search

openPMD / openPMD-example-datasets

:gift: HDF5 Example Files
Creative Commons Zero v1.0 Universal
8 stars 4 forks source link

Modify the shape of the constant datasets #9

Closed RemiLehe closed 8 years ago

RemiLehe commented 8 years ago

This shape used to be 1 instead of the actual number of particles.

@C0nsultant : Could you download this branch and check that the data works with your yt implementation (without manually changing the shape of the constant particle attributes). If it does, please click merge (assuming you have the proper permissions, otherwise let me know.)

C0nsultant commented 8 years ago

For the moment I can only verify the 3D dataset (dimensionality in the 2D case does not fully work). If the shape in 2D (and I assume theta) is constructed the same way, this fixes the problem.

data00000500 h5_2d-profile_particle_position_relative_x_particle_position_relative_y_particle_weighting

ax3l commented 8 years ago

@C0nsultant btw, the image above looks quite wrong, we can take a look on Friday.

ax3l commented 8 years ago

@RemiLehe actually the picture is correct and your new 3D data set looks broken in the particle positions. @C0nsultant will add an openPMD-viewer image in a few minutes.

RemiLehe commented 8 years ago

Ok. Should we have a phone conversation about it (see personal emails) ?

Btw, please remember that here (in the Warp simulation), the particle positions were initialized in a regularly-spaced manner (i.e. not randomly distributed). This is similar to what is done in Osiris for instance. Does that explain the features that seemed broken in the above picture?

C0nsultant commented 8 years ago

For completenes, here is the plot (x/y on iteration 500) done with openPMD-viewer and a manual histogram from pyplot. opmd-viewer histo

RemiLehe commented 8 years ago

Both look okay to me (bearing in mind the above remark on the distribution of particles in Warp). What feature seemed broken to you @ax3l ?