loading Gadget outputs
Hi folks, I am trying to read in a Gadget output that’s split into 16 hdf5 files (fwiw, it’s an EAGLE output, and so “should” be similar to the OWLS format). I need to be able to analyze / render the entire output, not just one of the pieces, but I haven’t been able to figure out how to actually load them (akin to how for Enzo yt figures out that there are a lot of cpu outputs in a directory). I have taken a look at http://yt-project.org/doc/cookbook/owls_notebook.html#owls-notebook, but even there in the example, it’s just reading in the one output and ignoring the other 7 in the example tarball. And there I was having trouble tracking down what n_ref and over_refine_factor refer to. Interestingly, if I load just the single hdf5s, I can get ProjectionPlots to work but not SlicePlots (which insist there’s “zero dynamic range” with “Min = Max = 0.00000”.) FWIW, yt.__version__ gives me 3.3.1. (And major thanks to Nathan for getting that installation to go smoothly!!) Suggestions? —Molly
Hey Molly,
I'm not sure about why the Projection plots work but not the Slice plot,
but I can answer at least some of the questions.
When you have gadget/gizmo snapshot split into multiples (i.e.
snapshot_100.0.hdf5, snapshot_100.1.hdf5), you should be able to just load
the '0.hdf5' file, and it will load all of the multiples.
n_ref tells you the maximum number of particles can be in a cell before the
octree stops splitting (so lower number means finer octree splitting).
over_refine_factor > 0 will refine each cell in the octree one extra time
even after the n_ref threshold is met. Increasing this to 1 or 2 will make
much prettier images, at the cost of memory.
-desika
On Wed, Jul 27, 2016 at 12:50 PM, Molly Peeples
Hi folks,
I am trying to read in a Gadget output that’s split into 16 hdf5 files (fwiw, it’s an EAGLE output, and so “should” be similar to the OWLS format). I need to be able to analyze / render the entire output, not just one of the pieces, but I haven’t been able to figure out how to actually load them (akin to how for Enzo yt figures out that there are a lot of cpu outputs in a directory).
I have taken a look at http://yt-project.org/doc/cookbook/owls_notebook.html#owls-notebook, but even there in the example, it’s just reading in the one output and ignoring the other 7 in the example tarball. And there I was having trouble tracking down what n_ref and over_refine_factor refer to.
Interestingly, if I load just the single hdf5s, I can get ProjectionPlots to work but not SlicePlots (which insist there’s “zero dynamic range” with “Min = Max = 0.00000”.)
FWIW, yt.__version__ gives me 3.3.1. (And major thanks to Nathan for getting *that* installation to go smoothly!!)
Suggestions? —Molly
_______________________________________________ yt-users mailing list yt-users@lists.spacepope.org http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org
On Wed, Jul 27, 2016 at 12:56 PM, Desika Narayanan < desika.narayanan@gmail.com> wrote:
Hey Molly,
I'm not sure about why the Projection plots work but not the Slice plot, but I can answer at least some of the questions.
When you have gadget/gizmo snapshot split into multiples (i.e. snapshot_100.0.hdf5, snapshot_100.1.hdf5), you should be able to just load the '0.hdf5' file, and it will load all of the multiples.
n_ref tells you the maximum number of particles can be in a cell before the octree stops splitting (so lower number means finer octree splitting).
over_refine_factor > 0 will refine each cell in the octree one extra time even after the n_ref threshold is met. Increasing this to 1 or 2 will make much prettier images, at the cost of memory.
-desika
On Wed, Jul 27, 2016 at 12:50 PM, Molly Peeples
wrote: Hi folks,
I am trying to read in a Gadget output that’s split into 16 hdf5 files (fwiw, it’s an EAGLE output, and so “should” be similar to the OWLS format). I need to be able to analyze / render the entire output, not just one of the pieces, but I haven’t been able to figure out how to actually load them (akin to how for Enzo yt figures out that there are a lot of cpu outputs in a directory).
I have taken a look at http://yt-project.org/doc/cookbook/owls_notebook.html#owls-notebook, but even there in the example, it’s just reading in the one output and ignoring the other 7 in the example tarball. And there I was having trouble tracking down what n_ref and over_refine_factor refer to.
Interestingly, if I load just the single hdf5s, I can get ProjectionPlots to work but not SlicePlots (which insist there’s “zero dynamic range” with “Min = Max = 0.00000”.)
Is this a zoom-in simulation? It might be that you're slicing through a region that has no gas particles.
What happens if you specify the center of the SlicePlot in a region where you know there are gas particles? Something like: v, c = ds.find_max('density') slc = SlicePlot(ds, 2, 'density', center=c) slc.save()
FWIW, yt.__version__ gives me 3.3.1. (And major thanks to Nathan for
getting *that* installation to go smoothly!!)
Suggestions? —Molly
_______________________________________________ yt-users mailing list yt-users@lists.spacepope.org http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org
_______________________________________________ yt-users mailing list yt-users@lists.spacepope.org http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org
Hey Desika, Nathan,
Thanks! That was more painless than I expected :-)
Here’s a celebration image:
[cid:95DEA218-4B08-4334-ACB3-3D3AB7656F01]
—Molly
On July 27, 2016 at 2:00:13 PM, Nathan Goldbaum (nathan12343@gmail.commailto:nathan12343@gmail.com) wrote:
On Wed, Jul 27, 2016 at 12:56 PM, Desika Narayanan
participants (3)
-
Desika Narayanan
-
Molly Peeples
-
Nathan Goldbaum