19:59:36 発信元: Dr Ayman Mohamed : please help me to go to room 3 19:59:52 発信元: *Tutor* Tatsuhiko OGAWA : Replying to "please help me to go..." OK. I did. 20:00:05 発信元: 028 Amiegbereta Edwin Ehis : Good day. please move me to session 3 20:00:20 発信元: *Tutor* Tatsuhiko OGAWA : Replying to "Good day. please mov..." Sure 20:00:55 発信元: 099 hyoriyaz : Good mornig to all! 20:01:45 発信元: *Tutor* Tatsuhiko OGAWA : Replying to "Good mornig to all!" Good Morning! 20:03:26 発信元: 178 Nur Aida : Replying to "Good mornig to all!" konbanwa 20:04:00 発信元: *Tutor* Tatsuhiko OGAWA : Reacted to "konbanwa" with 👍 20:09:51 発信元: 083 Firliyani : Good morning, Sir.. Yesterday, I sent an email regarding a certification request, but I haven’t received any response until today. I kindly ask for your assistance. Thank you 20:11:44 発信元: 083 Firliyani : ok thank you 20:15:24 発信元: 001 Hamed Khodadadi : A question about DUMP file. In general, is it possible to use a dump file generated using a model in another model? 20:15:54 発信元: 110 I. Casalbore : What are the main differences between all the e-mode options? 20:16:29 発信元: *Tutor* Hunter Ratliff : Replying to "A question about DUM..." What exactly do you mean? If you're talking about using a dump file as a source for a second simulation, then yes. The second simulation is blind to the physics/parameter settings used to generate the dump file provided to it. 20:19:59 発信元: *Tutor* Hunter Ratliff : e-mode stands for "event generator mode". The short explanation is that event generator mode is needed for event-by-event data to be "realistically" simulated for <20 MeV neutrons. Otherwise, with e-mode=0 (off, "simple sampling mode"), individual histories may not properly conserve energy/momentum, though aggregate values, like flux, over many histories will be accurate. 20:20:02 発信元: 001 Hamed Khodadadi : Replying to "A question about DUM..." I want to use a dump file generated using an entire reactor model as source in another model which includes only one reactor component with more details. 20:20:37 発信元: *Tutor* Hunter Ratliff : Replying to "What are the main di..." e-mode = 2 is "version 2" of event generator mode with improved charged particle yields and neutron energy distributions relative to "version 1" 20:23:25 発信元: 110 I. Casalbore : Replying to "What are the main di..." Thank you. If I am mainly interested in charged particles, which option is better? 20:25:39 発信元: 083 Firliyani : can phits simulation calculate dead time on detector? 20:26:19 発信元: *Tutor* Hunter Ratliff : Replying to "A question about DUM..." Yes, it should be no problem as long as the coordinates are the same. The second simulation only sees the data in the dump file, particle species, coordinates, directions, energies, etc. To not have a needlessly large dump file, I'd make sure your dump file is focused on the area of interest for the second simulation rather than covering a much larger area (unless it's actually important). 20:27:45 発信元: *Tutor* Tatsuhiko OGAWA : Replying to "can phits simulation..." No. Because PHITS does not care what happens before or after radiation transport. To consider dead time, please manually correct it by totfact or by off-line analysis. 20:28:10 発信元: *Tutor* Tatsuhiko OGAWA : Replying to "can phits simulation..." PHITS can take into account for dead layer of detectors though 20:28:31 発信元: 001 Hamed Khodadadi : Replying to "A question about DUM..." Thank you. Got it! 20:28:44 発信元: *Tutor* Hunter Ratliff : Replying to "What are the main di..." It really depends on your calculation, as in higher-energy environments, you get more secondary production which can then go on to spawn even more radiation. In those cases, I'd suggest e-mode=2. But, if you're in a really high-energy environment and don't care at all about <20 MeV things, then I think it's less important. 20:29:36 発信元: 083 Firliyani : Replying to "can phits simulation..." @*Tutor* Tatsuhiko OGAWA ok thank you 20:37:11 発信元: *Tutor* Tatsuhiko OGAWA : Reacted to "@*Tutor* Tatsuhiko O..." with 👍 20:38:36 発信元: 110 I. Casalbore : Reacted to "It really depends on..." with 👍 20:55:51 発信元: 083 Firliyani : How to write material code for HPGe detector, and plastic detector in PHITS? 20:57:55 発信元: *Tutor* Tatsuhiko OGAWA : m2 Ge 1.0 $ HPGe $ NE213 0.874 g/cm^3 m3 H 0.548 C 0.452 $ NE110 1.032 g/cm^3 m4 H 0.524 C 0.476 20:59:36 発信元: 083 Firliyani : Reacted to "m2 Ge 1.0 $ HPGe ..." with ❤️ 21:06:13 発信元: 224 Yuri Franken : hello, in the red part of the polot i see a sumup peak as well, but shifted to a bit lower energy, I ran the simulation with ncas 10.000 21:08:26 発信元: *Tutor* Hunter Ratliff : Yep, this is expected. Kinematically, this is when one of the gamma rays backscatters from one detector to another, meaning energy is lost/deposited in both detectors, while the other gamma ray fully deposits its energy in a single detector. 21:09:43 発信元: 249 zulfahmi : sensei, can we get the explain about the output like in the phits output file. there are coulum lower and upper, so basically it can determine the resolution? so what is about activity x yield? 21:13:06 発信元: *Tutor* Hunter Ratliff : Replying to "sensei, can we get t..." The e-lower and e-upper just specify the minimum and maximum energies covered by each bin of the tally. In this example, you adjust that via the e1-type section, with the number of energy bins over the range emin to emax controlled by ne (300 here). 21:15:18 発信元: *Tutor* Hunter Ratliff : Replying to "sensei, can we get t..." This is different from energy resolution. This is simply how fine we make the tally binning. Energy resolution is something you provide as an input via dresol and dfano. 21:17:31 発信元: *Tutor* Hunter Ratliff : Replying to "sensei, can we get t..." In our [Source], we were specifying activities of the Co-60 source (well, its emission rates) such that the results would be scaled to #/second. 21:17:40 発信元: 249 zulfahmi : Replying to "sensei, can we get t..." thank you 21:17:52 発信元: *Tutor* Hunter Ratliff : Reacted to "thank you" with 👍 21:25:01 発信元: 249 zulfahmi : Replying to "sensei, can we get t..." so to determine the efficiency we use t track vs t-deposit? 21:27:45 発信元: *Tutor* Hunter Ratliff : Replying to "sensei, can we get t..." Yes, that's one way to do it. More strictly, I'd use [T-Cross] to score the number of particles entering the detector. You can use [T-Deposit] to count the fraction of histories interacting and depositing energy. (The [T-Interact] tally can also be used for this if you're curious in more complex investigation of the specific reactions occurring, and/or with a [Counter] section.) 21:28:37 発信元: 257 Ranulfo Dias : I have a question, in case about detection of neutrons source 241 Am_Be with reaction in a ³He detector (quenching gás) the product should be mainly Proton, Trition and Deuteron? 21:32:29 発信元: *Tutor* Hunter Ratliff : Since the detection reaction is 3He + n --> p + t, I'd assume you'd be detecting primarily protons and tritons, plus some 3He recoils for scattering events from higher energy neutrons (since this reaction is more likely for thermalized neutrons if I recall). Any deuterons would be from 3He breakup, assuming the source neutrons are high-energy enough to do that. 21:34:22 発信元: 257 Ranulfo Dias : Replying to "I have a question,..." Ok, excellent! Thank You. 21:34:40 発信元: 138-Le Huu Loi-Main : Reacted to "m2 Ge 1.0 $ HPGe ..." with ❤️ 21:35:43 発信元: *Tutor* Hunter Ratliff : Replying to "I have a question, i..." Just a quick lookup on ENDF reveals the threshold energy for 3He(n,d) is a bit over 4 MeV, so you would see some deuterons from an AmBe source if unmoderated. 21:38:26 発信元: 257 Ranulfo Dias : Replying to "I have a question,..." In my project about ³He detector programming in Geant4 and Phits will review it. Again, thank you to support. 21:39:57 発信元: *Tutor* Hunter Ratliff : you're welcome! Also, if you're modelling an AmBe source, I'd suggest checking out the resources in the \phits\sample\source\NeutronSource\AmBe and \phits\sample\source\NeutronSource\Precise-model\AmBe folders, if you've not seen this already. 21:41:29 発信元: 027 A. Al-Qaaod : Transporting the low energy electron < 1 KeV is it possible in the upgraded PHITS version? if so, how can one set for that? 21:43:21 発信元: 257 Ranulfo Dias : I'm looking at these examples, thank you. Oh, I have another question: if neutron detectors are under pressure over the years, can they lose pressure, reduce it? For example, a detector with 50 years reduced pressure to half ( 6 atm --> 3 atm) 21:45:36 発信元: 249 zulfahmi : Replying to "sensei, can we get t..." thank you 21:46:08 発信元: 027 A. Al-Qaaod : Yes, Thank you 21:46:16 発信元: *Tutor* Tatsuhiko OGAWA : Reacted to "Yes, Thank you" with 👍 21:46:31 発信元: 224 Yuri Franken : nice lecture, thank you 21:46:46 発信元: 052 Dali Georgobiani : Thank you very much, Ogawa-san!!! 21:47:18 発信元: *Tutor* Tatsuhiko OGAWA : Reacted to "nice lecture, thank ..." with 👍 21:47:19 発信元: 224 Yuri Franken : arigato gozaimasu 21:47:24 発信元: *Tutor* Tatsuhiko OGAWA : Reacted to "Thank you very much,..." with 👍 21:47:28 発信元: *Tutor* Tatsuhiko OGAWA : Reacted to "arigato gozaimasu" with 👍 21:47:46 発信元: 178 Nur Aida : arigato gozaimasu, Ogawa Sensei 21:47:47 発信元: 150 Melih Tahsin Değer Main : Arigato Ogawa-san 21:47:59 発信元: *Tutor* Tatsuhiko OGAWA : Reacted to "arigato gozaimasu, O..." with 👍 21:48:01 発信元: *Tutor* Tatsuhiko OGAWA : Reacted to "Arigato Ogawa-san" with 👍 21:48:18 発信元: *Tutor* Hunter Ratliff : I'm not really an expert in these detectors (my experience is more with organics), so this is something you could research. I think this is going to be highly variable depending on the manufacturing quality, age, how much physical stress the detector experienced, etc.. But yes, I think this is something that could happen, with these detectors or any sort of sealed gas chamber. You could model the impact of this in PHITS by just modifying the material composition and/or cell's density. 21:49:18 発信元: *Tutor* Hunter Ratliff : Reacted to "thank you" with 👍 21:51:00 発信元: wakai : ) Pages 38-39 (PHITS-shielding –en text in the lecture on Oct. 30) Tesch =1.07? <--What equation do we use for the evaluation in comparison with PHITS?? 21:51:59 発信元: wakai : In the DCHAIN excise calculation it does not use the JENDL-5 data. If we use the JENDL-5, how much can the result accuracy improve? What is the difference in principle in the calculation? 21:56:24 発信元: 257 Ranulfo Dias : Replying to "I have a question,..." So, I have this problem that indicates a reduction in pressure by half of the “useful life.” However, this results in the need for more experimental and comparative simulation tests to support this thesis. Thank you for your collaboration in the discussion. 21:57:09 発信元: *Tutor* Hunter Ratliff : The DCHAIN lecture's example did use JENDL5 data: dmax(1) = 200.000000 # (D=emin(1)) data max. energy of proton (MeV) dmax(2) = 200.000000 # (D=20.0) data max. energy of neutron (MeV) was set in the PHITS input (this is only utilized on your system if you have the data files present though; owing to its large size JENDL-5 is installed separately as detailed in the /phits/XS/ folder), and the default libraries in DCHAIN prioritize JENDL-5 first. But, the basic difference is that without JENDL-5 in the PHITS calculation, the [T-Yield] tally spawned by [T-Dchain] is just relying on reaction physics models to determine what nuclides are being produced for all reactions aside from <20 MeV neutrons. With JENDL-5, [T-Yield] instead uses activation cross section data for these reactions where available. I'm not sure how much comparative benchmarking has been done on this front. 21:57:18 発信元: *Tutor* Tatsuhiko OGAWA : Replying to ") Pages 38-39 (PHITS..." This is based on the Tesch's formula explained in page 20. 21:57:43 発信元: *Tutor* Tatsuhiko OGAWA : Replying to ") Pages 38-39 (PHITS..." H=H_casc∙J∙1/r^2 exp⁡(-d/λ(E_p ) ) is the basic formula 21:59:16 発信元: 099 hyoriyaz : Thanks. And can ask students to send a request 21:59:28 発信元: wakai : Replying to "In the DCHAIN excise..." Thank you very much! 21:59:37 発信元: *Tutor* Hunter Ratliff : Replying to "I have a question, i..." You're welcome, and good luck with your studies! 22:00:18 発信元: 257 Ranulfo Dias : Reacted to "You're welcome, an..." with 🙌 22:00:22 発信元: wakai : Replying to ") Pages 38-39 (PHITS..." Thank you. I got it. 22:00:24 発信元: 257 Ranulfo Dias : Reacted to "You're welcome, an..." with ❤️ 22:00:25 発信元: 257 Ranulfo Dias : Removed a 🙌 from "You're welcome, an..." 22:01:06 発信元: *Tutor* Hunter Ratliff : Replying to "In the DCHAIN excise..." You're welcome! 😊 22:01:11 発信元: *Tutor* SHIGYO Nobuhiro : Reacted to "Thanks. And can ask ..." with 👍 22:03:04 発信元: *Tutor* Tatsuhiko OGAWA : Reacted to "Thank you. I got it." with 👍 22:04:30 発信元: *Tutor* Hunter Ratliff : cd C:\phits\lecture\advanced\autorun cd phits/lecture/advanced/autorun 22:05:03 発信元: 264 + Kebede Shogile : Do we have this ppt in our folder? 22:05:20 発信元: *Tutor* Hunter Ratliff : Replying to "Do we have this ppt ..." yes, this ppt file is in that folder 22:05:28 発信元: *Tutor* Hunter Ratliff : Replying to "Do we have this ppt ..." phits-lec-autorun-en.pptx 22:06:21 発信元: 264 + Kebede Shogile : Replying to "Do we have this ppt ..." 👍 22:16:14 発信元: 214 S.Schmidt : Could I repeat a simulation in the very same folder, but change the name of the output files automatically using the shell file? Like for the first run, all output files will end on *_1.out, and for the second run *_2.out? 22:16:49 発信元: *Tutor* Tatsuhiko OGAWA : Yes. You can use the same input files. 22:17:30 発信元: 060 Dewa Ngurah Yudhi Prasada : I think the bat file already in 2.3, an I opening the incorrect folder? 22:17:58 発信元: 060 Dewa Ngurah Yudhi Prasada : Replying to "I think the bat file..." thank u 22:17:59 発信元: 214 S.Schmidt : Replying to "Could I repeat a sim..." And how is the format of the output file name automatically changed? 22:18:41 発信元: wakai : In eachrun.bat file, as I open it, it run right now. How do I change it? 22:20:18 発信元: *Tutor* Tatsuhiko OGAWA : Replying to "Could I repeat a sim..." I'm afraid I took your question wrong. The output files are overwritten. Output file names are not changed. 22:20:26 発信元: *Tutor* Hunter Ratliff : We will get to this in Exercise 4 :) That is, copying the outputs from each run to preserve them. 22:20:35 発信元: 214 S.Schmidt : Reacted to "I'm afraid I took yo..." with 👍 22:20:38 発信元: 214 S.Schmidt : Reacted to "We will get to this ..." with 👍 22:20:41 発信元: *Tutor* Tatsuhiko OGAWA : Reacted to "We will get to this ..." with 👍 22:20:44 発信元: 214 S.Schmidt : Replying to "Could I repeat a sim..." Oh great, thank you! 22:22:11 発信元: *Tutor* Hunter Ratliff : Ah, double-clicking it will execute it. To edit/change it, right-click and Open With a text editor like Notepad++ (or open the file from within a text editor). 22:25:16 発信元: wakai : Replying to "In eachrun.bat file,..." It did using edit in notepad. Thank you. 22:25:31 発信元: *Tutor* Hunter Ratliff : Reacted to "It did using edit in..." with 👍 22:29:34 発信元: Anton Moroz : I get an error The syntax of the command is incorrect. For these commands in Windows: mkdir eachout/%1 copy *.out eachout/%1 copy *.eps eachout/%1 22:30:25 発信元: 103 A.HUGUES : Replying to "I get an error The s..." you need to use \ and not / 22:30:29 発信元: *Tutor* Tatsuhiko OGAWA : Replying to "I get an error The s..." What if you use backslash in place of slash 22:30:35 発信元: *Tutor* Tatsuhiko OGAWA : Reacted to "you need to use \ an..." with 👍 22:30:50 発信元: Anton Moroz : Replying to "I get an error The s..." Wow that worked 22:30:54 発信元: *Tutor* Tatsuhiko OGAWA : Reacted to "Wow that worked" with 👍 22:30:59 発信元: 103 A.HUGUES : Reacted to "Wow that worked" with 👍 22:33:10 発信元: 103 A.HUGUES : is it possible to have the 5 graphs in one same graph? 22:35:11 発信元: *Tutor* Tatsuhiko OGAWA : Replying to "is it possible to ha..." By the native function of PHITS, no. Since these graphs were made by different execution of PHITS, it is not possible. I would superimpose these graphs off-line by external programs like Excel. 22:35:37 発信元: 103 A.HUGUES : Reacted to "By the native functi..." with 👍 22:36:21 発信元: 214 S.Schmidt : Here, the simulations are conducted subsequently. Could I also use this multirun for parallel computation via a slurm script on different CPUs? 22:37:43 発信元: *Tutor* Tatsuhiko OGAWA : Replying to "Here, the simulation..." No but it sounds useful. I will discuss this idea with my teammates after this tutorial. 22:38:14 発信元: 224 T.DahnCostante : Reacted to "No but it sounds use..." with 👍 22:38:22 発信元: *Tutor* Takuya Furuta : Replying to "Here, the simulation..." It is possible, since each run uses phits.bat so that single run is computed parallel way. 22:39:01 発信元: *Tutor* Takuya Furuta : Replying to "Here, the simulation..." But different calculation done by different CPU is not possible. 22:40:22 発信元: 214 S.Schmidt : Replying to "Here, the simulation..." Okay, then in this case, I would need to use MPI to parallelize the calculation? 22:41:17 発信元: 214 S.Schmidt : Reacted to "No but it sounds use..." with 👍 22:41:20 発信元: wakai : I performed/copied them and rum in phits.inp as shown in page 11 (EX.5), but I could not see five figures(only one). How to do for it? 22:41:21 発信元: 214 S.Schmidt : Reacted to "It is possible, sinc..." with 👍 22:41:46 発信元: *Tutor* Takuya Furuta : Replying to "Here, the simulation..." If the MPI setting is completed, just include $MPI=4 at the beginning of the input file. It will automatically initiate MPI parallel calculation in single run. 22:42:15 発信元: 214 S.Schmidt : Reacted to "If the MPI setting i..." with 👍 22:42:47 発信元: *Tutor* Tatsuhiko OGAWA : Replying to "Here, the simulation..." It seems this command allows parallel execution of commands in a batch script. Since current lecture example uses the same batch file (eachrun.bat), we need to add extra commands to duplicate eachrun.bat and change them before commands like below. @echo off start /b batch1.bat start /b batch2.bat start /b batch3.bat 22:43:02 発信元: 060 Dewa Ngurah Yudhi Prasada : so we can run autorun.inp by simply using autorun.bat command? 22:43:03 発信元: *Tutor* OHNISHI Seiki : Replying to "Here, the simulation..." It is possible to execute in parallel but the output file is overwritten. 22:43:52 発信元: *Tutor* Hunter Ratliff : Replying to "Here, the simulation..." I've done something very similar with initiating many PHITS simulations at once in Slurm, but my approach ultimately involved just having a bunch of separate folders (one per PHITS run) with identical inputs but using infl to insert a file into each where that inserted file had different "set: c1[value]" variables of interest. I'd written a general shell script for sbatch submitting Slurm jobs that used passed variables to determine which folder went with each run, and then one big shell script to submit all these jobs in one go. 22:44:24 発信元: *Tutor* Tatsuhiko OGAWA : Replying to "so we can run autoru..." Yes. Just running autrun.bat, autorun.inp and phits.inp are read as default input files. 22:44:41 発信元: *Tutor* Hunter Ratliff : Replying to "Here, the simulation..." I'd also used a Python script to help me organize all of this, but you could do such things in whatever scripting language you're most comfortable in. 22:45:15 発信元: *Tutor* OHNISHI Seiki : Did you copy the first line, paste, and change the last parameters of the multirun.bat? 22:45:17 発信元: *Tutor* Tatsuhiko OGAWA : Replying to "I performed/copied t..." You need to run multirun.bat 22:46:50 発信元: *Tutor* Hunter Ratliff : Replying to "Here, the simulation..." How you run this is entirely dependent on your specific environment, but this is a simple example of how I've ran PHITS with Slurm: 22:46:55 発信元: *Tutor* Hunter Ratliff : Replying to "Here, the simulation..." #!/bin/bash #SBATCH --ntasks=5 ### Number of tasks (MPI processes) #SBATCH --cpus-per-task=10 ### Number of threads per task (OMP threads) #SBATCH --time=3-23:58:59 #SBATCH --output=logfile_name.log #SBATCH --gres=gpu:0 module purge module load intel/22.4 module load phits/3.341 cd /shared/path/to/directory/with/PHITS/input/file/ #Only set OMP_NUM_THREADS if using OpenMP shared-memory parallelization, should match --cpus-per-task export OMP_NUM_THREADS=10 #To run PHITS with combined OpenMP + MPI (here, 5 (4+1) MPI processes & 10 OMP threads per process) #Note: if using MPI, MPI's "-np" should match Slurm's "--ntasks" mpirun -np 5 phits3341_LinIfort_OMP_MPI #To run PHITS with MPI alone (here, 5 (4+1) MPI processes) mpirun -np 5 phits3341_LinIfort_MPI #To run PHITS with OpenMP alone (here, 10 OMP threads) phits3341_LinIfort_OMP < beam-on-target_phits-input.inp #To run PHITS without any parallelization phits3341_LinIfort < beam-on-target_phits-input.inp 22:47:12 発信元: 214 S.Schmidt : Replying to "Here, the simulation..." Okay, then I would need to automatically adjust rseed in each input file to start with a new seed, like 1, 2, 3 ...? 22:47:33 発信元: 214 S.Schmidt : Reacted to "#!/bin/bash #SBATCH ..." with 👍 22:47:37 発信元: *Tutor* Tatsuhiko OGAWA : Replying to "I performed/copied t..." I am afraid you ran just phits.inp. It is also possible that you have only one line in multirun.bat 22:49:40 発信元: *Tutor* Hunter Ratliff : Replying to "Here, the simulation..." Yes, otherwise all the simulations would just start with the default seed (it is a fixed number, not dependent on time or anything like that) and have absolutely identical results. 22:50:04 発信元: wakai : Thank you. I got the five figures. 22:50:30 発信元: *Tutor* Tatsuhiko OGAWA : Reacted to "Thank you. I got the..." with 👍 22:51:48 発信元: *Tutor* Hunter Ratliff : Replying to "Here, the simulation..." I don't know the full details of best practices for picking random seeds, so perhaps another developer can chime in, but I usually try to make them a large odd number. The default rseed is 6.647299061401E+12 for example. 22:52:42 発信元: 214 S.Schmidt : Reacted to "Yes, otherwise all t..." with 👍 22:53:57 発信元: 214 S.Schmidt : Replying to "Here, the simulation..." But in theory should already be sufficient to have rseed = 1 in the first file, rseed = 2 in the second and so on? 22:54:54 発信元: *Tutor* Tatsuhiko OGAWA : If you mean to run multiple files with different configurations (say different c1 value), I Don't find reasons to use different random number seeds. Did I take you wrong? 22:57:21 発信元: 214 S.Schmidt : I realize I wasn’t very clear earlier. My goal is to split the simulation into several smaller runs, each with a different input file (but different rseed), since the cluster limits jobs to 72 hours. By running around 100 input files, I can both stay within the time limit and improve statistics using my customized t-userdefined scoring. 23:01:19 発信元: *Tutor* Tatsuhiko OGAWA : If your 100 input files are not identical, you Don't have to care about random number seeds because the calculations are different due to the difference of input files. If you think of running identical 100 files to reduce statistical uncertainty, I would suggest to use MPI or OpenMP. When MPI and/or OpenMP is used, the random number seeds in threads/processors are varied inside PHITS not to let them overlap. 23:01:46 発信元: *Tutor* Hunter Ratliff : Replying to "Here, the simulation..." I think he means the inputs are identical except for varying rseed, right? 23:02:07 発信元: 214 S.Schmidt : Replying to "Here, the simulation..." Yes exactly. 23:02:12 発信元: 214 S.Schmidt : Reacted to "If your 100 input fi..." with 👍 23:02:14 発信元: 103 A.HUGUES : when we compare the original phits.inp and the version in the differents folder after running autorun.bat we can see the the line 44 "infl:{density.inp}" deleted and replace it with a comment of the density.inp. All other $ comments were deleted. This is how works the autorun.bat ? 23:02:15 発信元: 214 S.Schmidt : Reacted to "I think he means the..." with 👍 23:02:37 発信元: *Tutor* Hunter Ratliff : Replying to "Here, the simulation..." Does your [T-userdefined] work with OpenMP/MPI parallelization, or does it only work single-threaded? 23:03:33 発信元: 214 S.Schmidt : That is something I was trying to figure out, so far it was not really working. Would it be possible for you @*Tutor* Hunter Ratliff to quickly discuss that in a Breakout room? 23:03:55 発信元: *Tutor* Hunter Ratliff : Replying to "Here, the simulation..." Actually, I kind of have the same issue with my own [T-Userdefined] tally haha 23:04:15 発信元: *Tutor* Hunter Ratliff : Replying to "Here, the simulation..." I never got mine working with OMP/MPI, but I want to revisit that sometime. 23:04:31 発信元: *Tutor* Takuya Furuta : Replying to "Here, the simulation..." You need to code it very carefully. Otherwise it is impossible. 23:04:46 発信元: *Tutor* Hunter Ratliff : Replying to "Here, the simulation..." (Part of this was just Fortran compilation issues on Windows for me that I never completely debugged.) 23:04:52 発信元: *Tutor* Takuya Furuta : It is not meant to use it in parallel in first place. 23:05:14 発信元: *Tutor* Hunter Ratliff : Reacted to "You need to code it ..." with 👍 23:05:16 発信元: *Tutor* Hunter Ratliff : Reacted to "It is not meant to u..." with 👍 23:06:26 発信元: 214 S.Schmidt : Reacted to "T-userdefined is MPI..." with 👍 23:07:16 発信元: *Tutor* Tatsuhiko OGAWA : Replying to "Here, the simulation..." So, in your case, please consider manual parallelization by varying rseed. 23:07:39 発信元: *Tutor* Hunter Ratliff : Replying to "Here, the simulation..." I guess the easiest solution here is the "poor man's parallelization" that Schmidt was mentioning, just running a bunch of single-threaded simulations simultaneously and combining results in post-processing, with different rseed in each. 23:08:23 発信元: *Tutor* Tatsuhiko OGAWA : Reacted to "I guess the easiest ..." with 👍 23:10:08 発信元: wakai : I run the same way as shown in Ex.8, but I did not obtain the same result shown in page 19 (answer 8) with larger deviation error data. What is the problem, do you think in my calculation? 23:10:45 発信元: *Tutor* Hunter Ratliff : Do you know if there are any particular advice regarding rseed selection? I remember in my Monte Carlo university course many years ago there was mention that the selected seed value could affect the RNG's period length, where some seeds would go "stale" much faster than others. I don't know the full details of the Linear Congruential Generator / xorshift64 used by PHITS and whether those warnings from my old MC course are relevant for those generators. 23:11:06 発信元: 214 S.Schmidt : Okay thank you so much for all your help and sorry for bothering you! So all in all, I will try the following: 1.) Generate n number of input files with the same geometry, but just changing the rseed. -> For the choice of rseed, any recommendation? 2.) Write a shell and slurm scripts to automize the job submission. Then I could have some kind of manual parallelization, with output stored in different folders, right? 23:11:53 発信元: 214 S.Schmidt : Reacted to "So, in your case, pl..." with 👍 23:12:02 発信元: *Tutor* Tatsuhiko OGAWA : Replying to "I run the same way a..." I cannot say anything definitive from your explanation. Do you discuss this issue in a breakout session? 23:12:26 発信元: *Tutor* Hunter Ratliff : Reacted to "Okay thank you so mu..." with 👍 23:12:39 発信元: 214 S.Schmidt : Reacted to "Do you know if there..." with 😅 23:13:42 発信元: *Tutor* Hunter Ratliff : just make sure you're varying the tally output file path (either directory or filename) of your N different inputs to make sure you're also getting N different outputs to later recombine. 23:14:50 発信元: 214 S.Schmidt : Replying to "Here, the simulation..." This I could do according to Excercise 4, correct? 23:15:36 発信元: *Tutor* Tatsuhiko OGAWA : Replying to "Here, the simulation..." In general , I ageee. -> For the choice of rseed, any recommendation? I would suggest to use irskip in place of rseed. If you run 1000 histories per one run, insert irskip = 1000 to [parameters] section. 1000 random Numbers are skipped so as to avoid random number overlap. 23:16:02 発信元: 214 S.Schmidt : Reacted to "just make sure you'r..." with 👍 23:17:03 発信元: *Tutor* Tatsuhiko OGAWA : Replying to "Here, the simulation..." irskip is an option to skip random numbers. 23:17:19 発信元: 052 Dali Georgobiani : Thank you very much, Sato-san!!! 23:17:31 発信元: wakai : Replying to "I run the same way a..." Thank you. I will ask it again in the breakout session. 23:18:09 発信元: 001 Hamed Khodadadi : Thank you for the great workshop. 23:18:18 発信元: 083 Firliyani : Thank you for all tutors 23:18:44 発信元: *Tutor* Hunter Ratliff : Ex4 is kind of serial in a way, depending on how you implement it. I think it's better for PHITS to be writing to a unique path per instance while it runs instead of writing to a file many instances may be trying to write to simultaneously and then renaming it afterward (which would obviously cause problems). 23:18:45 発信元: 129 - K.Butalag : Thanks you to all the tutors ! 23:18:46 発信元: 052 Dali Georgobiani : Many thanks to the wonderful PHITS team!!! ❤️ 23:18:57 発信元: 199 Renato Carvalho : Thank very much to all the tutors. You are the best! 23:19:02 発信元: 103 A.HUGUES : Thank you for all tutors 👏 23:19:05 発信元: 178 Nur Aida : Arigato gozaimashu,. Otsukare samadesu,..Mata ne . 23:19:09 発信元: 060 Dewa Ngurah Yudhi Prasada : Thank You sensei! ありがとうございます先生！ 23:19:34 発信元: 214 S.Schmidt : Replying to "Here, the simulation..." Do I get you correctly? You recommend to always use the same rseed for all the input files, but to vary irskip for each run, like run #1 with irskip = 1, #2 with irskip = 2 and so on? 23:19:40 発信元: Meutia Rahmi Afifa : Thank you very much sensei🙏 23:19:59 発信元: 087 Gede S : Thank you all (tutor) 23:19:59 発信元: 058 Desi Listianti : Thank you very much 🙏 23:20:01 発信元: 110 I. Casalbore : Thank you very much!! 23:20:28 発信元: 214 S.Schmidt : Great workshop! 23:20:30 発信元: 088 G.A. Atregenio : Thank you very much Sensei! 23:20:47 発信元: 102 Hsiao-po Lau : Thanks you for the excellent work from PHITS team！ 23:20:55 発信元: 008 A.Shehada : Thank you. How we can ask you after that , if there is something important to ask ? Can I send questions to email or in phits Forum ? 23:21:05 発信元: 138-Le Huu Loi-Main : Thank you very much! Thank for my 4th PHITS CODE! 23:21:38 発信元: *Tutor* Hunter Ratliff : Replying to "Thank you. How we ca..." The PHITS forum is a great place to post questions and discussions you may have. https://meteor.nucl.kyushu-u.ac.jp/phitsforum/ 23:21:43 発信元: 188 Per Roos : Thank you tutors for excellent work -truly enjoyable! 23:22:04 発信元: 008 A.Shehada : Replying to "Thank you. How we ca..." Ok thanks 23:23:03 発信元: 060 Dewa Ngurah Yudhi Prasada : Let's organize another PHITS Workhsop in Tokai !! 😁 23:23:15 発信元: 052 Dali Georgobiani : Reacted to "Let's organize anoth..." with ❤️ 23:23:20 発信元: 060 Dewa Ngurah Yudhi Prasada : Reacted to "Let's organize anoth..." with ❤️ 23:23:20 発信元: 060 Dewa Ngurah Yudhi Prasada : Removed a ❤️ reaction from "Let's organize anoth..." 23:23:22 発信元: 060 Dewa Ngurah Yudhi Prasada : Reacted to "Let's organize anoth..." with ❤️ 23:23:42 発信元: Mohamed ABDELLI : Are co-autors of papers also allowed for the workshop ? 23:23:51 発信元: 014 ahid nurmanjaya : Replying to "Let's organize anoth..." Okinawa is also nice I think. :D 23:24:10 発信元: 047-Chalit Muanglay-(Main) : Thank you for the great course!!! 23:24:11 発信元: 060 Dewa Ngurah Yudhi Prasada : Reacted to "Okinawa is also nice..." with ❤️ 23:24:14 発信元: 043 A.Bekouale : Thanks Mr Sato on 23:24:27 発信元: *Tutor* Tatsu Sato : Replying to "Are co-autors of pap..." Depends on our budget status, but the first author has priority 23:24:27 発信元: 060 Dewa Ngurah Yudhi Prasada : Replying to "Let's organize anoth..." @014 ahid nurmanjaya or Sapporo 😁 23:24:44 発信元: 138-Le Huu Loi-Main : Reacted to "Depends on our budge..." with 👍 23:24:46 発信元: 060 Dewa Ngurah Yudhi Prasada : Reacted to "Depends on our budge..." with 👍 23:25:05 発信元: Bal Vikram Khatri : Thank you very much! 23:25:14 発信元: 004 Pin-Jing Cheng : Thank you very much for the excellent tutorials 23:25:19 発信元: 114 Gordeev Main : Thank you very much! 23:25:43 発信元: Mohamed ABDELLI : Replying to "Are co-autors of pap..." Alright, thank you very much ! also I can't be the main author because i'm not a PhD student, but I mainly did the work for that paper so I was wondering how it works. I hope I will be able to participate ! 23:25:45 発信元: 210 Takoukam Serge : Thank so much Dear Tutors 23:26:33 発信元: 052 Dali Georgobiani : Thank you, Ogawa-san! 23:26:39 発信元: 052 Dali Georgobiani : Looking forward to it! 23:26:45 発信元: Mohamed ABDELLI : Thank you very much for this course ! 23:27:17 発信元: *Tutor* Tatsu Sato : Reacted to "Alright, thank you v..." with 👍 23:27:50 発信元: 150 Melih Tahsin Değer Main : Thanks a lot for this great tutorial 23:28:33 発信元: 072 E. Musacchio Gonzalez : Thanks for everything! 23:28:45 発信元: 043 A.Bekouale : Really interesting.Thank for all 23:28:54 発信元: William Asztalos : Thank you for the tutorial! 23:29:13 発信元: David Ferreira : Congratulations to everyone who developed PHITS and the training. 23:29:19 発信元: *Tutor* Hunter Ratliff : I did a bit more digging, and I think maybe rseed selection concerns are actually trivial for these particular RNGs in PHITS, so I think your original idea of just picking different rseed values could be fine. 23:29:22 発信元: 083 Firliyani : Reacted to "The PHITS forum is a..." with ❤️ 23:30:00 発信元: 214 S.Schmidt : Reacted to "I did a bit more dig..." with 🙏 23:30:29 発信元: 210 Takoukam Serge : Thanks for This PHITS wiki Very useful 23:31:00 発信元: 060 Dewa Ngurah Yudhi Prasada : Maybe I need some ANGEL tutorial, since I just know some ANGEL function in this course 😁 23:31:12 発信元: 074 Enrique Páez Lovera : Thank you very much for everything! Looking forward to learn more and apply PHITS to my work. 23:32:18 発信元: *Tutor* Tatsuhiko OGAWA : 1.Please answer our survey https://forms.gle/htPwzn7eSbFMKoAr5 to improve PHITS tutorials. 2.All tutorial videos will be uploaded here http://phits.jaea.go.jp/tutorial/rec.html in the next week. 23:33:54 発信元: *Tutor* Tatsuhiko OGAWA : 1.https://meteor.nucl.kyushu-u.ac.jp/phitsforum/ 23:34:58 発信元: 107 Iaroslav Bielov : Thank you for great tutorial! 23:35:00 発信元: 052 Dali Georgobiani : Thank you very much, and hope to see you soon again!!! 23:35:11 発信元: Mohamed ABDELLI : Thank you very much ! 23:35:17 発信元: 129 - K.Butalag : Thank you all again. 23:35:24 発信元: 167 Nahuel Emiliano Ricart : thank you very much for this tutorial! 23:35:24 発信元: 129 - K.Butalag : Have a great weekend 23:35:27 発信元: 012 Agus Dwi Purnomo : thanks you 23:35:29 発信元: 138-Le Huu Loi-Main : Thank you very much!!!!! 23:35:29 発信元: 210 Takoukam Serge : Thank you very much Dear Tutors 23:35:37 発信元: 157 MOHD HARIS RIDZUAN OOI BIN ABDULLAH : Thank you very much! 23:35:41 発信元: 039 Ayman Mohamed : Thanks you for the excellent 23:35:44 発信元: 224 Yuri Franken : i have a question about multiruns 23:35:44 発信元: 251Kayo : thanks you 23:35:51 発信元: 264 + Kebede Shogile : Thank you! 23:35:52 発信元: 047-Chalit Muanglay-(Main) : Thank you all very much, and hope to see you soon again!!! 23:35:53 発信元: 039 Ayman Mohamed : Thanks you very much 23:36:21 発信元: 150 Melih Tahsin Değer Main : Kono subarashī torēningu no okage de, watashi wa ōkiku seichō shimashita. Anata wa hontōni nesshin ni torikunde kuremashita.