SciQLopPlots 0.26.0

SciQLop plot API based on QCustomPlot

SciQLopPlots

A high-performance scientific plotting library built on C++20/Qt6 with Python bindings via Shiboken6 (PySide6). Designed for the SciQLop data analysis platform but usable standalone.

Features

• Async resampling — render millions of points smoothly; data is downsampled in background threads via NeoQCP pipelines
• Multiple plot types — time series (line graphs), spectrograms (color maps), 2D histograms, parametric curves, N-D projection curves, waterfall plots
• Interactive — pan, zoom, data-driven callbacks, vertical/horizontal/rectangular spans, tracers, straight lines, text, shapes, pixmaps
• Multi-plot panels — synchronized axes, aligned margins, drag-and-drop from product trees
• Reactive pipelines — connect plot properties with >> to build live data flows
• Export — PDF (vector), PNG, JPG, BMP for both individual plots and panels
• Busy indicator — visual feedback when data is loading or being processed
• Runtime inspector — tree model/view for inspecting and editing plot properties
• Cross-platform — Linux, macOS, Windows

Architecture

graph TD
    subgraph Python["Python Layer"]
        PY[SciQLopPlots package]
        SB[Shiboken6 bindings]
    end

    subgraph Plots["Plot Hierarchy"]
        PI[SciQLopPlotInterface<br/><i>QFrame</i>]
        SP[SciQLopPlot]
        TSP[SciQLopTimeSeriesPlot]
        NDP[SciQLopNDProjectionPlot]
    end

    subgraph Plotables["Plotables"]
        PTI[SciQLopPlottableInterface<br/><i>QObject</i>]
        GI[SciQLopGraphInterface]
        CMI[SciQLopColorMapInterface]
        LG[SciQLopLineGraph]
        SLG[SciQLopSingleLineGraph]
        CRV[SciQLopCurve]
        WF[SciQLopWaterfallGraph]
        NDC[SciQLopNDProjectionCurves]
        CMB[SciQLopColorMapBase]
        CM[SciQLopColorMap]
        H2D[SciQLopHistogram2D]
    end

    subgraph Items["Overlay Items"]
        VS[VerticalSpan]
        HS[HorizontalSpan]
        RS[RectangularSpan]
        TR[Tracer]
        SL[StraightLine]
    end

    subgraph MultiPlot["Multi-Plot"]
        PPI[SciQLopPlotPanelInterface]
        MPP[SciQLopMultiPlotPanel]
        AX[AxisSynchronizer]
        VA[VPlotsAlign]
    end

    PY --> SB --> PI

    PI --> SP
    PI --> NDP
    SP --> TSP

    PTI --> GI
    PTI --> CMI
    GI --> LG
    GI --> SLG
    GI --> CRV
    GI --> WF
    GI --> NDC
    CMI --> CMB
    CMB --> CM
    CMB --> H2D

    PPI --> MPP
    MPP --> AX
    MPP --> VA

    SP -.->|contains| PTI
    SP -.->|contains| Items
    MPP -.->|contains| SP

Data flow

sequenceDiagram
    participant User
    participant Plot as SciQLopPlot
    participant Resampler as Resampler<br/>(worker thread)
    participant NeoQCP as NeoQCP

    User->>Plot: pan / zoom
    Plot->>Resampler: new visible range
    Resampler->>Resampler: downsample data
    Resampler->>NeoQCP: resampled points
    NeoQCP->>Plot: render

For callback-driven data, the plot invokes a Python callable with (start, stop) on range change, and the returned arrays flow through the same resampling pipeline.

Quick start

pip install SciQLopPlots

import numpy as np
from PySide6.QtWidgets import QApplication
from SciQLopPlots import SciQLopPlot

app = QApplication([])
plot = SciQLopPlot()

# Static data
x = np.arange(0, 1000, dtype=np.float64)
y = np.sin(x / 100) * np.cos(x / 10)
plot.plot(x, y, labels=["signal"])

# Data callback (called on pan/zoom with visible range)
def get_data(start, stop):
    x = np.arange(start, stop, dtype=np.float64)
    y = np.column_stack([np.sin(x / 100), np.cos(x / 100)])
    return x, y

plot.plot(get_data, labels=["sin", "cos"])

plot.show()
app.exec()

Reactive pipelines

Connect plot properties with the >> operator to build live data pipelines:

from SciQLopPlots import SciQLopPlot

plot = SciQLopPlot()
graph = plot.plot(lambda start, stop: ..., labels=["signal"])

# Axis range changes automatically feed a transform, which pushes data to the graph
plot.x_axis.on.range >> get_data >> graph.on.data

# Direct property forwarding (no transform)
span.on.range >> plot.x_axis.on.range

# Chain multiple steps
source.on.range >> transform >> target.on.data

Run the gallery for a full feature showcase:

python tests/manual-tests/gallery.py

Runtime tracing

SciQLopPlots ships with a built-in tracer that emits Chrome trace JSON, viewable in Perfetto, Speedscope, or chrome://tracing. Always compiled in, runtime-toggled, ~1 ns when off.

Enable for a session, then open the file in Perfetto:

from SciQLopPlots import tracing

with tracing.session("/tmp/sciqlop.json"):
    panel.zoom_in_a_lot()       # whatever's slow

Annotate Python work to land alongside the C++ zones (plot.replot, setdata.colormap, resample.async_2d, …):

with tracing.zone("speasy.fetch", cat="data", product="amda/mms_fgm"):
    data = speasy.get_data(...)

@tracing.traced("layer.eval", cat="layer")
def render_layer(...): ...

tracing.counter("queue_depth", q.size(), cat="fetch")

You can also auto-enable at process start with the SCIQLOP_TRACE env var:

SCIQLOP_TRACE=/tmp/sciqlop.json python my_script.py

When tracy_enable=true is also passed at build time, the same PROFILE_HERE_N sites feed both the Chrome JSON tracer and the Tracy live-streaming view.

Building from source

Requires Qt6, PySide6 == 6.11.0, a C++20 compiler, and Meson.

# Development build (recommended — plain `debug` disables optimizations
# and makes the resampling pipelines noticeably sluggish)
meson setup build --buildtype=debugoptimized
meson compile -C build

# Install as editable Python package
pip install -e . --no-build-isolation

Build options

Option	Type	Default	Description
trace_refcount	bool	false	Enable reference count tracing
tracy_enable	bool	false	Enable Tracy profiling
with_opengl	bool	true	Enable OpenGL support

Contributing

Fork the repository, make your changes and submit a pull request. Bug reports and feature requests are welcome.

Credits

Development is supported by CDPP. We acknowledge support from Plas@Par.

Thanks

• PySide6 — Qt bindings and Shiboken6 binding generator
• NeoQCP — fork of QCustomPlot with async pipelines, multi-dtype support, and QRhi rendering