Searching for Better Molecular Qubits with a Differentiable Physics Workflow
This case study examines molecular qubits, spin-bearing molecules under investigation as building blocks for quantum sensing and quantum information. Published vanadyl measurements support a reproducible workflow that ranks candidates, evaluates model performance on held-out splits, and records the claim level each result supports. A differentiable physics block raised held-out rank correlation from 0.13 to 0.51. A deterministic 125-candidate search returned five candidates for follow-up. The report separates a screening result from a verified molecular-qubit discovery.
A molecular-qubit screening workflow was run at full scale over approximately 150 sourced literature measurements of vanadium electron-spin properties. Surrogate models were trained end to end through public Toymodel nodes. A differentiable physics optimization assembles a semiempirical ligand-field Hamiltonian, diagonalizes it, and back-propagates through the eigensolve. Held-out Spearman rank correlation increased from 0.13 to 0.51. A deterministic 125-candidate search was evaluated under six physical-meaningfulness gates. Each reported claim is scoped to the level the public data supports.
- Spearman 0.13 -> 0.51
- 125-candidate gated search
- Differentiable physics
What the study set out to do
Molecular qubits are single molecules whose electron spin can act as a quantum bit. This study focuses on vanadyl complexes and asks which ligand environments and geometries best support coherence, controllable spin behavior, and future molecular-qubit or quantum-sensing applications.
The workflow was specified with explicit claim boundaries: ingest public measurements, train differentiable surrogates over physically meaningful descriptors, search a bounded candidate space under stated gates, and promote a claim only to the level the data supports.
What the public data allowed
The sourced corpus produced 149 ingested rows, of which 55 carry usable geometry and T1/T2 labels for the relaxation model. The anisotropic g/A readout path was blocked because the public corpus carried only one row with all four labels, so the artifact records that route as blocked, with training not attempted.
What the workflow found
On held-out ranking, a differentiable physics block raised Spearman rank correlation from 0.13 to 0.51, clearing the 0.5 gate for exploratory structure-computed relaxation ranking.
The fixed-topology search evaluated 125 candidates under six physical-meaningfulness gates and narrowed the set to five candidates for follow-up. The report separates a screening result from a verified molecular-qubit discovery.