View page source

warp.hash_grid_query#

warp.hash_grid_query(
id: uint64,
point: vec3f,
max_dist: float32,
) → HashGridQuery#

  • Kernel

Construct a point query against a warp.HashGrid.

Returns a query that iterates over candidate neighbors of point: every point in the grid cells overlapped by the box from point - max_dist to point + max_dist. These are candidates — the grid does not test distance, so some are farther than max_dist; filter by actual distance yourself. Advance the query and read each candidate’s index with hash_grid_query_next(). point must be in the same coordinate space as the points the grid was built from (see warp.HashGrid).

Parameters:

  • id – The warp.HashGrid identifier

  • point – The query point

  • max_dist – The query radius

Returns:

A hash-grid query object to pass to hash_grid_query_next().

Example

@wp.kernel
def count_neighbors(grid_id: wp.uint64, pts: wp.array[wp.vec3], radius: wp.float32, out_count: wp.array[wp.int32]):
    i = wp.tid()
    p = pts[i]
    query = wp.hash_grid_query(grid_id, p, radius)
    index = int(0)
    n = int(0)
    while wp.hash_grid_query_next(query, index):
        if wp.length(p - pts[index]) <= radius:
            n += 1
    out_count[i] = n

points = wp.array([[0,0,0],[0.1,0,0],[0.5,0,0],[2.0,0,0]], dtype=wp.vec3)
grid = wp.HashGrid(dim_x=8, dim_y=8, dim_z=8)
grid.build(points=points, radius=0.3)

out_count = wp.zeros(4, dtype=wp.int32)
wp.launch(count_neighbors, dim=4, inputs=[grid.id, points, 0.3], outputs=[out_count])
print(out_count.numpy())

[2 2 1 1]
warp.hash_grid_query(
id: uint64,
point: vec3f,
max_dist: float32,
group: int32,
) → HashGridQuery

  • Kernel

Construct a point query against a warp.HashGrid, restricted to one point group.

Returns a query that iterates over candidate neighbors of point: every point in the grid cells overlapped by the box from point - max_dist to point + max_dist. These are candidates — the grid does not test distance, so some are farther than max_dist; filter by actual distance yourself. Advance the query and read each candidate’s index with hash_grid_query_next(). point must be in the same coordinate space as the points the grid was built from (see warp.HashGrid).

If the grid was built with groups, only points whose group id equals group are returned as candidates; any int32 value is a valid group id. Omit the group argument to visit all groups, matching ungrouped behavior. Unlike grouped BVH queries, grouped hash-grid queries do not require a root lookup; pass the group id directly.

Parameters:

  • id – The warp.HashGrid identifier

  • point – The query point

  • max_dist – The query radius

  • group – Restrict candidates to points built with this group id

Returns:

A hash-grid query object to pass to hash_grid_query_next().

warp.hash_grid_query(
id: uint64,
point: vec3h,
max_dist: float16,
) → HashGridQuery

  • Kernel

Construct a point query against a warp.HashGrid (float16 precision).

The float16 overload of hash_grid_query(). Behavior and usage match the default float32 overload; see it for details and a usage example.

Parameters:

  • id – The warp.HashGrid identifier

  • point – The query point

  • max_dist – The query radius

Returns:

A hash-grid query object to pass to hash_grid_query_next().

warp.hash_grid_query(
id: uint64,
point: vec3h,
max_dist: float16,
group: int32,
) → HashGridQuery

  • Kernel

Construct a point query against a warp.HashGrid, restricted to one point group (float16 precision).

The float16 overload of hash_grid_query(). Behavior and usage match the default float32 overload; see it for details and a usage example.

If the grid was built with groups, only points whose group id equals group are returned as candidates; any int32 value is a valid group id. Omit the group argument to visit all groups, matching ungrouped behavior. Unlike grouped BVH queries, grouped hash-grid queries do not require a root lookup; pass the group id directly.

Parameters:

  • id – The warp.HashGrid identifier

  • point – The query point

  • max_dist – The query radius

  • group – Restrict candidates to points built with this group id

Returns:

A hash-grid query object to pass to hash_grid_query_next().

warp.hash_grid_query(
id: uint64,
point: vec3d,
max_dist: float64,
) → HashGridQuery

  • Kernel

Construct a point query against a warp.HashGrid (float64 precision).

The float64 overload of hash_grid_query(). Behavior and usage match the default float32 overload; see it for details and a usage example.

Parameters:

  • id – The warp.HashGrid identifier

  • point – The query point

  • max_dist – The query radius

Returns:

A hash-grid query object to pass to hash_grid_query_next().

warp.hash_grid_query(
id: uint64,
point: vec3d,
max_dist: float64,
group: int32,
) → HashGridQuery

  • Kernel

Construct a point query against a warp.HashGrid, restricted to one point group (float64 precision).

The float64 overload of hash_grid_query(). Behavior and usage match the default float32 overload; see it for details and a usage example.

If the grid was built with groups, only points whose group id equals group are returned as candidates; any int32 value is a valid group id. Omit the group argument to visit all groups, matching ungrouped behavior. Unlike grouped BVH queries, grouped hash-grid queries do not require a root lookup; pass the group id directly.

Parameters:

  • id – The warp.HashGrid identifier

  • point – The query point

  • max_dist – The query radius

  • group – Restrict candidates to points built with this group id

Returns:

A hash-grid query object to pass to hash_grid_query_next().