[yt-users] star_density slices
Matthew Turk
matthewturk at gmail.com
Mon Jun 13 16:10:34 PDT 2011
Hi Brian,
That's what the routine Elizabeth was calling does -- it utilizes a
CIC_Deposit routine that Britton wrote to deposit the particles. The
problem was that the particles were located in coarse grids inside
regions covered by finer grids. So their deposited densities were
overwritten in the projection routine because the were in cells that
were "child masked."
The solution to this problem I prefer is to have well-posed datasets,
where the fluid quantities and the particle quantities are expected to
respect the same implicit understanding that higher-resolution regions
have the better solution to the underlying equations. However, one
could imagine solving this through a set of solution-projection
routines that ensure consistency between coarse and fine grids.
Such routines will ultimately be part of yt.
-Matt
On Mon, Jun 13, 2011 at 4:06 PM, Brian O'Shea <bwoshea at gmail.com> wrote:
> Hi Matt,
>
> There is a mechanism within yt for the dark matter density to be
> recalculated for some (all?) Enzo simulations. Could this be extended to
> star particles as well? That way, the user could specify "recalculate this
> field even if it already exists. Just trust me, yt!"
>
> --Brian
>
> On Mon, Jun 13, 2011 at 6:25 PM, Matthew Turk <matthewturk at gmail.com> wrote:
>>
>> Hi all,
>>
>> Elizabeth, Britton and I worked on this off-list on IRC.
>>
>> What I have deduced is that the star particles in her simulation are
>> not located at the finest resolution element. After discussing it
>> with her, this seems to me to be likely related to how she generated
>> the data, by restarting data written by an older version of Enzo with
>> a recent version of Enzo, having it write the potential field, and
>> then outputting again. What happens with data in a projection, or a
>> slice for instance, is that higher-resolution grids are assumed to
>> contain *better* data than the lower-resolution grids. So while
>> actual plots of the *particles* looked fine (I verified this) the
>> fluid quantities, particle_density etc etc, were being generated for
>> coarse level grids and then overwritten by finer level grids.
>>
>> Unfortunately, those finer level grids did not contain any particles.
>> So their fluid values were zero, and were overwriting coarser level
>> information which was non-zero. This is desired behavior; but I
>> cannot think of any solution to this problem that will work in all
>> cases. So, this will continue to go 'unfixed' for now, as well-formed
>> outputs are assumed to have their particles located at the
>> locally-finest level of resolution, and this dataset does not meet
>> that criterion.
>>
>> Best,
>>
>> Matt
>>
>> On Mon, Jun 13, 2011 at 11:04 AM, Britton Smith <brittonsmith at gmail.com>
>> wrote:
>> > Ok, that at least tells us it's not a problem with the filter that's
>> > looking
>> > for just the star particles. Can you run this script:
>> > http://paste.enzotools.org/show/1704/
>> > on your data, pipe the output to a file and send that back? Just give
>> > the
>> > path to your dataset on the command line. This should check to see if
>> > there
>> > are any inconsistencies in the grid dx's and positions.
>> >
>> > Britton
>> >
>> > On Mon, Jun 13, 2011 at 1:38 PM, Elizabeth Tasker <taskere at mcmaster.ca>
>> > wrote:
>> >>
>> >> Hi,
>> >>
>> >> Doesn't dm_density require the particle creation time to be less than
>> >> or
>> >> equal to zero? That is:
>> >>
>> >>
>> >> field = pf.field_info["dm_density"]
>> >> ------> print(field.get_source())
>> >> def _dmpdensity(field, data):
>> >> blank = na.zeros(data.ActiveDimensions, dtype='float32')
>> >> if data.NumberOfParticles == 0: return blank
>> >> filter = data['creation_time'] <= 0.0
>> >> if not filter.any(): return blank
>> >>
>> >>
>> >> amr_utils.CICDeposit_3(data["particle_position_x"][filter].astype(na.float64),
>> >>
>> >> data["particle_position_y"][filter].astype(na.float64),
>> >>
>> >> data["particle_position_z"][filter].astype(na.float64),
>> >>
>> >> data["particle_mass"][filter].astype(na.float32),
>> >> na.int64(na.where(filter)[0].size),
>> >> blank,
>> >> na.array(data.LeftEdge).astype(na.float64),
>> >>
>> >> na.array(data.ActiveDimensions).astype(na.int32),
>> >> na.float64(data['dx']))
>> >> return blank
>> >>
>> >>
>> >> It is zero in my simulation; I checked.
>> >>
>> >> The particle_density field though works ok (well.... it looks the same
>> >> as
>> >> the stars!) and I've attached a projection of it.
>> >>
>> >> Elizabeth
>> >>
>> >>
>> >>
>> >>
>> >> Matthew Turk wrote:
>> >>>
>> >>> Hi Elizabeth,
>> >>>
>> >>> We're basically trying to break down where the problem is.
>> >>>
>> >>> Here are the definitions of star_density:
>> >>>
>> >>> http://hg.enzotools.org/yt/src/tip/yt/frontends/enzo/fields.py#cl-269
>> >>>
>> >>>
>> >>> http://hg.enzotools.org/yt/src/tip/yt/data_objects/universal_fields.py#cl-384
>> >>>
>> >>> Alternately to plotting the dm_density would just be particle_density:
>> >>>
>> >>>
>> >>>
>> >>> http://hg.enzotools.org/yt/src/tip/yt/data_objects/universal_fields.py#cl-906
>> >>>
>> >>> Additionally, trying without the grid overplot would also be helpful.
>> >>> The specific item we're focusing on with the particle_density etc is
>> >>> whether it's the grids that are causing a problem or the particle
>> >>> selection.
>> >>>
>> >>> Thanks,
>> >>>
>> >>> -Matt
>> >>>
>> >>> On Mon, Jun 13, 2011 at 9:11 AM, Elizabeth Tasker
>> >>> <taskere at mcmaster.ca>
>> >>> wrote:
>> >>>
>> >>>>
>> >>>> Hi Britton,
>> >>>>
>> >>>> I don't have any DM in my simulations, so I don't think that will
>> >>>> produce
>> >>>> anything?
>> >>>>
>> >>>> Elizabeth
>> >>>>
>> >>>>
>> >>>> Britton Smith wrote:
>> >>>>
>> >>>>>
>> >>>>> Hi Elizabeth,
>> >>>>>
>> >>>>> Matt and I are still puzzling over this. It's not clear right now
>> >>>>> if
>> >>>>> the
>> >>>>> issue is related to projecting particles in general in your data or
>> >>>>> projecting star particles specifically. Can you do another full box
>> >>>>> projection for us, but this time with the field dm_density? That
>> >>>>> should
>> >>>>> help us to narrow down the problem.
>> >>>>>
>> >>>>> Britton
>> >>>>>
>> >>>>> On Mon, Jun 13, 2011 at 10:31 AM, Elizabeth Tasker
>> >>>>> <taskere at mcmaster.ca
>> >>>>> <mailto:taskere at mcmaster.ca>> wrote:
>> >>>>>
>> >>>>> Hi Matt and Britton,
>> >>>>>
>> >>>>> I've attached the image for the stars over the whole disk; it
>> >>>>> shows the same issue, with the grids (and also any gas property)
>> >>>>> showing the region correctly, but the stars only appearing in a
>> >>>>> small patch.
>> >>>>>
>> >>>>> Matt, your script gives:
>> >>>>>
>> >>>>> Min: [('x', 5.6774955112245706), ('y', 5.7878015648296586), ('z',
>> >>>>> 15.07668303734844)]
>> >>>>> Max: [('x', 26.537083788811703), ('y', 25.802074691296504), ('z',
>> >>>>> 17.201602159672426)]
>> >>>>>
>> >>>>>
>> >>>>> Which are the values I expect.
>> >>>>>
>> >>>>> It could be, as Britton suggests, a problem with the box size not
>> >>>>> being 0-1. Although, that is a particularly random patch it's
>> >>>>> decided to fill in.
>> >>>>>
>> >>>>> Elizabeth
>> >>>>>
>> >>>>>
>> >>>>>
>> >>>>> Matthew Turk wrote:
>> >>>>>
>> >>>>> Hi Elizabeth,
>> >>>>>
>> >>>>> Weird.
>> >>>>>
>> >>>>> This script should help us find out where the stars live.
>> >>>>> What's it
>> >>>>> give you? (Replace the loaded dataset with your own.)
>> >>>>>
>> >>>>> http://paste.enzotools.org/show/1696/
>> >>>>>
>> >>>>> Also, it'd be nice to see that image Britton requested.
>> >>>>>
>> >>>>> -Matt
>> >>>>>
>> >>>>> On Fri, Jun 10, 2011 at 2:02 PM, Elizabeth Tasker
>> >>>>> <taskere at mcmaster.ca <mailto:taskere at mcmaster.ca>> wrote:
>> >>>>>
>> >>>>> Oddly, neither of those make any difference.
>> >>>>>
>> >>>>> Elizabeth
>> >>>>>
>> >>>>>
>> >>>>>
>> >>>>> Britton Smith wrote:
>> >>>>>
>> >>>>> I would also suggest trying some projections instead
>> >>>>> of slices as well.
>> >>>>> Both of those fields are created using the
>> >>>>> CIC_deposit interpolation method
>> >>>>> with just the star particles. I could see a field
>> >>>>> like that being
>> >>>>> problematic when doing slices.
>> >>>>>
>> >>>>> Britton
>> >>>>>
>> >>>>> On Fri, Jun 10, 2011 at 4:10 PM, Cameron Hummels
>> >>>>> <chummels at astro.columbia.edu
>> >>>>> <mailto:chummels at astro.columbia.edu>
>> >>>>> <mailto:chummels at astro.columbia.edu
>> >>>>> <mailto:chummels at astro.columbia.edu>>> wrote:
>> >>>>>
>> >>>>> Try using the field StarMassMsun instead of
>> >>>>> star_density and see
>> >>>>> what it yields. I've had better luck with that
>> >>>>> field.
>> >>>>>
>> >>>>> Cameron
>> >>>>>
>> >>>>>
>> >>>>> On 6/10/11 4:00 PM, Elizabeth Tasker wrote:
>> >>>>>
>> >>>>> Hi,
>> >>>>>
>> >>>>> Slices of the star_density field are only
>> >>>>> showing a tiny patch of
>> >>>>> the area in the corner, rather than the whole
>> >>>>> region. My script
>> >>>>> looks like:
>> >>>>>
>> >>>>> pc = PlotCollection(pf, center=[17.,11.5, 16])
>> >>>>> p = pc.add_slice("star_density", 2)
>> >>>>> pc.set_width(10, 'kpc')
>> >>>>> p.modify["grids"]()
>> >>>>>
>> >>>>> Density comes out fine and the grids are drawn
>> >>>>> in but then only a
>> >>>>> few of them get stars.
>> >>>>>
>> >>>>> Elizabeth
>> >>>>>
>> >>>>>
>> >>>>> _______________________________________________
>> >>>>> yt-users mailing list
>> >>>>> yt-users at lists.spacepope.org
>> >>>>> <mailto:yt-users at lists.spacepope.org>
>> >>>>> <mailto:yt-users at lists.spacepope.org
>> >>>>> <mailto: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
>> >>>>> <mailto:yt-users at lists.spacepope.org>
>> >>>>> <mailto:yt-users at lists.spacepope.org
>> >>>>> <mailto: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
>> >>>>> <mailto: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
>> >>>>> <mailto: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
>> >>>>> <mailto: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 <mailto: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
>> >>>>>
>> >>>>>
>> >>>>
>> >>>> _______________________________________________
>> >>>> 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
>> >>>
>> >>
>> >>
>> >> _______________________________________________
>> >> 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
>> >
>> >
>> _______________________________________________
>> 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
>
>
More information about the yt-users
mailing list