Selected edges are shown below with both ligands drawn side by side. The warm highlight is the soft-core region that changes during the transformation; the cool highlight is the common core that stays fixed.
16
Ligands
20
Edges
4
Repaired
0
Rejected
mst
Planner
Diagnostics
Network-level metrics. Machine time is estimated from published per-edge measurements and is a report only -- it played no part in choosing these edges.
Metric
Value
Total cost
21.793 scorer units
Estimated machine time
79.4 GPU-hours (about $31.76)
Mean cost per edge
1.090
Degree
min 2 · mean 2.50 · max 4
Isolated ligands
none
Diameter
5
Short cycles
3 of length ≤ 4
Robustness
88% of 100 trials stay connected at 5% edge failure; 15.7 ligands retained on average
20 edges over 16 ligands is 25 below the n*ln(n) precision floor of 45 (Pitman 2023). Below that floor precision degrades faster as the series grows; raise --edges-per-ligand or --n-edges to buy it back.
Network
Thicker edges are cheaper. Hollow red nodes are unconnected. Click a network edge or the index below to jump to its transformation card.
(orientation flipped: sides below refer to ejm44~ejm54)
initial: 2 soft-core region(s) on side 1, 2 on side 2
iter 1 side 1: bridged 2 regions by demoting 1 atom(s) [29]
iter 1 side 2: bridged 2 regions by demoting 1 atom(s) [30]
iter 1 side 1: absorbed 2 core atom(s) [30, 31] stranded behind the soft-core
iter 1 side 2: absorbed 2 core atom(s) [31, 32] stranded behind the soft-core
final: 1/1 region(s), soft-core 13/9 atom(s)
closure: normalised raw soft-core 9/12 -> 11/14 atoms
initial: 1 soft-core region(s) on side 1, 1 on side 2
final: 1/1 region(s), soft-core 11/14 atom(s)