Technical University of Munich: Researchers Quantify Fermionic Complexity Using Occupation Number Entropies

Researchers at Technical University of Munich, in collaboration with the Munich Centre for Quantum Science and Technology (MCQST), have developed a new framework for quantifying fermionic non-Gaussianity, a crucial resource for understanding the computational power inherent in fermionic quantum systems. Poetri Sonya Tarabunga and colleagues introduce two computable measures, both derived from the covariance matrix of pure fermionic states, which address the longstanding challenge of lacking tractable quantification methods for this key property. By establishing one measure as a convex resource monotone and demonstrating a connection between both measures and state compressibility, alongside classical simulation cost, the framework provides a valuable toolkit for assessing the complexity of quantum many-body systems and exploring the potential for quantum advantage across diverse fields including quantum information science, condensed-matter physics, and quantum chemistry.

Covariance matrix analysis unlocks polynomial scaling for fermionic quantum state complexity

A significant five-fold improvement in quantifying fermionic non-Gaussianity has been achieved, effectively reducing the upper bound on classical simulation cost. Previously, representing a fermionic quantum state required an exponential number of Gaussian basis states for accurate simulation; this new framework demonstrates the potential to reduce this requirement to a polynomial number. This breakthrough leverages occupation number entropies, calculated from the covariance matrix, to overcome the previous inability to efficiently assess the complexity of fermionic quantum states for practical applications. The covariance matrix, a second-order moment of the fermionic operators, fully characterises Gaussian states and provides a convenient starting point for analysing deviations from Gaussianity. Fermionic systems, governed by anti-commutation relations, necessitate this specific approach as opposed to the bosonic case where commutation relations apply. The Williamson normal form, a canonical transformation of the covariance matrix, is central to the derivation of these new measures, allowing for a simplified representation of the quantum state and facilitating efficient computation of the relevant entropies. This is particularly important as the number of fermionic modes increases, quickly leading to computational bottlenecks with traditional methods.

The framework establishes a rigorous resource theory, a mathematical formalism for quantifying the resources needed to perform a given quantum task. Within this theory, the newly developed measures serve as lower bounds on the number of non-Gaussian operations required to create a given quantum state. This direct link between resource quantification and state preparation was previously absent, hindering progress in understanding the fundamental limits of quantum computation with fermionic systems. Non-Gaussian operations are those that cannot be implemented using only Gaussian transformations, and are therefore essential for creating states with non-classical correlations. The ability to quantify the ‘cost’ of creating such states is vital for determining whether a particular quantum algorithm is likely to offer a genuine advantage over classical algorithms. The resource theory is constructed to be convex, meaning that the cost of preparing a mixed state can be determined from the costs of preparing its constituent pure states, simplifying the analysis and ensuring mathematical consistency.

Occupation number entropies calculate the Tsallis entropy of occupation numbers, providing a measure of the distribution of particles across single-particle orbitals. This reveals the extent to which a state deviates from a fully occupied Gaussian state, the simplest possible quantum state, and provides a lower bound on the number of non-Gaussian operations needed for its creation. Introducing natural-orbital participation entropies as a complementary measure has further refined the quantification of fermionic non-Gaussianity. These entropies assess how broadly a state spreads across a specific mathematical basis known as the natural orbital basis, directly upper-bounding the cost of simulating the state on a classical computer. Natural orbitals represent the most efficient basis for describing the electron correlation in a many-body system. This approach is particularly useful for analysing complex systems like random SWAP-doped matchgate circuits, which model the effects of noise in quantum computations, and the bond-modulated XXZ model, a paradigmatic model in condensed-matter physics exhibiting rich quantum behaviour. These analyses offer insights into many-body phenomena and enable more precise comparisons between different quantum states and circuits, accelerating progress in fields like materials science and drug discovery by allowing researchers to better understand and predict the properties of complex quantum systems.

Quantifying quantum state complexity predicts simulation tractability

Quantifying complexity in quantum systems is paramount to realising practical quantum technologies, but assessing “fermionic non-Gaussianity”, a measure of how far a quantum state deviates from simple, easily simulated ones, has proven remarkably difficult. The challenge arises from the exponential scaling of the Hilbert space with the number of fermionic modes. A robust resource theory was established alongside two new computational measures, derived from the covariance matrix, providing a comprehensive set of tools for understanding relationships within the quantum state. These measures link state compressibility, a measure of how efficiently a quantum state can be represented, to the cost of classical simulation, offering a pathway to evaluate the feasibility of quantum computation. Higher non-Gaussianity generally indicates greater complexity and potential for quantum advantage, as quantified using calculations involving Tsallis-α and Rényi-α entropy, both established information-theoretic measures. The value of α controls the sensitivity of the entropy to different parts of the probability distribution, allowing for a nuanced characterisation of the state’s complexity.

The ability to predict simulation tractability is crucial for guiding the development of quantum algorithms and hardware. If a quantum state exhibits high non-Gaussianity, it is likely to be intractable for classical computers, suggesting that a quantum algorithm exploiting this state could potentially outperform classical algorithms. Conversely, if a state is nearly Gaussian, it may be possible to simulate it efficiently on a classical computer, rendering a quantum algorithm unnecessary. This framework therefore provides a valuable tool for identifying promising candidates for quantum computation and for optimising quantum algorithms to maximise their potential for achieving quantum advantage. Furthermore, the connection between state compressibility and simulation cost opens up the possibility of developing new data compression techniques for quantum information, potentially reducing the resources required to store and transmit quantum data.

The research successfully quantified fermionic non-Gaussianity using a new resource theory and two computable measures derived from the covariance matrix. This is important because it provides a way to assess the complexity of quantum states and predict how difficult they are for classical computers to simulate. The resulting measures, based on Tsallis-α and Rényi-α entropy, link a state’s compressibility to its classical simulation cost. Researchers analysed these measures for specific states, including those generated by SWAP-doped matchgate circuits and the bond-modulated XXZ model, to better understand many-body phenomena.