Archive

The 2014 archive is the public OptBench site at the commit named in the oracle. Chill, Stevenson, Ruehle, Shang, Xiao, Farrell, Wales, and Henkelman set those problems in J. Chem. Theory Comput. 2014, 10, 5476-5482 (doi:10.1021/ct5008718). A run in that archive counts as finished when it stops under 10000 force calls.

Each problem below leads with its lowest-cost row. The full tables, including archived rows, are on the problem page.

Each problem leads with its lowest-cost row. The full tables are on the problem page. Archived rows sit in those tables.

Minimization

Lennard-Jones 38 clusters

Newtons Method maxstep=0.25

PAR-279

Row	Family	Family rank	Unit	PAR-2	Published	Median calls	Mean calls	Hessians	FD calls	Failures	Runs
Newtons Method maxstep=0.25	Minimizers	1	force calls	79	79		79	0	0	0 failed	1000

A run short of the full set, or a run that is not the stock method, stays off the ranked table.

Files with no per-start harness record stay off the ranked table.

Family tables, overall table, and performance profile.

Lennard-Jones 38 clusters with Gaussian force noise

rgmin (0.4.1) · DOGLEG

PAR-281.881

Row	Family	Family rank	Unit	PAR-2	Published	Median calls	Mean calls	Hessians	FD calls	Failures	Runs
rgmin (0.4.1) · DOGLEG	Minimizers	1	force calls	81.881		79	81.881	80.881	9220.43	crash 0, divergence 0, cap 0, wrong_stationary 0	1000

A run short of the full set, or a run that is not the stock method, stays off the ranked table.

Family tables, overall table, and performance profile.

Morse bulk

rgmin (0.4.1) · NEWTON

PAR-28.35

Row	Family	Family rank	Unit	PAR-2	Published	Median calls	Mean calls	Hessians	FD calls	Failures	Runs
rgmin (0.4.1) · NEWTON	Minimizers	1	force calls	8.35		9	8.35	7.35		crash 0, divergence 0, cap 0, wrong_stationary 0	100

A run short of the full set, or a run that is not the stock method, stays off the ranked table.

Files with no per-start harness record stay off the ranked table.

Family tables, overall table, and performance profile.

Saddle search

Pt heptamer island

Sella (2.6.0) · Sella

PAR-25201.62

Row	Family	Family rank	Unit	PAR-2	Published	Median calls	Mean calls	Hessians	FD calls	Failures	Runs
Sella (2.6.0) · Sella	Single-ended saddle	1	force calls per search	5201.62		225	247.812	0	0	crash 0, divergence 0, cap 0, wrong_stationary 4	16

Eon rows are not yet on the harness counter.

A run short of the full set, or a run that is not the stock method, stays off the ranked table.

Family tables, overall table, and performance profile.

Double-ended saddle search

LJ38 double-ended saddle

OPTIM · DNEB + hybrid eigenvector following

PAR-2131

Row	Family	Family rank	Unit	PAR-2	Published	Median calls	Mean calls	Per-image mean	Hessians	FD calls	Failures	Runs
OPTIM · DNEB + hybrid eigenvector following	Connect / double-ended	1	force calls	131	131	120	131		0	0	0 failed	50

eOn band rows are not yet on the harness counter.

A run short of the full set, or a run that is not the stock method, stays off the ranked table.

Files with no per-start harness record stay off the ranked table.

Family tables, overall table, and performance profile.

Lennard-Jones 38 transition-state separation

Eon svn (r2549) · Lanczos improved L-BFGS

PAR-2502.784

Row	Family	Family rank	Unit	PAR-2	Published	Median calls	Mean calls	Per-image mean	Hessians	FD calls	Failures	Runs
Eon svn (r2549) · Lanczos improved L-BFGS	Minimum mode	1	force calls	502.784		418	502.784	100.557	0	0	crash 0, divergence 0, cap 0, wrong_stationary 0	51

A run short of the full set, or a run that is not the stock method, stays off the ranked table.

Family tables, overall table, and performance profile.

Pt heptamer island

OPTIM · DNEB + hybrid eigenvector following

PAR-2103

Row	Family	Family rank	Unit	PAR-2	Published	Median calls	Mean calls	Per-image mean	Hessians	FD calls	Failures	Runs
OPTIM · DNEB + hybrid eigenvector following	Connect / double-ended	1	force calls	103	103	98	103		0	0	0 failed	59

eOn band rows are not yet on the harness counter.

A run short of the full set, or a run that is not the stock method, stays off the ranked table.

Family tables, overall table, and performance profile.

Global optimization

Binary Lennard-Jones 100

GMIN · basin-hopping

Published-589.507

Row	Family	Family rank	Unit	PAR-2	Published	Median calls	Mean calls	Hessians	FD calls	Failures	Runs
GMIN · basin-hopping	Basin hopping / global	1	lowest energy (epsilon)		-589.507			0	0	not recorded	100

Family tables, overall table, and performance profile.

Lennard-Jones 38 global minimum

GMIN (df8da599) · basin-hopping with continuous symmetry measures

PAR-25883.83

Row	Family	Family rank	Unit	PAR-2	Published	Median calls	Mean calls	Hessians	FD calls	Failures	Runs
GMIN (df8da599) · basin-hopping with continuous symmetry measures	Basin hopping / global	1	force calls	5883.83		3980.5	5883.83	0	0	crash 0, divergence 0, cap 0, wrong_stationary 0	100

Family tables, overall table, and performance profile.

Lennard-Jones 150

GMIN (df8da599) · basin-hopping

PAR-21.75804e+06

Row	Family	Family rank	Unit	PAR-2	Published	Median calls	Mean calls	Hessians	FD calls	Failures	Runs
GMIN (df8da599) · basin-hopping	Basin hopping / global	1	force calls	1.75804e+06		1.14412e+06	1.65804e+06	0	0	crash 0, divergence 0, cap 1, wrong_stationary 0	100

Family tables, overall table, and performance profile.

Lennard-Jones 75

GMIN · basin-hopping with symmetrise csm moves

PAR-260692

Row	Family	Family rank	Unit	PAR-2	Published	Median calls	Mean calls	Hessians	FD calls	Failures	Runs
GMIN · basin-hopping with symmetrise csm moves	Basin hopping / global	1	force calls	60692	60692	38118	60692	0	0	0 failed	200

A run short of the full set, or a run that is not the stock method, stays off the ranked table.

Family tables, overall table, and performance profile.

Lennard-Jones 98

GMIN (df8da599) · basin-hopping with core-orbit symmetrisation

PAR-2282815

Row	Family	Family rank	Unit	PAR-2	Published	Median calls	Mean calls	Hessians	FD calls	Failures	Runs
GMIN (df8da599) · basin-hopping with core-orbit symmetrisation	Basin hopping / global	1	force calls	282815		173058	282815	0	0	crash 0, divergence 0, cap 0, wrong_stationary 0	100

Family tables, overall table, and performance profile.

TIP4P 10

GMIN · linear temperature basin-paving

PAR-216099

Row	Family	Family rank	Unit	PAR-2	Published	Median calls	Mean calls	Hessians	FD calls	Failures	Runs
GMIN · linear temperature basin-paving	Basin hopping / global	1	force calls	16099	16099	11387	16099	0	0	0 failed	10000

Family tables, overall table, and performance profile.

TIP4P 15

GMIN · linear temperature basin-paving

PAR-2625362

Row	Family	Family rank	Unit	PAR-2	Published	Median calls	Mean calls	Hessians	FD calls	Failures	Runs
GMIN · linear temperature basin-paving	Basin hopping / global	1	force calls	625362	625362	416929	625362	0	0	0 failed	500

Family tables, overall table, and performance profile.

TIP4P 20

GMIN · linear temperature basin-paving

PAR-22.55667e+07

Row	Family	Family rank	Unit	PAR-2	Published	Median calls	Mean calls	Hessians	FD calls	Failures	Runs
GMIN · linear temperature basin-paving	Basin hopping / global	1	force calls	2.55667e+07	2.55667e+07	1.81821e+07	2.55667e+07	0	0	0 failed	100

Family tables, overall table, and performance profile.

Lowest eigenvector

Lennard-Jones 38

Sella (0.1.1) · lowest eigenvector

PAR-217.8

Row	Family	Family rank	Unit	PAR-2	Published	Median calls	Mean calls	Hessians	FD calls	Failures	Runs
Sella (0.1.1) · lowest eigenvector	Minimum mode	1	force calls	17.8		17	17.8	0	0	crash 0, divergence 0, cap 0, wrong_stationary 0	200

A run short of the full set, or a run that is not the stock method, stays off the ranked table.

Family tables, overall table, and performance profile.

Transition state optimization

H2 dissociation on a fixed Cu substrate

Row	Family	Family rank	Unit	PAR-2	Published	Median calls	Mean calls	Hessians	FD calls	Failures	Runs
No ranked row.

No run is committed for this problem.

Family tables, overall table, and performance profile.

Lennard-Jones 38

Sella (2.6.0) · Sella

PAR-268.74

Row	Family	Family rank	Unit	PAR-2	Published	Median calls	Mean calls	Hessians	FD calls	Failures	Runs
Sella (2.6.0) · Sella	Single-ended saddle	1	force calls	68.74		59.5	68.74	0	0	crash 0, divergence 0, cap 0, wrong_stationary 0	200

eOn rows are not yet on the harness counter.

A run short of the full set, or a run that is not the stock method, stays off the ranked table.

Files with no per-start harness record stay off the ranked table.

Family tables, overall table, and performance profile.

Pt heptamer island

Sella (0.1.1) · sella

PAR-253

Row	Family	Family rank	Unit	PAR-2	Published	Median calls	Mean calls	Hessians	FD calls	Failures	Runs
Sella (0.1.1) · sella	Single-ended saddle	1	force calls	53	53	51	53	0	0	0 failed	49

eOn rows are not yet on the harness counter.

A run short of the full set, or a run that is not the stock method, stays off the ranked table.

Family tables, overall table, and performance profile.

Machine learning optimization

Ethene

n2p2 (2.3.0) · Kalman energy

PAR-217397.1

Row	Family	Family rank	Unit	PAR-2	Published	Median calls	Mean calls	Hessians	FD calls	Failures	Runs
n2p2 (2.3.0) · Kalman energy	Minimizers	1	loss-gradient evaluations	17397.1		10000	9397.12	0	0	crash 0, divergence 0, cap 40, wrong_stationary 0	50

Family tables, overall table, and performance profile.

EMT Pt 38 nanoparticle

n2p2 (2.3.0) · Kalman energy

PAR-225.02

Row	Family	Family rank	Unit	PAR-2	Published	Median calls	Mean calls	Hessians	FD calls	Failures	Runs
n2p2 (2.3.0) · Kalman energy	Minimizers	1	loss-gradient evaluations	25.02		14	25.02	0	0	crash 0, divergence 0, cap 0, wrong_stationary 0	50

Family tables, overall table, and performance profile.

Nudged elastic band

Pt heptamer island

geodesic (1.0.0) · CI-NEB-LBFGS geodesic

PAR-2318.923

Row	Family	Family rank	Unit	PAR-2	Published	Median calls	Mean calls	Per-image mean	Hessians	FD calls	Failures	Runs
geodesic (1.0.0) · CI-NEB-LBFGS geodesic	Band / NEB	1	force calls per band	318.923		314	318.923	39.8654	0	0	crash 0, divergence 0, cap 0, wrong_stationary 0	13

eOn band rows are not yet on the harness counter.

A run short of the full set, or a run that is not the stock method, stays off the ranked table.

Files with no per-start harness record stay off the ranked table.

Family tables, overall table, and performance profile.

Harmonic transition state theory

EAM aluminum

eOn (r2006) · Dimer-CG rotation: max=10 tol=1

Published21070

Row	Family	Family rank	Unit	PAR-2	Published	Median calls	Mean calls	Hessians	FD calls	Failures	Runs
eOn (r2006) · Dimer-CG rotation: max=10 tol=1	Minimum mode	1	force calls		21070		21070	0	0	not recorded	432

A run short of the full set, or a run that is not the stock method, stays off the ranked table.

Family tables, overall table, and performance profile.

Pt heptamer island

TSASE · quenchDimer+neb-QM

Published293200

Row	Family	Family rank	Unit	PAR-2	Published	Median calls	Mean calls	Hessians	FD calls	Failures	Runs
TSASE · quenchDimer+neb-QM	Band / NEB	1	force calls		293200		293200	0	0	not recorded	10

A run short of the full set, or a run that is not the stock method, stays off the ranked table.

Family tables, overall table, and performance profile.

Accelerated molecular dynamics

Al adatom on Al(100) at 150 K

eOn · parallel replica

Published835900

Row	Family	Family rank	Unit	PAR-2	Published	Median calls	Mean calls	Hessians	FD calls	Failures	Runs
eOn · parallel replica	Minimizers	1	escape rate (1/s)		835900			0	0	not recorded	1

eOn rows are not yet on the harness counter.

Family tables, overall table, and performance profile.

Al adatom on Al(100) at 250 K

eOn · parallel replica

Published2.54294e+08

Row	Family	Family rank	Unit	PAR-2	Published	Median calls	Mean calls	Hessians	FD calls	Failures	Runs
eOn · parallel replica	Minimizers	1	escape rate (1/s)		2.54294e+08			0	0	not recorded	1

eOn rows are not yet on the harness counter.

A run short of the full set, or a run that is not the stock method, stays off the ranked table.

Family tables, overall table, and performance profile.

The sidebar lists the suites. This page keeps the 2014 provenance and links each problem in the text.