Update datasets PIG

docs/development/pigs/pig-008.rst

@@ -82,16 +82,13 @@ The `MapDataset` implements at least the following methods:
 
     dataset.npred()
     dataset.likelihood_per_bin()
-    dataset.likelihood(mask)
+    dataset.likelihood()
 
 The `.setup()` method creates cutouts of the exposure map for the individual
 model components, assigns the correspoding IRFs to the model component and
-bundles all into `SourceIRFModel` or `ModelEvaluator()` classes. The list of model
-evaluators is cached on the `MapDataset` and used later to efficiently compute
-the predicted number of counts.
-
-The `mask` argument to the `.likelihood` method allows to set a global fitting
-mask, when mutiple `MapDataset` are fitted jointly.
+bundles all into `MapEvaluator()` object. The list of model evaluators is
+cached on the `MapDataset` and used later to efficiently compute the predicted
+number of counts.
 
 
 `MapDatasetOnOff`
@@ -112,7 +109,7 @@ For on-off based analyses a `MapDatasetOnOff` class could be introduced.
     psf = Map.read("stacked/psf.fits")
 
     dataset_onoff = MapDatasetOnOff(
-        counts_on=counts_on,
+        counts=counts,
         counts_off=counts_off,
         alpha=alpha,
         edisp=edisp,
@@ -155,37 +152,43 @@ For spectral analysis we propose to introduce a `SpectrumDataset`:
 
     dataset.npred()
     dataset.likelihoop_per_bin()
-    dataset.likelihood(mask)
+    dataset.likelihood()
 
-The `SpectrumDatasets` with a parametric background model will be introduced first.
+The `SpectrumDataset` with a parametric background model will be introduced first.
 
 
 `SpectrumDatasetOnOff`
 ----------------------
 
-For ON / OFF based spectral analysis we propose to introduce a `SpectrumDatasetOnOff`:
+For ON / OFF based spectral analysis we propose to refactor the existing `SpectrumObservation`
+into a`SpectrumDatasetsOnOff` class:
 
 .. code ::
 
     from gammapy.spectrum import SpectrumDatasetOnOff
 
-    model = SpectralModel()
+    model = PowerLaw()
     edisp = EnergyDispersion.read()
 
-    counts = Spectrum.read()
+    counts = CountsSpectrum.read()
+    counts_off = CountsSpectrum.read()
 
-    dataset = SpectrumDataset(
+    aeff = EffectiveAreaTable.read()
+
+    dataset = SpectrumDatasetOnOff(
         counts=counts,
+        counts_off=counts_off,
         model=model,
-        background_model=background_model,
         edisp=edisp,
+        livetime="1 h".
+        aeff=aeff,
         )
 
     dataset.npred()
     dataset.likelihoop_per_bin()
     dataset.likelihood()
 
-The existing `SpectrumObservation` will be refactored into the `SpectrumDatasetsOnOff` class.
+.
 
 
 `FluxPointsDataset`
@@ -215,7 +218,7 @@ The `FluxPointsDataset` should define at least the following methods:
     dataset = FluxPointsDataset()
 
     dataset.likelihood_per_bin()
-    dataset.likelihood(mask)
+    dataset.likelihood()
 
     # predicted flux form the spectral model
     dataset.flux_pred()
@@ -256,7 +259,7 @@ is supposed to work:
     # check whether all contained datasets are of the same type
     datasets.is_all_same_type
 
-    datasets.likelihood(mask)
+    datasets.likelihood()
 
     fit = Fit(datasets)
 
@@ -266,7 +269,6 @@ is supposed to work:
     fit.optimize()
 
 
-
 The linking of parameters of the spectral model is natively achieved, as the same
 instance of the spectral model is passed to two different datasets, while the
 background model is created for every dataset separately.
@@ -344,6 +346,9 @@ dataset. Here is an example:
         background-model: "obs-123/background.fits"
         model: "model.yaml" # optionally
 
+The `SpectrumDatasetOnOff` in addition implemnets serialisation to the standard
+OGIP format described on the gadf_pha_ page:
+
 
 Addtionally one could introduce a single FITS file serializiaton for quickly writing /
 reading datasets to disk:
@@ -503,4 +508,5 @@ Decision
 
 
 .. _gammapy: https://github.com/gammapy/gammapy
-.. _gammapy-web: https://github.com/gammapy/gammapy-webpage
+.. _gammapy-web: https://github.com/gammapy/gammapy-webpage
+.. _gadf_pha: https://gamma-astro-data-formats.readthedocs.io/en/latest/spectra/ogip/index.html