sampling.py

  1
  2
  3
  4
  5
  6
  7
  8
  9
 10
 11
 12
 13
 14
 15
 16
 17
 18
 19
 20
 21
 22
 23
 24
 25
 26
 27
 28
 29
 30
 31
 32
 33
 34
 35
 36
 37
 38
 39
 40
 41
 42
 43
 44
 45
 46
 47
 48
 49
 50
 51
 52
 53
 54
 55
 56
 57
 58
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
# SPDX-License-Identifier: AGPL-3.0-or-later
# Copyright (C) 2025 SWGY, Inc
#
# This program is free software: you can redistribute it and/or modify
# it under the terms of the GNU Affero General Public License as published by
# the Free Software Foundation, either version 3 of the License, or
# (at your option) any later version.
#
# This program is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
# GNU Affero General Public License for more details.
#
# You should have received a copy of the GNU Affero General Public License
# along with this program.  If not, see <https://www.gnu.org/licenses/>.
"""
Host-side sampling prep (pure NumPy, float64).

Builds the filtered explosion grid, per-explosion orthonormal frames, the padded
azimuth/elevation ray-fan offsets, and the Friedlander impulse constant. The
Warp device core (:mod:`warp_core`) consumes these arrays and generates the rays
in-kernel, so nothing large is ever materialised here.
"""

from __future__ import annotations

from dataclasses import dataclass
from typing import Optional

import numpy as np

from config import Config

# Arc-minute -> radian conversion: 1 arc-minute = (pi/180)/60 = pi/10800.
_ARCMIN_TO_RAD = np.pi / 10800.0


@dataclass(frozen=True)
class Sensor:
    origin: np.ndarray  # (3,) world coordinates
    radius: float


@dataclass(frozen=True)
class GridArrays:
    """Compact per-explosion ray-fan description (the device builds the rays)."""
    blast_os: np.ndarray   # (E, 3) kept blast points
    frames: np.ndarray     # (E, 3, 3) rows = (forward, right, up)
    az_off: np.ndarray     # (E, Gaz) azimuth angles
    el_off: np.ndarray     # (E, Gel) elevation angles
    valid_az: np.ndarray   # (E, Gaz) bool
    valid_el: np.ndarray   # (E, Gel) bool
    res_rad: float         # angular ray spacing


def _unit_rows(v: np.ndarray) -> np.ndarray:
    """Normalise each row; rows with ~zero norm become zero."""
    norm = np.linalg.norm(v, axis=-1, keepdims=True)
    safe = np.where(norm >= 1e-12, norm, 1.0)
    return np.where(norm >= 1e-12, v / safe, 0.0)


def _frames(forward: np.ndarray) -> np.ndarray:
    """Orthonormal (forward, right, up) rows per explosion (gimbal-guarded)."""
    fwd = _unit_rows(forward)
    up0 = np.array([0.0, 0.0, 1.0])
    alt = np.array([0.0, 1.0, 0.0])
    near_parallel = np.abs(fwd @ up0) > 0.99
    up = np.where(near_parallel[:, None], alt, up0)
    right = _unit_rows(np.cross(fwd, up))
    up = _unit_rows(np.cross(right, fwd))
    return np.stack([fwd, right, up], axis=1)  # (E, 3, 3)


def precompute_waveform_constant(t_d: float, alpha: float) -> float:
    """Impulse per unit peak pressure: K = trapz(friedlander(., 1.0, t_d, alpha)).

    The Friedlander waveform is linear in peak pressure, so a ray's impulse is
    just ``p_eff * K``; K is that fixed shape integral, evaluated once as a
    1000-point trapezoid over the positive (clipped) phase.
    """
    t = np.linspace(0.0, 3.0 * t_d, 1000)
    p = np.zeros_like(t)
    m = t >= 0
    p[m] = 1.0 * (1.0 - t[m] / t_d) * np.exp(-alpha * t[m] / t_d)
    p[p < 0] = 0.0
    return float(np.sum(0.5 * (p[1:] + p[:-1]) * np.diff(t)))


def _axis_samples(center: float, extents: float, segments: int) -> np.ndarray:
    """Sample coordinates along one axis, centred correctly (single offset)."""
    return np.linspace(center - extents / 2.0, center + extents / 2.0, segments)


def _counts(lo: np.ndarray, hi: np.ndarray, res: float) -> np.ndarray:
    """Sample count per axis, matching ``len(np.arange(lo, hi + res, res))``."""
    return np.maximum(1, np.ceil((hi - lo) / res + 1.0 - 1e-9)).astype(np.int64)


def build_grid(cfg: Config, sensor: Sensor,
               vertices_all: Optional[np.ndarray] = None) -> GridArrays:
    """Build the filtered explosion grid and its padded per-explosion ray fans.

    `vertices_all` (all armor vertices) is only needed for
    ``rays.fan_mode == "geometry"``.
    """
    b = cfg.blast
    xs = _axis_samples(b.cube_center_m[0], b.cube_extents_m, b.cube_segments)
    ys = _axis_samples(b.cube_center_m[1], b.cube_extents_m, b.cube_segments)
    zs = _axis_samples(b.cube_center_m[2], b.cube_extents_m, b.cube_segments)
    grid = np.stack(np.meshgrid(xs, ys, zs, indexing="ij"), axis=-1).reshape(-1, 3)

    dist = np.linalg.norm(grid - sensor.origin, axis=-1)
    keep = dist >= b.min_dist_m
    blast_os = grid[keep]
    dist = dist[keep]
    if blast_os.shape[0] == 0:
        raise ValueError("No blast points survive the min-distance filter; "
                         "check blast.cube_* and blast.min_dist_m.")

    frames = _frames(sensor.origin - blast_os)      # (E, 3, 3)
    fwd, right, up = frames[:, 0, :], frames[:, 1, :], frames[:, 2, :]

    ratio = np.clip(sensor.radius / dist, -1.0, 1.0)
    half = np.where(dist <= sensor.radius, np.pi / 2.0, np.arcsin(ratio))
    az_lo, az_hi, el_lo, el_hi = -half, half, -half, half

    if cfg.rays.fan_mode == "geometry":
        if vertices_all is None:
            raise ValueError("fan_mode 'geometry' requires vertices_all")
        rel = vertices_all[None, :, :] - blast_os[:, None, :]     # (E, V, 3)
        proj_fwd = np.einsum("evc,ec->ev", rel, fwd)
        h_ang = np.arctan2(np.einsum("evc,ec->ev", rel, right), proj_fwd)
        v_ang = np.arctan2(np.einsum("evc,ec->ev", rel, up), proj_fwd)
        az_lo = np.minimum(az_lo, h_ang.min(axis=1))
        az_hi = np.maximum(az_hi, h_ang.max(axis=1))
        el_lo = np.minimum(el_lo, v_ang.min(axis=1))
        el_hi = np.maximum(el_hi, v_ang.max(axis=1))

    res = float(cfg.rays.resolution_arcmin) * _ARCMIN_TO_RAD
    n_az = _counts(az_lo, az_hi, res)
    n_el = _counts(el_lo, el_hi, res)
    gaz, gel = int(n_az.max()), int(n_el.max())

    k_az, k_el = np.arange(gaz), np.arange(gel)
    az_off = az_lo[:, None] + res * k_az[None, :]     # (E, Gaz)
    el_off = el_lo[:, None] + res * k_el[None, :]     # (E, Gel)
    valid_az = k_az[None, :] < n_az[:, None]
    valid_el = k_el[None, :] < n_el[:, None]

    return GridArrays(blast_os, frames, az_off, el_off, valid_az, valid_el, res)