Quantum computers promise to describe molecules with perfect accuracy, but their main nemesis is noise. Every extra qubit and every extra operation add errors. Chemists and physicists have now devised a way to compute molecules with half the qubits: the HSQD method. It takes a quantum sample, like a soil core from a borehole, and hands it over to a classical computer for precise processing.

Using the nitrogen molecule (10 electrons, 26 orbitals) as a test case, HSQD delivered the same accuracy as the previous SQD method but used half the qubits and 40% fewer measurements. It's like getting the same yield from a field half the size and with less water. The sampling was then refined by selecting only the most important states, like a chef picking the best ingredients. This variant, called HCI-HSQD, reduced the error to sub-atomic precision (less than one thousandth of a hartree), and the subspaces became 39% more compact than those of the classical HCI method.

The method was tested on clusters with active spaces up to 54 electrons and 36 orbitals — sizes where classical computers start to struggle. HSQD proves to be a noise-resilient and economical path toward quantum advantage in strongly correlated chemistry. The quantum calculator is becoming more real: fewer qubits, less noise, more accuracy.