Issue with visualization

[ShokraniEZ]

To whom it may concern,

I have installed your package. I ran this example with periodic boundary condition: [FixedDynamicLinkNetwork.py].(https://github.com/stochasticHydroTools/SlenderBody/blob/master/Python/Examples/FixedDynamicLinkNetwork.py)
Then I ran the example [StrainDynamicNetwork.py]:
https://github.com/stochasticHydroTools/SlenderBody/blob/master/Python/Examples/StrainDynamicNetwork.py

Now I have some results. However, when I tried to visualize the results, your py file did not work ( I know that it has been written to plot with plotoptix).
https://github.com/stochasticHydroTools/SlenderBody/blob/master/Python/plots/BundledFibers.py
In fact, I modified that file somehow to dump the output files in a way that can be used for visualisation with any other packages. However, it does not look good. I would be wondering if you could kindly guide me on this issue.

Yours sincerely
Ezat

New_File_6.py

[omaxian]

Hi Ezat,

Do you want to visualize the filaments, or the cross linkers/motors as well? For the filaments, what I have been doing with PlotOptix is to read in a file with the filament positions (Chebyshev pts), and then construct a matrix that will interpolate those to 1/epsilon uniformly-spaced points. Specifically, the code will output a file 'Locs'+str(FileName) which is a matrix of dimension FNxNt x 3, where F is the number of filaments, Nx is the number of Chebyshev points, and Nt is the number of time steps. From that file, you want to separate the locations at each time step (into an F*Nx x 3 matrix), and then the loop on line 70 here computes the uniformly-spaced points for you. You can then mod those to be in the periodic box, and it should look good.

If you have that working, then I can help you figure out the cross linkers if you need them.

Thanks,

Ondrej Maxian

[ShokraniEZ]

Dear Professor Ondrej Maxian,

I hope you are doing well.
Thanks for your reply and the info which were quite useful.
I want to visualize both fibers and cross linkers/motors with any other packages like MATLAB, Paraview, Tecplot (I prefer to use this one) etc. To do this, I need the data to be structured for both fibers and cross linkers/motors at each time instant separately (which I have implemented in my python script: 1 file (Fibers.txt) for all fibers and Nt files (Step_0.txt, Step_1.txt, Step_2.txt, …, Step_ Nt.txt) for all cross linkers/motors at all time instants). I do not want to structure the data out of your package (I just want to dump the structed data and then simply import that data in any visualizer (like Tecplot) and just plot them without any further postprocessing), which I have done by use of this python script.
Initially, I have plotted them using your own code and it look like this:
(Please look at images 1 and 2)
It seems that some points of some fibers are stretched across the domain artificially.
Then I modified your python code in way that (As you can see) to be plotted within the domain when no fibers is stretched across the domain artificially and all of them are inside the domain as well when Brownian motion is also activated, which look like this:
(Please look at images 3 and 4)

However, there are some faults in visualization. Some crosslinkers are within the domain without any connection to any fiber (since the fibers have been shifted but not with own crosslinker) and also some of them are stretched artificially across the domain like yours.

I would be wondering if you could kindly help me solve this issue.

Best regards,
Ezat

[omaxian]

Hi Ezat,

If I understand your issue correctly, you have figured out how to plot the fibers, but the cross linkers are the issue. Here is what I recommend: for the cross linkers, instead of writing the coordinates on the fiber to a file, you can write the coordinates in three dimensional space. You can do this by changing the method writeLinks (e.g., here, line 152), to the method writeLinkLocations (which is written down here, line 606). This will give you the coordinates of the head and tail of each link. You can then treat those as you would any other filament, shifting it appropriately to be inside the domain.

By the way, my approach to visualization is to just plot everything as a bunch of spheres (instead of fibers/Bezier curves). This avoids headaches when you have to periodic shifts, because you can just shift the individual sphere locations and keep everything inside the box. If you choose to take a fiber-centric view, then you need to worry about shifting the entire fiber appropriately (you will have to make multiple copies, and some will be falling out of the box, as in the images you shared with me).

Hope this is helpful.

Ondrej Maxian