Loading Ramses data#
[1]:
import numpy as np
import osyris
Loading a Ramses output#
We load a data output from a star formation simulation
[2]:
path = "osyrisdata/starformation"
data = osyris.RamsesDataset(8, path=path).load()
Processing 12 files in osyrisdata/starformation/output_00008
16% : read 65623 cells, 0 particles
25% : read 90140 cells, 956 particles
33% : read 118232 cells, 956 particles
41% : read 147100 cells, 956 particles
50% : read 170244 cells, 2109 particles
66% : read 235859 cells, 2109 particles
75% : read 260384 cells, 3065 particles
83% : read 288476 cells, 3065 particles
91% : read 312840 cells, 4217 particles
Loaded: 340488 cells, 4218 particles.
and show a 2D histogram of the gas density versus the magnetic field strength
[3]:
osyris.hist2d(data["mesh"]["density"], data["mesh"]["B_field"], norm="log", loglog=True)
[3]:
<osyris.core.plot.Plot at 0x717247c627d0>
Selective loading#
Filtering on cell values#
It is possible to load only a subset of the cells, by using custom functions to perform the selection.
As an example, to load all the cells with \(\rho > 10^{-15}~{\rm g~cm}^{-3}\), we use a selection criterion
[4]:
data = osyris.RamsesDataset(8, path=path).load(
select={"mesh": {"density": lambda d: d > 1.0e-15 * osyris.units("g/cm**3")}}
)
Processing 12 files in osyrisdata/starformation/output_00008
16% : read 23 cells, 0 particles
25% : read 5064 cells, 956 particles
33% : read 6550 cells, 956 particles
41% : read 7635 cells, 956 particles
50% : read 13259 cells, 2109 particles
66% : read 13282 cells, 2109 particles
75% : read 18323 cells, 3065 particles
83% : read 19809 cells, 3065 particles
91% : read 23824 cells, 4217 particles
Loaded: 26518 cells, 4218 particles.
and we can now see in the resulting histogram that all the low-density cells have been left out:
[5]:
osyris.hist2d(data["mesh"]["density"], data["mesh"]["B_field"], norm="log", loglog=True)
[5]:
<osyris.core.plot.Plot at 0x717245896140>
Multiple selection criteria are ANDed:
[6]:
data = osyris.RamsesDataset(8, path=path).load(
select={
"mesh": {
"density": lambda d: d > 1.0e-16 * osyris.units("g/cm**3"),
"position_x": lambda x: x > 1500.0 * osyris.units("au"),
},
}
)
Processing 12 files in osyrisdata/starformation/output_00008
16% : read 2180 cells, 0 particles
25% : read 16026 cells, 956 particles
33% : read 24666 cells, 956 particles
41% : read 31428 cells, 956 particles
50% : read 47636 cells, 2109 particles
66% : read 51327 cells, 2109 particles
75% : read 65647 cells, 3065 particles
83% : read 74636 cells, 3065 particles
91% : read 89198 cells, 4217 particles
Loaded: 99892 cells, 4218 particles.
[7]:
osyris.map(
data["mesh"].layer("density"),
dx=1000 * osyris.units("au"),
origin=data["mesh"]["position"][np.argmax(data["mesh"]["density"])],
direction="z",
norm="log",
)
[7]:
<osyris.core.plot.Plot at 0x71723e7f92a0>
Only loading certain groups or variables#
It is also possible to select which groups to load. The different groups for the present simulation are mesh, part, and sink. For example, to load only the mesh and sink groups, use
[8]:
data = osyris.RamsesDataset(8, path=path).load(["mesh", "sink"])
data.keys()
Processing 12 files in osyrisdata/starformation/output_00008
16% : read 65623 cells, 0 particles
25% : read 90140 cells, 0 particles
33% : read 118232 cells, 0 particles
41% : read 147100 cells, 0 particles
50% : read 170244 cells, 0 particles
66% : read 235859 cells, 0 particles
75% : read 260384 cells, 0 particles
83% : read 288476 cells, 0 particles
91% : read 312840 cells, 0 particles
Loaded: 340488 cells, 0 particles.
[8]:
dict_keys(['sink', 'mesh'])
It is also possible to load only certain variables in a group by listing them in the same way:
[9]:
data = osyris.RamsesDataset(8, path=path).load(
select={"mesh": ["density", "position_x", "velocity_y"]}
)
data["mesh"].keys()
Processing 12 files in osyrisdata/starformation/output_00008
16% : read 65623 cells, 0 particles
25% : read 90140 cells, 956 particles
33% : read 118232 cells, 956 particles
41% : read 147100 cells, 956 particles
50% : read 170244 cells, 2109 particles
66% : read 235859 cells, 2109 particles
75% : read 260384 cells, 3065 particles
83% : read 288476 cells, 3065 particles
91% : read 312840 cells, 4217 particles
Loaded: 340488 cells, 4218 particles.
[9]:
dict_keys(['position_x', 'density', 'velocity_y'])
Selecting AMR levels#
Selecting AMR levels uses the same syntax as selecting other variables, but works slightly differently under the hood. A maximum level will be found by testing the selection function provided, and only the required levels will be traversed by the loader, thus speeding up the loading process.
Hence,
[10]:
data = osyris.RamsesDataset(8, path=path).load(
select={"mesh": {"level": lambda l: l < 7}}
)
data["mesh"]["level"]
Processing 12 files in osyrisdata/starformation/output_00008
16% : read 29402 cells, 0 particles
25% : read 29842 cells, 956 particles
33% : read 43887 cells, 956 particles
41% : read 48004 cells, 956 particles
50% : read 53124 cells, 2109 particles
66% : read 82526 cells, 2109 particles
75% : read 82966 cells, 3065 particles
83% : read 97011 cells, 3065 particles
91% : read 98600 cells, 4217 particles
Loaded: 106248 cells, 4218 particles.
[10]:
'level' Min: 5 Max: 6 [] (106248,)
will only read levels 1 to 6, while
[11]:
data = osyris.RamsesDataset(8, path=path).load(
select={"mesh": {"level": lambda l: np.logical_and(l > 5, l < 9)}}
)
data["mesh"]["level"]
Processing 12 files in osyrisdata/starformation/output_00008
16% : read 60265 cells, 0 particles
25% : read 76654 cells, 956 particles
33% : read 96290 cells, 956 particles
41% : read 121166 cells, 956 particles
50% : read 133758 cells, 2109 particles
66% : read 194015 cells, 2109 particles
75% : read 210412 cells, 3065 particles
83% : read 230904 cells, 3065 particles
91% : read 247060 cells, 4217 particles
Loaded: 267516 cells, 4218 particles.
[11]:
'level' Min: 6 Max: 8 [] (267516,)
will read levels 1 to 8, but will then discard all cells with level < 6.
Loading only selected CPU outputs#
It is also possible to feed a list of CPU numbers to read from.
[12]:
data = osyris.RamsesDataset(8, path=path).load(cpu_list=[1, 2, 10, 4, 5])
Processing 5 files in osyrisdata/starformation/output_00008
20% : read 32808 cells, 0 particles
40% : read 65623 cells, 0 particles
60% : read 93715 cells, 0 particles
80% : read 121807 cells, 0 particles
Loaded: 150675 cells, 0 particles.
[13]:
osyris.map(
data["mesh"].layer("density"),
dx=2000 * osyris.units("au"),
origin=data["mesh"]["position"][np.argmax(data["mesh"]["density"])],
direction="z",
norm="log",
)
[13]:
<osyris.core.plot.Plot at 0x71723ddc1270>
Note
When performing a selection (using select) on spatial position x, y, or z, and if the ordering of the cells in the AMR mesh is using the Hilbert curve, then a pre-selection is automatically made on the CPU files to load to speed-up the loading.
Example: loading a region around a sink particle#
Combining some of the methods illustrated above, we show here how to load only a small region 400 AU wide around a sink particle.
We begin by loading only the sink particle data, by using select=['sink'] in the load method:
[14]:
data = osyris.RamsesDataset(8, path=path).load(select=["sink"])
data
Loaded: 0 cells, 0 particles.
[14]:
Dataset: osyrisdata/starformation/output_00008: 128.00 B
Datagroup: sink 128.00 B
'id' Min: 1.0 Max: 2.0 [] (2,)
'msink' Min: 3.482e+32 Max: 3.485e+32 [g] (2,)
'position' Min: 5.061e+16 Max: 5.192e+16 [cm] (2,), {x,y,z}
'v' Min: 7.464e+04 Max: 7.577e+04 [cm / s] (2,), {x,y,z}
The sink data group contains the positions of the sink particles. We wish to load the region around the first sink particle, which will be the center of our domain.
[15]:
center = data["sink"]["position"][0]
dx = 200 * osyris.units("au")
We are now in a position to load the rest of the data (amr, hydro, etc.) and perform a spatial selection based on the sink’s position (note how the loading below is only looking through 6 files, as opposed to 12 at the top of the notebook, because it uses the knowledge from the Hilbert space-filling curve to skip CPUs that are not connected to the domain of interest).
[16]:
data.load(
select={
"mesh": {
"position_x": lambda x: (x > center.x - dx) & (x < center.x + dx),
"position_y": lambda y: (y > center.y - dx) & (y < center.y + dx),
"position_z": lambda z: (z > center.z - dx) & (z < center.z + dx),
}
}
)
Processing 6 files in osyrisdata/starformation/output_00008
16% : read 6816 cells, 0 particles
33% : read 20520 cells, 1153 particles
50% : read 20520 cells, 1153 particles
66% : read 32842 cells, 2109 particles
83% : read 40543 cells, 2109 particles
Loaded: 40543 cells, 2109 particles.
[16]:
Dataset: osyrisdata/starformation/output_00008: 8.60 MB
Datagroup: sink 128.00 B
'id' Min: 1.0 Max: 2.0 [] (2,)
'msink' Min: 3.482e+32 Max: 3.485e+32 [g] (2,)
'position' Min: 5.061e+16 Max: 5.192e+16 [cm] (2,), {x,y,z}
'v' Min: 7.464e+04 Max: 7.577e+04 [cm / s] (2,), {x,y,z}
Datagroup: mesh 8.43 MB
'level' Min: 8 Max: 9 [] (40543,)
'cpu' Min: 3 Max: 6 [] (40543,)
'dx' Min: 1.151e+14 Max: 2.303e+14 [cm] (40543,)
'density' Min: 4.702e-17 Max: 8.047e-12 [g / cm ** 3] (40543,)
'thermal_pressure' Min: 1.703e-08 Max: 0.057 [erg / cm ** 3] (40543,)
'internal_energy' Min: 2.555e-08 Max: 0.085 [erg / cm ** 3] (40543,)
'temperature' Min: 10.060 Max: 196.393 [K] (40543,)
'grav_potential' Min: -6.219e+12 Max: -5.543e+11 [cm ** 2 / s ** 2] (40543,)
'position' Min: 4.569e+16 Max: 5.561e+16 [cm] (40543,), {x,y,z}
'velocity' Min: 1.024e+04 Max: 4.216e+05 [cm / s] (40543,), {x,y,z}
'B_left' Min: 1.174e-04 Max: 0.162 [G] (40543,), {x,y,z}
'B_right' Min: 1.155e-04 Max: 0.162 [G] (40543,), {x,y,z}
'grav_acceleration' Min: 2.263e-06 Max: 2.028e-04 [cm / s ** 2] (40543,), {x,y,z}
'B_field' Min: 1.197e-04 Max: 0.157 [G] (40543,), {x,y,z}
'mass' Min: 6.851e-08 Max: 0.006 [M_sun] (40543,)
Datagroup: part 168.72 KB
'mass' Min: 1.652e+29 Max: 1.652e+29 [g] (2109,)
'identity' Min: -1.0 Max: -1.0 [] (2109,)
'levelp' Min: 5.0 Max: 5.0 [] (2109,)
'birth_time' Min: 8.888e-05 Max: 8.888e-05 [] (2109,)
'position' Min: 5.017e+16 Max: 5.105e+16 [cm] (2109,), {x,y,z}
'velocity' Min: 7.464e+04 Max: 7.464e+04 [cm / s] (2109,), {x,y,z}
Finally, we can make a map of the gas density around our sink particle (note that no spatial limits are specified in the map call, it is making a map using all the cells that have been loaded)
[17]:
osyris.map(
data["mesh"].layer("density"),
data["sink"].layer(
"position",
mode="scatter",
c="white",
s=20.0 * osyris.units("au"),
alpha=0.7,
),
norm="log",
direction="z",
origin=center,
)
[17]:
<osyris.core.plot.Plot at 0x71723e712d40>
