TopolearnWidgets turn a notebook into a small exploratory app: tweak a parameter and see results update without editing code.
ipywidgets provides interactive controls rendered in the notebook and synchronized with Python objects in the kernel. Install it in the kernel environment (pip install ipywidgets); JupyterLab 4 and Notebook 7 support it without extra steps.
The fastest route is interact, which builds controls from a function's arguments:
from ipywidgets import interact
@interact(n=(1, 100), kind=["linear", "log"])
def show(n=10, kind="linear"): ...A tuple becomes a slider, a list becomes a dropdown, a bool a checkbox, a string a text box. The function reruns every time a value changes; interact_manual adds a Run button for slow functions.
For more control, create widgets directly: widgets.IntSlider(min=0, max=10), Dropdown, Text, Button, Checkbox, DatePicker, combined with layout containers HBox, VBox, Tab and Accordion. Each widget has a value trait. React to changes with:
w.observe(callback, names="value"): called with a change dict holdingoldandnew;button.on_click(callback);widgets.link((a, "value"), (b, "value"))to keep two widgets in sync.
Printed output and plots from callbacks should go into a widgets.Output() area, used as a context manager, or they may vanish.
Widgets need a live kernel; in exported HTML they show their last state at best. For shareable apps, consider Voila, Streamlit or Dash.
Example
import ipywidgets as widgets
price = widgets.FloatSlider(value=10, min=0, max=50, step=0.5, description="Price")
qty = widgets.IntSlider(value=3, min=1, max=20, description="Qty")
total = widgets.Label()
log = []
def update(change=None):
total.value = f"Total: {price.value * qty.value:.2f}"
price.observe(update, names="value")
qty.observe(update, names="value")
qty.observe(lambda ch: log.append((ch["old"], ch["new"])), names="value")
update()
ui = widgets.VBox([price, qty, total]) # display(ui) in a notebook
qty.value = 5 # simulate the user moving a slider
print(total.value) # Total: 50.00
print(log) # [(3, 5)]Remember
interactbuilds controls from function arguments in one line- Widgets have a
value; react withobserve(..., names='value')oron_click - Send callback output to a
widgets.Output()area
Watch out for
- Printing from a widget callback without an Output widget, so the output appears in the log instead of under the widget.