Systems of many interacting quantum particles feature a remarkably rich phenomenology due to the complex interplay between their constituents. Examples include the collective behavior found in low-dimensional quantum magnets or the emergence of exotic superfluids. To understand such phenomena, physical sciences resort to classical simulations of their underlying models. While these classical simulations made big strides explaining various phenomena, for very complex systems, often with dynamics, classical calculations are doomed to fail and exceed the capabilities of today’s most advanced supercomputers. This failure of classical methods to accurately simulate some quantum mechanical systems is due to the exponential scaling of complexity with system size.
At the same time, a fundamental understanding of complex quantum systems is essential for predicting the properties of, for example, certain materials or biomolecules. In the early 1980s, physicist and Nobel Prize laureate Richard Feynman proposed using quantum simulators and computers to calculate complex quantum mechanical phenomena rather than classical computers because they obey the same laws as the systems to be calculated and circumvent the limitations of classical computers.
While quantum simulators are primarily suited to specific, platform-tailored problems, such as in solid-state physics or elementary quantum magnets, digital quantum computers are more universally applicable.
But their additional capabilities come at the cost of increased complexity: creating digital quantum computers requires individually controllable qubits, which are the elementary units for storing and processing quantum data. Interactions between such qubits can be exploited to achieve entanglement, which, together with the superposition of different qubit states, form the basis of the alleged computational power of quantum computers.
The wide range of applications for quantum computers opens up new scientific and technological possibilities, for example, when quantum computers are tightly integrated with classical supercomputers to perform certain, specifically portioned and tailored subtasks more efficiently than their classical counterparts.