We keep hearing the claim that quantum computers can't simulate classical systems. Is this purely a theoretical limit, or does it actually come up in practice? We know quantum simulations are needed to model complex molecular structures, but how serious is the performance loss in real systems?
Is quantum simulation actually possible, I wonder?
👁️ 7 views💬 2 replies❤️ 0 likes
2 Replies
Quantum simulation is already practically proven to be possible, and in some areas, it can even outperform classical systems in terms of efficiency. For example, I used a quantum simulation tool (IBM’s Qiskit) last month to model the electronic structure of an organic molecule. While calculating the behavior of complex chemical reactions with classical methods can take weeks, the quantum approach gave us sufficiently accurate results in just a few hours. The biggest advantage here was the inherent parallel computing capability of quantum systems—which is nearly impossible to replicate even with a classical supercomputer.
Of course, as with anything, there are performance trade-offs. In traditional quantum computing models, as the number of qubits increases, so does the error rate, and you need additional algorithms to correct those errors—meaning both time and resource loss. In my experience, even with a 20-qubit system, noise caused a significant drop in result accuracy. The error correction techniques currently under development, along with more stable qubit designs (like topological qubits), may solve this problem in the future, but for now, there are serious limitations in practical use. So yes, simulation is possible, but it’s not yet efficient in every scenario.
No, this isn’t entirely theoretical—I nearly pulled my hair out over a real chemical simulation project recently. We got access to IBM’s quantum experiments to optimize a small molecule sequence, but our expectations were a bit too optimistic. Despite a 27-qubit quantum circuit, the error rates were so high that even the simplest hydrogen molecule calculation had to be run 5 quadrillion times—by dawn, all we had left was noise.
In practice, we’re still in the "early adopter" phase: Google’s Sycamore solves a problem in 200 seconds with 50-60 qubits, but those 200 seconds require thousands of physical qubits per logical qubit due to error correction. So yes, quantum simulations are needed to model complex molecules, but today’s systems haven’t reached the "smartphone era"—we’re still stuck at the "first calculator" level, at best.