search

topology-tool-kit / ttk

TTK - Topological Data Analysis and Visualization - Source Code
https://topology-tool-kit.github.io/
Other
413 stars 122 forks source link

vertex indices missing after simplification #130

Closed kamakshidasan closed 6 years ago

kamakshidasan commented 6 years ago

hi,

I'm currently using TTK's master branch for running experiments. I perform the same tasks as issue #126

There, you can verify that for all the nodes of the split tree, there exists a unique corresponding point in the diagram such that:

  • the PointData field "VertexId" of the split tree equals the PointData field "VertexIdentifier" in the diagram

I cannot verify the above statement. There are vertex indices that are present in the contour tree as nodes but not present in the thresholded persistent diagram. I'm attaching a screenshot and the state file.

missing-issue

bug.pvsm.zip

Cheers,

Adhitya

julien-tierny commented 6 years ago

hi, can you please provide the file "tv_1.vtk" mentioned in your state file. thanks -- Dr Julien Tierny CNRS Researcher Sorbonne Universite http://lip6.fr/Julien.Tierny

On Saturday, July 7, 2018 8:30:14 PM CEST Kamakshidasan wrote:

hi,

I'm currently using TTK's master branch for running experiments. I perform the same tasks as issue #126

There, you can verify that for all the nodes of the split tree, there exists a unique corresponding point in the diagram such that:

  • the PointData field "VertexId" of the split tree equals the PointData field "VertexIdentifier" in the diagram

I cannot verify the above statement. There are vertex indices that are present in the contour tree as nodes but not present in the thresholded persistent diagram. I'm attaching a screenshot and the state file.

<img width="1440" alt="missing-issue" src="https://user-images.githubusercontent.com/9362064/42413618-4a4710c8-82 24-11e8-8b8e-3ca196d7e265.png">

bug.pvsm.zip

Cheers,

Adhitya

kamakshidasan commented 6 years ago

Sorry about that. Here you go.

tv_1.vtk.zip

julien-tierny commented 6 years ago

hi, there's a mismatch indeed for a few of them. the mismatches seem to involve boundary saddles which are not constrained by the algorithm. a work around would consist in computing a second persistence diagram after the topological simplification, there you should have the fixed matching by construction (which seems to do the trick here). thanks for letting us know if that fixes your issue. best, -- Dr Julien Tierny CNRS Researcher Sorbonne Universite http://lip6.fr/Julien.Tierny

On Saturday, July 7, 2018 8:51:54 PM CEST Kamakshidasan wrote:

Sorry about that. Here you go.

tv_1.vtk.zip

kamakshidasan commented 6 years ago

there's a mismatch indeed for a few of them.

the mismatches seem to involve boundary saddles which are not constrained by

the algorithm.

Thanks for acknowledging this. Will this be fixed in the future?

a work around would consist in computing a second persistence

diagram after the topological simplification, there you should have the fixed matching by construction (which seems to do the trick here).

This is exactly the work around I use as well!

On Sat, Jul 7, 2018 at 9:23 PM, Julien Tierny notifications@github.com wrote:

hi, there's a mismatch indeed for a few of them. the mismatches seem to involve boundary saddles which are not constrained by the algorithm. a work around would consist in computing a second persistence diagram after the topological simplification, there you should have the fixed matching by construction (which seems to do the trick here). thanks for letting us know if that fixes your issue. best,

Dr Julien Tierny CNRS Researcher Sorbonne Universite http://lip6.fr/Julien.Tierny

On Saturday, July 7, 2018 8:51:54 PM CEST Kamakshidasan wrote:

Sorry about that. Here you go.

tv_1.vtk.zip

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/topology-tool-kit/ttk/issues/130#issuecomment-403237717, or mute the thread https://github.com/notifications/unsubscribe-auth/AI7akLRqcCouynIrsVyON14JU9q4RKPkks5uEQqmgaJpZM4VGc3C .

julien-tierny commented 6 years ago

Will this be fixed in the future? if you come up with an algorithm that enforces constraints on boundary saddles and you're willing to share its implementation, then probably :)

This is exactly the work around I use as well! good. I'll go ahead and close this one then.

best, -- Dr Julien Tierny CNRS Researcher Sorbonne Universite http://lip6.fr/Julien.Tierny

On Saturday, July 7, 2018 9:26:58 PM CEST Kamakshidasan wrote:

there's a mismatch indeed for a few of them.

the mismatches seem to involve boundary saddles which are not constrained by

the algorithm.

Thanks for acknowledging this. Will this be fixed in the future?

a work around would consist in computing a second persistence

diagram after the topological simplification, there you should have the fixed matching by construction (which seems to do the trick here).

This is exactly the work around I use as well!

On Sat, Jul 7, 2018 at 9:23 PM, Julien Tierny notifications@github.com

wrote:

hi, there's a mismatch indeed for a few of them. the mismatches seem to involve boundary saddles which are not constrained by the algorithm. a work around would consist in computing a second persistence diagram after the topological simplification, there you should have the fixed matching by construction (which seems to do the trick here). thanks for letting us know if that fixes your issue. best,

Dr Julien Tierny CNRS Researcher Sorbonne Universite http://lip6.fr/Julien.Tierny

On Saturday, July 7, 2018 8:51:54 PM CEST Kamakshidasan wrote:

Sorry about that. Here you go.

[tv_1.vtk.zip](https://github.com/topology-tool-kit/ttk/

files/2173004/tv_1.v

tk.zip)

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/topology-tool-kit/ttk/issues/130#issuecomment-40323771 7, or mute the thread https://github.com/notifications/unsubscribe-auth/AI7akLRqcCouynIrsVyON14 JU9q4RKPkks5uEQqmgaJpZM4VGc3C .