!LQrdVpOJEohPSWMlmf:matrix.org

Smilei users

90 Members
github.com/SmileiPIC/Smilei1 Servers

Load older messages


SenderMessageTime
8 Jul 2021
@fredpz:matrix.orgfredpzand number of processes / threads ?11:26:18
@paulten:matrix.orgpaultenequal "SLURM/jobstart script". It will probably be restarted on different nodes than the first run was. That I cant control (I think). 11:27:25
@paulten:matrix.orgpaultenI can debug using just 1 mpi and 1 omp to see if that makes a difference. I pulled the latest and greatest version from GIT - I can try the stable version. I will keep you updated. Thanks for the suggestions. 11:31:41
@fredpz:matrix.orgfredpzYou need to rerun the first simulation though. Restarts require the same MPI/thread numbers11:40:39
@paulten:matrix.orgpaulten
In reply to @fredpz:matrix.org
You need to rerun the first simulation though. Restarts require the same MPI/thread numbers
Will do :)
11:50:39
@cecilia_n:matrix.orgcecilia_n joined the room.12:18:30
9 Jul 2021
@paulten:matrix.orgpaultenHi again everyone. I made some progress in debugging the restart-issue. My system has a number of nodes with 2 sockets with 64 cores each - I am therefore running 2 MPIs (-ntasks-per-node) * 64 threads (-cpus-per-task). To distribute the threads over the sockets and use the available resources I execute using "mpirun -map-by ppr:1:socket --bind-to socket" . -- If I dont include "-map-by ppr:1:socket" it distributes 128 threads to 1 socket (instead of 64 on each socket). Restarting fails when using this flag with mpirun. I can successfully start the code using the map-by flag, and restart the same simulation with the flag removed. Any other mpirun flags that would ensure distributing the threads over both sockets? Thanks everyone. Appreciate any suggestions. 13:14:20
@backereth:matrix.orgbackereth joined the room.14:06:35
@backereth:matrix.orgbackerethHello! Could you please tell me how to control and change the CEP of the laser pulse in LaserGaussian2D configuration?14:07:56
@beck-llr:matrix.orgbeck-llr paulten: I usually run -map-by ppr:1:socket:pe=64 to tell mpi there are 64 processing elements. 14:15:05
@beck-llr:matrix.orgbeck-llr I think the --bind-to-socket is useless at best, harmful at worse so you should try to remove it. 14:15:59
@beck-llr:matrix.orgbeck-llr backereth: If you want to fine tune the phase of the laser you will have to use the laser block in the namelist. The LaserGaussian2D is a convenient wrapper but does not allow as much flexibility. You can refer to the documentation. 14:19:35
@z10f:matrix.orgz10f Does changing phase_zero work? Inside the LaserGaussian2D block. 14:27:15
@z10f:matrix.orgz10f
In reply to @backereth:matrix.org
Hello! Could you please tell me how to control and change the CEP of the laser pulse in LaserGaussian2D configuration?
* Changing phase_zero works? Inside the LaserGaussian2D block
14:27:46
@z10f:matrix.orgz10f * Does changing phase_zero work? Inside the LaserGaussian2D block. 14:31:33
13 Jul 2021
@iustinouatu:matrix.orgiustinouatuHello, can I please check with you something wrt Ionization Rates and how they are implemented in SMILEI please? I see that beta_tunell[Z] has at its end the multiplication * au_to_w0 . Does it mean that the ADK rate expressed in atomic units is converted into SMILEI units by this multiplication? Then this rate is used in (1 - exp(IonizRate_Tunnel[z]*dt)). The dt in there is in SMILEI units as well so the SMILEI units of time and frequency cancel in the product? Correct? So the idea is that in the multiplication of rate * dt inside the exp there is a must that the rate is in SMILEI units, and this is achieved through the multiplication of the ionization rate expressed in Atomic Units by au_to_w0. Correct ?14:17:09
@fredpz:matrix.orgfredpz iustinouatu: correct 14:21:57
@iustinouatu:matrix.orgiustinouatuThank you14:24:24
@robintimmis:matrix.orgRobin TimmisDownload extract.txt21:22:06
@robintimmis:matrix.orgRobin TimmisHi there, I have been having difficulty with creating my own deposited quantity for the particle binning diagnostic. Please see attached for what I tried. Think perhaps I have misunderstood the docs?21:23:17
14 Jul 2021
@backereth:matrix.orgbackerethRedacted or Malformed Event15:23:37
@fredpz:matrix.orgfredpz Robin Timmis I am worried that `weights = particles.weight` might not make a copy of the array so that actual particle weights are really put to zero two lines after. Instead, try `weights = particles.weight.copy()` instead 15:46:12
15 Jul 2021
@rezaei.m.p:matrix.orgmohammad rezaei joined the room.09:02:35
@rezaei.m.p:matrix.orgmohammad rezaeiI have three questions: 1-Can I implement a Laser with radial polarization in Smilei? 2-Can I simulate a laser-ion accelerator that ions can ionization more than? 3- Is there any limitation in ions movement between grids (I read in the OSIRIS manual when I wanna simulate ions with more than one ionization occur during interaction I must assume ions can not move between cells)?09:10:16
@beck-llr:matrix.orgbeck-llrHello. 1) Yes with the generic "Laser" block. 2) Multiple ionizations can occur in a single iteration if that's the question. 3) I'm not aware of such a limitation.10:12:24
@rezaei.m.p:matrix.orgmohammad rezaei
In reply to @beck-llr:matrix.org
Hello. 1) Yes with the generic "Laser" block. 2) Multiple ionizations can occur in a single iteration if that's the question. 3) I'm not aware of such a limitation.
thanks so much
11:34:57
@robintimmis:matrix.orgRobin Timmis fredpz: That is a very good point I should definitely do that for my function but that does not explain why the second thing I tried did not work, please can you look at that? 15:38:54
@backereth:matrix.orgbackereth
In reply to @z10f:matrix.org
Does changing phase_zero work? Inside the LaserGaussian2D block.
In seems that phaseZero doesn't change CEP and waveform, it only shifts it along the time-axis.
16:02:04
@fredpz:matrix.orgfredpz Robin Timmis ah I didn't see the second part. Did you test with "weight" (not with a function)? 16:18:19
16 Jul 2021
@robint:matrix.orgrobint joined the room.14:08:23

Show newer messages


Back to Room List