Hi Michael, Unfortunately the rendering framework currently relies upon being able to create a smoothed covering grid to calculate vertex centered data, which has not been implemented for non-periodic boundary conditions. If the quality of the rendering using no_ghost=True is not good enough, then perhaps try creating a region that is the entire volume except the outer cell: root_dds = pf.domain_width/pf.domain_dimensions half_w = pf.domain_width/2. - root_dds reg = pf.region(pf.domain_center, pf.domain_center-half_w, pf.domain_center+half_w) then supply that reg to the camera as data_source=reg in the constructor. This requires being on yt-3.0, which I think you are. I *think* that should allow the no_ghost=False to get the vertex centered data using a smoothed covering grid the size of the full domain (but not beyond). The longer term fix is to use meaningful boundary conditions to calculate covering grids that extend beyond the domain. I'm not sure what the roadmap for that is... Best, Sam On Mon, Apr 14, 2014 at 7:05 PM, Nathan Goldbaum wrote:

On Monday, April 14, 2014, Michael Zingale wrote:

...

Recently I made the Maestro BoxLib frontend set the correct boundary periodicity for a data set, and I can confirm this when I load the data in yt and look at periodicity. However, when I try to volume render, with no_ghost=False, I get an error in the non-periodic direction (we are periodic in x and y, and non-periodic in z).

The error is:

Error: bad Region in non-periodic domain: -24.0 code_length 0.0 code_length 408.0 code_length 4608.0 code_length

for reference, the domain size is:

Parameters: domain_left_edge = [ 0. 0. 0.] Parameters: domain_right_edge = [ 1536. 1536. 4608.]

(aside, these should be cm, but the error reports them as code_length -- should I be setting that somwhere in the frontend? we are always CGS).

I can't speak to the volume rendering issue but this is the as-designed behavior. This has to be code_length because other codes do not have CGS as the internal unit system and this code has to work for all frontends, irrespective of the internal unit system.

...

Any thoughts on what I may be missed setting up for non-periodic BCs is appreciated.

Mike

-- Michael Zingale Associate Professor

Dept. of Physics & Astronomy • Stony Brook University • Stony Brook, NY 11794-3800 *phone*: 631-632-8225 *e-mail*: Michael.Zingale@stonybrook.edu *web*: http://www.astro.sunysb.edu/mzingale

_______________________________________________ yt-dev mailing list yt-dev@lists.spacepope.org http://lists.spacepope.org/listinfo.cgi/yt-dev-spacepope.org

Thanks Nathan, I'll look at that YTEP. Unfortunately the region fix you suggested did not work -- the error was the same regardless of the region (even clipping off more than one zone didn't work). I'll take a deeper look at this after I teach this morning. As a stopgap I may decide to go back to assuming we are triply periodic, even though that is rarely the case in our simulations. On Tue, Apr 15, 2014 at 12:48 AM, Nathan Goldbaum wrote:

There is an open YTEP about this issue: http://ytep.readthedocs.org/en/latest/YTEPs/YTEP-0006.html

On Mon, Apr 14, 2014 at 9:38 PM, Sam Skillman wrote:

...

Hi Michael,

Unfortunately the rendering framework currently relies upon being able to create a smoothed covering grid to calculate vertex centered data, which has not been implemented for non-periodic boundary conditions. If the quality of the rendering using no_ghost=True is not good enough, then perhaps try creating a region that is the entire volume except the outer cell:

root_dds = pf.domain_width/pf.domain_dimensions half_w = pf.domain_width/2. - root_dds reg = pf.region(pf.domain_center, pf.domain_center-half_w, pf.domain_center+half_w)

then supply that reg to the camera as data_source=reg in the constructor. This requires being on yt-3.0, which I think you are.

I *think* that should allow the no_ghost=False to get the vertex centered data using a smoothed covering grid the size of the full domain (but not beyond).

The longer term fix is to use meaningful boundary conditions to calculate covering grids that extend beyond the domain. I'm not sure what the roadmap for that is...

Best, Sam

On Mon, Apr 14, 2014 at 7:05 PM, Nathan Goldbaum wrote:

...

On Monday, April 14, 2014, Michael Zingale < michael.zingale@stonybrook.edu> wrote:

...

Recently I made the Maestro BoxLib frontend set the correct boundary periodicity for a data set, and I can confirm this when I load the data in yt and look at periodicity. However, when I try to volume render, with no_ghost=False, I get an error in the non-periodic direction (we are periodic in x and y, and non-periodic in z).

The error is:

Error: bad Region in non-periodic domain: -24.0 code_length 0.0 code_length 408.0 code_length 4608.0 code_length

for reference, the domain size is:

Parameters: domain_left_edge = [ 0. 0. 0.] Parameters: domain_right_edge = [ 1536. 1536. 4608.]

(aside, these should be cm, but the error reports them as code_length -- should I be setting that somwhere in the frontend? we are always CGS).

I can't speak to the volume rendering issue but this is the as-designed behavior. This has to be code_length because other codes do not have CGS as the internal unit system and this code has to work for all frontends, irrespective of the internal unit system.

...

Any thoughts on what I may be missed setting up for non-periodic BCs is appreciated.

Mike

-- Michael Zingale Associate Professor

Dept. of Physics & Astronomy * Stony Brook University * Stony Brook, NY 11794-3800 *phone*: 631-632-8225 *e-mail*: Michael.Zingale@stonybrook.edu *web*: http://www.astro.sunysb.edu/mzingale

_______________________________________________ yt-dev mailing list yt-dev@lists.spacepope.org http://lists.spacepope.org/listinfo.cgi/yt-dev-spacepope.org

_______________________________________________ yt-dev mailing list yt-dev@lists.spacepope.org http://lists.spacepope.org/listinfo.cgi/yt-dev-spacepope.org

_______________________________________________ yt-dev mailing list yt-dev@lists.spacepope.org http://lists.spacepope.org/listinfo.cgi/yt-dev-spacepope.org

-- Michael Zingale Associate Professor Dept. of Physics & Astronomy * Stony Brook University * Stony Brook, NY 11794-3800 *phone*: 631-632-8225 *e-mail*: Michael.Zingale@stonybrook.edu *web*: http://www.astro.sunysb.edu/mzingale