view_hits.py

"""Load hits_xy.npy + dataset.npz from this Hugging Face dataset repo and
save a PNG of the rasterized hit image for the closest real event to a
requested (PID, momentum, eta, phi).

The data sits on a fixed grid (momentum in 0.1 GeV/c steps, eta/phi in
0.1 steps) and each particle species was independently randomly sampled
from it, so an exact match to your input essentially never exists --
this finds the closest real event instead and prints its true kinematics.

Usage:
    python view_hits.py --pid 2212 --mom 30 --eta 2.0 --phi 3.14 --out proton.png
"""
import argparse

import matplotlib
matplotlib.use("Agg")
import matplotlib.pyplot as plt
import numpy as np
from huggingface_hub import hf_hub_download

REPO = "deepaksamuel-cuk/simhits"
PID2LABEL = {11: 0, 211: 1, 321: 2, 2212: 3}
NAMES = ["electron", "pion", "kaon", "proton"]
FULL_IMG = 384
FULL_WINDOW = 3540.0  # mm -- full detector extent, matches the real dRICH sensor plane


def rasterize(pts, img_size=FULL_IMG, window=FULL_WINDOW):
    """pixel value = log1p(hit count) -- multiple hits in the same cell
    are summed (np.add.at), never overwritten or capped."""
    img = np.zeros((img_size, img_size), dtype=np.float32)
    if len(pts) == 0:
        return img
    rel = (pts - (-window / 2)) * (img_size / window)
    ij = np.floor(rel).astype(np.int64)
    ok = (ij[:, 0] >= 0) & (ij[:, 0] < img_size) & (ij[:, 1] >= 0) & (ij[:, 1] < img_size)
    ij = ij[ok]
    np.add.at(img, (ij[:, 1], ij[:, 0]), 1.0)
    return np.log1p(img)


def find_nearest_event(label, mom, eta, phi, pid, p, eta_q, phi_q):
    """Nearest event of the requested species in normalized (p, eta, phi)
    space -- phi handled circularly so 0 and 2*pi are adjacent."""
    c = PID2LABEL[abs(pid)]
    cand = np.where(label == c)[0]
    dphi = np.angle(np.exp(1j * (phi[cand] - phi_q)))
    d2 = (((mom[cand] - p) / 60.0) ** 2
          + ((eta[cand] - eta_q) / 2.0) ** 2
          + (dphi / np.pi) ** 2)
    return int(cand[np.argmin(d2)])


def main():
    ap = argparse.ArgumentParser(description=__doc__,
                                 formatter_class=argparse.RawDescriptionHelpFormatter)
    ap.add_argument("--pid", type=int, required=True, choices=list(PID2LABEL),
                    help="11=electron, 211=pion, 321=kaon, 2212=proton")
    ap.add_argument("--mom", type=float, required=True, help="momentum [GeV/c]")
    ap.add_argument("--eta", type=float, required=True, help="pseudorapidity")
    ap.add_argument("--phi", type=float, required=True, help="azimuth [rad]")
    ap.add_argument("--out", default="hit_image.png", help="output PNG path")
    args = ap.parse_args()

    dataset_path = hf_hub_download(REPO, "dataset.npz", repo_type="dataset")
    hits_path = hf_hub_download(REPO, "hits_xy.npy", repo_type="dataset")
    d = np.load(dataset_path)
    label, mom, eta, phi, offsets = (d["label"], d["mom"], d["eta"],
                                     d["phi"], d["offsets"])
    hits = np.load(hits_path, mmap_mode="r")

    j = find_nearest_event(label, mom, eta, phi, args.pid, args.mom, args.eta, args.phi)
    a, b = offsets[j], offsets[j + 1]
    pts = np.asarray(hits[a:b], dtype=np.float32)
    img = rasterize(pts)

    name = NAMES[PID2LABEL[abs(args.pid)]]
    print(f"nearest match: {name} (PID {args.pid}) -- "
          f"p={mom[j]:.2f} GeV/c, eta={eta[j]:.2f}, phi={phi[j]:.2f} rad, "
          f"{len(pts)} hits (event idx {j})")

    half = FULL_WINDOW / 2
    fig, ax = plt.subplots(figsize=(6, 6))
    ax.imshow(img, origin="lower", cmap="Blues", extent=[-half, half, -half, half])
    ax.set_xlabel("x [mm]")
    ax.set_ylabel("y [mm]")
    ax.set_title(f"{name}: p={mom[j]:.2f} GeV/c, eta={eta[j]:.2f}, phi={phi[j]:.2f} rad")
    fig.tight_layout()
    fig.savefig(args.out, dpi=150)
    print("wrote", args.out)


if __name__ == "__main__":
    main()