Aplicación
Técnicas de IA y quantum machine learning para la identificación de procesos cuánticos
Información Cuántica y Óptica Cuántica
Descripción del grupo:
Los grupos de Información Cuántica y Óptica Cuántica del departamento de Física de la UAB trabajan desde hace años en temas afines al uso de AI y machine learning para la optimización de los recursos cuánticos y dar solución a algunas de las preguntas abiertas en la física de altas energías y materia condensada. En la página web del grupo (https://grupsderecerca.uab.cat/giq/publications) se pueden encontrar más de 400 publicaciones científicas en el ámbito de la información cuántica, varias de ellas centradas en el uso de AI y machine learning para entender mejor los sistemas cuánticos, es decir, entre otras cosas, métodos de certificación y verificación de estados cuánticos. Esta experiencia será fundamental para poder discernir si lo generado en un ordenador cuántico es realmente el estado que queremos estudiar y no otro que se haya generado por error.
Descripción de la actividad:
El objetivo de esta tarea es estudiar de la frontera entre la algoritmia cuántica y la inteligencia artificial, y su aplicación a la mejora de tareas específicamente cuánticas.
En concreto, se abordarán los siguientes aspectos:
1. Clasificación de estados cuánticos a partir de datos experimentales.
2. Discriminación de mapas o procesos cuánticos mediante técnicas de IA.
3. Certificación y verificación de los recursos cuánticos mediante aprendizaje cuántico.
4. Desarrollo de algoritmos con utilidad para la física de altas energías y evaluación de los mismos en diversas plataformas cuánticas y clásicas.
Resultados
Rout, S.; Sankar Bhattacharya, S.; Horodecki, P.
Randomness-free Detection of Non-projective Measurements: Qubits & beyond Sin publicar
Preprint, 2024.
Resumen | Enlaces | BibTeX | Etiquetas: UAB
@unpublished{nokey,
title = {Randomness-free Detection of Non-projective Measurements: Qubits & beyond},
author = {Rout, S. and Sankar Bhattacharya, S. and Horodecki, P.},
url = {https://doi.org/10.48550/arXiv.2412.00213},
doi = {doi.org/10.48550/arXiv.2412.00213},
year = {2024},
date = {2024-11-29},
abstract = {Non-projective measurements are resourceful in several information-processing protocols. In this work, we propose an operational task involving space-like separated parties to detect measurements that are neither projective nor a classical post-processing of data obtained from a projective measurement. In the case of qubits, we consider a bipartite scenario and different sets of target correlations. While some correlations in each of these sets can be obtained by performing non-projective measurements on some shared two-qubit state it is impossible to simulate correlation in any of them using projective simulable measurements on bipartite qubit states or equivalently one bit of shared randomness. While considering certain sets of target correlations we show that the detection of qubit non-projective measurement is robust under arbitrary depolarising noise (except in the limiting case). For qutrits, while considering a similar task we show that some correlations obtained from local non-projective measurements are impossible to be obtained while performing the same qutrit projective simulable measurements by both parties. We provide numerical evidence of its robustness under arbitrary depolarising noise. For a more generic consideration (bipartite and tripartite scenario), we provide numerical evidence for a projective-simulable bound on the reward function for our task. We also show a violation of this bound by using qutrit POVMs. From a foundational perspective, we extend the notion of non-projective measurements to general probabilistic theories (GPTs) and use a randomness-free test to demonstrate that a class of GPTs, called square-bits or box-world are unphysical.},
howpublished = {Preprint},
keywords = {UAB},
pubstate = {published},
tppubtype = {unpublished}
}
Non-projective measurements are resourceful in several information-processing protocols. In this work, we propose an operational task involving space-like separated parties to detect measurements that are neither projective nor a classical post-processing of data obtained from a projective measurement. In the case of qubits, we consider a bipartite scenario and different sets of target correlations. While some correlations in each of these sets can be obtained by performing non-projective measurements on some shared two-qubit state it is impossible to simulate correlation in any of them using projective simulable measurements on bipartite qubit states or equivalently one bit of shared randomness. While considering certain sets of target correlations we show that the detection of qubit non-projective measurement is robust under arbitrary depolarising noise (except in the limiting case). For qutrits, while considering a similar task we show that some correlations obtained from local non-projective measurements are impossible to be obtained while performing the same qutrit projective simulable measurements by both parties. We provide numerical evidence of its robustness under arbitrary depolarising noise. For a more generic consideration (bipartite and tripartite scenario), we provide numerical evidence for a projective-simulable bound on the reward function for our task. We also show a violation of this bound by using qutrit POVMs. From a foundational perspective, we extend the notion of non-projective measurements to general probabilistic theories (GPTs) and use a randomness-free test to demonstrate that a class of GPTs, called square-bits or box-world are unphysical.
Rout, S.; N. Bhattacharya Sakharwade, S. S.; Ramanathan, R.; Horodecki, P.
Unbounded Quantum Advantage in Communication with Minimal Input Scaling Sin publicar
Preprint, 2024.
Resumen | Enlaces | BibTeX | Etiquetas: UAB
@unpublished{nokey,
title = {Unbounded Quantum Advantage in Communication with Minimal Input Scaling},
author = {Rout,S. and Sakharwade, N. Bhattacharya, S.S. and Ramanathan, R. and Horodecki, P.},
url = {https://arxiv.org/pdf/2305.10372},
doi = {doi.org/10.48550/arXiv.2305.10372},
year = {2024},
date = {2024-11-29},
urldate = {2024-11-29},
abstract = {In communication complexity-like problems, previous studies have shown either an exponential quantum advantage or an unbounded quantum advantage with an exponentially large input set Θ(2n) bits with respect to classical communication Θ(n) bits. In the former, the quantum and classical separation grows exponentially in input while the latter's quantum communication resource is a constant. Remarkably, it was still open whether an unbounded quantum advantage exists while the inputs do not scale exponentially. Here we answer this question affirmatively using an input size of optimal order. Considering two variants as tasks: 1) distributed computation of relation and 2) {it relation reconstruction}, we study the one-way zero-error communication complexity of a relation induced by a distributed clique labelling problem for orthogonality graphs. While we prove no quantum advantage in the first task, we show an {it unbounded quantum advantage} in relation reconstruction without public coins. Specifically, for a class of graphs with order m, the quantum complexity is Θ(1) while the classical complexity is Θ(logm). Remarkably, the input size is Θ(logm) bits and the order of its scaling with respect to classical communication is {it minimal}. This is exponentially better compared to previous works. Additionally, we prove a lower bound (linear in the number of maximum cliques) on the amount of classical public coin necessary to overcome the separation in the scenario of restricted communication and connect this to the existence of Orthogonal Arrays. Finally, we highlight some applications of this task to semi-device-independent dimension witnessing as well as to the detection of Mutually Unbiased Bases.},
howpublished = {Preprint},
keywords = {UAB},
pubstate = {published},
tppubtype = {unpublished}
}
In communication complexity-like problems, previous studies have shown either an exponential quantum advantage or an unbounded quantum advantage with an exponentially large input set Θ(2n) bits with respect to classical communication Θ(n) bits. In the former, the quantum and classical separation grows exponentially in input while the latter's quantum communication resource is a constant. Remarkably, it was still open whether an unbounded quantum advantage exists while the inputs do not scale exponentially. Here we answer this question affirmatively using an input size of optimal order. Considering two variants as tasks: 1) distributed computation of relation and 2) {it relation reconstruction}, we study the one-way zero-error communication complexity of a relation induced by a distributed clique labelling problem for orthogonality graphs. While we prove no quantum advantage in the first task, we show an {it unbounded quantum advantage} in relation reconstruction without public coins. Specifically, for a class of graphs with order m, the quantum complexity is Θ(1) while the classical complexity is Θ(logm). Remarkably, the input size is Θ(logm) bits and the order of its scaling with respect to classical communication is {it minimal}. This is exponentially better compared to previous works. Additionally, we prove a lower bound (linear in the number of maximum cliques) on the amount of classical public coin necessary to overcome the separation in the scenario of restricted communication and connect this to the existence of Orthogonal Arrays. Finally, we highlight some applications of this task to semi-device-independent dimension witnessing as well as to the detection of Mutually Unbiased Bases.
Rout, S.; Sakharwade, N.; Bhattacharya, S. S.; Ramanathan, R.; Horodecki, P.
Unbounded Quantum Advantage in Communication with Minimal Input Scaling Sin publicar
Preprint, 0000.
Resumen | Enlaces | BibTeX | Etiquetas: UAB
@unpublished{nokey,
title = {Unbounded Quantum Advantage in Communication with Minimal Input Scaling},
author = {Rout, S. and Sakharwade, N. and Bhattacharya, S.S. and Ramanathan, R. and Horodecki, P.},
url = {https://arxiv.org/pdf/2305.10372},
doi = {doi.org/10.48550/arXiv.2305.10372},
abstract = {In communication complexity-like problems, previous studies have shown either an exponential quantum advantage or an unbounded quantum advantage with an exponentially large input set Θ(2n) bits with respect to classical communication Θ(n) bits. In the former, the quantum and classical separation grows exponentially in input while the latter's quantum communication resource is a constant. Remarkably, it was still open whether an unbounded quantum advantage exists while the inputs do not scale exponentially. Here we answer this question affirmatively using an input size of optimal order. Considering two variants as tasks: 1) distributed computation of relation and 2) {it relation reconstruction}, we study the one-way zero-error communication complexity of a relation induced by a distributed clique labelling problem for orthogonality graphs. While we prove no quantum advantage in the first task, we show an {it unbounded quantum advantage} in relation reconstruction without public coins. Specifically, for a class of graphs with order m, the quantum complexity is Θ(1) while the classical complexity is Θ(logm). Remarkably, the input size is Θ(logm) bits and the order of its scaling with respect to classical communication is {it minimal}. This is exponentially better compared to previous works. Additionally, we prove a lower bound (linear in the number of maximum cliques) on the amount of classical public coin necessary to overcome the separation in the scenario of restricted communication and connect this to the existence of Orthogonal Arrays. Finally, we highlight some applications of this task to semi-device-independent dimension witnessing as well as to the detection of Mutually Unbiased Bases.},
howpublished = {Preprint},
keywords = {UAB},
pubstate = {published},
tppubtype = {unpublished}
}
In communication complexity-like problems, previous studies have shown either an exponential quantum advantage or an unbounded quantum advantage with an exponentially large input set Θ(2n) bits with respect to classical communication Θ(n) bits. In the former, the quantum and classical separation grows exponentially in input while the latter's quantum communication resource is a constant. Remarkably, it was still open whether an unbounded quantum advantage exists while the inputs do not scale exponentially. Here we answer this question affirmatively using an input size of optimal order. Considering two variants as tasks: 1) distributed computation of relation and 2) {it relation reconstruction}, we study the one-way zero-error communication complexity of a relation induced by a distributed clique labelling problem for orthogonality graphs. While we prove no quantum advantage in the first task, we show an {it unbounded quantum advantage} in relation reconstruction without public coins. Specifically, for a class of graphs with order m, the quantum complexity is Θ(1) while the classical complexity is Θ(logm). Remarkably, the input size is Θ(logm) bits and the order of its scaling with respect to classical communication is {it minimal}. This is exponentially better compared to previous works. Additionally, we prove a lower bound (linear in the number of maximum cliques) on the amount of classical public coin necessary to overcome the separation in the scenario of restricted communication and connect this to the existence of Orthogonal Arrays. Finally, we highlight some applications of this task to semi-device-independent dimension witnessing as well as to the detection of Mutually Unbiased Bases.
