[Lang] Add ti.Vector.ndarray and ti.Matrix.ndarray #2808
Conversation
|
/format |
|
/format |
python/taichi/lang/ndarray.py
Outdated
| """Class for accessing VectorNdarray/MatrixNdarray in Python scope. | ||
| Args: | ||
| arr (Union[VectorNdarray, MatrixNdarray]): See above. | ||
| indices_first (Tuple[Int]): Indices of first-level access. |
There was a problem hiding this comment.
What are first level and second level?
There was a problem hiding this comment.
First level refers to coordinates in the field, while second level refers to indices inside the vector/matrix. Maybe I should make it clearer?
There was a problem hiding this comment.
Using first/second is to tell users this is what they write in the first/second pair of brackets.
There was a problem hiding this comment.
Explanations are added.
| @@ -57,8 +58,25 @@ def __init__(self, element_shape=(), layout=Layout.AOS): | |||
| self.layout = layout | |||
|
|
|||
| def extract(self, x): | |||
| shape = tuple(x.shape) | |||
| from taichi.lang.matrix import MatrixNdarray, VectorNdarray | |||
There was a problem hiding this comment.
move to top? (or will it cause circular import?)
There was a problem hiding this comment.
It will cause circular import :-(
| assert v[4][9] == 9 | ||
|
|
||
|
|
||
| # number of compiled functions |
There was a problem hiding this comment.
is this comment not longer needed?
There was a problem hiding this comment.
It's used to separate three different parts of tests of this file.
Related issue = #2771
This PR adds
ti.Vector.ndarrayandti.Matrix.ndarray, which support Python-scope and Taichi-scope access. When passing them into kernels, we need to useti.any_arr()annotation withelement_shapeand optionallayout(default value:ti.Layout.AOS) to enforce type consistency.