[yt-users] Rockstar and RAMSES (DM only run). Rockstar appears to run, but no halo information is written.

Britton Smith brittonsmith at gmail.com
Fri Oct 3 15:29:41 PDT 2014


My guess is that this is a units issue.  It might be worth it to stick some
debugging statements into
yt/analysis_modules/halo_finding/rockstar/rockstar.py where the particles
are loaded up to be sent to rockstar.  I believe the units that rockstar is
expecting are positions in Mpc/h, masses in Msun/h, and velocities in km/s.

Britton

On Fri, Oct 3, 2014 at 11:39 AM, Ben Thompson <bthompson2090 at gmail.com>
wrote:

> I am using Ramses v3.11
>
> The current version and changeset of YT I am using is:
> ---
> Version = 3.0.1
> Changeset = f66765a58a61 (stable) @
> ---
>
> And yes, I have produced a projection plot of the simulation, and halo
> structures exist (in z=0). The halos have also been identified by running
> another halo finder independently (AHF)
>
> I will run the updates to see that helps.
>
> Will be back later.
>
> Ben.
>
>
> On Fri, Oct 3, 2014 at 5:30 PM, Cameron Hummels <chummels at gmail.com>
> wrote:
>
>> That's odd.  Could you produce a projection of the dark matter particles
>> to identify if halos actually exist? Try something like this:
>>
>> http://paste.yt-project.org/show/5145/
>>
>> Also, what simulation code are you using?
>>
>> And what version of yt are you using?  To determine your yt version, type:
>>
>> $ yt version
>>
>> I know that recently there were some changes in the output format of
>> rockstar that broke yt's interaction with it.  Perhaps that is related to
>> the problem you're experiencing.  You could try upgrading to the newest
>> version of yt and rockstar with this:
>>
>> $ yt update --all
>>
>> This may resolve your problems.
>>
>> Cameron
>>
>>
>> On Fri, Oct 3, 2014 at 9:22 AM, Ben Thompson <bthompson2090 at gmail.com>
>> wrote:
>>
>>> Hello.
>>>
>>> My simulation runs to z=0. Only with 10 timesteps though (purely a test
>>> run). It identifies no halos in halos_0 (expected with it being the first
>>> snapshot) and halos_9 (the final snapshot).. All files are the same size
>>> too with no data.
>>>
>>> My code is (with the filters taken out... it's a dark matter only one
>>> anyway)
>>>
>>>
>>> from mpi4py import MPI
>>> import yt, sys, os
>>> yt.enable_parallelism()
>>> from yt.config import ytcfg
>>> from yt.analysis_modules.halo_finding.rockstar.api import \
>>>         RockstarHaloFinder
>>> from yt.data_objects.particle_filters import \
>>>         particle_filter
>>> from yt.data_objects.time_series import DatasetSeries
>>> import numpy as np
>>>
>>> ncpu = MPI.COMM_WORLD.Get_size()
>>> ytcfg.set('yt', '__global_parallel_size', str(ncpu))
>>>
>>> outputs = np.arange(1, 11)
>>> dirs = []
>>> #Add the datasets
>>> for ioutput in outputs:
>>>     ds = yt.load('../output_%05d/info_%05d.txt'%(ioutput, ioutput))
>>>     dirs.append(ds)
>>>
>>> es = DatasetSeries(dirs)
>>> readers = int(ncpu/4.)
>>> writers = ncpu - readers - 1
>>> rh = RockstarHaloFinder(es, num_readers=readers, num_writers=writers)
>>> rh.run()
>>>
>>>
>>> Ben
>>>
>>>
>>> On Fri, Oct 3, 2014 at 5:09 PM, Cameron Hummels <chummels at gmail.com>
>>> wrote:
>>>
>>>> Hi Ben,
>>>>
>>>> No, you don't need to have halos in the first timestep.  Rockstar will
>>>> find the halos when they exist, so if you run it on all of the outputs and
>>>> you get to a later redshift then 5 or so, rockstar will start to identify
>>>> halos.  Does your simulation run to z=0?  If so, just look at the output
>>>> file associated with that redshift and see if it identifies any halos.  It
>>>> doesn't matter your starting redshift.
>>>>
>>>> Cameron
>>>>
>>>> On Fri, Oct 3, 2014 at 9:02 AM, Ben Thompson <bthompson2090 at gmail.com>
>>>> wrote:
>>>>
>>>>> Hi Cameron.
>>>>>
>>>>> So it is dependent as of what timestep I start my run at? I start my
>>>>> run (i.e there needs to be halos in the snapshot that you start with)? This
>>>>> is the first time I have used Rockstar before (figured that since the
>>>>> package is there, for free, I should use it) I have DM only runs that start
>>>>> at z=50 (of which I am not expecting to find halos there, as part of the
>>>>> loading script I loop over all the snapshots)
>>>>>
>>>>> This is what I have
>>>>>
>>>>> cat halos_0.0.ascii
>>>>> #id num_p mvir mbound_vir rvir vmax rvmax vrms x y z vx vy vz Jx Jy Jz
>>>>> E Spin PosUncertainty VelUncertainty bulk_vx bulk_vy bulk_vz BulkVelUnc
>>>>> n_core m200b m200c m500c m2500c Xoff Voff spin_bullock b_to_a c_to_a A[x]
>>>>> A[y] A[z] b_to_a(500c) c_to_a(500c) A[x](500c) A[y](500c) A[z](500c) Rs
>>>>> Rs_Klypin T/|U| M_pe_Behroozi M_pe_Diemer idx i_so i_ph num_cp mmetric
>>>>> #a = 0.019608
>>>>> #Bounds: (0.000000, 0.000000, 0.000000) - (99.816002, 99.816002,
>>>>> 19.807144)
>>>>> #Om = 0.300000; Ol = 0.700000; h = 0.700000
>>>>> #FOF linking length: 0.280000
>>>>> #Unbound Threshold: 0.500000; FOF Refinement Threshold: 0.700000
>>>>> #Particle mass: 2.87164e+10 Msun/h
>>>>> #Box size: 99.816005 Mpc/h
>>>>> #Total particles processed: 419422
>>>>> #Force resolution assumed: 0.780498 Mpc/h
>>>>> #Units: Masses in Msun / h
>>>>> #Units: Positions in Mpc / h (comoving)
>>>>> #Units: Velocities in km / s (physical, peculiar)
>>>>> #Units: Halo Distances, Lengths, and Radii in kpc / h (comoving)
>>>>> #Units: Angular Momenta in (Msun/h) * (Mpc/h) * km/s (physical)
>>>>> #Units: Spins are dimensionless
>>>>> #Units: Total energy in (Msun/h)*(km/s)^2 (physical)
>>>>> #Note: idx, i_so, and i_ph are internal debugging quantities
>>>>> #Np is an internal debugging quantity.
>>>>> #Rockstar Version: 0.99.9-RC3
>>>>>
>>>>> (yt-x86_64)[bthompson1 at leopard rockstar_halos]$ cat halos_9.0.ascii
>>>>> #id num_p mvir mbound_vir rvir vmax rvmax vrms x y z vx vy vz Jx Jy Jz
>>>>> E Spin PosUncertainty VelUncertainty bulk_vx bulk_vy bulk_vz BulkVelUnc
>>>>> n_core m200b m200c m500c m2500c Xoff Voff spin_bullock b_to_a c_to_a A[x]
>>>>> A[y] A[z] b_to_a(500c) c_to_a(500c) A[x](500c) A[y](500c) A[z](500c) Rs
>>>>> Rs_Klypin T/|U| M_pe_Behroozi M_pe_Diemer idx i_so i_ph num_cp mmetric
>>>>> #a = 1.000879
>>>>> #Bounds: (0.000000, 0.000000, 0.000000) - (99.816002, 99.816002,
>>>>> 19.667080)
>>>>> #Om = 0.300000; Ol = 0.700000; h = 0.700000
>>>>> #FOF linking length: 0.280000
>>>>> #Unbound Threshold: 0.500000; FOF Refinement Threshold: 0.700000
>>>>> #Particle mass: 2.87164e+10 Msun/h
>>>>> #Box size: 99.816005 Mpc/h
>>>>> #Total particles processed: 419437
>>>>> #Force resolution assumed: 0.779813 Mpc/h
>>>>> #Units: Masses in Msun / h
>>>>> #Units: Positions in Mpc / h (comoving)
>>>>> #Units: Velocities in km / s (physical, peculiar)
>>>>> #Units: Halo Distances, Lengths, and Radii in kpc / h (comoving)
>>>>> #Units: Angular Momenta in (Msun/h) * (Mpc/h) * km/s (physical)
>>>>> #Units: Spins are dimensionless
>>>>> #Units: Total energy in (Msun/h)*(km/s)^2 (physical)
>>>>> #Note: idx, i_so, and i_ph are internal debugging quantities
>>>>> #Np is an internal debugging quantity.
>>>>> #Rockstar Version: 0.99.9-RC3
>>>>>
>>>>> Both exactly the same pretty much.
>>>>>
>>>>> If it is because I am starting at an early timestep, then I will cut
>>>>> out the earlier ones from the initial setup.
>>>>>
>>>>> Thanks.
>>>>>
>>>>> Ben,
>>>>>
>>>>> On Fri, Oct 3, 2014 at 4:55 PM, Cameron Hummels <chummels at gmail.com>
>>>>> wrote:
>>>>>
>>>>>> Hi Ben,
>>>>>>
>>>>>> Oftentimes, rockstar cannot find any halos for the first few
>>>>>> snapshots of a cosmology DM-only sim because things have not collapsed
>>>>>> enough.  In that case, you'll get files like this:
>>>>>>
>>>>>> $ % cat halos_0.0.ascii
>>>>>> #id num_p mvir mbound_vir rvir vmax rvmax vrms x y z vx vy vz Jx Jy
>>>>>> Jz E Spin PosUncertainty VelUncertainty bulk_vx bulk_vy bulk_vz BulkVelUnc
>>>>>> n_core m200b m200c m500c m2500c Xoff Voff spin_bullock b_to_a c_to_a A[x]
>>>>>> A[y] A[z] b_to_a(500c) c_to_a(500c) A[x](500c) A[y](500c) A[z](500c) Rs
>>>>>> Rs_Klypin T/|U| M_pe_Behroozi M_pe_Diemer idx i_so i_ph num_cp mmetric
>>>>>> #a = 0.009901
>>>>>> #Bounds: (0.000000, 0.000000, 0.000000) - (40.000000, 40.000000,
>>>>>> 3.099222)
>>>>>> #Om = 0.282000; Ol = 0.718000; h = 0.697000
>>>>>> #FOF linking length: 0.280000
>>>>>> #Unbound Threshold: 0.500000; FOF Refinement Threshold: 0.700000
>>>>>> #Particle mass: 2.98630e+08 Msun/h
>>>>>> #Box size: 40.000000 Mpc/h
>>>>>> #Total particles processed: 1290100
>>>>>> #Force resolution assumed: 0.15625 Mpc/h
>>>>>> #Units: Masses in Msun / h
>>>>>> #Units: Positions in Mpc / h (comoving)
>>>>>> #Units: Velocities in km / s (physical, peculiar)
>>>>>> #Units: Halo Distances, Lengths, and Radii in kpc / h (comoving)
>>>>>> #Units: Angular Momenta in (Msun/h) * (Mpc/h) * km/s (physical)
>>>>>> #Units: Spins are dimensionless
>>>>>> #Units: Total energy in (Msun/h)*(km/s)^2 (physical)
>>>>>> #Note: idx, i_so, and i_ph are internal debugging quantities
>>>>>> #Np is an internal debugging quantity.
>>>>>> #Rockstar Version: 0.99.9-RC3
>>>>>>
>>>>>>
>>>>>> Is this what you're seeing or are your files truly empty?  If you're
>>>>>> seeing this, you may just need to run Rockstar on data files that are more
>>>>>> evolved in time, so as to allow for more non-linear collapse of the halos.
>>>>>> For my DM-only runs, depending on the mass resolution that I use, I may not
>>>>>> start to see halos until after z=10 to z=5, with the first outputs all
>>>>>> looking like the stuff above: empty.
>>>>>>
>>>>>> Cameron
>>>>>>
>>>>>> On Fri, Oct 3, 2014 at 8:46 AM, Ben Thompson <bthompson2090 at gmail.com
>>>>>> > wrote:
>>>>>>
>>>>>>> P.S sorry if I sent the same email twice.. this is the actual
>>>>>>> email... accidental press send before I sent the email out.
>>>>>>>
>>>>>>> Hello everyone.
>>>>>>>
>>>>>>> I am currently using YT to run rockstar on my simulation.
>>>>>>>
>>>>>>> I have been following the instructions which are around and about
>>>>>>> and have managed to write a script to run rockstar via MPI
>>>>>>>
>>>>>>> readers = int(ncpu/4.)
>>>>>>> #Reserve one cpu for the server
>>>>>>> writers = ncpu - readers - 1
>>>>>>> print 'Running rockstar with %i writers and %i readers'%(writers,
>>>>>>> readers)
>>>>>>> rh = RockstarHaloFinder(es, num_readers=readers, num_writers=writers,
>>>>>>>                         particle_type="dark_matter", dm_only=True)
>>>>>>> rh.run()
>>>>>>>
>>>>>>> using a particle filter on the datasets
>>>>>>>
>>>>>>> @yt.particle_filter("dark_matter", requires=[('particle_mass')])
>>>>>>> def dark_matter(pfilter, data):
>>>>>>>         if ('all', 'particle_age') in data.ds.field_list:
>>>>>>>                 return data[("all", "particle_age")] == 0.0
>>>>>>>         else:
>>>>>>>                 arr = np.zeros(len(data['particle_mass']))
>>>>>>>                 return arr == 0.0
>>>>>>>
>>>>>>> def setup_ds(ds):
>>>>>>>         #Return only dark matter particles, and assert that the
>>>>>>> filter holds
>>>>>>>         print 'Here'
>>>>>>>         assert(ds.add_particle_filter("dark_matter"))
>>>>>>>
>>>>>>>
>>>>>>> It seems to run fine.... e.g
>>>>>>>
>>>>>>>
>>>>>>> [   111s] Sending projection requests...
>>>>>>> [   112s] Transferring particles to writers...
>>>>>>> [   116s] Analyzing for FoF groups...
>>>>>>> [   116s] Transferring boundary particles between writers...
>>>>>>> [   117s] Linking boundary particles...
>>>>>>> [   117s] Analyzing for halos / subhalos...
>>>>>>> [   119s] Loading merger tree information...
>>>>>>> [   120s] Constructing merger tree...
>>>>>>> [   121s] [Success] Done with snapshot 7.
>>>>>>> [   121s] Reading 6 blocks for snapshot 8...
>>>>>>> P001 yt : [INFO     ] 2014-10-03 16:19:28,101 Loading field plugins.
>>>>>>> P002 yt : [INFO     ] 2014-10-03 16:19:28,101 Loading field plugins.
>>>>>>> P001 yt : [INFO     ] 2014-10-03 16:19:28,101 Loaded
>>>>>>> angular_momentum (8 new fields)
>>>>>>> P001 yt : [INFO     ] 2014-10-03 16:19:28,101 Loaded astro (14 new
>>>>>>> fields)
>>>>>>> P001 yt : [INFO     ] 2014-10-03 16:19:28,102 Loaded cosmology (20
>>>>>>> new fields)
>>>>>>> P001 yt : [INFO     ] 2014-10-03 16:19:28,102 Loaded fluid (56 new
>>>>>>> fields)
>>>>>>> P001 yt : [INFO     ] 2014-10-03 16:19:28,103 Loaded fluid_vector
>>>>>>> (88 new fields)
>>>>>>> P001 yt : [INFO     ] 2014-10-03 16:19:28,103 Loaded geometric (102
>>>>>>> new fields)
>>>>>>> P001 yt : [INFO     ] 2014-10-03 16:19:28,103 Loaded local (102 new
>>>>>>> fields)
>>>>>>> P001 yt : [INFO     ] 2014-10-03 16:19:28,104 Loaded magnetic_field
>>>>>>> (108 new fields)
>>>>>>> P001 yt : [INFO     ] 2014-10-03 16:19:28,104 Loaded species (108
>>>>>>> new fields)
>>>>>>> P006 yt : [INFO     ] 2014-10-03 16:19:28,127 Loading field plugins.
>>>>>>> P005 yt : [INFO     ] 2014-10-03 16:19:28,128 Loading field plugins.
>>>>>>> P003 yt : [INFO     ] 2014-10-03 16:19:28,128 Loading field plugins.
>>>>>>> P004 yt : [INFO     ] 2014-10-03 16:19:28,135 Loading field plugins.
>>>>>>> [   129s] Sending projection requests...
>>>>>>> [   130s] Transferring particles to writers...
>>>>>>> [   132s] Analyzing for FoF groups...
>>>>>>> [   133s] Transferring boundary particles between writers...
>>>>>>> [   133s] Linking boundary particles...
>>>>>>> [   134s] Analyzing for halos / subhalos...
>>>>>>> [   137s] Loading merger tree information...
>>>>>>> [   138s] Constructing merger tree...
>>>>>>> [   139s] [Success] Done with snapshot 8.
>>>>>>> [   139s] Reading 6 blocks for snapshot 9...
>>>>>>> [   143s] Sending projection requests...
>>>>>>> [   144s] Transferring particles to writers...
>>>>>>> [   146s] Analyzing for FoF groups...
>>>>>>> [   147s] Transferring boundary particles between writers...
>>>>>>> [   147s] Linking boundary particles...
>>>>>>> [   148s] Analyzing for halos / subhalos...
>>>>>>> [   150s] Loading merger tree information...
>>>>>>> [   151s] Constructing merger tree...
>>>>>>> [   152s] Constructing merger tree...
>>>>>>> [   153s] [Success] Done with snapshot 9.
>>>>>>> [   153s] [Finished]
>>>>>>>
>>>>>>> generates the outputs
>>>>>>>
>>>>>>>
>>>>>>> halos_0.0.ascii    halos_1.14.bin    halos_2.4.ascii
>>>>>>> halos_3.9.bin     halos_5.14.ascii  halos_6.3.bin     halos_7.9.ascii
>>>>>>> halos_9.13.bin
>>>>>>> halos_0.0.bin      halos_1.15.ascii  halos_2.4.bin
>>>>>>> halos_4.0.ascii   halos_5.14.bin    halos_6.4.ascii   halos_7.9.bin
>>>>>>> halos_9.14.ascii
>>>>>>> halos_0.10.ascii   halos_1.15.bin    halos_2.5.ascii
>>>>>>> halos_4.0.bin     halos_5.15.ascii  halos_6.4.bin     halos_8.0.ascii
>>>>>>> halos_9.14.bin
>>>>>>> halos_0.10.bin     halos_1.16.ascii  halos_2.5.bin
>>>>>>> halos_4.10.ascii  halos_5.15.bin    halos_6.5.ascii   halos_8.0.bin
>>>>>>> halos_9.15.ascii
>>>>>>> halos_0.11.ascii   halos_1.16.bin    halos_2.6.ascii
>>>>>>> halos_4.10.bin    halos_5.16.ascii  halos_6.5.bin     halos_8.10.ascii
>>>>>>> halos_9.15.bin
>>>>>>>
>>>>>>>
>>>>>>> but they are all empty.... yet within profiling, it seems to imply
>>>>>>> that there were some particles submitted to be analised
>>>>>>>
>>>>>>>
>>>>>>> [Prof] S0,C0 0s: 10191 fofs, 123365 particles, 0s for conf.
>>>>>>> [Prof] S0,C0: 0p,0h,0w; wt:0s; rcv:0s,0s; snd:0s; wk:0s; idl:1s
>>>>>>> [Prof] S1,C0 0s: 10377 fofs, 123363 particles, 0s for conf.
>>>>>>> [Prof] S1,C0: 1814p,73h,1w; wt:1s; rcv:0s,0s; snd:0s; wk:0s; idl:1s
>>>>>>> [Prof] S2,C0 0s: 9843 fofs, 123353 particles, 0s for conf.
>>>>>>> [Prof] S2,C0: 5115p,139h,1w; wt:1s; rcv:0s,0s; snd:0s; wk:0s; idl:1s
>>>>>>> [Prof] S3,C0 0s: 9616 fofs, 123359 particles, 0s for conf.
>>>>>>> [Prof] S3,C0: 9637p,221h,1w; wt:1s; rcv:0s,0s; snd:0s; wk:0s; idl:1s
>>>>>>> [Prof] S4,C0 1s: 9256 fofs, 123367 particles, 0s for conf.
>>>>>>> [Prof] S4,C0: 14849p,314h,1w; wt:0s; rcv:0s,1s; snd:0s; wk:0s; idl:0s
>>>>>>> [Prof] S5,C0 0s: 9233 fofs, 123361 particles, 0s for conf.
>>>>>>> [Prof] S5,C0: 17771p,369h,1w; wt:0s; rcv:0s,0s; snd:1s; wk:0s; idl:0s
>>>>>>> [Prof] S6,C0 0s: 8964 fofs, 123365 particles, 0s for conf.
>>>>>>> [Prof] S6,C0: 19888p,388h,1w; wt:0s; rcv:0s,1s; snd:0s; wk:0s; idl:0s
>>>>>>> [Prof] S7,C0 0s: 8996 fofs, 123356 particles, 0s for conf.
>>>>>>> [Prof] S7,C0: 22352p,426h,1w; wt:0s; rcv:0s,1s; snd:0s; wk:0s; idl:1s
>>>>>>> [Prof] S8,C0 0s: 8801 fofs, 123361 particles, 0s for conf.
>>>>>>> [Prof] S8,C0: 22553p,441h,1w; wt:0s; rcv:0s,1s; snd:0s; wk:0s; idl:1s
>>>>>>> [Prof] S9,C0 1s: 8806 fofs, 123361 particles, 0s for conf.
>>>>>>> [Prof] S9,C0: 22685p,440h,1w; wt:0s; rcv:0s,1s; snd:0s; wk:0s; idl:1s
>>>>>>>
>>>>>>>
>>>>>>> Is there anything I should consider looking at to debug this
>>>>>>> problem? Any suggestions of what to look at?
>>>>>>>
>>>>>>> Thanks
>>>>>>>
>>>>>>> Ben
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> _______________________________________________
>>>>>>> yt-users mailing list
>>>>>>> yt-users at lists.spacepope.org
>>>>>>> http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org
>>>>>>>
>>>>>>>
>>>>>>
>>>>>>
>>>>>> --
>>>>>> Cameron Hummels
>>>>>> Postdoctoral Researcher
>>>>>> Steward Observatory
>>>>>> University of Arizona
>>>>>> http://chummels.org
>>>>>>
>>>>>> _______________________________________________
>>>>>> yt-users mailing list
>>>>>> yt-users at lists.spacepope.org
>>>>>> http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org
>>>>>>
>>>>>>
>>>>>
>>>>> _______________________________________________
>>>>> yt-users mailing list
>>>>> yt-users at lists.spacepope.org
>>>>> http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org
>>>>>
>>>>>
>>>>
>>>>
>>>> --
>>>> Cameron Hummels
>>>> Postdoctoral Researcher
>>>> Steward Observatory
>>>> University of Arizona
>>>> http://chummels.org
>>>>
>>>> _______________________________________________
>>>> yt-users mailing list
>>>> yt-users at lists.spacepope.org
>>>> http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org
>>>>
>>>>
>>>
>>> _______________________________________________
>>> yt-users mailing list
>>> yt-users at lists.spacepope.org
>>> http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org
>>>
>>>
>>
>>
>> --
>> Cameron Hummels
>> Postdoctoral Researcher
>> Steward Observatory
>> University of Arizona
>> http://chummels.org
>>
>> _______________________________________________
>> yt-users mailing list
>> yt-users at lists.spacepope.org
>> http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org
>>
>>
>
> _______________________________________________
> yt-users mailing list
> yt-users at lists.spacepope.org
> http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.spacepope.org/pipermail/yt-users-spacepope.org/attachments/20141003/5262bab3/attachment.htm>


More information about the yt-users mailing list