pyrene+CH3 C17H12

A2_2b693cf6e7 · SMILES Cc1cc2ccc3cccc4ccc(c1)c2c34 · InChIKey VIRFPLJXRDHVEI-UHFFFAOYSA-N

Cheap level done

analytic Hessian

finite-difference Hessian was noisy → release uses the analytic Hessian. The finite-difference Hessian and the analytic Hessian of the same geometry differ by more than 15 cm⁻¹ on at least one frequency; the numerical route is the noisy one. In the released training set this molecule carries the analytic Hessian instead of the finite-difference one.

imaginary mode healed

imaginary mode healed by the analytic Hessian. The finite-difference Hessian showed an imaginary frequency that the analytic Hessian does not: numerical noise, not a saddle point.

Harmonic frequencies at the cheap level (deck v1: B3LYP and ωB97X, 6-31G*, at the B3LYP geometry)

81 vibrational modes; imaginary modes negative. Shift = ωB97X − B3LYP.

Band heights are not computed for this molecule yet (no dipole derivatives); positions only, at unit height. High wavenumbers on the left; the table below is the same data.

#	B3LYP (cm⁻¹)	ωB97X (cm⁻¹)	shift
1	32.5	-41.6	-74.1
2	75.9	85.0	+9.1
3	155.3	173.0	+17.7
4	176.9	181.3	+4.3
5	223.7	228.7	+5.0
6	242.8	244.7	+1.8
7	253.1	260.6	+7.6
8	331.3	342.3	+11.0
9	379.6	381.5	+1.9
10	400.7	405.3	+4.6
11	404.5	419.3	+14.8
12	474.9	477.9	+3.1
13	505.6	510.3	+4.7
14	509.1	514.1	+5.0
15	513.5	525.0	+11.4
16	537.0	548.5	+11.4
17	544.8	549.9	+5.1
18	569.3	571.6	+2.4
19	570.4	575.6	+5.2
20	601.7	616.0	+14.3
21	621.8	626.9	+5.1
22	690.7	703.8	+13.0
23	727.0	739.5	+12.5
24	769.0	773.5	+4.5
25	772.3	777.3	+5.0
26	772.5	784.2	+11.7
27	820.2	833.1	+12.9
28	823.8	836.6	+12.8
29	830.9	839.3	+8.4
30	852.0	867.7	+15.8
31	888.3	907.2	+19.0
32	895.5	923.2	+27.6
33	913.5	940.3	+26.8
34	966.0	990.6	+24.6
35	972.0	994.1	+22.1
36	978.9	998.9	+20.0
37	980.7	1007.3	+26.5
38	991.3	1013.4	+22.1
39	1012.9	1022.5	+9.6
40	1045.0	1047.8	+2.8
41	1074.6	1081.7	+7.1
42	1110.0	1116.9	+6.9
43	1138.8	1141.4	+2.6
44	1169.1	1157.7	-11.4
45	1186.7	1172.6	-14.1
46	1205.9	1207.3	+1.5
47	1212.5	1213.3	+0.8
48	1241.0	1239.8	-1.2
49	1273.4	1272.9	-0.5
50	1276.9	1285.1	+8.2
51	1313.6	1318.3	+4.7
52	1359.4	1328.7	-30.7
53	1364.2	1376.1	+11.9
54	1416.4	1398.3	-18.2
55	1426.4	1431.3	+4.9
56	1442.5	1451.4	+8.9
57	1447.6	1470.3	+22.7
58	1463.9	1476.4	+12.5
59	1482.2	1497.6	+15.5
60	1499.7	1520.4	+20.7
61	1502.3	1521.4	+19.2
62	1519.0	1527.5	+8.5
63	1531.1	1556.0	+24.9
64	1546.6	1571.0	+24.4
65	1623.3	1647.4	+24.1
66	1636.0	1678.5	+42.5
67	1656.3	1686.6	+30.3
68	1659.2	1694.9	+35.7
69	1685.0	1733.6	+48.6
70	3041.3	3076.9	+35.6
71	3098.4	3147.4	+48.9
72	3125.1	3172.7	+47.6
73	3170.6	3215.4	+44.9
74	3171.7	3216.5	+44.8
75	3174.7	3219.0	+44.3
76	3175.8	3219.6	+43.8
77	3178.1	3221.3	+43.2
78	3185.2	3229.1	+43.9
79	3193.9	3237.2	+43.3
80	3196.2	3239.7	+43.5
81	3205.4	3248.2	+42.7

Structure

2D depiction from the SMILES.

Status ladder

Listed
Cheap level done — 2026-09-22, ubuntu-32gb-hel1-9
Correction predicted — not yet: the learned layer is training
Spectrum predicted — not yet: the learned layer is training
Anchored
Validated

Request a run

Not available on this site yet: the pipeline (a molecule in, its spectrum and certificate out) runs on the project's own computer, with three of the four mode families licensed at the DFT stand-in level. Sign-in will be needed only to request runs; reading stays open.

Notify me when this molecule climbs the ladder: open a GitHub issue — you are subscribed to it and we close it with a link when the page changes. No e-mail address is stored by this site.

Second route

Analytic Hessian (pyscf, grid 99/590) against the corpus finite differences: b3lyp max |Δω| 6 cm⁻¹; wb97x max |Δω| 110 cm⁻¹.

Provenance

Releases: layerA2_2026-09-24
Entered the corpus as: substituted three- and four-ring cores, enumerated from the parents (layer A′) (layer A2)
Run: ubuntu-32gb-hel1-9, 2026-09-22 08:44:11 → 2026-09-22 10:56:57, 7966 s, peak 7.38 GB, deck 97ed2d67c66b
Imaginary modes: B3LYP 0, ωB97X 1
Source files: corpus factory · manifest