fix rb_analysis (#1457)

### Summary Fixes a counting error in rb_analysis because it doesn't take into account the inverse
qiskit-community · May 28, 2024 · d7df5f0 · d7df5f0
1 parent ce485f8
commit d7df5f0
Show file tree

Hide file tree

Showing 2 changed files with 8 additions and 1 deletion.
diff --git a/qiskit_experiments/library/randomized_benchmarking/rb_analysis.py b/qiskit_experiments/library/randomized_benchmarking/rb_analysis.py
@@ -244,7 +244,7 @@ def _initialize(
                         "This analysis cannot compute error per gates. "
                         "Please disable this with 'gate_error_ratio=False'."
                     ) from ex
-                nclif = circ_result["metadata"]["xval"]
+                nclif = circ_result["metadata"]["xval"] + 1
                 for (qinds, gate), count in count_ops:
                     formatted_key = tuple(sorted(qinds)), gate
                     avg_gpc[formatted_key] += count / nclif / n_circs

diff --git a/releasenotes/notes/fix-epg-gatecount-60777f7a3f3566bc.yaml b/releasenotes/notes/fix-epg-gatecount-60777f7a3f3566bc.yaml
@@ -0,0 +1,7 @@
+---
+fixes:
+  - |
+    The gate counting for epg in the RB analysis code was not including the 
+    inverse so that the total number of operations per Clifford was incorrect
+    leading to incorrect reporting of EPG from EPC. Fixed by adding +1 for the
+    inverse gate