Illustrative image: an IBM Quantum System One cryostat. It is NOT the ibm-quebec Heron r2 processor that ran the VGSQD experiments — VGSQD is an algorithm, and no photograph of the specific experiment exists in the primary sources. Image: Wikimedia Commons / OJB Quantum, CC BY 4.0.
TL;DR — IBM VGSQD 124 qubits molecular reactions explained in one line: IBM researchers have extended sample-based quantum diagonalization to moving molecules, reusing quantum samples from one molecular shape at other shapes instead of starting each calculation from scratch, and demonstrated it on circuits of up to 124 qubits on the ibm-quebec Heron r2 processor — finding equilibrium shapes for beryllium dimer and water, and mapping a hydrogen exchange reaction, at accuracy comparable to leading classical methods.
📌 KEY HIGHLIGHTS
- The algorithm: variable-geometry sample-based quantum diagonalization (VGSQD) extends IBM's SQD method from fixed-shape energy calculations to geometry optimisation and reaction pathways.
- The trick: quantum samples computed at one reference molecular geometry are reused at displaced geometries, cutting the quantum-computing cost of scanning many shapes.
- The hardware: circuits of up to 124 qubits on ibm-quebec, an IBM Heron r2 superconducting processor, with classical high-performance computing for the diagonalization step.
- The results: equilibrium geometry and energy for beryllium dimer and water, plus the minimum energy path and barrier of a hydrogen exchange reaction — comparable to CCSD(T) and DMRG classical benchmarks.
- The catch: the paper is a preprint, not yet peer-reviewed; the molecules are small, and the team has not quantified the cost saving versus simply repeating SQD at every geometry.
IBM VGSQD 124 qubits molecular reactions explained: what happened?
Matthieu Fortin-Deschenes and Ibrahim Shehzad of IBM's Thomas J. Watson Research Center posted "Sample-Based Quantum Diagonalization for Variable Molecular Geometries" to arXiv on 28 October 2026 (arXiv:2610.38701). It introduces VGSQD, a hybrid quantum-classical algorithm that takes sample-based quantum diagonalization — a method IBM has already used for large molecular energy calculations — and makes it work for molecules whose shapes change.
That distinction matters. Knowing a molecule's energy at one fixed shape is only half of chemistry. Real chemistry is about shapes changing: bonds stretching, angles bending, atoms swapping partners during a reaction. Until now, SQD-style methods mostly answered "how much energy does this exact arrangement have?" VGSQD answers the follow-up questions: "what shape is most stable, and what path does a reaction take?"
How VGSQD reuses quantum samples
Sample-based quantum diagonalization works as a partnership. A quantum processor prepares a circuit representing the molecule and measures it, producing a set of candidate electron configurations — the "samples". A classical computer then diagonalizes the molecular Hamiltonian within that sampled subspace to estimate the ground-state energy. The quantum machine does the part classical machines find hardest (exploring the quantum state space); the classical machine does the precise bookkeeping.
The problem with changing shapes is that, in principle, every new shape needs its own fresh set of quantum samples — a full circuit execution per geometry. VGSQD's idea is simple to state and hard to make work: take the samples computed at one reference geometry and reuse them at nearby, displaced geometries. Because neighbouring shapes have similar electronic structure, the same samples still span a useful subspace, so the classical diagonalization step can be repeated cheaply while the expensive quantum step is done far less often. The paper reports that this significantly reduces quantum computational cost, though — honestly — it does not put a single number on that saving, and the authors acknowledge that repeating standard SQD at each geometry could reach similar results. Quantifying the real saving is the open question for this method's practical value.
Illustrative image: cryogenic quantum hardware inside a dilution refrigerator — the kind of extreme-cold environment superconducting processors like Heron r2 run in. It is not the ibm-quebec system itself. Image: Wikimedia Commons / OJB Quantum, CC BY 4.0.
The results, in numbers
| Test | What VGSQD computed | Accuracy reported |
|---|---|---|
| Beryllium dimer (Be2) | Equilibrium geometry and energy of a weakly bound, strongly correlated molecule | Relaxed bond lengths comparable with experimental values and with CCSD(T) (cc-pVDZ basis) |
| Water (H2O) | Equilibrium geometry and energy | Bond lengths within 0.002 Ã… and bond angles within 0.5° (larger cc-pVTZ basis) |
| Hydrogen exchange reaction | Minimum energy path and energy barrier — the "valley route" a reaction follows | Barrier heights within 0.2 kcal/mol of the highly accurate FCI reference |
| Hardware | Circuits of up to 124 qubits on ibm-quebec (IBM Heron r2), plus classical HPC for diagonalization | Results comparable with state-of-the-art classical methods CCSD(T) and DMRG |
Two details in that table deserve emphasis. First, beryllium dimer is deliberately awkward: weakly bound and strongly correlated, the kind of system where textbook approximations break down. Second, the hydrogen exchange barrier landing within 0.2 kcal/mol of the full configuration interaction (FCI) reference is a demanding standard — chemical accuracy is usually defined as about 1 kcal/mol, so this is well inside it.
Why it matters for World Tech readers
Quantum chemistry is quantum computing's most plausible early payoff: drug discovery, catalysts, batteries and fertilisers all depend on understanding molecules that classical supercomputers can only approximate. But most quantum chemistry demonstrations so far have been single-point energy calculations — one shape, one number. Chemistry, as practised, needs whole landscapes: optimised shapes, transition states, reaction barriers. VGSQD is a step from "quantum computers can compute an energy" toward "quantum computers can do a chemist's workflow".
For India, the angle is structural rather than a local announcement: India runs a National Quantum Mission and a growing quantum-chemistry research community, and methods like VGSQD — hybrid, runnable on today's noisy processors rather than waiting for fault tolerance — are exactly the kind of technique that community can adopt and test on cloud-accessible hardware. The honest read is that this is an algorithms result on small molecules, not a drug or catalyst discovery; scaling to the complex systems industry cares about still needs to be demonstrated, along with error mitigation strong enough for larger active spaces.
Related reading on the blog: IBM's earlier modular-hardware milestone, IBM Quantum L-Coupler 99 Percent Explained, and the latest quantum-gate result, Princeton Cryogenic Atom Array Quantum Gates Explained.
Should you care?
✅ Yes, as an algorithms milestone — not as a chemistry breakthrough. VGSQD genuinely widens what near-term quantum hardware can do, from single energies to shapes and reaction paths, with classical-grade accuracy on its test systems. ⏳ But it is a preprint on three small systems, with the key efficiency claim still unquantified. Watch for peer review, larger molecules, and a head-to-head cost comparison against repeated SQD — those three will decide whether sample reuse becomes standard practice.
FAQs
What is sample-based quantum diagonalization (SQD)?
A hybrid quantum-classical method: a quantum processor samples candidate electron configurations for a molecule, and a classical computer diagonalizes the Hamiltonian in that sampled subspace to estimate the ground-state energy. IBM has previously used it for large systems, including proteins with thousands of atoms.
How is VGSQD different from SQD?
SQD computes energies at fixed molecular geometries. VGSQD extends it to variable geometries by reusing quantum samples from a reference geometry at displaced ones, so geometry optimisation and reaction-path calculations need far fewer fresh quantum circuit executions.
What molecules did IBM simulate with VGSQD?
Beryllium dimer and water for equilibrium geometry and energy, plus a hydrogen exchange reaction for its minimum energy path and barrier — all on circuits of up to 124 qubits on the ibm-quebec Heron r2 processor.
Can quantum computers design drugs now?
No. VGSQD ran on small test molecules, and the paper is a preprint awaiting peer review. The route to drug or catalyst design needs larger, messier systems, quantified cost savings and stronger error mitigation — this result is a step on that road, not the destination.
Takeaway: IBM's VGSQD lets a quantum computer reuse its homework across molecular shapes — equilibrium geometries and a reaction path, at classical-grade accuracy, on 124 qubits — but the efficiency payoff still needs its own proof.
Follow World Tech on TechByCountry for tomorrow's update.
📸 Get tech in the last 24 hours on Instagram — follow @_techinlast24hrs for daily posts (9 AM–9 PM) and reels at 12:30 PM & 8:30 PM.
Sources
- arXiv:2610.38701 — Sample-Based Quantum Diagonalization for Variable Molecular Geometries, Matthieu Fortin-Deschenes & Ibrahim Shehzad, IBM Thomas J. Watson Research Center (submitted 28 Oct 2026)
- Quantum Zeitgeist — IBM Quantum Watson Team Simulates Molecular Reactions Using 124 Qubits (31 Oct 2026)
- IBM Research — What is sample-based quantum diagonalization? (background on SQD)
Comments
Post a Comment