Typo fix

docs/performance.rst

@@ -129,7 +129,7 @@ On random universal circuits, Qrack's optimized QUnit type appears to achieve ef
 
 .. image:: performance/random_universal.png
 
-This might be an extraordinary claim, that Qrack's optimization layer achieves apparently linear performance in the test domain. We should be clear that this is average performance, whereas worst case iterations in the spread of the benchmark tests still approach exponential resource usage. To validate that QUnit is actually reproducing the equivalent measurement distribution of the other libraries, we have implemented a cross entropy benchmark test between Qrack's QUnit and its Schrödinger method "QEngine" types. This test is available as "test_universal_circuit_digital_cross_entropy" in the vm6502q/qrack repository standard benchmark suit. This benchmark can be run on demand, and its width, depth, and iteration count parameters can be varied as desired, (by changing these constants declared in code). This is an example of an actual output from the benchmark:
+This might be an extraordinary claim, that Qrack's optimization layer achieves apparently linear performance in the test domain. We should be clear that this is average performance, whereas worst case iterations in the spread of the benchmark tests still approach exponential resource usage. To validate that QUnit is actually reproducing the equivalent measurement distribution of the other libraries, we have implemented a cross entropy benchmark test between Qrack's QUnit and its Schrödinger method "QEngine" types. This test is available as "test_universal_circuit_digital_cross_entropy" in the vm6502q/qrack repository standard benchmark suite. This benchmark can be run on demand, and its width, depth, and iteration count parameters can be varied as desired, (by changing these constants declared in code). This is an example of an actual output from the benchmark:
 
 .. code-block:: bash