Refactoring trait pipeline (1/2)

[linwang9926]

• Reorganizing the trait_map dictionary in the format: "trait_name": (function, ["input_trait1", "input_trait2", ...], {"additional_kwarg1": True, "additional_kwarg2": 0.5})
• modify ellipse, network functions by adding an Optional argument(s). Take ellipse for example:
  • def get_ellipse_a(pts_all_array: np.ndarray,ellipse: Optional[Tuple[float, float, float]],).
  • The argument "ellipse" is ellipse feature, the values will be saved to the traits dictionary once we did trait calculation and BFS for the edges (traits).

==================================================================================

• Update ellipse and network functions argument. If the arguments have different types (Tuple vs. np.ndarray), then use a Union to make this arguments inputting as pts_all_array or ellipse Tuple. If they are the same type, them make two arguments. for example:
  • get_ellipse_a(pts_all_array: Union[np.ndarray, Tuple[float, float, float]])
  • get_network_solidity(primary_pts: np.ndarray,lateral_pts: np.ndarray,
    chull_area: float = None,pts_all_array: Optional[np.ndarray] = None,monocots: bool = False,)
• Modify the ellipse and network test functions
• Reorganize the arguments in trait_map dictionary in get_traits_value_frame function.

==================================================================================

• Update get_root_angle function by adding node_ind as an arugment.
• Update base-related functions (get_base_ct_density, get_base_length_ratio) by assuming the primary_pts as primary_length_max.
• Update get_base_median_ratio function by using lateral_base_ys and primary_tip_pt_y.
• Modify argument pts as Optional in get_base_tip_dist and get_grav_index functions
• Draft the trait_definitions using the defined TraitDef class
• Redo the get_node_ind function:
  • if proximal, get the nearest non-Nan node index.
  • If proximal, just use the first half root (if odd number node, the centre node is included).
  • If not proximal, just use the last half root (if odd number node, the centre node is included).
  • if there is no non-Nan node detected, just return an index of 0 (base node).

[eberrigan]

Uppercase "P" in primary :)

[eberrigan]

looks good!

[eberrigan]

the input traits always have to match the trait definition. This function is too complicated. We should simplify it to just assume taking in the expected input traits from from the trait graph.

[linwang9926]

The base_pts and tip_pts can be used for our pipeline to quickly calculate the base-tip distance. The optional pts is designed for general purpose.

[eberrigan]

Same here: only input traits in the trait definition can be arguments to the function.

[linwang9926]

I redo the get_node_ind function:

• if proximal, get the nearest non-Nan node index.
• If proximal, just use the first half root (if odd number node, the centre node is included).
• If not proximal, just use the last half root (if odd number node, the centre node is included).
• if there is no non-Nan node detected, just return an index of 0 (base node).

[linwang9926]

Network-related functions need further test after testing get_root_lengths function.

[eberrigan]

sorry I tried to vectorize this but looks like I did it incorrectly. If it was better before you can change it back!

[linwang9926]

No problem! The current version also works and uses vectors instead of a for loop.