Carlos A. Riofrio
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Carlos A. Riofrio has a total of 30 co-authors. Here are the top co-authors by number of shared publications:
- Andre Luckow
BMW Group, Munich, Germany and LMU Munich, Munich, Germany
3 shared publications - Caitlin Jones
QUTAC, Quantum Technology and Application Consortium Germany
3 shared publications - Florian Krellner
SAP SE Walldorf Germany
3 shared publications - Johannes Klepsch
BMW Group
3 shared publications - Joseph Doetsch
QUTAC, Quantum Technology and Application Consortium Germany
3 shared publications - Aleksandar Vuckovic
QUTAC, Quantum Technology and Application Consortium Germany
2 shared publications
Latest Publications
- Open Access
- 17 June 2026
- Published by IOP Publishing in Quantum Science and Technology
- Vol. 11 (3) , 035042
Scaling Quantum Machine Learning without Tricks: Full-Resolution and Diverse Image Generation
Quantum generative modeling is a rapidly evolving discipline at the intersection of quantum computing and machine learning. Contemporary quantum machine learning is generally limited to toy examples or heavily restricted datasets with few elements. This is not only due to the current limitations of available quantum hardware but also due to the absence of inductive biases arising from application-agnostic designs. Current quantum solutions must resort to tricks to scale down high-resolution images, such as relying heavily on dimensionality reduction or utilizing multiple quantum models for low-resolution image patches. Building on recent developments in classical image loading to quantum computers, we circumvent these limitations and train quantum Wasserstein GANs on the established classical MNIST and Fashion-MNIST datasets. Using the complete datasets, our system generates full-resolution images across all ten classes and establishes a new state-of-the-art performance with a single end-to-end quantum generator without tricks. As a proof-of-principle, we also demonstrate that our approach can be extended to color images, exemplified on the Street View House Numbers dataset. We analyze how the choice of variational circuit architecture introduces inductive biases, which crucially unlock this performance. Furthermore, enhanced noise input techniques enable highly diverse image generation while maintaining quality. Finally, we show promising results even under quantum shot noise conditions.
- 17 June 2026
- Published by IOP Publishing in Quantum Science and Technology
- Vol. 11 (3) , 035042
Open AccessScaling Quantum Machine Learning without Tricks: Full-Resolution and Diverse Image Generation
Quantum generative modeling is a rapidly evolving discipline at the intersection of quantum computing and machine learning. Contemporary quantum machine learning is generally limited to toy examples or heavily restricted datasets with few elements. This is not only due to the current limitations of available quantum hardware but also due to the absence of inductive biases arising from application-agnostic designs. Current quantum solutions must resort to tricks to scale down high-resolution images, such as relying heavily on dimensionality reduction or utilizing multiple quantum models for low-resolution image patches. Building on recent developments in classical image loading to quantum computers, we circumvent these limitations and train quantum Wasserstein GANs on the established classical MNIST and Fashion-MNIST datasets. Using the complete datasets, our system generates full-resolution images across all ten classes and establishes a new state-of-the-art performance with a single end-to-end quantum generator without tricks. As a proof-of-principle, we also demonstrate that our approach can be extended to color images, exemplified on the Street View House Numbers dataset. We analyze how the choice of variational circuit architecture introduces inductive biases, which crucially unlock this performance. Furthermore, enhanced noise input techniques enable highly diverse image generation while maintaining quality. Finally, we show promising results even under quantum shot noise conditions.
- Open Access
- 19 May 2026
- Published by IOP Publishing in Quantum Science and Technology
- Vol. 11 (2) , 025054
Towards solving industrial integer linear programs with decoded quantum interferometry
Optimization via decoded quantum interferometry (DQI) has recently gained a great deal of attention as a promising avenue for solving optimization problems using quantum computers. In this paper, we apply DQI to an industrial optimization problem in the automotive industry: the vehicle option-package pricing problem. Our main contributions are (1) formulating the industrial problem as an integer linear program (ILP), (2) converting the ILP into instances of max-XORSAT, and (3) developing a detailed quantum circuit implementation for belief propagation, a heuristic algorithm for decoding low density parity-check codes. Thus, we provide a full implementation of the DQI algorithm using Belief Propagation, which can be applied to any industrially relevant ILP by first transforming it into a max-XORSAT instance. We also evaluate the effectiveness of our implementation by benchmarking it against both Gurobi and a random sampling baseline.
- 19 May 2026
- Published by IOP Publishing in Quantum Science and Technology
- Vol. 11 (2) , 025054
Open AccessTowards solving industrial integer linear programs with decoded quantum interferometry
Optimization via decoded quantum interferometry (DQI) has recently gained a great deal of attention as a promising avenue for solving optimization problems using quantum computers. In this paper, we apply DQI to an industrial optimization problem in the automotive industry: the vehicle option-package pricing problem. Our main contributions are (1) formulating the industrial problem as an integer linear program (ILP), (2) converting the ILP into instances of max-XORSAT, and (3) developing a detailed quantum circuit implementation for belief propagation, a heuristic algorithm for decoding low density parity-check codes. Thus, we provide a full implementation of the DQI algorithm using Belief Propagation, which can be applied to any industrially relevant ILP by first transforming it into a max-XORSAT instance. We also evaluate the effectiveness of our implementation by benchmarking it against both Gurobi and a random sampling baseline.
- Open Access
- 1 September 2025
- Published by IOP Publishing in Quantum Science and Technology
- Vol. 10 (4)
Typical Machine Learning Datasets as Low-Depth Quantum Circuits
Quantum machine learning (QML) is an emerging field that investigates the capabilities of quantum computers for learning tasks. While QML models can theoretically offer advantages such as exponential speed-ups, challenges in data loading and the ability to scale to relevant problem sizes have prevented demonstrations of such advantages on practical problems. In particular, the encoding of arbitrary classical data into quantum states usually comes at a high computational cost, either in terms of qubits or gate count. However, real-world data typically exhibits some inherent structure (such as image data) which can be leveraged to load them with a much smaller cost on a quantum computer. This work further develops an efficient algorithm for finding low-depth quantum circuits to load classical image data as quantum states. To evaluate its effectiveness, we conduct systematic studies on the MNIST, Fashion-MNIST, CIFAR-10, and Imagenette datasets. The corresponding circuits for loading the full large-scale datasets are available publicly as PennyLane datasets and can be used by the community for their own benchmarks. We further analyze the performance of various quantum classifiers, such as quantum kernel methods, parameterized quantum circuits, and tensor-network classifiers, and we compare them to convolutional neural networks. In particular, we focus on the performance of the quantum classifiers as we introduce nonlinear functions of the input state, e.g., by letting the circuit parameters depend on the input state.
- 1 September 2025
- Published by IOP Publishing in Quantum Science and Technology
- Vol. 10 (4)
Open AccessTypical Machine Learning Datasets as Low-Depth Quantum Circuits
Quantum machine learning (QML) is an emerging field that investigates the capabilities of quantum computers for learning tasks. While QML models can theoretically offer advantages such as exponential speed-ups, challenges in data loading and the ability to scale to relevant problem sizes have prevented demonstrations of such advantages on practical problems. In particular, the encoding of arbitrary classical data into quantum states usually comes at a high computational cost, either in terms of qubits or gate count. However, real-world data typically exhibits some inherent structure (such as image data) which can be leveraged to load them with a much smaller cost on a quantum computer. This work further develops an efficient algorithm for finding low-depth quantum circuits to load classical image data as quantum states. To evaluate its effectiveness, we conduct systematic studies on the MNIST, Fashion-MNIST, CIFAR-10, and Imagenette datasets. The corresponding circuits for loading the full large-scale datasets are available publicly as PennyLane datasets and can be used by the community for their own benchmarks. We further analyze the performance of various quantum classifiers, such as quantum kernel methods, parameterized quantum circuits, and tensor-network classifiers, and we compare them to convolutional neural networks. In particular, we focus on the performance of the quantum classifiers as we introduce nonlinear functions of the input state, e.g., by letting the circuit parameters depend on the input state.
- Open Access
Path Matters: Industrial Data Meet Quantum Optimization
Real-world optimization problems must undergo a series of transformations before becoming solvable on current quantum hardware. Even for a fixed problem, the number of possible transformation paths -- from industry-relevant formulations through binary constrained linear programs (BILPs), to quadratic unconstrained binary optimization (QUBO), and finally to a hardware-executable representation -- is remarkably large. Each step introduces free parameters, such as Lagrange multipliers, encoding strategies, slack variables, rounding schemes or algorithmic choices -- making brute-force exploration of all paths intractable. In this work, we benchmark a representative subset of these transformation paths using a real-world industrial production planning problem with industry data: the optimization of work allocation in a press shop producing vehicle parts. We focus on QUBO reformulations and algorithmic parameters for both quantum annealing (QA) and the Linear Ramp Quantum Approximate Optimization Algorithm (LR-QAOA). Our goal is to identify a reduced set of effective configurations applicable to similar industrial settings. Our results show that QA on D-Wave hardware consistently produces near-optimal solutions, whereas LR-QAOA on IBM quantum devices struggles to reach comparable performance. Hence, the choice of hardware and solver strategy significantly impacts performance. The problem formulation and especially the penalization strategy determine the solution quality. Most importantly, mathematically-defined penalization strategies are equally successful as hand-picked penalty factors, paving the way for automated QUBO formulation. Moreover, we observe a strong correlation between simulated and quantum annealing performance metrics, offering a scalable proxy for predicting QA behavior on larger problem instances.
Open AccessPath Matters: Industrial Data Meet Quantum Optimization
Real-world optimization problems must undergo a series of transformations before becoming solvable on current quantum hardware. Even for a fixed problem, the number of possible transformation paths -- from industry-relevant formulations through binary constrained linear programs (BILPs), to quadratic unconstrained binary optimization (QUBO), and finally to a hardware-executable representation -- is remarkably large. Each step introduces free parameters, such as Lagrange multipliers, encoding strategies, slack variables, rounding schemes or algorithmic choices -- making brute-force exploration of all paths intractable. In this work, we benchmark a representative subset of these transformation paths using a real-world industrial production planning problem with industry data: the optimization of work allocation in a press shop producing vehicle parts. We focus on QUBO reformulations and algorithmic parameters for both quantum annealing (QA) and the Linear Ramp Quantum Approximate Optimization Algorithm (LR-QAOA). Our goal is to identify a reduced set of effective configurations applicable to similar industrial settings. Our results show that QA on D-Wave hardware consistently produces near-optimal solutions, whereas LR-QAOA on IBM quantum devices struggles to reach comparable performance. Hence, the choice of hardware and solver strategy significantly impacts performance. The problem formulation and especially the penalization strategy determine the solution quality. Most importantly, mathematically-defined penalization strategies are equally successful as hand-picked penalty factors, paving the way for automated QUBO formulation. Moreover, we observe a strong correlation between simulated and quantum annealing performance metrics, offering a scalable proxy for predicting QA behavior on larger problem instances.
- Open Access
Generative-enhanced optimization for knapsack problems: an industry-relevant study
Optimization is a crucial task in various industries such as logistics, aviation, manufacturing, chemical, pharmaceutical, and insurance, where finding the best solution to a problem can result in significant cost savings and increased efficiency. Tensor networks (TNs) have gained prominence in recent years in modeling classical systems with quantum-inspired approaches. More recently, TN generative-enhanced optimization (TN-GEO) has been proposed as a strategy which uses generative modeling to efficiently sample valid solutions with respect to certain constraints of optimization problems. Moreover, it has been shown that symmetric TNs (STNs) can encode certain constraints of optimization problems, thus aiding in their solution process. In this work, we investigate the applicability of TN- and STN-GEO to an industry relevant problem class, a multi-knapsack problem, in which each object must be assigned to an available knapsack. We detail a prescription for practitioners to use the TN-and STN-GEO methodology and study its scaling behavior and dependence on its hyper-parameters. We benchmark 60 different problem instances and find that TN-GEO and STN-GEO produce results of similar quality to simulated annealing.
Open AccessGenerative-enhanced optimization for knapsack problems: an industry-relevant study
Optimization is a crucial task in various industries such as logistics, aviation, manufacturing, chemical, pharmaceutical, and insurance, where finding the best solution to a problem can result in significant cost savings and increased efficiency. Tensor networks (TNs) have gained prominence in recent years in modeling classical systems with quantum-inspired approaches. More recently, TN generative-enhanced optimization (TN-GEO) has been proposed as a strategy which uses generative modeling to efficiently sample valid solutions with respect to certain constraints of optimization problems. Moreover, it has been shown that symmetric TNs (STNs) can encode certain constraints of optimization problems, thus aiding in their solution process. In this work, we investigate the applicability of TN- and STN-GEO to an industry relevant problem class, a multi-knapsack problem, in which each object must be assigned to an available knapsack. We detail a prescription for practitioners to use the TN-and STN-GEO methodology and study its scaling behavior and dependence on its hyper-parameters. We benchmark 60 different problem instances and find that TN-GEO and STN-GEO produce results of similar quality to simulated annealing.
- Open Access
- 3 December 2024
- Published by Verein zur Forderung des Open Access Publizierens in den Quantenwissenschaften in Quantum
- Vol. 8, 1544
Efficient MPS representations and quantum circuits from the Fourier modes of classical image data
Machine learning tasks are an exciting application for quantum computers, as it has been proven that they can learn certain problems more efficiently than classical ones. Applying quantum machine learning algorithms to classical data can have many important applications, as qubits allow for dealing with exponentially more data than classical bits. However, preparing the corresponding quantum states usually requires an exponential number of gates and therefore may ruin any potential quantum speedups. Here, we show that classical data with a sufficiently quickly decaying Fourier spectrum after being mapped to a quantum state can be well-approximated by states with a small Schmidt rank (i.e., matrix-product states) and we derive explicit error bounds. These approximated states can, in turn, be prepared on a quantum computer with a linear number of nearest-neighbor two-qubit gates. We confirm our results numerically on a set of -pixel images taken from the `Imagenette' and DIV2K datasets. Additionally, we consider different variational circuit ansätze and demonstrate numerically that one-dimensional sequential circuits achieve the same compression quality as more powerful ansätze.- 3 December 2024
- Published by Verein zur Forderung des Open Access Publizierens in den Quantenwissenschaften in Quantum
- Vol. 8, 1544
Open AccessEfficient MPS representations and quantum circuits from the Fourier modes of classical image data
Machine learning tasks are an exciting application for quantum computers, as it has been proven that they can learn certain problems more efficiently than classical ones. Applying quantum machine learning algorithms to classical data can have many important applications, as qubits allow for dealing with exponentially more data than classical bits. However, preparing the corresponding quantum states usually requires an exponential number of gates and therefore may ruin any potential quantum speedups. Here, we show that classical data with a sufficiently quickly decaying Fourier spectrum after being mapped to a quantum state can be well-approximated by states with a small Schmidt rank (i.e., matrix-product states) and we derive explicit error bounds. These approximated states can, in turn, be prepared on a quantum computer with a linear number of nearest-neighbor two-qubit gates. We confirm our results numerically on a set of -pixel images taken from the `Imagenette' and DIV2K datasets. Additionally, we consider different variational circuit ansätze and demonstrate numerically that one-dimensional sequential circuits achieve the same compression quality as more powerful ansätze. - Open Access
- 17 June 2024
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Quantum Computing
- Vol. 5 (2) , 1-34
A Characterization of Quantum Generative Models
Quantum generative modeling is a growing area of interest for industry-relevant applications. This work systematically compares a broad range of techniques to guide quantum computing practitioners when deciding which models and methods to use in their applications. We compare fundamentally different architectural ansatzes of parametric quantum circuits: (1) A continuous architecture, which produces continuous-valued data samples, and (2) a discrete architecture, which samples on a discrete grid. We also compare the performance of different data transformations: the min-max and the probability integral transforms. We use two popular training methods: (1) quantum circuit Born machines (QCBM), and (2) quantum generative adversarial networks (QGAN). We study their performance and tradeoffs as the number of model parameters increases, with a baseline comparison of similarly trained classical neural networks. The study is performed on six low-dimensional synthetic and two real financial data sets. Our two key findings are that: (1) For all data sets, our quantum models require similar or fewer parameters than their classical counterparts. In the extreme case, the quantum models require two orders of magnitude less parameters. (2) We empirically find that a variant of the discrete architecture, which learns the copula of the probability distribution, outperforms all other methods.
- 17 June 2024
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Quantum Computing
- Vol. 5 (2) , 1-34
Open AccessA Characterization of Quantum Generative Models
Quantum generative modeling is a growing area of interest for industry-relevant applications. This work systematically compares a broad range of techniques to guide quantum computing practitioners when deciding which models and methods to use in their applications. We compare fundamentally different architectural ansatzes of parametric quantum circuits: (1) A continuous architecture, which produces continuous-valued data samples, and (2) a discrete architecture, which samples on a discrete grid. We also compare the performance of different data transformations: the min-max and the probability integral transforms. We use two popular training methods: (1) quantum circuit Born machines (QCBM), and (2) quantum generative adversarial networks (QGAN). We study their performance and tradeoffs as the number of model parameters increases, with a baseline comparison of similarly trained classical neural networks. The study is performed on six low-dimensional synthetic and two real financial data sets. Our two key findings are that: (1) For all data sets, our quantum models require similar or fewer parameters than their classical counterparts. In the extreme case, the quantum models require two orders of magnitude less parameters. (2) We empirically find that a variant of the discrete architecture, which learns the copula of the probability distribution, outperforms all other methods.
- Open Access
A performance characterization of quantum generative models
Quantum generative modeling is a growing area of interest for industry-relevant applications. With the field still in its infancy, there are many competing techniques. This work is an attempt to systematically compare a broad range of these techniques to guide quantum computing practitioners when deciding which models and techniques to use in their applications. We compare fundamentally different architectural ansatzes of parametric quantum circuits used for quantum generative modeling: 1. A continuous architecture, which produces continuous-valued data samples, and 2. a discrete architecture, which samples on a discrete grid. We compare the performance of different data transformations: normalization by the min-max transform or by the probability integral transform. We learn the underlying probability distribution of the data sets via two popular training methods: 1. quantum circuit Born machines (QCBM), and 2. quantum generative adversarial networks (QGAN). We study their performance and trade-offs as the number of model parameters increases, with the baseline of similarly trained classical neural networks. The study is performed on six low-dimensional synthetic and two real financial data sets. Our two key findings are that: 1. For all data sets, our quantum models require similar or fewer parameters than their classical counterparts. In the extreme case, the quantum models require two of orders of magnitude less parameters. 2. We empirically find that a variant of the discrete architecture, which learns the copula of the probability distribution, outperforms all other methods.
Open AccessA performance characterization of quantum generative models
Quantum generative modeling is a growing area of interest for industry-relevant applications. With the field still in its infancy, there are many competing techniques. This work is an attempt to systematically compare a broad range of these techniques to guide quantum computing practitioners when deciding which models and techniques to use in their applications. We compare fundamentally different architectural ansatzes of parametric quantum circuits used for quantum generative modeling: 1. A continuous architecture, which produces continuous-valued data samples, and 2. a discrete architecture, which samples on a discrete grid. We compare the performance of different data transformations: normalization by the min-max transform or by the probability integral transform. We learn the underlying probability distribution of the data sets via two popular training methods: 1. quantum circuit Born machines (QCBM), and 2. quantum generative adversarial networks (QGAN). We study their performance and trade-offs as the number of model parameters increases, with the baseline of similarly trained classical neural networks. The study is performed on six low-dimensional synthetic and two real financial data sets. Our two key findings are that: 1. For all data sets, our quantum models require similar or fewer parameters than their classical counterparts. In the extreme case, the quantum models require two of orders of magnitude less parameters. 2. We empirically find that a variant of the discrete architecture, which learns the copula of the probability distribution, outperforms all other methods.