On the optimality of quantum circuit initial mapping using reinforcement learning

EPJ Quantum Technology

Table 4 Time comparison in seconds between heuristics approach and RL agent inference per circuit

Benchmark	Qubits #	Logical Depth #	Logical CNOTs#	Time Per Circuit in Seconds
Benchmark	Qubits #	Logical Depth #	Logical CNOTs#	Cheng et al. [9]	RL Without Routing	RL With Routing
clip_206	14	12,028	14,772	6.15	5.56	53.31
cm85a_209	14	4256	4986	1.94	1.96	20.05
cycle10_2_110	12	2276	2648	1.08	1.11	11.62
dist_223	13	13,274	16,624	6.77	6.61	56.55
hwb6_56	7	2559	2952	1.32	1.85	10.9
hwb7_59	8	9112	10,681	4.4	4.77	37.04
hwb8_113	9	26,041	30,372	12.38	11.21	108.66
mlp4_245	16	6930	8232	4.02	3.33	34.02
radd_250	13	1210	1405	0.64	0.55	5.9
rd73_252	10	1963	2319	1	0.77	8.4
rd84_253	12	4917	5960	2.69	2.46	22.09
root_255	13	5965	7493	3.02	3.33	26.39
sao2_257	14	13,209	16,864	7.6	6.24	56.19
sym10_262	12	23,736	28,084	12.9	11.59	102.45
sym9_148	10	8062	9408	4.02	3.66	28.84
sym9_193	11	12,849	15,232	6.11	5.63	50.17
Urf1_278	9	22,307	26,692	10.5	10.1	96.32
Urf2_277	8	8312	10,066	3.85	4.18	39.31
Urf5_280	9	19,888	23,764	9.25	9.43	94.7