While quantum algorithms offer a promising route to solving complex electronic structure problems, practical implementations on near-term hardware are severely bottlenecked by the scaling of system size. Specifically, the number of optimization iterations required for the variational quantum eigensolver to converge grows exponentially with system size. This scaling crisis can be attributed to two main factors: (1) the vastness of the search space, which makes it more likely to get trapped in local minima or barren plateaus, and (2) the huge number of ansatz parameters, which makes each iteration more expensive.

In this work, we introduce a hybrid classical-quantum workflow to systematically overcome these limitations. First, we generate a high-quality initial ground-state wavefunction using classical pre-computation with OpenMolcas software, which is then refined using a quantum emulator with customized Qiskit libraries. Second, we accelerate the convergence of the quantum optimization by dynamically discriminating and updating only the most impactful ansatz parameters. This presentation will detail our methodology and provide quantitative results for the speed-up achieved by our hybrid algorithm, highlighting strongly correlated molecular systems.