25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268 | def plot_lightcurve(
source: Source,
vs_abs_min: float = 4.3,
m_abs_min: float = 0.26,
use_peak_flux: bool = True,
) -> Row:
"""Create the lightcurve and 2-epoch metric graph for a source with Bokeh.
Args:
source (Source): Source object.
vs_abs_min (float, optional): MeasurementPair objects with an absolute vs metric
greater than `vs_abs_min` and m metric greater than `m_abs_min` will be connected
in the metric graph. Defaults to 4.3.
m_abs_min (float, optional): See `vs_abs_min`. Defaults to 0.26.
use_peak_flux (bool, optional): If True, use peak fluxes, otherwise use integrated
fluxes. Defaults to True.
Returns:
Row: Bokeh Row layout object containing the lightcurve and graph plots.
"""
PLOT_WIDTH = 800
PLOT_HEIGHT = 300
flux_column = "flux_peak" if use_peak_flux else "flux_int"
metric_suffix = "peak" if use_peak_flux else "int"
measurements_qs = (
Measurement.objects.filter(source__id=source.id)
.annotate(
taustart_ts=F("image__datetime"),
flux=F(flux_column),
flux_err_lower=F(flux_column) - F(f"{flux_column}_err"),
flux_err_upper=F(flux_column) + F(f"{flux_column}_err"),
)
.values(
"id",
"pk",
"taustart_ts",
"flux",
"flux_err_upper",
"flux_err_lower",
"forced",
)
.order_by("taustart_ts")
)
candidate_measurement_pairs_qs = (
source.measurementpair_set.annotate(
m_abs=Abs(f"m_{metric_suffix}"), vs_abs=Abs(f"vs_{metric_suffix}")
)
.filter(vs_abs__gte=vs_abs_min, m_abs__gte=m_abs_min)
.values("measurement_a_id", "measurement_b_id", "vs_abs", "m_abs")
)
candidate_measurement_pairs_df = pd.DataFrame(candidate_measurement_pairs_qs)
# lightcurve required cols: taustart_ts, flux, flux_err_upper, flux_err_lower, forced
lightcurve = pd.DataFrame(measurements_qs)
# remap method values to labels to make a better legend
lightcurve["method"] = lightcurve.forced.map({True: "Forced", False: "Selavy"})
source = ColumnDataSource(lightcurve)
method_mapper = factor_cmap(
"method", palette="Colorblind3", factors=["Selavy", "Forced"]
)
min_y = min(0, lightcurve.flux_err_lower.min())
max_y = lightcurve.flux_err_upper.max()
y_padding = (max_y - min_y) * 0.1
fig_lc = figure(
plot_width=PLOT_WIDTH,
plot_height=PLOT_HEIGHT,
sizing_mode="stretch_width",
x_axis_type="datetime",
x_range=DataRange1d(default_span=timedelta(days=1)),
y_range=DataRange1d(start=min_y, end=max_y + y_padding),
)
# line source must be a COPY of the data for the scatter source for the hover and
# selection to work properly, using the same ColumnDataSource will break it
fig_lc.line("taustart_ts", "flux", source=lightcurve)
lc_scatter = fig_lc.scatter(
"taustart_ts",
"flux",
marker="circle",
size=6,
color=method_mapper,
nonselection_color=method_mapper,
selection_color="red",
nonselection_alpha=1.0,
hover_color="red",
alpha=1.0,
source=source,
legend_group="method",
)
fig_lc.add_layout(
Whisker(
base="taustart_ts",
upper="flux_err_upper",
lower="flux_err_lower",
source=source,
)
)
fig_lc.xaxis.axis_label = "Datetime"
fig_lc.xaxis[0].formatter = DatetimeTickFormatter(days="%F", hours='%H:%M')
fig_lc.yaxis.axis_label = (
"Peak flux (mJy/beam)" if use_peak_flux else "Integrated flux (mJy)"
)
# determine legend location: either bottom_left or top_left
legend_location = (
"top_left"
if lightcurve.sort_values("taustart_ts").iloc[0].flux < (max_y - min_y) / 2
else "bottom_left"
)
fig_lc.legend.location = legend_location
# TODO add vs and m metrics to graph edges
# create plot
fig_graph = figure(
plot_width=PLOT_HEIGHT,
plot_height=PLOT_HEIGHT,
x_range=Range1d(-1.1, 1.1),
y_range=Range1d(-1.1, 1.1),
x_axis_type=None,
y_axis_type=None,
sizing_mode="fixed",
)
hover_tool_lc_callback = None
if len(candidate_measurement_pairs_df) > 0:
g = nx.Graph()
for _row in candidate_measurement_pairs_df.itertuples(index=False):
g.add_edge(_row.measurement_a_id, _row.measurement_b_id)
node_layout = nx.circular_layout(g, scale=1, center=(0, 0))
# add node positions to dataframe
for suffix in ["a", "b"]:
pos_df = pd.DataFrame(
candidate_measurement_pairs_df[f"measurement_{suffix}_id"]
.map(node_layout)
.to_list(),
columns=[f"measurement_{suffix}_x", f"measurement_{suffix}_y"],
)
candidate_measurement_pairs_df = candidate_measurement_pairs_df.join(pos_df)
candidate_measurement_pairs_df["measurement_x"] = list(
zip(
candidate_measurement_pairs_df.measurement_a_x.values,
candidate_measurement_pairs_df.measurement_b_x.values,
)
)
candidate_measurement_pairs_df["measurement_y"] = list(
zip(
candidate_measurement_pairs_df.measurement_a_y.values,
candidate_measurement_pairs_df.measurement_b_y.values,
)
)
node_positions_df = pd.DataFrame.from_dict(
node_layout, orient="index", columns=["x", "y"]
)
node_positions_df["lc_index"] = node_positions_df.index.map(
{v: k for k, v in lightcurve.id.to_dict().items()}
).values
node_source = ColumnDataSource(node_positions_df)
edge_source = ColumnDataSource(candidate_measurement_pairs_df)
# add edges to plot
edge_renderer = fig_graph.multi_line(
"measurement_x",
"measurement_y",
line_width=5,
hover_color="red",
source=edge_source,
name="edges",
)
# add nodes to plot
node_renderer = fig_graph.circle(
"x",
"y",
size=20,
hover_color="red",
selection_color="red",
nonselection_alpha=1.0,
source=node_source,
name="nodes",
)
# create hover tool for node edges
edge_callback_code = """
// get edge index
let indices_a = cb_data.index.indices.map(i => edge_data.data.measurement_a_id[i]);
let indices_b = cb_data.index.indices.map(i => edge_data.data.measurement_b_id[i]);
let indices = indices_a.concat(indices_b);
let lightcurve_indices = indices.map(i => lightcurve_data.data.id.indexOf(i));
lightcurve_data.selected.indices = lightcurve_indices;
"""
hover_tool_edges = HoverTool(
tooltips=None,
renderers=[edge_renderer],
callback=CustomJS(
args={
"lightcurve_data": lc_scatter.data_source,
"edge_data": edge_renderer.data_source,
},
code=edge_callback_code,
),
)
fig_graph.add_tools(hover_tool_edges)
# create labels for nodes
graph_source = ColumnDataSource(node_positions_df)
labels = LabelSet(
x="x",
y="y",
text="lc_index",
source=graph_source,
text_align="center",
text_baseline="middle",
text_font_size="1em",
text_color="white",
)
fig_graph.renderers.append(labels)
# prepare a JS callback for the lightcurve hover tool to mark the associated nodes
hover_tool_lc_callback = CustomJS(
args={
"node_data": node_renderer.data_source,
"lightcurve_data": lc_scatter.data_source,
},
code="""
let ids = cb_data.index.indices.map(i => lightcurve_data.data.id[i]);
let node_indices = ids.map(i => node_data.data.index.indexOf(i));
node_data.selected.indices = node_indices;
""",
)
# create hover tool for lightcurve
hover_tool_lc = HoverTool(
tooltips=[
("Index", "@index"),
("Date", "@taustart_ts{%F}"),
(f"Flux {metric_suffix}", "@flux mJy"),
],
formatters={"@taustart_ts": "datetime", },
mode="vline",
callback=hover_tool_lc_callback,
)
fig_lc.add_tools(hover_tool_lc)
plot_row = row(fig_lc, fig_graph, sizing_mode="stretch_width")
plot_row.css_classes.append("mx-auto")
return plot_row
|