File size: 5,859 Bytes
23e78eb
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
#!/usr/bin/env python3
"""Download N random samples from the U-bend HuggingFace dataset and visualize them.

For each sample, generates individual PNGs:
  - <id>_geometry.png  : fluid + solid mesh
  - <id>_U.png         : velocity magnitude
  - <id>_p.png         : pressure
  - <id>_T.png         : temperature (fluid + solid)
  - <id>_k.png         : turbulent kinetic energy
  - <id>_nut.png       : turbulent viscosity

Additionally, an overview grid (N rows x 6 columns) is saved as `overview.png`.

Usage:
    python visualize_sample.py --n 5
"""

import argparse
import os
import random
import numpy as np
import matplotlib.pyplot as plt
from matplotlib.patches import Patch
from huggingface_hub import hf_hub_download
from safetensors.numpy import load_file

REPO_ID = "JensDe/ubend-cfd"

parser = argparse.ArgumentParser()
parser.add_argument("--n", type=int, default=5, help="Number of random samples")
parser.add_argument("--seed", type=int, default=42, help="Random seed")
parser.add_argument("--out_dir", type=str, default="visualizations", help="Output directory")
args = parser.parse_args()

os.makedirs(args.out_dir, exist_ok=True)
random.seed(args.seed)
sample_ids = random.sample(range(10000), args.n * 3)  # buffer for failed downloads

FIELD_SPECS = [
    ("U", "Velocity magnitude", "viridis", "|U| [m/s]"),
    ("p", "Pressure", "coolwarm", "p [Pa]"),
    ("T", "Temperature", "hot", "T [K]"),
    ("k", "Turbulent kinetic energy", "magma", "k [m²/s²]"),
    ("nut", "Turbulent viscosity", "plasma", "ν_t [m²/s]"),
]


def render_geometry(ax, x, y, sx, sy):
    fluid_field = np.ones_like(x)
    solid_field = np.ones_like(sx) * 2
    ax.pcolormesh(x, y, fluid_field, cmap="Blues", shading="auto", vmin=0, vmax=3)
    ax.pcolormesh(sx, sy, solid_field, cmap="Oranges", shading="auto", vmin=0, vmax=3)
    ax.set_aspect("equal")


def render_field(ax, x, y, field, cmap, sx=None, sy=None, sfield=None):
    vmin, vmax = field.min(), field.max()
    if sfield is not None:
        vmin = min(vmin, sfield.min())
        vmax = max(vmax, sfield.max())
    pcm = ax.pcolormesh(x, y, field, cmap=cmap, shading="auto", vmin=vmin, vmax=vmax)
    if sfield is not None:
        ax.pcolormesh(sx, sy, sfield, cmap=cmap, shading="auto", vmin=vmin, vmax=vmax)
    ax.set_aspect("equal")
    return pcm


def get_field(data, name):
    if name == "U":
        U = data["U"]
        return np.sqrt(U[0]**2 + U[1]**2 + U[2]**2)
    return data[name]


# Download samples
samples = []
i = 0
while len(samples) < args.n and i < len(sample_ids):
    sid = sample_ids[i]
    i += 1
    try:
        print(f"[{len(samples)+1}/{args.n}] Downloading sample {sid}...")
        file_path = hf_hub_download(
            repo_id=REPO_ID,
            filename=f"fields/sample_{sid}.safetensors",
            repo_type="dataset",
        )
        samples.append((sid, load_file(file_path)))
    except Exception as e:
        print(f"  Skipping {sid}: {e.__class__.__name__}")

# Individual PNGs
for sid, data in samples:
    x, y = data["coords"][0], data["coords"][1]
    sx, sy = data["solid_coords"][0], data["solid_coords"][1]
    base = os.path.join(args.out_dir, f"sample_{sid:05d}")

    # Geometry
    fig, ax = plt.subplots(figsize=(5, 7))
    render_geometry(ax, x, y, sx, sy)
    ax.set_title(f"Sample {sid} — Geometry")
    ax.set_xlabel("x [m]"); ax.set_ylabel("y [m]")
    ax.legend(handles=[Patch(facecolor="steelblue", label="Fluid"),
                       Patch(facecolor="orange", label="Solid")],
              loc="upper center", bbox_to_anchor=(0.5, -0.1), ncol=2)
    plt.tight_layout()
    plt.savefig(f"{base}_geometry.png", dpi=150)
    plt.close(fig)

    # Fields
    for fname, ftitle, cmap, label in FIELD_SPECS:
        field = get_field(data, fname)
        fig, ax = plt.subplots(figsize=(5, 7))
        sfield = data["solid_T"] if fname == "T" else None
        pcm = render_field(ax, x, y, field, cmap,
                           sx=sx if sfield is not None else None,
                           sy=sy if sfield is not None else None,
                           sfield=sfield)
        plt.colorbar(pcm, ax=ax, label=label, orientation="horizontal", location="bottom", pad=0.08)
        ax.set_title(f"Sample {sid}{ftitle}")
        ax.set_xlabel("x [m]"); ax.set_ylabel("y [m]")
        plt.tight_layout()
        plt.savefig(f"{base}_{fname}.png", dpi=150)
        plt.close(fig)

# Overview grid: N rows x 6 columns
print(f"\nCreating overview grid...")
ncols = 1 + len(FIELD_SPECS)  # geometry + fields
fig, axes = plt.subplots(args.n, ncols, figsize=(3.5 * ncols, 4.5 * args.n))
if args.n == 1:
    axes = axes[None, :]

col_titles = ["Geometry"] + [s[1] for s in FIELD_SPECS]

for row, (sid, data) in enumerate(samples):
    x, y = data["coords"][0], data["coords"][1]
    sx, sy = data["solid_coords"][0], data["solid_coords"][1]

    # Geometry
    ax = axes[row, 0]
    render_geometry(ax, x, y, sx, sy)
    ax.set_xticks([]); ax.set_yticks([])
    if row == 0:
        ax.set_title(col_titles[0])
    ax.set_ylabel(f"Sample {sid}", fontsize=10)

    # Fields
    for col, (fname, _, cmap, _) in enumerate(FIELD_SPECS, start=1):
        ax = axes[row, col]
        field = get_field(data, fname)
        sfield = data["solid_T"] if fname == "T" else None
        render_field(ax, x, y, field, cmap,
                     sx=sx if sfield is not None else None,
                     sy=sy if sfield is not None else None,
                     sfield=sfield)
        ax.set_xticks([]); ax.set_yticks([])
        if row == 0:
            ax.set_title(col_titles[col])

plt.tight_layout()
overview_path = os.path.join(args.out_dir, "overview.png")
plt.savefig(overview_path, dpi=150)
plt.close(fig)
print(f"Saved overview to {overview_path}")
print(f"\nDone! {len(samples)} samples visualized in {args.out_dir}/")