|Hi i dont know if it does stand in the forum allready but what is the difference between all the simulations? i mean PhaseRotD , C, SolenoidsTo15cm and so on?|
|some reference or anything would be nice|
|well, easiest thing to do is open the lattice file, and read the parameters.|
I did start having a go at writing down the lattices in plain english, but, to be honest, I couldn't do it justice.
|hm i am not interested in the differenc in the parameters but if it is another simulation or does it somehow simulate something other?|
i mean if you would build it would it be something other or are this just some different optimising algorithms?
|well, chicanelinac is different from phaserot, for instance|
they have the same overall method, and are equivilent (in the overall accelerator, they're different designs for what would go in the one place).
Chicanelinac, for instance has a magnetic chicane to pick certain energy particles,and then a linac to control their speed. The phaserot uses phase rotation to do the same thing. Different methods for the same overall result. Which is used depends on yield. That help any?
|yap thats better=) thanks|
but is one of the methods better to do that or have they equal possibility to have a higher effectiveness?
|If you want to "see" what's happening, run the visible version (just double click muon1.exe).|
If you get a black screen, it is usually because of the Resolution setting in config.txt
Auto doesn't always work great, so look up your screen resolution and put those numbers into config.txt (1024x768) for example.
|velociraptor - the versions are all experimental. This is the ragged edge of design. Stephen (as far as I know) doesn't have any idea what the possible highest yield will be when he writes the lattice. Some of the designs (like the change from 2.2GeV to 5.something or whatever GeV was the result of a confrence and some work he did, thenwrote the file and looked at the results this project gave.|
Often, the 'computed' value given on the main page above the top 10 for that design, for comparison, is calculated the old way, if possible. Other times, it's the best result from a related lattice, run to see the difference. With even a 'small variable' lattice (200 or so variables) giving something like 10^600 possible dseigns (thats 1 with 600 zeros) there's no way other than with a project like this, to see which is the best.
but i thought that the simulator works evolutionary? so that he gets allway better and better? Or is it just a search with random start values?
|While it does use old results to build on, it also occasionally does random seeding for variation as far as I can tell. Otherwise there's the risk of getting stuck in one design and ignoring other possibilities.|
|yeah. It needs to go random. the problem otherwise is, since there's often 10^600 that it can focus on trying to evolve a certain area of only a million (10^6) or so simulations.|
The way I think of it is like an ocean floor. You can try and find the deepest spot all you like, but if you keep looking around the deepest spot you've currently found, you might miss that 4 mile deep crevace over there, that drops suddently (deptch being yield)
Thats the best metaphore I can think of for it anyway. You need the random to make sure you cvover the entire space. Thats also why it is also nice to have people who use the samplefiles, as well as those who don't
(I dunno, all these things, so easy to see in my head, but i'm so bad at getting the ideas out)