def write_fit_report(result_entries: List[AnalysisResultData]) -> str:
"""A function that generates fit reports documentation from list of data.
Args:
result_entries: List of data entries.
Returns:
Documentation of fit reports.
"""
analysis_description = ""
def format_val(float_val: float) -> str:
if np.abs(float_val) < 1e-3 or np.abs(float_val) > 1e3:
return f"{float_val: .4e}"
return f"{float_val: .4g}"
for res in result_entries:
if isinstance(res.value, uncertainties.UFloat):
fitval = res.value
unit = res.extra.get("unit", None)
if unit:
# unit is defined. do detaching prefix, i.e. 1000 Hz -> 1 kHz
val, val_prefix = detach_prefix(fitval.nominal_value,
decimal=3)
val_unit = val_prefix + unit
value_repr = f"{val: .3g}"
# write error bar if it is finite value
if fitval.std_dev is not None and np.isfinite(fitval.std_dev):
# with stderr
err, err_prefix = detach_prefix(fitval.std_dev, decimal=3)
err_unit = err_prefix + unit
if val_unit == err_unit:
# same value scaling, same prefix
value_repr += f" \u00B1 {err: .2f} {val_unit}"
else:
# different value scaling, different prefix
value_repr += f" {val_unit} \u00B1 {err: .2f} {err_unit}"
else:
# without stderr, just append unit
value_repr += f" {val_unit}"
else:
# unit is not defined. raw value formatting is performed.
value_repr = format_val(fitval.nominal_value)
if np.isfinite(fitval.std_dev):
# with stderr
value_repr += f" \u00B1 {format_val(fitval.std_dev)}"
analysis_description += f"{res.name} = {value_repr}\n"
return analysis_description
def test_detach_prefix(self):
"""Test detach prefix from the value."""
ref_values = [
(1e12, (1.0, "T")),
(1e11, (100.0, "G")),
(1e10, (10.0, "G")),
(1e9, (1.0, "G")),
(1e8, (100.0, "M")),
(1e7, (10.0, "M")),
(1e6, (1.0, "M")),
(1e5, (100.0, "k")),
(1e4, (10.0, "k")),
(1e3, (1.0, "k")),
(100, (100.0, "")),
(10, (10.0, "")),
(1.0, (1.0, "")),
(0.1, (100.0, "m")),
(0.01, (10.0, "m")),
(1e-3, (1.0, "m")),
(1e-4, (100.0, "µ")),
(1e-5, (10.0, "µ")),
(1e-6, (1.0, "µ")),
(1e-7, (100.0, "n")),
(1e-8, (10.0, "n")),
(1e-9, (1.0, "n")),
(1e-10, (100.0, "p")),
(1e-11, (10.0, "p")),
(1e-12, (1.0, "p")),
]
for arg, ref_rets in ref_values:
self.assertTupleEqual(detach_prefix(arg), ref_rets)
def test_get_same_value_after_attach_detach(self, value: float):
"""Test if same value can be obtained."""
unit = "Hz"
for prefix in ["P", "T", "G", "k", "m", "µ", "n", "p", "f"]:
scaled_val = apply_prefix(value, prefix + unit)
test_val, ret_prefix = detach_prefix(scaled_val)
self.assertAlmostEqual(test_val, value)
self.assertEqual(prefix, ret_prefix)
def _format_val(value):
# Return value with unit with prefix, i.e. 1000 Hz -> 1 kHz.
if unit:
try:
val, val_prefix = detach_prefix(value, decimal=3)
except ValueError:
val = value
val_prefix = ""
return f"{val: .3g}", f" {val_prefix}{unit}"
if np.abs(value) < 1e-3 or np.abs(value) > 1e3:
return f"{value: .4e}", ""
return f"{value: .4g}", ""
def draw(
cls,
series_defs: List[SeriesDef],
raw_samples: List[CurveData],
fit_samples: List[CurveData],
tick_labels: Dict[str, str],
fit_data: FitData,
result_entries: List[AnalysisResultData],
style: Optional[PlotterStyle] = None,
axis: Optional["matplotlib.axes.Axes"] = None,
) -> "pyplot.Figure":
"""Create a fit result of all curves in the single canvas.
Args:
series_defs: List of definition for each curve.
raw_samples: List of raw sample data for each curve.
fit_samples: List of formatted sample data for each curve.
tick_labels: Dictionary of axis label information. Axis units and label for x and y
value should be explained.
fit_data: fit data generated by the analysis.
result_entries: List of analysis result data entries.
style: Optional. A configuration object to modify the appearance of the figure.
axis: Optional. A matplotlib Axis object.
Returns:
A matplotlib figure of the curve fit result.
"""
if axis is None:
axis = get_non_gui_ax()
# update image size to experiment default
figure = axis.get_figure()
figure.set_size_inches(*style.figsize)
else:
figure = axis.get_figure()
# draw all curves on the same canvas
for series_def, raw_samp, fit_samp in zip(series_defs, raw_samples,
fit_samples):
draw_single_curve_mpl(
axis=axis,
series_def=series_def,
raw_sample=raw_samp,
fit_sample=fit_samp,
fit_data=fit_data,
style=style,
)
# add legend
if len(series_defs) > 1:
axis.legend(loc=style.legend_loc)
# get axis scaling factor
for this_axis in ("x", "y"):
sub_axis = getattr(axis, this_axis + "axis")
unit = tick_labels[this_axis + "val_unit"]
label = tick_labels[this_axis + "label"]
if unit:
maxv = np.max(np.abs(sub_axis.get_data_interval()))
scaled_maxv, prefix = detach_prefix(maxv, decimal=3)
prefactor = scaled_maxv / maxv
# pylint: disable=cell-var-from-loop
sub_axis.set_major_formatter(
FuncFormatter(lambda x, p: f"{x * prefactor: .3g}"))
sub_axis.set_label_text(f"{label} [{prefix}{unit}]",
fontsize=style.axis_label_size)
else:
sub_axis.set_label_text(label, fontsize=style.axis_label_size)
axis.ticklabel_format(axis=this_axis,
style="sci",
scilimits=(-3, 3))
if tick_labels["xlim"]:
axis.set_xlim(tick_labels["xlim"])
if tick_labels["ylim"]:
axis.set_ylim(tick_labels["ylim"])
# write analysis report
if fit_data:
report_str = write_fit_report(result_entries)
report_str += r"Fit $\chi^2$ = " + f"{fit_data.reduced_chisq: .4g}"
report_handler = axis.text(
*style.fit_report_rpos,
report_str,
ha="center",
va="top",
size=style.fit_report_text_size,
transform=axis.transAxes,
)
bbox_props = dict(boxstyle="square, pad=0.3",
fc="white",
ec="black",
lw=1,
alpha=0.8)
report_handler.set_bbox(bbox_props)
axis.tick_params(labelsize=style.tick_label_size)
axis.grid(True)
return figure
def draw(
cls,
series_defs: List[SeriesDef],
raw_samples: List[CurveData],
fit_samples: List[CurveData],
tick_labels: Dict[str, str],
fit_data: FitData,
result_entries: List[AnalysisResultData],
style: Optional[PlotterStyle] = None,
axis: Optional["matplotlib.axes.Axes"] = None,
) -> "pyplot.Figure":
"""Create a fit result of all curves in the single canvas.
Args:
series_defs: List of definition for each curve.
raw_samples: List of raw sample data for each curve.
fit_samples: List of formatted sample data for each curve.
tick_labels: Dictionary of axis label information. Axis units and label for x and y
value should be explained.
fit_data: fit data generated by the analysis.
result_entries: List of analysis result data entries.
style: Optional. A configuration object to modify the appearance of the figure.
axis: Optional. A matplotlib Axis object.
Returns:
A matplotlib figure of the curve fit result.
"""
if axis is None:
axis = get_non_gui_ax()
# update image size to experiment default
figure = axis.get_figure()
figure.set_size_inches(*style.figsize)
else:
figure = axis.get_figure()
# get canvas number
n_subplots = max(series_def.canvas for series_def in series_defs) + 1
# use inset axis. this allows us to draw multiple canvases on a given single axis object
inset_ax_h = (1 - (0.05 * (n_subplots - 1))) / n_subplots
inset_axes = [
axis.inset_axes(
[
0, 1 -
(inset_ax_h + 0.05) * n_axis - inset_ax_h, 1, inset_ax_h
],
transform=axis.transAxes,
zorder=1,
) for n_axis in range(n_subplots)
]
# show x label only in the bottom canvas
for inset_axis in inset_axes[:-1]:
inset_axis.set_xticklabels([])
inset_axes[-1].get_shared_x_axes().join(*inset_axes)
# remove original axis frames
axis.spines.right.set_visible(False)
axis.spines.left.set_visible(False)
axis.spines.top.set_visible(False)
axis.spines.bottom.set_visible(False)
axis.set_xticks([])
axis.set_yticks([])
# collect data source per canvas
plot_map = defaultdict(list)
for curve_ind, series_def in enumerate(series_defs):
plot_map[series_def.canvas].append(curve_ind)
y_labels = tick_labels["ylabel"].split(",")
if len(y_labels) == 1:
y_labels = y_labels * n_subplots
for ax_ind, curve_inds in plot_map.items():
inset_axis = inset_axes[ax_ind]
for curve_ind in curve_inds:
draw_single_curve_mpl(
axis=inset_axis,
series_def=series_defs[curve_ind],
raw_sample=raw_samples[curve_ind],
fit_sample=fit_samples[curve_ind],
fit_data=fit_data,
style=style,
)
# add legend to each inset axis
if len(curve_inds) > 1:
inset_axis.legend(loc=style.legend_loc)
# format y axis tick value of each inset axis
yaxis = getattr(inset_axis, "yaxis")
unit = tick_labels["yval_unit"]
label = y_labels[ax_ind]
if unit:
maxv = np.max(np.abs(yaxis.get_data_interval()))
scaled_maxv, prefix = detach_prefix(maxv, decimal=3)
prefactor = scaled_maxv / maxv
# pylint: disable=cell-var-from-loop
yaxis.set_major_formatter(
FuncFormatter(lambda x, p: f"{x * prefactor: .3g}"))
yaxis.set_label_text(f"{label} [{prefix}{unit}]",
fontsize=style.axis_label_size)
else:
inset_axis.ticklabel_format(axis="y",
style="sci",
scilimits=(-3, 3))
yaxis.set_label_text(label, fontsize=style.axis_label_size)
if tick_labels["ylim"]:
inset_axis.set_ylim(tick_labels["ylim"])
# format x axis
xaxis = getattr(inset_axes[-1], "xaxis")
unit = tick_labels["xval_unit"]
label = tick_labels["xlabel"]
if unit:
maxv = np.max(np.abs(xaxis.get_data_interval()))
scaled_maxv, prefix = detach_prefix(maxv, decimal=3)
prefactor = scaled_maxv / maxv
# pylint: disable=cell-var-from-loop
xaxis.set_major_formatter(
FuncFormatter(lambda x, p: f"{x * prefactor: .3g}"))
xaxis.set_label_text(f"{label} [{prefix}{unit}]",
fontsize=style.axis_label_size)
else:
axis.ticklabel_format(axis="x", style="sci", scilimits=(-3, 3))
xaxis.set_label_text(label, fontsize=style.axis_label_size)
if tick_labels["xlim"]:
inset_axes[-1].set_xlim(tick_labels["xlim"])
# write analysis report
if fit_data:
report_str = write_fit_report(result_entries)
report_str += r"Fit $\chi^2$ = " + f"{fit_data.reduced_chisq: .4g}"
report_handler = axis.text(
*style.fit_report_rpos,
report_str,
ha="center",
va="top",
size=style.fit_report_text_size,
transform=axis.transAxes,
)
bbox_props = dict(boxstyle="square, pad=0.3",
fc="white",
ec="black",
lw=1,
alpha=0.8)
report_handler.set_bbox(bbox_props)
axis.tick_params(labelsize=style.tick_label_size)
axis.grid(True)
return figure
def format_canvas(self):
if self._axis.child_axes:
# Multi canvas mode
all_axes = self._axis.child_axes
else:
all_axes = [self._axis]
# Add data labels if there are multiple labels registered per sub_ax.
for sub_ax in all_axes:
_, labels = sub_ax.get_legend_handles_labels()
if len(labels) > 1:
sub_ax.legend()
# Format x and y axis
for ax_type in ("x", "y"):
# Get axis formatter from drawing options
if ax_type == "x":
lim = self.options.xlim
unit = self.options.xval_unit
else:
lim = self.options.ylim
unit = self.options.yval_unit
# Compute data range from auto scale
if not lim:
v0 = np.nan
v1 = np.nan
for sub_ax in all_axes:
if ax_type == "x":
this_v0, this_v1 = sub_ax.get_xlim()
else:
this_v0, this_v1 = sub_ax.get_ylim()
v0 = np.nanmin([v0, this_v0])
v1 = np.nanmax([v1, this_v1])
lim = (v0, v1)
# Format axis number notation
if unit:
# If value is specified, automatically scale axis magnitude
# and write prefix to axis label, i.e. 1e3 Hz -> 1 kHz
maxv = max(np.abs(lim[0]), np.abs(lim[1]))
try:
scaled_maxv, prefix = detach_prefix(maxv, decimal=3)
prefactor = scaled_maxv / maxv
except ValueError:
prefix = ""
prefactor = 1
formatter = MplCurveDrawer.PrefixFormatter(prefactor)
units_str = f" [{prefix}{unit}]"
else:
# Use scientific notation with 3 digits, 1000 -> 1e3
formatter = ScalarFormatter()
formatter.set_scientific(True)
formatter.set_powerlimits((-3, 3))
units_str = ""
for sub_ax in all_axes:
if ax_type == "x":
ax = getattr(sub_ax, "xaxis")
tick_labels = sub_ax.get_xticklabels()
else:
ax = getattr(sub_ax, "yaxis")
tick_labels = sub_ax.get_yticklabels()
if tick_labels:
# Set formatter only when tick labels exist
ax.set_major_formatter(formatter)
if units_str:
# Add units to label if both exist
label_txt_obj = ax.get_label()
label_str = label_txt_obj.get_text()
if label_str:
label_txt_obj.set_text(label_str + units_str)
# Auto-scale all axes to the first sub axis
if ax_type == "x":
all_axes[0].get_shared_x_axes().join(*all_axes)
all_axes[0].set_xlim(lim)
else:
all_axes[0].get_shared_y_axes().join(*all_axes)
all_axes[0].set_ylim(lim)
def test_detach_prefix_with_value_too_small(self):
"""Test detach prefix by input too small value."""
with self.assertRaises(Exception):
self.assertTupleEqual(detach_prefix(1e-20), (1e-20, ""))
def test_detach_prefix_with_negative(self):
"""Test detach prefix by input negative values."""
self.assertTupleEqual(detach_prefix(-1.234e7), (-12.34, "M"))
def test_detach_prefix_with_zero(self):
"""Test detach prefix by input zero."""
self.assertTupleEqual(detach_prefix(0.0), (0.0, ""))
def test_get_symbol_mu(self):
"""Test if µ is returned rather than u."""
_, prefix = detach_prefix(3e-6)
self.assertEqual(prefix, "µ")
