Allow to modify slider range in interactive widget (#1038)

Close #1033

---------

Co-authored-by: Hans Dembinski <[email protected]>
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>

.gitignore

@@ -6,6 +6,9 @@
 /root-*
 /dev
 /venv
+/out
+/.venv
+/.vs
 
 /build
 /dist

doc/notebooks/interactive.ipynb

@@ -393,7 +393,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.12.4"
+   "version": "3.12.6"
   },
   "vscode": {
    "interpreter": {

src/iminuit/ipywidget.py

@@ -0,0 +1,294 @@
+"""Interactive fitting widget for Jupyter notebooks."""
+
+import warnings
+import numpy as np
+from typing import Dict, Any, Callable
+import sys
+
+with warnings.catch_warnings():
+    # ipywidgets produces deprecation warnings through use of internal APIs :(
+    warnings.simplefilter("ignore")
+    try:
+        import ipywidgets as widgets
+        from ipywidgets.widgets.interaction import show_inline_matplotlib_plots
+        from IPython.display import clear_output
+        from matplotlib import pyplot as plt
+    except ModuleNotFoundError as e:
+        e.msg += (
+            "\n\nPlease install ipywidgets, IPython, and matplotlib to "
+            "enable interactive"
+        )
+        raise
+
+
+def make_widget(
+    minuit: Any,
+    plot: Callable[..., None],
+    kwargs: Dict[str, Any],
+    raise_on_exception: bool,
+):
+    """Make interactive fitting widget."""
+    # Implementations makes heavy use of closures,
+    # we frequently use variables which are defined
+    # near the end of the function.
+    original_values = minuit.values[:]
+    original_limits = minuit.limits[:]
+
+    def plot_with_frame(from_fit, report_success):
+        trans = plt.gca().transAxes
+        try:
+            with warnings.catch_warnings():
+                minuit.visualize(plot, **kwargs)
+        except Exception:
+            if raise_on_exception:
+                raise
+
+            import traceback
+
+            plt.figtext(
+                0,
+                0.5,
+                traceback.format_exc(limit=-1),
+                fontdict={"family": "monospace", "size": "x-small"},
+                va="center",
+                color="r",
+                backgroundcolor="w",
+                wrap=True,
+            )
+            return
+
+        fval = minuit.fmin.fval if from_fit else minuit._fcn(minuit.values)
+        plt.text(
+            0.05,
+            1.05,
+            f"FCN = {fval:.3f}",
+            transform=trans,
+            fontsize="x-large",
+        )
+        if from_fit and report_success:
+            plt.text(
+                0.95,
+                1.05,
+                f"{'success' if minuit.valid and minuit.accurate else 'FAILURE'}",
+                transform=trans,
+                fontsize="x-large",
+                ha="right",
+            )
+
+    def fit():
+        if algo_choice.value == "Migrad":
+            minuit.migrad()
+        elif algo_choice.value == "Scipy":
+            minuit.scipy()
+        elif algo_choice.value == "Simplex":
+            minuit.simplex()
+            return False
+        else:
+            assert False  # pragma: no cover, should never happen
+        return True
+
+    class OnParameterChange:
+        # Ugly implementation notes:
+        # We want the plot when the user moves the slider widget, but not when
+        # we update the slider value manually from our code. Unfortunately,
+        # the latter also calls OnParameterChange, which leads to superfluous plotting.
+        # I could not find a nice way to prevent that (and I tried many), so as a workaround
+        # we optionally skip a number of calls, when the slider is updated.
+        def __init__(self, skip: int = 0):
+            self.skip = skip
+
+        def __call__(self, change: Dict[str, Any] = {}):
+            if self.skip > 0:
+                self.skip -= 1
+                return
+
+            from_fit = change.get("from_fit", False)
+            report_success = change.get("report_success", False)
+            if not from_fit:
+                for i, x in enumerate(parameters):
+                    minuit.values[i] = x.slider.value
+
+            if any(x.fit.value for x in parameters):
+                saved = minuit.fixed[:]
+                for i, x in enumerate(parameters):
+                    minuit.fixed[i] = not x.fit.value
+                from_fit = True
+                report_success = do_fit(None)
+                minuit.fixed = saved
+
+            # Implementation like in ipywidegts.interaction.interactive_output
+            with out:
+                clear_output(wait=True)
+                plot_with_frame(from_fit, report_success)
+                with warnings.catch_warnings():
+                    warnings.simplefilter("ignore")
+                    show_inline_matplotlib_plots()
+
+    def do_fit(change):
+        report_success = fit()
+        for i, x in enumerate(parameters):
+            x.reset(minuit.values[i])
+        if change is None:
+            return report_success
+        OnParameterChange()({"from_fit": True, "report_success": report_success})
+
+    def on_update_button_clicked(change):
+        for x in parameters:
+            x.slider.continuous_update = not x.slider.continuous_update
+
+    def on_reset_button_clicked(change):
+        minuit.reset()
+        minuit.values = original_values
+        minuit.limits = original_limits
+        for i, x in enumerate(parameters):
+            x.reset(minuit.values[i], minuit.limits[i])
+        OnParameterChange()()
+
+    class Parameter(widgets.HBox):
+        def __init__(self, minuit, par):
+            val = minuit.values[par]
+            vmin, vmax = minuit.limits[par]
+            step = _guess_initial_step(val, vmin, vmax)
+            vmin2 = vmin if np.isfinite(vmin) else val - 100 * step
+            vmax2 = vmax if np.isfinite(vmax) else val + 100 * step
+
+            tlabel = widgets.Label(par, layout=widgets.Layout(width=f"{longest_par}em"))
+
+            tmin = widgets.BoundedFloatText(
+                _round(vmin2),
+                min=_make_finite(vmin),
+                max=vmax2,
+                step=1e-1 * (vmax2 - vmin2),
+                layout=widgets.Layout(width="4.1em"),
+            )
+
+            tmax = widgets.BoundedFloatText(
+                _round(vmax2),
+                min=vmin2,
+                max=_make_finite(vmax),
+                step=1e-1 * (vmax2 - vmin2),
+                layout=widgets.Layout(width="4.1em"),
+            )
+
+            self.slider = widgets.FloatSlider(
+                val,
+                min=vmin2,
+                max=vmax2,
+                step=step,
+                continuous_update=True,
+                readout_format=".3g",
+                layout=widgets.Layout(min_width="50%"),
+            )
+            self.slider.observe(OnParameterChange(), "value")
+
+            def on_min_change(change):
+                self.slider.min = change["new"]
+                tmax.min = change["new"]
+                lim = minuit.limits[par]
+                minuit.limits[par] = (self.slider.min, lim[1])
+
+            def on_max_change(change):
+                self.slider.max = change["new"]
+                tmin.max = change["new"]
+                lim = minuit.limits[par]
+                minuit.limits[par] = (lim[0], self.slider.max)
+
+            tmin.observe(on_min_change, "value")
+            tmax.observe(on_max_change, "value")
+
+            self.fix = widgets.ToggleButton(
+                minuit.fixed[par],
+                description="Fix",
+                tooltip="Fix",
+                layout=widgets.Layout(width="3.1em"),
+            )
+
+            self.fit = widgets.ToggleButton(
+                False,
+                description="Fit",
+                tooltip="Fit",
+                layout=widgets.Layout(width="3.5em"),
+            )
+
+            def on_fix_toggled(change):
+                minuit.fixed[par] = change["new"]
+                if change["new"]:
+                    self.fit.value = False
+
+            def on_fit_toggled(change):
+                self.slider.disabled = change["new"]
+                if change["new"]:
+                    self.fix.value = False
+                OnParameterChange()()
+
+            self.fix.observe(on_fix_toggled, "value")
+            self.fit.observe(on_fit_toggled, "value")
+            super().__init__([tlabel, tmin, self.slider, tmax, self.fix, self.fit])
+
+        def reset(self, value, limits=None):
+            self.slider.unobserve_all("value")
+            self.slider.value = value
+            if limits:
+                self.slider.min, self.slider.max = limits
+            # Installing the observer actually triggers a notification,
+            # we skip it. See notes in OnParameterChange.
+            self.slider.observe(OnParameterChange(1), "value")
+
+    longest_par = max(len(par) for par in minuit.parameters)
+    parameters = [Parameter(minuit, par) for par in minuit.parameters]
+
+    button_layout = widgets.Layout(max_width="8em")
+
+    fit_button = widgets.Button(
+        description="Fit",
+        button_style="primary",
+        layout=button_layout,
+    )
+    fit_button.on_click(do_fit)
+
+    update_button = widgets.ToggleButton(
+        True,
+        description="Continuous",
+        layout=button_layout,
+    )
+    update_button.observe(on_update_button_clicked)
+
+    reset_button = widgets.Button(
+        description="Reset",
+        button_style="danger",
+        layout=button_layout,
+    )
+    reset_button.on_click(on_reset_button_clicked)
+
+    algo_choice = widgets.Dropdown(
+        options=["Migrad", "Scipy", "Simplex"],
+        value="Migrad",
+        layout=button_layout,
+    )
+
+    ui = widgets.VBox(
+        [
+            widgets.HBox([fit_button, update_button, reset_button, algo_choice]),
+            widgets.VBox(parameters),
+        ]
+    )
+    out = widgets.Output()
+    OnParameterChange()()
+    return widgets.HBox([out, ui])
+
+
+def _make_finite(x: float) -> float:
+    sign = -1 if x < 0 else 1
+    if abs(x) == np.inf:
+        return sign * sys.float_info.max
+    return x
+
+
+def _guess_initial_step(val: float, vmin: float, vmax: float) -> float:
+    if np.isfinite(vmin) and np.isfinite(vmax):
+        return 1e-2 * (vmax - vmin)
+    return 1e-2
+
+
+def _round(x: float) -> float:
+    return float(f"{x:.1g}")