v0.12.0.md

@@ -112,6 +112,48 @@ As you can see the extracted samples match the original.
 
 * https://git.ligo.org/lscsoft/pesummary/-/merge_requests/518
 
+## Review tests
+
+To test the `summarycombine_posteriors` executable, we test that the output from `summarycombine_posteriors` matches the output from `cbcBayesCombinePosteriors`. First we extract the `PublicationSamples` dataset from the public GW190814.h5 file:
+
+```bash
+summaryextract --outdir ./ --samples GW190814.h5 --filename extracted_PublicationSamples.dat --file_format dat --label PublicationSamples
+```
+
+We then run the following:
+
+```python
+summarycombine_posteriors --outdir ./ --samples extracted_PublicationSamples.dat extracted_PublicationSamples.dat --weights 0.2 0.8 --filename pesummary_combined.dat --labels one two
+cbcBayesCombinePosteriors -p extracted_PublicationSamples.dat -p extracted_PublicationSamples.dat -o lal_combined.dat -w 0.2 -w 0.8
+```
+
+We then compare the contents with:
+
+```python
+import numpy as np
+
+pesummary_generated = np.genfromtxt("extracted_PublicationSamples.dat", names=True)
+lal_generated = np.genfromtxt("lal_combined.dat", names=True)
+assert all(param in pesummary_generated.dtype.names for param in lal_generated.dtype.names)
+for param in lal_generated.dtype.names:
+    np.testing.assert_almost_equal(
+        lal_generated[param], pesummary_generated[param]
+    )
+```
+
+Note, we made a small edit to both executables to ensure both use the same random seed. The diffs can be seen below:
+
+```bash
+diff cbcBayesCombinePosteriors cbcBayesCombinePosteriors_original
+38,39d37
+< np.random.seed(123456789)
+<
+diff summarycombine_posteriors.py summarycombine_posteriors_original.py
+18,19d17
+< import numpy as np
+< np.random.seed(123456789)
+```
+
 ## Reviewer comments
 
 # `summarytgr` :heavy_check_mark: