problem with disk data container and related fields in yt-3.0
Hi all, Some issues seem to have cropped up in the disk data container in yt-3.0. I made a disk object in a periodic simulation in the following way: http://paste.yt-project.org/show/5021/ I then made a project of the "cylindrical_z" field, which yielded the following: http://i.imgur.com/0XSv5Zs.png For reference, the domain is ~45 Mpc in length on a side, so the sphere should be just the object at the bottom, extending a little less than halfway up. I don't know what that top part is about. My guess is that it's an issue with periodicity. Secondly, I made a similar projection of the "cylindrical_r" field, which gave this: http://i.imgur.com/h8SEJp8.png As you can see, it is not symmetric as one would expect. For reference, this is what I get from yt-2.x: http://i.imgur.com/O1xT9MI.png This is what I would expect. Does anyone have any insight on this? Britton
Hi Britton,
On Tue, Aug 12, 2014 at 7:30 AM, Britton Smith
Hi all,
Some issues seem to have cropped up in the disk data container in yt-3.0. I made a disk object in a periodic simulation in the following way: http://paste.yt-project.org/show/5021/
I then made a project of the "cylindrical_z" field, which yielded the following: http://i.imgur.com/0XSv5Zs.png
For reference, the domain is ~45 Mpc in length on a side, so the sphere should be just the object at the bottom, extending a little less than halfway up. I don't know what that top part is about. My guess is that it's an issue with periodicity.
Yes, looks like it to me.
Secondly, I made a similar projection of the "cylindrical_r" field, which gave this: http://i.imgur.com/h8SEJp8.png
As you can see, it is not symmetric as one would expect. For reference, this is what I get from yt-2.x: http://i.imgur.com/O1xT9MI.png This is what I would expect.
Are you doing a MIP projection here? Could you do a MIP projection so we can compare? It almost looks to me like the center is set incorrectly.
Does anyone have any insight on this?
Britton
_______________________________________________ yt-dev mailing list yt-dev@lists.spacepope.org http://lists.spacepope.org/listinfo.cgi/yt-dev-spacepope.org
Hi Matt, Britton,
Given a periodic box, a center at the bottom edge of the box, and a height that is nearly half the box height, I would expect to see some cells appearing at the top of your plot. That may have been a change from 2.X (since YTDisk accounts for periodicity), but I'm not sure it's incorrect behavior.
There is definitely something strange going on that eats away that portion of the lower part of the cylinder, which must be something to do with the way the code computes the periodic radial distance, then subtracts the portion projected onto the cylinder normal to obtain the cylindrical radius.
Douglas Rudd
Scientific Computing Consultant
Research Computing Center
drudd@uchicago.edu
On Aug 12, 2014, at 7:32 AM, Matthew Turk
Hi Britton,
On Tue, Aug 12, 2014 at 7:30 AM, Britton Smith
wrote: Hi all,
Some issues seem to have cropped up in the disk data container in yt-3.0. I made a disk object in a periodic simulation in the following way: http://paste.yt-project.org/show/5021/
I then made a project of the "cylindrical_z" field, which yielded the following: http://i.imgur.com/0XSv5Zs.png
For reference, the domain is ~45 Mpc in length on a side, so the sphere should be just the object at the bottom, extending a little less than halfway up. I don't know what that top part is about. My guess is that it's an issue with periodicity.
Yes, looks like it to me.
Secondly, I made a similar projection of the "cylindrical_r" field, which gave this: http://i.imgur.com/h8SEJp8.png
As you can see, it is not symmetric as one would expect. For reference, this is what I get from yt-2.x: http://i.imgur.com/O1xT9MI.png This is what I would expect.
Are you doing a MIP projection here? Could you do a MIP projection so we can compare?
It almost looks to me like the center is set incorrectly.
Does anyone have any insight on this?
Britton
_______________________________________________ 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
Hi Doug,
On Tue, Aug 12, 2014 at 7:13 PM, Douglas Harvey Rudd
Hi Matt, Britton,
Given a periodic box, a center at the bottom edge of the box, and a height that is nearly half the box height, I would expect to see some cells appearing at the top of your plot. That may have been a change from 2.X (since YTDisk accounts for periodicity), but I'm not sure it's incorrect behavior.
I looked into it, and I believe that this particular behavior -- selecting on to the top -- *is* correct now in yt 3.0. I believe that, as you noted in the PR, the change I made *only* allowed all grids to be selected, and the other change was irrelevant.
There is definitely something strange going on that eats away that portion of the lower part of the cylinder, which must be something to do with the way the code computes the periodic radial distance, then subtracts the portion projected onto the cylinder normal to obtain the cylindrical radius.
My understanding from looking into it is that the only problem with the disk bbox selection is in selecting bboxes that have no corners within the cylinder. All cell/point selection is correct, I believe.
Douglas Rudd Scientific Computing Consultant Research Computing Center drudd@uchicago.edu
On Aug 12, 2014, at 7:32 AM, Matthew Turk
wrote: Hi Britton,
On Tue, Aug 12, 2014 at 7:30 AM, Britton Smith
wrote: Hi all,
Some issues seem to have cropped up in the disk data container in yt-3.0. I made a disk object in a periodic simulation in the following way: http://paste.yt-project.org/show/5021/
I then made a project of the "cylindrical_z" field, which yielded the following: http://i.imgur.com/0XSv5Zs.png
For reference, the domain is ~45 Mpc in length on a side, so the sphere should be just the object at the bottom, extending a little less than halfway up. I don't know what that top part is about. My guess is that it's an issue with periodicity.
Yes, looks like it to me.
Secondly, I made a similar projection of the "cylindrical_r" field, which gave this: http://i.imgur.com/h8SEJp8.png
As you can see, it is not symmetric as one would expect. For reference, this is what I get from yt-2.x: http://i.imgur.com/O1xT9MI.png This is what I would expect.
Are you doing a MIP projection here? Could you do a MIP projection so we can compare?
It almost looks to me like the center is set incorrectly.
Does anyone have any insight on this?
Britton
_______________________________________________ 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
There is definitely something strange going on that eats away that portion of the lower part of the cylinder, which must be something to do with the way the code computes the periodic radial distance, then subtracts the portion projected onto the cylinder normal to obtain the cylindrical radius.
My understanding from looking into it is that the only problem with the disk bbox selection is in selecting bboxes that have no corners within the cylinder. All cell/point selection is correct, I believe.
I'm not sure how that's possible, given the shape of the carved out region. That appears to have a refined region, so most of those grids should have all 8 corners inside the cylinder. I've reproduced the problem and will probably throw an hour or two at it today to see if I can find a simple fix. Either way, I agree with Britton that the no-op bbox selector should be swapped in immediately. Doug
On Aug 12, 2014, at 7:32 AM, Matthew Turk
wrote: Hi Britton,
On Tue, Aug 12, 2014 at 7:30 AM, Britton Smith
wrote: Hi all,
Some issues seem to have cropped up in the disk data container in yt-3.0. I made a disk object in a periodic simulation in the following way: http://paste.yt-project.org/show/5021/
I then made a project of the "cylindrical_z" field, which yielded the following: http://i.imgur.com/0XSv5Zs.png
For reference, the domain is ~45 Mpc in length on a side, so the sphere should be just the object at the bottom, extending a little less than halfway up. I don't know what that top part is about. My guess is that it's an issue with periodicity.
Yes, looks like it to me.
Secondly, I made a similar projection of the "cylindrical_r" field, which gave this: http://i.imgur.com/h8SEJp8.png
As you can see, it is not symmetric as one would expect. For reference, this is what I get from yt-2.x: http://i.imgur.com/O1xT9MI.png This is what I would expect.
Are you doing a MIP projection here? Could you do a MIP projection so we can compare?
It almost looks to me like the center is set incorrectly.
Does anyone have any insight on this?
Britton
_______________________________________________ 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
_______________________________________________ yt-dev mailing list yt-dev@lists.spacepope.org http://lists.spacepope.org/listinfo.cgi/yt-dev-spacepope.org
On Wed, Aug 13, 2014 at 10:32 AM, Douglas Harvey Rudd
There is definitely something strange going on that eats away that portion of the lower part of the cylinder, which must be something to do with the way the code computes the periodic radial distance, then subtracts the portion projected onto the cylinder normal to obtain the cylindrical radius.
My understanding from looking into it is that the only problem with the disk bbox selection is in selecting bboxes that have no corners within the cylinder. All cell/point selection is correct, I believe.
I'm not sure how that's possible, given the shape of the carved out region. That appears to have a refined region, so most of those grids should have all 8 corners inside the cylinder. I've reproduced the problem and will probably throw an hour or two at it today to see if I can find a simple fix. Either way, I agree with Britton that the no-op bbox selector should be swapped in immediately.
Thank you, Doug -- I really appreciate it.
Doug
On Aug 12, 2014, at 7:32 AM, Matthew Turk
wrote: Hi Britton,
On Tue, Aug 12, 2014 at 7:30 AM, Britton Smith
wrote: Hi all,
Some issues seem to have cropped up in the disk data container in yt-3.0. I made a disk object in a periodic simulation in the following way: http://paste.yt-project.org/show/5021/
I then made a project of the "cylindrical_z" field, which yielded the following: http://i.imgur.com/0XSv5Zs.png
For reference, the domain is ~45 Mpc in length on a side, so the sphere should be just the object at the bottom, extending a little less than halfway up. I don't know what that top part is about. My guess is that it's an issue with periodicity.
Yes, looks like it to me.
Secondly, I made a similar projection of the "cylindrical_r" field, which gave this: http://i.imgur.com/h8SEJp8.png
As you can see, it is not symmetric as one would expect. For reference, this is what I get from yt-2.x: http://i.imgur.com/O1xT9MI.png This is what I would expect.
Are you doing a MIP projection here? Could you do a MIP projection so we can compare?
It almost looks to me like the center is set incorrectly.
Does anyone have any insight on this?
Britton
_______________________________________________ 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
_______________________________________________ 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
participants (3)
-
Britton Smith
-
Douglas Harvey Rudd
-
Matthew Turk