On the optimality of quantum circuit initial mapping using reinforcement learning

EPJ Quantum Technology

Table 3 Benchmark test circuits for the 20-qubit model

Benchmark	Logical CNOTs#	Δ CNOTs after mapping and routing				Improv %
Benchmark	Logical CNOTs#	Huang et al. [13]	Cheng et al. [9]	Zhu et al. [25]	RL	Improv %
clip_206	14,772	7959	9339	9444	7572	1.7
cm85a_209	4986	2706	3075	2907	2322	4.99
cycle10_2_110	2648	1293	1440	1611	1449	−3.99
dist_223	16,624	8082	10,668	10,569	7791	1.17
hwb6_56	2952	1143	1665	1740	1668	−12.82
hwb7_59	10,681	4950	5226	5619	5129	−1.14
hwb8_113	30,372	17,319	17,886	20,007	16,905	0.86
mlp4_245	8232	5511	4881	5862	4932	4.21
radd_250	1405	900	993	1065	834	2.86
rd73_252	2319	1323	1344	1272	1239	2.30
rd84_253	5960	3381	3912	3804	2940	4.72
root_255	7493	3531	4554	4164	3315	1.95
sao2_257	16,864	7812	11,268	9999	7527	1.15
sym10_262	28,084	13,833	16,899	17,079	14,340	−1.2
sym9_148	9408	3033	4200	3366	2306	5.55
sym9_193	15,232	8676	9252	9462	8496	0.75
Urf1_278	26,692	17,391	17,106	17,217	16,959	0.97
Urf2_277	10,066	6129	6168	6108	6805	−4.35
Urf5_280	23,764	13,791	15,261	15,201	13,845	−0.14
Geo. Mean		1	1.145	1.148	0.978
of Addit. Gates		0.873	1	1.002	0.854
of Addit. Gates		0.870	0.997	1	0.852
Avg. # of		6777	7638.7	7710.3	6651.2
Addit. Gates		±5347.24	±5753.6	±5979.35	±5330.34