Xanadu Quantum Technologies (Canada)

Access to clean, safe water is vital for both environmental sustainability and human health. Water quality assessment and management can benefit greatly from the use of predictive models, which is made possible by the development of sophisticated technology such as machine learning. This study makes use of a dataset that includes a variety of metrics that were gathered from several water sources, including turbidity, dissolved oxygen, pH levels, and other contaminants. In order to deal with missing values, outliers, and normalization, data preparation techniques are first used. The most pertinent variables influencing water quality are then found using feature selection techniques. The effectiveness of a number of well-known ML classifiers, such as Random Forest, Support Vector Machines, Decision Trees, and XGBoost, in predicting water quality is assessed and contrasted. Cross-validation techniques are used to train, validate, and test the models in order to guarantee their generalizability and robustness. The experimental outcomes show how well the suggested method works to forecast the levels of water quality. In particular, the XGBoost performs better with low overfitting and great precision. Furthermore, feature importance analysis identifies important variables offering environmentalists and policymakers insightful information.

The discovery of the backpropagation algorithm ranks among one of the most important moments in the history of machine learning, and has made possible the training of large-scale neural networks through its ability to compute gradients at roughly the same computational cost as model evaluation. Despite its importance, a similar backpropagation-like scaling for gradient evaluation of parameterised quantum circuits has remained elusive. Currently, the most popular method requires sampling from a number of circuits that scales with the number of circuit parameters, making training of large-scale quantum circuits prohibitively expensive in practice. Here we address this problem by introducing a class of structured circuits that are not known to be classically simulable and admit gradient estimation with significantly fewer circuits. In the simplest case – for which the parameters feed into commuting quantum gates – these circuits allow for fast estimation of the gradient, higher order partial derivatives and the Fisher information matrix. Moreover, specific families of parameterised circuits exist for which the scaling of gradient estimation is in line with classical backpropagation, and can thus be trained at scale. In a toy classification problem on 16 qubits, such circuits show competitive performance with other methods, while reducing the training cost by about two orders of magnitude.

Quantum metrology promises to revolutionize precision measurements in quantum networks, particularly for accurate channel transmission loss estimation, which is critical for loss-sensitive and security protocols. Traditional methods using coherent states are limited by shot noise. However, quantum states like Two-Mode Squeezed Vacuum (TMSV) states can surpass this limit. Our theoretical and experimental study shows that measurement with TMSV states with no post selection provide up to 3.9 dB higher Fisher Information (FI) than coherent states, allowing for superior transmission loss characterization over fibers up to 35 km. This demonstrates TMSV's potential in enhancing quantum network applications.

Quantum algorithms for ground-state energy estimation of chemical systems require a high-quality initial state. However, initial state preparation is commonly either neglected entirely, or assumed to be solved by a simple product state like Hartree-Fock. Even if a nontrivial state is prepared, strong correlations render ground-state overlap inadequate for quality assessment. In this work, we address the initial state preparation problem with an end-to-end algorithm that and the quality of initial states, accomplishing the latter with a new metric—the energy distribution. To be able to prepare more complicated initial states, we introduce an implementation technique for states in the form of a sum of Slater determinants that exhibits significantly better scaling than all prior approaches. We also propose low-precision quantum phase estimation (QPE) for further state quality refinement. The complete algorithm is capable of generating high-quality states for energy estimation, and is shown in select cases to lower the overall estimation cost by several orders of magnitude when compared with the best single product state ansatz. More broadly, the energy distribution picture suggests that the goal of QPE should be reinterpreted as generating improvements compared to the energy of the initial state and other classical estimates: such an improvement can still be achieved even if QPE does not project directly onto the ground state. Finally, we show how the energy distribution can help in identifying potential quantum advantage. Published by the American Physical Society 2024

As the scale of programmable photonic circuits increases, solutions are needed to address the increasing complexity of electronic readout circuits that monitor the optical functionality. Here we propose two scalable methods for controlling programmable chips with a single readout circuit, without significant performance penalties.

Variational quantum algorithms, which combine highly expressive parameterized quantum circuits (PQCs) and optimization techniques in machine learning, are one of the most promising applications of a near-term quantum computer. Despite their huge potential, the utility of variational quantum algorithms beyond tens of qubits is still questioned. One of the central problems is the trainability of PQCs. The cost function landscape of a randomly initialized PQC is often too flat, asking for an exponential amount of quantum resources to find a solution. This problem, dubbed barren plateaus, has gained lots of attention recently, but a general solution is still not available. In this paper, we solve this problem for the Hamiltonian variational ansatz (HVA), which is widely studied for solving quantum many-body problems. After showing that a circuit described by a time-evolution operator generated by a local Hamiltonian does not have exponentially small gradients, we derive parameter conditions for which the HVA is well approximated by such an operator. Based on this result, we propose an initialization scheme for the variational quantum algorithms and a parameter-constrained ansatz free from barren plateaus.

Quantum signal processing provides an optimal procedure for simulating Hamiltonian evolution on a quantum computer using calls to a block encoding of the Hamiltonian. In many situations it is possible to control between forward and reverse steps with almost identical cost to a simple controlled operation. We show that it is then possible to reduce the cost of Hamiltonian simulation by a factor of 2 using the recent results of generalised quantum signal processing.

We experimentally demonstrate an EAM-based photonic-electronic multiply-accumulate neuron that performs inference in a health monitoring task with 1350 trainable parameters, achieving an f1 score of 85.9 % at 10 Gbaud compute rate.

Linear optical quantum circuits with photon number resolving (PNR) detectors are used for both Gaussian Boson Sampling (GBS) and for the preparation of non-Gaussian states such as Gottesman-Kitaev-Preskill (GKP), cat and NOON states. They are crucial in many schemes of quantum computing and quantum metrology. Classically optimizing circuits with PNR detectors is challenging due to their exponentially large Hilbert space, and quadratically more challenging in the presence of decoherence as state vectors are replaced by density matrices. To tackle this problem, we introduce a family of algorithms that calculate detection probabilities, conditional states (as well as their gradients with respect to circuit parametrizations) with a complexity that is comparable to the noiseless case. As a consequence we can simulate and optimize circuits with twice the number of modes as we could before, using the same resources. More precisely, for an M-mode noisy circuit with detected modes D and undetected modes U, the complexity of our algorithm is O(M2∏i∈UCi2∏i∈DCi), rather than O(M2∏i∈D∪UCi2), where Ci is the Fock cutoff of mode i. As a particular case, our approach offers a full quadratic speedup for calculating detection probabilities, as in that case all modes are detected. Finally, these algorithms are implemented and ready to use in the open-source photonic optimization library MrMustard.