[Paraview] [EXTERNAL] Time Synchronization for Case and Geometry Sequences
Huangrui Mo
huangrui.mo at uwaterloo.ca
Fri Dec 2 10:22:28 EST 2016
Hi Ken,
Yes, it works properly now. I would follow your suggestion that using
dt/2 shift to fix this problem for future cases.
Thank you very much for your help.
Best wishes,
Huangrui
On 2016-12-01 07:40 PM, Moreland, Kenneth wrote:
> I may have told you the wrong parameters. Try using 0.005 for the temporal shift rather than -0.005.
>
> -Ken
>
> On 12/1/16, 3:10 PM, "Huangrui Mo" <huangrui.mo at uwaterloo.ca> wrote:
>
> Hello Ken,
>
> Thank you for the provided solution, however, it seems like the solution
> does not work well: using the Paraview 5.2.0
> (I just downloaded the latest version, and realizing that there is a
> modification on showing the max counter of time steps),
> after applying the method you provided, it artificially works for the
> first 9 counts, but when you input "10", then you would realize that
> the mismatch is still there :D
>
> I have found the following weired situation:
>
> 1) The time values in field.case, field.pvd, geo_stl.pvd are all exactly
> the same and are all in ASCII format.
> 2) When paraview read in the field.pvd or geo_stl.pvd, it will change
> the floating point values slightly, however, it maintains consistence
> if they are ".pvd" files. Therefore, it correctly synchronizes these two
> data sets.
> 3) When paraview read in the field.case, the floating point
> representation of the time values are changed slightly and are not
> consistently with ".pvd" files.
> However, in this mixed case, the "0" value is the same, so paraview
> merges field.case and geo_stl.pvd for time "0", and reduces the total
> number of cases
> from 11 + 11 to 21. (sorry for previous miscalculation of the counter,
> it's 11 rather than 10 for each data set.)
>
> Is it because when reading Ensight format, paraview choose single
> precision, but using double precision for .pvd file? Is there a way to
> fix this problem, since
> the time values are all in ASCII format.
>
> Thanks,
> Huangrui
>
> On 2016-12-01 03:35 PM, Moreland, Kenneth wrote:
> > Huangrui,
> >
> > I found a work around that I think does what you want. After you load in your two data sets, right click on one of them in the Pipeline Browser and check on the “Ignore Time” option. This will tell ParaView to ignore the time in that data set, so will only visit the time steps in the other data set.
> >
> > Actually, this _almost_ works. There is in fact a problem (at least with the example data sets you sent us). The imprecision is not uniform. Sometimes the time for field.case is ahead, other times the time for geo_stl.pvd is ahead. So, when you play through the time steps, sometimes the data for which you are ignoring time are skipped because the time is first just before one step and then just after another.
> >
> > To fix this problem, you can use the Temporal Shift Scale filter to adjust the time of one of the data sets to subtract a bit to ensure that every time hits in between time steps. Here is specifically what I did with the test data that you sent to fix the problem.
> >
> > 1. Load in the field.case and geo_stl.pvd files. Apply both.
> > 2. Add the Temporal Shift Scale filter to geo_stl.pvd. Set the Pre Shift to -0.005. Apply.
> > 3. Right click on TemporalShiftScale1 in the pipeline browser and select “Ignore Time”
> >
> > Now when you hit play you should get the right amount of time steps with the data you expect loaded.
> >
> > -Ken
> >
> > On 12/1/16, 1:58 PM, "Huangrui Mo" <huangrui.mo at uwaterloo.ca> wrote:
> >
> > Hello Alan and Ken,
> >
> > Thank you very much for your prompt response of my question.
> >
> > After testing your suggestions, I found both the Real Time Mode and
> > Sequence Mode can correctly synchronize the
> > two data sets with mixed formats, if a correct number of steps is given.
> > However, both of the above two methods
> > share a common subproblem: they give artificial time steps that are
> > different with the "true" time moments contained in
> > the case (or pvd) files, when a annotated time filter is used, even the
> > specified number of steps matches the true number of case files.
> >
> > Regarding the Snap to Timesteps mode and the numerical imprecision of
> > the different times problem raised by Ken, I provided an example
> > data set attached in this email. This data set contains the simulated
> > flow field of three airfoils running on a fixed Cartesian background grid.
> > Since the Cartesian grid is fixed, therefore, using Ensight format for
> > the field data has the advantage to avoid outputting the grid
> > coordinates in each
> > time step but only the physical quantities. To facilitate the testing, I
> > solved this problem exactly twice, the only difference is that one uses
> > Ensight for field data
> > and VTK for geometry, the other uses both VTK. Each data sets has 10
> > time sequences, and the time steps in each data sets are all the same.
> >
> > When loading the mixed data sets, the field.case + geo_stl.pvd gives 20
> > sequences, and the field.pvd + geo_stl.pvd gives 10 synchronized sequences.
> >
> > Due to the artificial times steps in the Real Time Mode and Sequence
> > Mod, is there a way to synchronize the mixed case in the Snap to
> > Timesteps mode?
> >
> > Thanks a lot,
> > Huangrui
> >
> >
> >
> > On 2016-12-01 01:54 PM, Moreland, Kenneth wrote:
> > > Actually, I think Sequence mode is more appropriate than Real Time mode in this case, but I too think that is the answer.
> > >
> > > To explain more what (we think) is going on: When you load data with time in ParaView, it goes to a Snap to Timesteps mode where it will visit each unique timestep once regardless of how far apart they are. Thus, ParaView is treating the numerical imprecision of the different times as unique time steps. If you switch to Sequence or Real Time mode, ParaView uses its own time units and will provide an even temporal spacing.
> > >
> > > There is more information about the different animation modes in the ParaView tutorial.
> > >
> > > -Ken
> > >
> > > Sent from my iPad so blame autocorrect.
> > >
> > >> On Dec 1, 2016, at 11:35 AM, Scott, W Alan <wascott at sandia.gov> wrote:
> > >>
> > >> Not sure if this is the correct answer, but try view/ animation View. Then, change the Mode to Real Time. Enter your start time and end time, and number of time steps of interest (i.e., 500).
> > >>
> > >> Is that what you are looking for?
> > >>
> > >> alan
> > >>
> > >> -----Original Message-----
> > >> From: ParaView [mailto:paraview-bounces at paraview.org] On Behalf Of Huangrui Mo
> > >> Sent: Thursday, December 1, 2016 9:54 AM
> > >> To: paraview at paraview.org
> > >> Subject: [EXTERNAL] [Paraview] Time Synchronization for Case and Geometry Sequences
> > >>
> > >> Dear Paraview Developer,
> > >>
> > >> May you please help me with the following issue:
> > >>
> > >> Suppose when a solver writes data out, the field data is written in Ensight format, and the time sequence is streamed in "ensight.case".
> > >> Meanwhile, the geometry data is written in VTK format, and the time sequence is streamed in "paraview.pvd".
> > >> At last, let's assume the number of time sequences is 500.
> > >>
> > >> Then, when loading into paraview, it does not synchronize the cases in a perfect sense and would show up with 1000 cases. When animating, there is a slight mismatch in time for the field data and geometry data.
> > >> However, if I use the VTK format for both the field and geometry data, then paraview does a great job to synchronize the data sets.
> > >>
> > >> Therefore, my question would be that is it possible to handle time synchronization even for different formats?
> > >>
> > >> Thank you very much for your time and help, Huangrui
> > >>
> > >> --
> > >> *****************************************************
> > >> Huangrui Mo, PhD Candidate
> > >> Fluid Mechanics
> > >> Department of Mechanical & Mechatronics Engineering University of Waterloo Waterloo, Ontario N2L 3G1, Canada
> > >> E-mail: huangrui.mo at uwaterloo.ca
> > >> *****************************************************
> > >>
> > >> _______________________________________________
> > >> Powered by www.kitware.com
> > >>
> > >> Visit other Kitware open-source projects at http://www.kitware.com/opensource/opensource.html
> > >>
> > >> Please keep messages on-topic and check the ParaView Wiki at: http://paraview.org/Wiki/ParaView
> > >>
> > >> Search the list archives at: http://markmail.org/search/?q=ParaView
> > >>
> > >> Follow this link to subscribe/unsubscribe:
> > >> http://public.kitware.com/mailman/listinfo/paraview
> > >> _______________________________________________
> > >> Powered by www.kitware.com
> > >>
> > >> Visit other Kitware open-source projects at http://www.kitware.com/opensource/opensource.html
> > >>
> > >> Please keep messages on-topic and check the ParaView Wiki at: http://paraview.org/Wiki/ParaView
> > >>
> > >> Search the list archives at: http://markmail.org/search/?q=ParaView
> > >>
> > >> Follow this link to subscribe/unsubscribe:
> > >> http://public.kitware.com/mailman/listinfo/paraview
> >
> > --
> > *****************************************************
> > Huangrui Mo, PhD Candidate
> > Fluid Mechanics
> > Department of Mechanical & Mechatronics Engineering
> > University of Waterloo
> > Waterloo, Ontario N2L 3G1, Canada
> > E-mail: huangrui.mo at uwaterloo.ca
> > *****************************************************
> >
> >
> >
>
> --
> *****************************************************
> Huangrui Mo, PhD Candidate
> Fluid Mechanics
> Department of Mechanical & Mechatronics Engineering
> University of Waterloo
> Waterloo, Ontario N2L 3G1, Canada
> E-mail: huangrui.mo at uwaterloo.ca
> *****************************************************
>
>
>
--
*****************************************************
Huangrui Mo, PhD Candidate
Fluid Mechanics
Department of Mechanical & Mechatronics Engineering
University of Waterloo
Waterloo, Ontario N2L 3G1, Canada
E-mail: huangrui.mo at uwaterloo.ca
*****************************************************
More information about the ParaView
mailing list