[yt-dev] advice on implementing halo member particle loading

[Hilary Egan]

I think this is a good idea and would be helpful to a lot of people! I have
a couple (minor) concerns/questions

1. The difference between accessing the attributes of a halo catalog and
accessing the attributes of a halo's particles needs to be really well
documented and continuously restated throughout any docs examples. In those
two lines (below), it took me a little bit to think about whether I'd
expect that to return the halo's mass (considering the halo as a
"particle") or a list of the masses of the particles that make up the halo
- and I've used this machinery a lot!

my_halo = ds.halo(some_id)
my_halo["particle_mass"]

2. On a similar note, we should be more deliberate with the naming
conventions of the various fields in a halo catalog. At the moment, the
default mass of the particle is stored as 'particle_mass', which seems
misleading if thats also what we want to call the mass of constituent
particle field. Maybe something like "particle_mass" vs "halo_mass" and
"particle_position_x" vs "halo_position_x"?  I haven't completely thought
this through, but its at least a question we should consider.

By a similar (but pretty unrelated) token, we also should be more careful
about what we store as the halo radius. Right now we store a field
"virial_radius" in the halo catalog by default, which I think confuses a
lot of users, because of its similarity to the virial_quantities callback.

3. Is a data container the right way to go? Is there enough overlap between
the functionality we would want from a halo and the functionality of a
(generally) geometric subset of some original dataset? Given my lack of
familiarity with that infrastructure, I have absolutely no idea.

-Hilary

On Thu, May 7, 2015 at 9:15 AM, Ben Thompson <bthompson2090 at gmail.com>
wrote:

> Hi Britton.
>
> Probably should not be too difficult to port across my code for that.
>
> Sorry to also distract from the original topic too. I think the idea is
> good, but I don't have much experience on the data selector side of things
> myself, so I do not know how much help I would be on that part in
> particular.
>
> Ben.
>
> On Thu, May 7, 2015 at 12:11 PM, Britton Smith <brittonsmith at gmail.com>
> wrote:
>
>> Hi Ben,
>>
>> I have added frontends for the outputs of a couple different halo finders
>> now and found it to be pretty straightforward.  If you're interested in
>> doing this, you should have a look at either the rockstar or owls_subfind
>> frontends.
>>
>> Britton
>>
>> On Thu, May 7, 2015 at 11:08 AM, Ben Thompson <bthompson2090 at gmail.com>
>> wrote:
>>
>>> +1 for this idea. I have essentially been using yt.sphere data contains
>>> with the virial radius as my way of querying the particles within the halo.
>>> When in fact depending on the halo finder, halos do not necessary have to
>>> be spherical. Likewise rockstar and other halo finders also have machinery
>>> to tell you what particles are currently in the halo too.
>>>
>>> On topic of halo finders, how easy do you guys think would it be to
>>> implement other halo finders? Or at least be able to read in the results of
>>> other halo finders? With machinery like this, it would potentially be
>>> easier to "at least" include the results of other halo finders too. I have
>>> been using AHF for a lot of my work recently but am currently working with
>>> a custom version of the pynbody halo catalog (pretty much a direct rip but
>>> using YT to handle the units upon initialisation). I imagine with machinery
>>> like this it would be easier to do so.
>>>
>>> Ben.
>>>
>>> On Thu, May 7, 2015 at 11:57 AM, Britton Smith <brittonsmith at gmail.com>
>>> wrote:
>>>
>>>> Hi everyone,
>>>>
>>>> Myself and one other who should be joining this list soon are
>>>> interested in implementing some new functionality and I would like to
>>>> solicit some advice.
>>>>
>>>> What we have are halo catalog datasets that, for the most part, are
>>>> already loadable in yt.  They are loadable in that each individual halo is
>>>> a particle such that querying the mass field returns the mass of each
>>>> halo.  However, these datasets also store the particles from the simulation
>>>> snapshot that are the members of each halo.  We currently do not have the
>>>> machinery to query them and this is what we would like to add.
>>>>
>>>> It seems the most sensible idea is to create a halo data container such
>>>> that something like the following returns the mass of all the particles in
>>>> a halo:
>>>>
>>>> my_halo = ds.halo(some_id)
>>>> my_halo["particle_mass"]
>>>>
>>>> This is somewhat different from the existing data containers as it is
>>>> not geometric.  The member particles are explicitly known.  I don't have
>>>> much experience with making data containers so I can't tell if this makes
>>>> things easier or harder.  Anyway, this is the plan.
>>>>
>>>> If anyone has reasons this may be too difficult or simply not work, I
>>>> would appreciate hearing them.  Any general advice or other ideas for
>>>> making this functionality work would also be welcome.
>>>>
>>>> Thanks!
>>>>
>>>> Britton
>>>>
>>>> _______________________________________________
>>>> yt-dev mailing list
>>>> yt-dev at lists.spacepope.org
>>>> http://lists.spacepope.org/listinfo.cgi/yt-dev-spacepope.org
>>>>
>>>>
>>>
>>> _______________________________________________
>>> yt-dev mailing list
>>> yt-dev at lists.spacepope.org
>>> http://lists.spacepope.org/listinfo.cgi/yt-dev-spacepope.org
>>>
>>>
>>
>> _______________________________________________
>> yt-dev mailing list
>> yt-dev at lists.spacepope.org
>> http://lists.spacepope.org/listinfo.cgi/yt-dev-spacepope.org
>>
>>
>
> _______________________________________________
> yt-dev mailing list
> yt-dev at lists.spacepope.org
> http://lists.spacepope.org/listinfo.cgi/yt-dev-spacepope.org
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.spacepope.org/pipermail/yt-dev-spacepope.org/attachments/20150507/4a20b853/attachment.html>
-------------- next part --------------
_______________________________________________
yt-dev mailing list
yt-dev at lists.spacepope.org
http://lists.spacepope.org/listinfo.cgi/yt-dev-spacepope.org