arXiv subjects

Li Qian

Publications and source records attributed to Li Qian.

At least 37 records · Page 2Linked to original sources

arXiv2023

UniMC: A Unified Framework for Long-Term Memory Conversation via Relevance Representation Learning

Open-domain long-term memory conversation can establish long-term intimacy with humans, and the key is the ability to understand and memorize long-term dialogue history information. Existing works integrate multiple models for modelling through a pipeline, which ignores the coupling between different stages. In this paper, we propose a Unified framework for Long-term Memory Conversations (UniMC), which increases the connection between different stages by learning relevance representation. Specifically, we decompose the main task into three subtasks based on probability graphs: 1) conversation summarization, 2) memory retrieval, 3) memory-augmented generation. Each subtask involves learning a representation for calculating the relevance between the query and memory, which is modelled by inserting a special token at the beginning of the decoder input. The relevance representation learning strengthens the connection across subtasks through parameter sharing and joint training. Extensive experimental results show that the proposed method consistently improves over strong baselines and yields better dialogue consistency and engagingness.

cs.CL↗

arXiv2023

Distinguishability Calibration to In-Context Learning

Recent years have witnessed increasing interests in prompt-based learning in which models can be trained on only a few annotated instances, making them suitable in low-resource settings. When using prompt-based learning for text classification, the goal is to use a pre-trained language model (PLM) to predict a missing token in a pre-defined template given an input text, which can be mapped to a class label. However, PLMs built on the transformer architecture tend to generate similar output embeddings, making it difficult to discriminate between different class labels. The problem is further exacerbated when dealing with classification tasks involving many fine-grained class labels. In this work, we alleviate this information diffusion issue, i.e., different tokens share a large proportion of similar information after going through stacked multiple self-attention layers in a transformer, by proposing a calibration method built on feature transformations through rotation and scaling to map a PLM-encoded embedding into a new metric space to guarantee the distinguishability of the resulting embeddings. Furthermore, we take the advantage of hyperbolic embeddings to capture the hierarchical relations among fine-grained class-associated token embedding by a coarse-to-fine metric learning strategy to enhance the distinguishability of the learned output embeddings. Extensive experiments on the three datasets under various settings demonstrate the effectiveness of our approach. Our code can be found at https://github.com/donttal/TARA.

cs.CL↗

arXiv2023

Machine-Learning-Derived Entanglement Witnesses

In this work, we show a correspondence between linear support vector machines (SVMs) and entanglement witnesses, and use this correspondence to generate entanglement witnesses for bipartite and tripartite qubit (and qudit) target entangled states. An SVM allows for the construction of a hyperplane that clearly delineates between separable states and the target entangled state; this hyperplane is a weighted sum of observables ('features') whose coefficients are optimized during the training of the SVM. We demonstrate with this method the ability to obtain witnesses that require only local measurements even when the target state is a non-stabilizer state. Furthermore, we show that SVMs are flexible enough to allow us to rank features, and to reduce the number of features systematically while bounding the inference error. This allows us to derive W state witnesses capable of detecting entanglement with fewer measurement terms than the fidelity method dominant in today's literature. The utility of this approach is demonstrated on quantum hardware furnished through the IBM Quantum Experience.

quant-ph↗

arXiv2023

On the spectral radius of the $(L,κ)$-lazy Markov chain

We consider an $(L,κ)$-lazy operation on an irreducible Markov transition probability $P$ with state space $S$ where $L \subset S$ and $κ\in[0,1)$. For each $x \in L$ and $y\in S$, this $(L,κ)$-operation replaces $P(x,y)$, the transition probability from $x$ to $y$, by $κ1_{\{x=y\}} + (1-κ)P(x,y)$. We are interested in how $L$ and $κ$ influence the spectral radius $ρ^L_κ$ of this new transition probability. We first show that $ρ^L_κ$ is non-decreasing and continuous in $κ$. We then show that: (1) If $L$ is nonempty and finite, then $P$ being rho-transient is equivalent to that the growth of $\displaystyle (ρ^L_κ)_{κ\in[0,1)}$ exhibits a phase transition: There exists a critical value $κ_c(L) \in (0,1)$ such that $κ\mapsto ρ^L_κ$ is a constant on $[0,κ_c(L)]$ and increases strictly on $[κ_c(L),1)$; (2) For every $κ\in(0,1)$, if $S\setminus L$ is nonempty and finite, then $ρ^L_κ=ρ^S_κ$ if and only if $P$ is not strictly rho-recurrent.

math.PR↗

arXiv2022

Fully-Passive Quantum Key Distribution

Passive implementations of quantum key distribution (QKD) sources are highly desirable as they eliminate side-channels that active modulators might introduce. Up till now, passive decoy-state and passive encoding BB84 schemes have both been proposed. Nonetheless, passive decoy-state generation and passive encoding have never been simultaneously implemented with linear optical elements before, which greatly limits the practicality of such passive QKD schemes. In this work, we overcome this limitation and propose a fully-passive QKD source with linear optics that eliminates active modulators for both decoy-state choice and encoding. This allows for highly practical QKD systems that avoid side-channels from the source modulators. The passive source we propose (combined with the decoy-state analysis) can create any arbitrary state on a qubit system and is protocol-independent. That is, it can be used for various protocols such as BB84, reference-frame-independent QKD, or the six-state protocol. It can also in principle be combined with e.g. measurement-device-independent QKD, to build a system without side-channels in either detectors or modulators.

quant-ph↗

arXiv2022

Sub-MHz spectral dip in a resonator-free twisted gain medium

Ultra-narrow optical spectral features resulting from highly dispersive light-matter interactions are essential for a broad range of applications such as spectroscopy, slow-light, and high-precision sensing. Features approaching sub-MHz, or equivalently, Q-factors approaching ~1 billion and beyond, are challenging to obtain in solid-state systems, ultimately limited by loss. We present a novel approach to achieve tunable sub-MHz spectral features, at room temperature, without resonators. We exploit gain-enhanced polarization pulling in a twisted birefringent medium where polarization eigenmodes are frequency-dependent. Using Brillouin gain in a commercial spun fiber, we experimentally achieve a 0.72 MHz spectral dip, the narrowest backward Brillouin scattering feature ever reported. Further optimization can potentially reduce the linewidth to <0.1 MHz. Our approach is simple and broadly applicable, offering on-demand tunability and high sensitivity, with a wide range of applications such as microwave photonic filters, slow and fast light, and optical sensing.

physics.optics↗

arXiv2022

TMR transition and highly sensitive pressure sensors based on magnetic tunnel junctions with black phosphorus barrier

Black phosphorus is a promising material to serve as the barrier of magnetic tunnel junctions (MTJs) due to the weak van der Waals interlayer interactions. In particular, the special band features of black phosphorus may bring intriguing physical characteristics. Here, we study theoretically the effect of band gap tunability of black phosphorus on the MTJs with black phosphorus barrier. It is found that, the tunneling magnetoresistance (TMR) may achieve a transition from finite value to infinity owing to the variation of the band gap of black phosphorus. Combining with the latest experimental results of the pressure-induced band gap tunability, we further investigate the pressure effect of TMR in the MTJs with black phosphorus barrier. The calculations show that the pressure sensitivity can be quite high under appropriate parameters. Physically, the high sensitivity originates from the TMR transition phenomenon. To take advantage of the high pressure sensitivity, we propose and design a detailed structure of highly sensitive pressure sensors based on MTJs with black phosphorus barrier, whose working mechanism is basically different from the convential pressure sensors. The present pressure sensors possess four advantages and benifits: (1) high sensitivity, (2) well anti-interference, (3) high spatial resolution, and (4) fast response speed. Our study may advance new research area for both the MTJs and pressure sensors.

cond-mat.mes-hall↗

arXiv2021

Comprehensive Performance Evaluation of LID Practices for the Sponge City Construction: A Case Study in Guangxi, China

Sponge city construction is a new concept of urban stormwater management, which can effectively relieve urban flooding, reduce non-point source pollution, and promote the usage of rainwater resources, often including the application of Low Impact Development (LID) techniques. Although 30 cities in China have been chosen to implement sponge city construction, there is a lack of a quantitative evaluation method to evaluate the environmental, economic, and social benefits of LID practices. This paper develops a comprehensive evaluation system to quantify the benefits of different combinations of LID units using the Storm Water Management Model (SWMM) and the Analytical Hierarchy Process (AHP) method. The performance of five LID design scenarios with different locations and sizes of the bio-retention facility, the grassed swale, the sunken green space, the permeable pavement, and the storage tank were analyzed for a sports center project in Guangxi, China. Results indicated that the green scenario that contains 34.5% of bio-retention facilities and 46.0% of sunken green spaces had the best comprehensive performance regarding meeting the requirements of 75% annual total runoff reduction and the attainment of good operation performance, rainwater utilization, landscape promotion, and ecological service functions, mainly because they are micro-scale and decentralized facilities that can manage stormwater at the source through the natural process. The optimal scenario was adopted to construct the project, and the proposed evaluation system can also be applied to optimal selection and performance effect evaluation of LID practices in other sponge city projects.

stat.AP↗

arXiv2021

Telecom-band Hyperentangled Photon Pairs from a Fiber-based Source

Hyperentanglement, the simultaneous and independent entanglement of quantum particles in multiple degrees of freedom, is a powerful resource that can be harnessed for efficient quantum information processing. In photonic systems, the two degrees of freedom (DoF) often used to carry quantum and classical information are polarization and frequency, thanks to their robustness in transmission, both in free space and in optical fibers. Telecom-band hyperentangled photons generated in optical fibers are of particular interest because they are compatible with existing fiber-optic infrastructure, and can be distributed over fiber networks with minimal loss. Here, we experimentally demonstrate the generation of telecom-band biphotons hyperentangled in both the polarization and frequency DoFs using a periodically-poled silica fiber and observe entanglement concurrences above 0.95 for both polarization and frequency DOFs. Furthermore, by concatenating a Hong-Ou-Mandel interference test for frequency entanglement and full state tomography for polarization entanglement in a single experiment, we can demonstrate simultaneous entanglement in both the polarization and frequency DOFs. The states produced by our hyperentanglement source can enable protocols such as dense coding and high-dimensional quantum key distribution.

quant-ph↗

arXiv2021

Simple security proofs for continuous-variable quantum key distribution with intensity fluctuating sources

Quantum Key Distribution (QKD) in principle offers unconditional security based on the laws of physics. Continuous variable (CV) quantum key distribution has the potential for high-key-rate and low-cost implementations using standard telecom components. Despite tremendous theoretical and experimental progress in continuous-variable quantum key distribution, the security has not been rigorously established for most current continuous-variable quantum key distribution systems that have imperfections. Among the imperfections, intensity fluctuation is a significant principal problem affecting security. In this paper, we provide simple security proofs for continuous-variable quantum key distribution systems with an intensity fluctuating sources. Specifically, depending on device assumptions in the source, the imperfect systems are divided into two cases for security proofs. Our proofs are simple to implement without any hardware adjustment for the current continuous-variable quantum key distribution system. Besides, we show some compensation schemes for the decoding flaws for an outlook.

quant-ph↗

arXiv2021

Measurement device-independent quantum key distribution with passive, time-dependent source side-channels

While measurement-device-independent (MDI) quantum key distribution (QKD) allows two trusted parties to establish a shared secret key from a distance without needing to trust a central detection node, their quantum sources must be well-characterized, with side-channels at the source posing the greatest loophole to the protocol's security. In this paper, we identify a time-dependent side-channel in a common polarization-based QKD source that employs a Faraday mirror for phase stabilization. We apply the recently developed numerical proof technique from [Phys. Rev. A 99, 062332 (2019)] to quantify the sensitivity of the secret key rate to the quantum optical model for the side-channel, and to develop strategies to mitigate the information leakage. In particular, we find that the MDI three-state and BB84 protocols, while yielding the same key rate under ideal conditions, have diverging results in the presence of a side-channel, with BB84 proving more advantageous. While we consider only a representative case example, we expect the strategies developed and key rate analysis method to be broadly applicable to other leaky sources.

quant-ph↗

arXiv2021

Real-time ultrasound sensing with a mode-optimized photonic crystal slab

Integrated photonic sensors can provide large scale, flexible detection schemes. Photonic crystal slabs (PCS) offer a miniaturized platform for wideband, sensitive ultrasound detection by exploiting the photoelastic effect in water. However, poor modal overlap with the sensing medium and non-negligible absorption loss of the aqueous medium have previously limited PCS sensor performance. In this study, we detail the development and optimization of a PCS-based acoustic sensor, by adding to it a low-loss high-index polymer cladding layer. Exploiting a mode-optimized TM-like optical resonance present in a PCS, with high bulk index sensitivity (>600 nm/RIU) and quality factor Q (>8000), we demonstrate real-time ultrasound-sensing at a noise equivalent pressure (NEP) of 170 Pa (1.9 Pa/rt Hz). The PCS sensor is backside-coupled to optical fiber which, along with its intensity-based ultrasound-sensing architecture, will allow us to scale up easily to a 2D array. This work paves the way to a sensitive compact ultrasound detector for photoacoustic-based diagnostics and monitoring of tissue.

physics.app-ph↗

arXiv2021

Experiment on scalable multi-user twin-field quantum key distribution network

Twin-field quantum key distribution (TFQKD) systems have shown great promise for implementing practical long-distance secure quantum communication due to its measurement-device-independent nature and its ability to offer fundamentally superior rate-loss scaling than point-to-point QKD systems. A surge of research and development effort in the last two years has produced many variants of protocols and experimental demonstrations. In terms of hardware topology, TFQKD systems interfering quantum signals from two remotely phase-locked laser sources are in essence giant Mach-Zehnder interferometers (MZIs) requiring active phase stabilization. Such configurations are inherently unsuitable for a TFQKD network, where more than one user-pair share the common quantum measurement station, because it is practically extremely difficult, if not impossible, to stabilize MZIs of largely disparate path lengths, a situation that is inevitable in a multi-user-pair TFQKD network. On the other hand, Sagnac interferometer based TFQKD systems exploiting the inherent phase stability of the Sagnac ring can implement asymmetric TFQKD, and are therefore eminently suitable for implementing a TFQKD network. In this work, we experimentally demonstrate a proof-of-principle multi-user-pair Sagnac TFQKD network where three user pairs sharing the same measurement station can perform pair-wise TFQKD through time multiplexing, with channel losses up to 58 dB, and channel loss asymmetry up to 15 dB. In some cases, the secure key rates still beat the rate-loss bounds for point-to-point repeaterless QKD systems, even in this network configuration. It is to our knowledge the first multi-user-pair TFQKD network demonstration, an important step in advancing quantum communication network technologies.

quant-ph↗

arXiv2021

Broadband Fiber-based Entangled Photon Pair Source at Telecom O-band

In this letter, we report a polarization-entangled photon-pair source based on type-II spontaneous parametric down conversion at telecom O-band in periodically poled silica fiber (PPSF). The photon-pair source exhibits more than 130 nm (~24 THz) emission bandwidth centered at 1306.6 nm. The broad emission spectrum results in a short biphoton correlation time and we experimentally demonstrate a Hong-Ou-Mandel interference dip with a full width of 26.6 fs at half maximum. Owing to the low birefringence of the PPSF, the biphotons generated from type-II SPDC are polarization-entangled over the entire emission bandwidth, with a measured fidelity to a maximally entangled state greater than 95.4%. The biphoton source provides the broadest bandwidth entangled biphotons at O-band to our knowledge.

quant-ph↗

arXiv2020

Tagging idea in continuous variable quantum key distribution and its application

Despite tremendous theoretical and experimental progress in continuous variable (CV) quantum key distribution (QKD), its security has not been rigorously established for practical systems with multiple imperfections. The idea of tagging is widely used in security proofs of discrete variable quantum key distribution with imperfect devices. In this paper, we generalize the tagging idea from discrete variable to continuous variable. Based on untagged signals, we prove the security of the imperfect quantum key distribution system in the most conservative case. By introducing a generic imperfection model, we can evaluate and further incorporate multiple imperfections in the different stages such as modulation, channel and detection. Finally, with this generic model and tagging idea, we can prove the security of continuous variable key distribution system with multiple imperfections. Our case study shows our proofs are able to provide secure keys in the presence of both modulation and detection flaws.

quant-ph↗

arXiv2020

Efficient experimental quantum fingerprinting with wavelength division multiplexing

Quantum communication complexity studies the efficiency of information communication (that is, the minimum amount of communication required to achieve a certain task) using quantum states. One representative example is quantum fingerprinting, in which the simultaneous message passing model is considered and the minimum amount of communication could be exponentially smaller than the classical fingerprinting protocol. Experimental demonstrations based on a practical quantum fingerprinting protocol where coherent states are used to construct the fingerprints, have successfully shown the superiority of quantum fingerprinting. However, as a consequence of using coherent states, the communication time in this coherent quantum fingerprinting (CQF) protocol is increased. Moreover, the minimum amount of information communicated in these experimental demonstrations is largely limited by the dark counts of the single photon detectors. Here, we propose to enhance the performance of the existing CQF protocol through applying wavelength division multiplexing (WDM) and simultaneous detection of multiple wavelength channels. We show that the new WDM-CQF protocol can potentially reduce the communication time significantly. More importantly, with the same experimental parameters, the amount of communication is much reduced in the new scheme compared with the original CQF protocol. The more wavelength channels are used, the less communication is required by WDM-CQF protocol. We also perform a proof-of-concept experimental demonstration of the new WDM-CQF protocol with 6 wavelength channels. The experimental results further validate that the new scheme with WDM not only beats the classical protocol, but also reduces the amount of communication required by the original CQF protocol by more than half.

quant-ph↗

arXiv2020

Proof-of-principle experimental demonstration of twin-field quantum key distribution over optical channels with asymmetric losses

Twin-field (TF) quantum key distribution (QKD) is highly attractive because it can beat the fundamental limit of secret key rate for point-to-point QKD without quantum repeaters. Many theoretical and experimental studies have shown the superiority of TFQKD in long-distance communication. All previous experimental implementations of TFQKD have been done over optical channels with symmetric losses. But in reality, especially in a network setting, the distances between users and the middle node could be very different. In this paper, we perform a first proof-of-principle experimental demonstration of TFQKD over optical channels with asymmetric losses. We compare two compensation strategies, that are (1) applying asymmetric signal intensities and (2) adding extra losses, and verify that strategy (1) provides much better key rate. Moreover, the higher the loss, the more key rate enhancement it can achieve. By applying asymmetric signal intensities, TFQKD with asymmetric channel losses not only surpasses the fundamental limit of key rate of point-to-point QKD for 50 dB overall loss, but also has key rate as high as $2.918\times10^{-6}$ for 56 dB overall loss. Whereas no keys are obtained with strategy (2) for 56 dB loss. The increased key rate and enlarged distance coverage of TFQKD with asymmetric channel losses guarantee its superiority in long-distance quantum networks.

quant-ph↗

arXiv2019

Recovering the full dimensionality of hyperentanglement in collinear photon pairs

Exploiting hyperentanglement of photon pairs, that is, simultaneous entanglement in multiple degrees of freedom(DOFs), increases the dimensionality of Hilbert spaces for quantum information processing. However, generation of hyperentangled photon pairs collinearlly, while produces high brightness, results in a smaller Hilbert space due to the two photons being in the same spatial mode. In this letter, we point out that one can recover the full dimensionality of such hyperentanglement through a simple interference set up, similar to the time-reversed Hong-Ou-Mandel (TR-HOM) process. Different from the standard TR-HOM, we point out a critical phase condition has to be satisfied in order to recover the hyperentanglement. We theoretically analyze the realization of this approach and discuss the feasibility of generating truly hyperentangled photon pairs. Our proposed approach does not require post-selection and hence enables efficient hyper-entangled photon pairs generation for high-dimensional quantum applications.

quant-ph↗