Surface extraction in FLASH
Hi all, I'm trying to calculate the total amount of photons crossing a spherical surface using FLASH simulation data. I noticed yt has a surface extraction method, and I used this to extract the spherical surface. I only have the flux of photons in each cell and therefore need to mulitply by the cross-sectional area of each cell to get what I need. I have a couple questions about this surface extraction method: - What exactly does it return? - Can I get the cross sectional area of each cell on the surface (as seen from the center of the simulation volume) - If not, is there a better way to do this using yt? Cheers and thanks for the help, Corey
On Mon, Feb 27, 2017 at 12:34 PM, C.S. Howard <howardcs@mcmaster.ca> wrote:
Hi all,
I'm trying to calculate the total amount of photons crossing a spherical surface using FLASH simulation data. I noticed yt has a surface extraction method, and I used this to extract the spherical surface. I only have the flux of photons in each cell and therefore need to mulitply by the cross-sectional area of each cell to get what I need.
I have a couple questions about this surface extraction method: - What exactly does it return?
It's not clear what you're referring to in this question. What does what return?
- Can I get the cross sectional area of each cell on the surface (as seen from the center of the simulation volume)
Sure: ad = ds.all_data() # create surface at an isodensity of 5x10^-27 g/cm^3 surf = ds.surface(ad, "density", 5e-27) # sample the "density" field at the isodensity surface surf['density'] # The locations of all of the vertices of the surface surf.vertices # Coordinates of the vertices of the triangles that make up the surface surf.triangles For the dataset I'm looking at to help write this e-mail, the surf.vertices.shape is (3, 295197) (i.e. one 3D coordinate for every vertex) and surf.triangles.shape is (98399, 3, 3) (i.e. for each triangle [the first axis] there are three vertices [the second axis], each with three coordinates [the last axis]. Note also that there are exactly a third as many triangles as vertices.
- If not, is there a better way to do this using yt?
yt can calculate the flux of a vector field across a surface: http://yt-project.org/docs/dev/reference/api/generated/yt.data_objects.data_... This method is defined on any 3D data object. If you already have a surface object, there's also the calculate_flux method: http://yt-project.org/docs/dev/reference/api/generated/yt.data_objects.const...
Cheers and thanks for the help, Corey
_______________________________________________ yt-users mailing list yt-users@lists.spacepope.org http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org
Hi Nathan, Thanks for the reply! As I mentioned earlier, I'm not interested in the total flux of photons across the surface, but the total number of photons crossing it. The flux in each cell can vary and is not spherically symmetric so I cannot find the total flux and multiply by 4*pi*r^2, for example. This means that, for each cell, I need to take the dot product between the flux and the area vector of the small bit of the surface in that cell (to get the outwards flux), and multiply by the area of the surface in that cell. This is all fine, as long as I know the shape of the surface within each cell. I think I'm just confused what the returned triangles actually are. Looking at the info for the marching cubes algorithm didn't clear things up because it seems there can be multiply triangles within one cell depending on the shape of the surface. Thanks again, Corey On Mon, Feb 27, 2017 at 1:44 PM, Nathan Goldbaum <nathan12343@gmail.com> wrote:
On Mon, Feb 27, 2017 at 12:34 PM, C.S. Howard <howardcs@mcmaster.ca> wrote:
Hi all,
I'm trying to calculate the total amount of photons crossing a spherical surface using FLASH simulation data. I noticed yt has a surface extraction method, and I used this to extract the spherical surface. I only have the flux of photons in each cell and therefore need to mulitply by the cross-sectional area of each cell to get what I need.
I have a couple questions about this surface extraction method: - What exactly does it return?
It's not clear what you're referring to in this question. What does what return?
- Can I get the cross sectional area of each cell on the surface (as seen from the center of the simulation volume)
Sure:
ad = ds.all_data()
# create surface at an isodensity of 5x10^-27 g/cm^3 surf = ds.surface(ad, "density", 5e-27)
# sample the "density" field at the isodensity surface surf['density']
# The locations of all of the vertices of the surface surf.vertices
# Coordinates of the vertices of the triangles that make up the surface surf.triangles
For the dataset I'm looking at to help write this e-mail, the surf.vertices.shape is (3, 295197) (i.e. one 3D coordinate for every vertex) and surf.triangles.shape is (98399, 3, 3) (i.e. for each triangle [the first axis] there are three vertices [the second axis], each with three coordinates [the last axis]. Note also that there are exactly a third as many triangles as vertices.
- If not, is there a better way to do this using yt?
yt can calculate the flux of a vector field across a surface:
http://yt-project.org/docs/dev/reference/api/generated/ yt.data_objects.data_containers.YTSelectionContainer3D. calculate_isocontour_flux.html#yt.data_objects.data_containers. YTSelectionContainer3D.calculate_isocontour_flux
This method is defined on any 3D data object. If you already have a surface object, there's also the calculate_flux method:
http://yt-project.org/docs/dev/reference/api/generated/ yt.data_objects.construction_data_containers.YTSurface. calculate_flux.html#yt.data_objects.construction_data_ containers.YTSurface.calculate_flux
Cheers and thanks for the help, Corey
_______________________________________________ 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
On Mon, Feb 27, 2017 at 3:37 PM, C.S. Howard <howardcs@mcmaster.ca> wrote:
Hi Nathan,
Thanks for the reply! As I mentioned earlier, I'm not interested in the total flux of photons across the surface, but the total number of photons crossing it. The flux in each cell can vary and is not spherically symmetric so I cannot find the total flux and multiply by 4*pi*r^2, for example.
This means that, for each cell, I need to take the dot product between the flux and the area vector of the small bit of the surface in that cell (to get the outwards flux), and multiply by the area of the surface in that cell. This is all fine, as long as I know the shape of the surface within each cell.
I think I'm just confused what the returned triangles actually are. Looking at the info for the marching cubes algorithm didn't clear things up because it seems there can be multiply triangles within one cell depending on the shape of the surface.
The union of all of the triangles constitutes the surface. With the worked example I was using earlier: In [4]: surf['density'].shape Out[4]: (98399,) In [5]: surf.triangles.shape Out[5]: (98399, 3, 3) Note how the shape of surf['density'] is the same as the number of triangles, surf['density'] is just the density value sampled "at" each of the triangles. Does that help?
Thanks again, Corey
On Mon, Feb 27, 2017 at 1:44 PM, Nathan Goldbaum <nathan12343@gmail.com> wrote:
On Mon, Feb 27, 2017 at 12:34 PM, C.S. Howard <howardcs@mcmaster.ca> wrote:
Hi all,
I'm trying to calculate the total amount of photons crossing a spherical surface using FLASH simulation data. I noticed yt has a surface extraction method, and I used this to extract the spherical surface. I only have the flux of photons in each cell and therefore need to mulitply by the cross-sectional area of each cell to get what I need.
I have a couple questions about this surface extraction method: - What exactly does it return?
It's not clear what you're referring to in this question. What does what return?
- Can I get the cross sectional area of each cell on the surface (as seen from the center of the simulation volume)
Sure:
ad = ds.all_data()
# create surface at an isodensity of 5x10^-27 g/cm^3 surf = ds.surface(ad, "density", 5e-27)
# sample the "density" field at the isodensity surface surf['density']
# The locations of all of the vertices of the surface surf.vertices
# Coordinates of the vertices of the triangles that make up the surface surf.triangles
For the dataset I'm looking at to help write this e-mail, the surf.vertices.shape is (3, 295197) (i.e. one 3D coordinate for every vertex) and surf.triangles.shape is (98399, 3, 3) (i.e. for each triangle [the first axis] there are three vertices [the second axis], each with three coordinates [the last axis]. Note also that there are exactly a third as many triangles as vertices.
- If not, is there a better way to do this using yt?
yt can calculate the flux of a vector field across a surface:
http://yt-project.org/docs/dev/reference/api/generated/yt. data_objects.data_containers.YTSelectionContainer3D.calculat e_isocontour_flux.html#yt.data_objects.data_containers.Y TSelectionContainer3D.calculate_isocontour_flux
This method is defined on any 3D data object. If you already have a surface object, there's also the calculate_flux method:
http://yt-project.org/docs/dev/reference/api/generated/yt. data_objects.construction_data_containers.YTSurface.calculat e_flux.html#yt.data_objects.construction_data_containers. YTSurface.calculate_flux
Cheers and thanks for the help, Corey
_______________________________________________ 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
_______________________________________________ yt-users mailing list yt-users@lists.spacepope.org http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org
participants (2)
-
C.S. Howard -
Nathan Goldbaum