Collection of ideas

[stanfish06]

• Found an interesting idea about geodesic entropy, which can be used to tell where branching happens. Might be useful? SLICER

[stanfish06]

MAPPER algorithm is interesting as I can use it to get a coarse-grained graph, which could make my package more scalable

[stanfish06]

potentially use waypoints to make diffmap computation more efficient

[stanfish06]

select diffusion plane for better loop visualization

[stanfish06]

downsampling can likely be down hierarchically if groupby not provided, potentially doing some robust clustering?

[stanfish06]

try mellon's functional estimator to compute slope field of around the loop

[stanfish06]

rotational decomposition on the velocity field and figure out gene regulatory network

[stanfish06]

replace cross-species matching with image based registration

[stanfish06]

implement image-based registration as a fast fail test

[stanfish06]

Trajectory embedding

[stanfish06]

for loop matching through boundary check, you need some level of tolerance based on Hausdorff distance and jittering width. for cross matching with continuous mapping function, Lipschitz constant should be taken into account.