 |
| Stephen Brooks 2004-05-05 04:41:43 | I need to set some parameters - how high does your particle count go on a simulation near or above 10%? The current simulation allows each particle 10 decays. The new version will also have a "timespread" because the incident proton beam is not instantaneous and we now need this because there are time-dependent components included further on. Here, I'd been running with 5x timespread (each particle replicated 5 times with different delays from the distribution) and also 10x decays, but that gets you up to 230`000 particles, which is rather too much RAM to expect from people. I was going to have a mixture like 2x from timespread, 5x from decays, or maybe 3x3x, to keep the current accuracy without introducing too many more, but as they appear at differnt points in the machine (the decay particles tend to rack up in number later on, whereas the timespread ones are at the beginning), I need to balance it. Hence the above question. |
| [TA]z 2004-05-05 06:33:35 | just from a quick sample of a fair amount of 10.00 thru 10.04: ~61,500 seems to be average but I do see some instances where that may be as high as 61,702 |
| Stephen Brooks 2004-05-05 07:05:29 | Right, that means I can expect the "10x" decays to give as much as 3x the number of particles in the later decay channel (you start off with ~20k). So if I release 3x the number of particles, that'll make 60k at the beginning also, but if I only let each one decay 3 times, that'll maintain the ~60k we've already got in the late decay channel. Having them equal is also nice because it balances out the artefacts a bit. |
| Stephen Brooks 2004-05-05 08:13:47 | What about the second number? I.e. it's 60000/something, which should be larger. |
| [TA]z 2004-05-05 08:32:38 | The second number is the number I gave you. So what we see for example is 785/61826 |
| Stephen Brooks 2004-05-05 08:57:47 | Oh right. In that case, how high does the first number ever get? |
| Pollock[Romulus2] 2004-05-05 10:22:17 | This is from a 10.01 run in command-line. Searching for auto-saved file... No file present Testing result from queue file Interpreting lattice file 'SolenoidsTo15cm'... Beamline consists of 59 units Adding components to simulation space Tantalum rod source data loaded. Building proximity grid 8x8x50 (3200 cells)... Done Done adding components Determining nearby components... - Starting - t = 17.79ns (15691/36758 particles) 22.9 Mpts Auto-saving... t = 18.50ns (16045/37163 particles) 24.0 Mpts Auto-saving... t = 19.31ns (16425/37571 particles) 25.3 Mpts Auto-saving... t = 20.08ns (16775/37948 particles) 26.6 Mpts Auto-saving... t = 20.73ns (17050/38241 particles) 27.7 Mpts Auto-saving... t = 21.39ns (17360/38600 particles) 28.8 Mpts Auto-saving... t = 22.18ns (17692/38995 particles) 30.2 Mpts Auto-saving... t = 22.99ns (18007/39396 particles) 31.7 Mpts Auto-saving... t = 27.20ns (19628/41319 particles) 39.6 Mpts Auto-saving... t = 46.03ns (25689/48409 particles) 82.5 Mpts Auto-saving... t = 61.10ns (29449/52742 particles) 124.1 Mpts Auto-saving... t = 74.61ns (32111/55845 particles) 165.8 Mpts Auto-saving... t = 87.06ns (34162/58201 particles) 207.1 Mpts Auto-saving... t = 98.27ns (35731/60038 particles) 246.3 Mpts Auto-saving... t = 112.90ns (10630/61420 particles) 286.3 Mpts Auto-saving... t = 200.00ns (624/61504 particles) 301.9 Mpts |
| [TA]z 2004-05-05 12:03:45 | quote: It has been my experience that the less efficient transfers produce a higher particle count on each side of the final result. Are you only interested in the ~10% transfers? I'll start watching and post my observations... -z |
| [TA]z 2004-05-05 18:35:19 | [transfer: 9.940772] t = 200.00ns (598/61608 particles) 300.6 Mpts [transfer: 9.964489] t = 200.00ns (576/61479 particles) 298.7 Mpts [transfer: 9.886478] t = 200.00ns (631/61556 particles) 299.7 Mpts [transfer: 9.854142] t = 200.00ns (617/61333 particles) 299.9 Mpts [transfer: 9.937522] t = 200.00ns (629/61808 particles) 302.5 Mpts / run 1 t = 200.00ns (618/61611 particles) 300.0 Mpts / run 2 t = 200.00ns (645/61565 particles) 300.6 Mpts / run 3 t = 200.00ns (635/61616 particles) 300.5 Mpts / run 4 t = 200.00ns (622/61639 particles) 300.2 Mpts / run 5 [transfer: 9.874082] t = 200.00ns (594/61491 particles) 300.7 Mpts [transfer: 10.055404] t = 200.00ns (680/61727 particles) 302.8 Mpts [transfer: 10.028430] t = 200.00ns (636/61601 particles) 300.2 Mpts [transfer: 10.063251] t = 200.00ns (657/61763 particles) 301.0 Mpts |
