases. And the signal is much stronger in the zCopper data set.If you want to save time, we could start by looking ONLY at the plane that has "Y" as normal (the x z plane).Please post csv files forExy, Eyy, Ezy, Hxy, Hyy, Hzy, (I need all 6 to compute the Poynting vector)for 3 time steps immediately preceding the last time step for which you already have the csvYes, I will look at the zCopper files when i have a chance. ThanksOk- That will give you 4 tenths of a full cycle which I believe will be from the final half cycle of the 32 cycle run. I'm sure it will be actually. The h5 files output starts "after" time = 30.7 periods, the run time is set fo 32 periods and Meep stops as soon as the simulated time exceeds the run time. That should be the same time step that the 14th time slice is written to the h5 file.