Points that grow when zooming

Explore 200,000 generated points without losing them as you zoom in.

Zoom horizontally with the mouse wheel. Point sizes increase as the view narrows, making individual points easier to see. A size limit prevents the points from growing indefinitely.

Data use and provenance

NumPy creates 200,000 positions and noise values with a fixed random seed. GenomeSpy calculates the wave using the upstream formula. The gallery loads the table from an Arrow file to keep the specification small.

Adapted from the official GenomeSpy example.

Code

"""Points that grow when zooming.

Explore 200,000 generated points without losing them as you zoom in.
"""

import numpy as np
import pandas as pd

import genome_spy as gs


# A fixed seed keeps the noisy wave the same each time.
rng = np.random.default_rng(42)
data = pd.DataFrame({"x": np.arange(200000), "u": rng.random(200000) * 0.682})
zoom = gs.Expression("zoomLevel")

chart = (
    gs.Chart(data)
    # Calculate the wave from the prepared positions and noise.
    .transform_calculate(
        y=(
            gs.expr.if_(gs.datum.u % 1e-8 > 5e-9, 1, -1)
            * (gs.expr.sqrt(-gs.expr.log(gs.expr.max(1e-9, gs.datum.u))) - 0.618)
        )
        * 1.618
        + gs.expr.sin(gs.datum.x / 10000)
    )
    # Increase point size with zoom, up to a readable maximum.
    .mark_point(size=gs.expr(gs.expr.min(0.5 * gs.expr.pow(zoom, 1.5), 200)))
    .encode(
        x=gs.X("x:Q").scale(zoom=True).axis(format="~s", labelFlush=True),
        y=gs.Y("y:Q"),
        opacity=gs.value(0.6),
    )
)

Show generated specification