Binding energies, reaction barriers, material properties — they all reduce to one question: what is the lowest energy state of this system? That question is native to quantum hardware, and it is the one place where classical simulation provably hits a wall.
Every candidate you simulate classically costs exponentially more as the system entangles. The variational eigensolver is the near-term answer — and we show it working, and failing, honestly.
A product-state ansatz — the best 'classical' guess — stalls at −1.4998 on our frustrated spin system. The true ground state is −3.0000. That gap is not marketing; it is entanglement.
When hardware matures, phase estimation reads energies to arbitrary precision. Our iterative version runs today, one control qubit at a time, so your team learns the real algorithm early.
We map one system you care about — a spin model, a small molecule fragment — to qubits, and choose ansätze your chemists can interrogate, not black boxes.
VQE training runs with real shot noise, warm starts, and convergence you can watch. Where the product ansatz ceilings out and entanglement breaks through, you see it in your own data.
System sizes that work today, the crossover we project, and what your team should build now to be ready — pursue, park, or drop.
These programs are published in our algorithm library. The numbers below come from recorded executions we can reproduce on demand.
Variational eigensolver on a frustrated Heisenberg spin system.
Measured: Exact ground state −3.0000; product ansatz stalls at −1.4998; the entangled ansatz reaches −2.9915 from a cold, shot-noisy start.
Reads eigenphases one control qubit at a time — honest on today's hardware.
Measured: Converges to the exact phase using hardware-realistic circuits.
Results are from the library items' own recorded runs on our simulator — the same one your browser uses.
Pharma-relevant molecules need error-corrected machines that do not exist yet. What exists is the method, running end to end at small scale, and a measured gap that classical guesses cannot close.
A fragment, a lattice, a materials question — tell us what you'd screen and an engineer replies.