Decrease subtitles by one level in gallery examples to get a coherent

gallery TOC

docs/gallery/plot_parallel_computing.py

@@ -9,7 +9,7 @@
 
 ##############################################################
 # Confirgure multithreading options of linear algebra library
-# ===========================================================
+# -----------------------------------------------------------
 #
 # Most of the low level computation/numerical optimization occuring during model
 # fitting with nrt relies on a linear algebra library. These libraries often implement
@@ -35,7 +35,7 @@
 
 ##############################################################
 # Create benchmark data
-# =====================
+# ---------------------
 #
 # Using the synthetic data generation functionalities of the package, we can create
 # an xarray DataArray for benchmark. Note that in order to keep the compilation time
@@ -57,7 +57,7 @@
 
 ##############################################################
 # Benchmark fitting time of all methods 
-# =====================================
+# -------------------------------------
 # 
 # Note that we are only interested in fitting time and therefore use a single
 # class instance for the benchmark. The time required for any subsequent .monitor()
@@ -117,7 +117,7 @@
 
 ##############################################################
 # Further considerations
-# ======================
+# ----------------------
 #
 # A deployment at scale may involve several levels of parallelization. The multi-threaded
 # example illustrated above is made possible thanks to the numba parallel accelerator.

docs/gallery/plot_simulation_data.py

@@ -8,7 +8,7 @@
 
 #############################################################
 # Synthetic data generation
-# =========================
+# -------------------------
 #
 # The data module of the nrt package contains functionalities to create synthetic
 # data with controlled parameters such as position of structural change, phenological
@@ -70,7 +70,7 @@
 
 ###################################################################
 # Simulation of an NRT monitoring scenario
-# ==================================================
+# --------------------------------------------------
 # For the simulating a near real time monitoring scenario, we consider all the pixels
 # of the datacube (no mask) and define the 2018-01-01 to 2019-12-31 period as the
 # stable history period and all subsequent dates as monitoring. We know from the
@@ -109,7 +109,7 @@
 
 ############################################################################
 # Monitoring performances evaluation
-# ==================================
+# ----------------------------------
 # Assessing the performance of a time-series monitoring algorithm can be a complex
 # task that depends on the specific use case and what the user wants to emphasize.
 # A user valuing rapid detection will chose an assessment approach that puts extra
@@ -156,7 +156,7 @@ def accuracy(nrtInstance, params_ds, dates, delta=180):
 
 ####################################################################
 # White noise sensitivity analysis
-# ================================
+# --------------------------------
 # To go one step further we can assess and visualize how these accuracy measures
 # vary with the amount of noise present in the synthetic data.
 # For that we define a new function encompassing all the steps of data generation,