arXiv subjects

Nan Fang

Publications and source records attributed to Nan Fang.

18 recordsLinked to original sources

arXiv2026

Poisson-Gamma Dynamical Systems with Time-varying Transition Dynamics

Bayesian methodologies for handling count-valued time series have gained prominence due to their ability to infer interpretable latent structures and to estimate uncertainties. Among these Bayesian models, Poisson-Gamma Dynamical Systems (PGDSs) are proven to be effective in capturing the evolving dynamics underlying observed count sequences. However, the state-of-the-art PGDS still falls short in capturing the transition dynamics that are commonly observed in real-world count time series. To mitigate this limitation, a PGDS with time-varying transition kernel (TV-PGDS), is proposed to allow the underlying transition matrices to evolve over time. Three specifically-designed Dirichlet Markov chains (Dir-Dir, Dir-Gam-Dir, PR-Gam-Dir) are constructed to accommodate heterogeneous structural mutations within these dependencies. Leveraging Dirichlet-Multinomial-Beta data augmentation techniques, a fully-conjugate and efficient Gibbs sampler is developed to perform posterior simulation. Experiments show that, in comparison with related models, the proposed PGDS achieves improved predictive performance due to its capacity to learn time-varying dependency structure captured by the time-evolving transition matrices.

cs.LG↗

arXiv2026

Poisson-Gamma Modeling of Inter-Relational Dependencies in Dynamic Knowledge Graphs

Dynamic knowledge graphs are ubiquitous in today's AI applications, as we represent molecular structures, social relationships, and language information using these graph models. As knowledge graphs evolve over time and are often noisy and incomplete, modeling their temporal and relational dependencies becomes crucial for downstream tasks. To address these challenges, this paper proposes PGRE (Poisson-Gamma Relational Evolution), a probabilistic model for modeling inter-relational dependencies in dynamic knowledge graphs. PGRE represents multi-relational temporal links via a Poisson-Bernoulli formulation. It introduces Gamma-distributed latent variables to capture entity-factor associations and cross-relation dependencies mediated by shared latent communities. A Gamma Markov process further models the temporal evolution of these latent variables, enabling principled characterization of relational dynamics. Experiments on benchmark datasets show that PGRE achieves competitive performance in link prediction, particularly in sparse settings, while revealing meaningful relational evolution patterns in dynamic knowledge graphs.

cs.LG↗

arXiv2026

Speedability of computably approximable reals and their approximations

An approximation of a real is a sequence of rational numbers that converges to the real. An approximation is left-c.e. if it is computable and nondecreasing and is d.c.e. if it is computable and has bounded variation. A real is computably approximable if it has some computable approximation, and left-c.e. and d.c.e. reals are defined accordingly. An approximation $\{a_s\}_{s \in \omega}$ is speedable if there exists a nondecreasing computable function $f$ such that the approximation $\{a_{f(s)}\}_{s \in \omega}$ converges in a certain formal sense faster than $\{a_s\}_{s \in \omega}$. This leads to various notions of speedability for reals, e.g., one may require for a computably approximable real that either all or some of its approximations of a specific type are speedable. Merkle and Titov established the equivalence of several speedability notions for left-c.e. reals that are defined in terms of left-c.e. approximations. We extend these results to d.c.e. reals and d.c.e. approximations, and we prove that in this setting, being speedable is equivalent to not being Martin-L\"{o}f random. Finally, we demonstrate that every computably approximable real has a computable approximation that is speedable.

math.LO↗

arXiv2026

vEcho: A Paradigm Shift from Vulnerability Verification to Proactive Discovery with Large Language Models

Static Application Security Testing (SAST) tools often suffer from high false positive rates, leading to alert fatigue that consumes valuable auditing resources. Recent efforts leveraging Large Language Models (LLMs) as filters offer limited improvements; however, these methods treat LLMs as passive, stateless classifiers, which lack project-wide context and the ability to learn from analyses to discover unknown, similar vulnerabilities. In this paper, we propose vEcho, a novel framework that transforms the LLM from a passive filter into a virtual security expert capable of learning, memory, and reasoning. vEcho equips its core reasoning engine with a robust developer tool suite for deep, context-aware verification. More importantly, we introduce a novel Echoic Vulnerability Propagation (EVP) mechanism. Driven by a Cognitive Memory Module that simulates human learning, EVP enables vEcho to learn from verified vulnerabilities and proactively infer unknown, analogous flaws, achieving a paradigm shift from passive verification to active discovery. Extensive experiments on the CWE-Bench-Java dataset demonstrate vEcho's dual advantages over the state-of-the-art baseline, IRIS. Specifically, vEcho achieves a 65% detection rate, marking a 41.8% relative improvement over IRIS's 45.83%. Crucially, it simultaneously addresses alert fatigue by reducing the false positive rate to 59.78%, a 28.3% relative reduction from IRIS's 84.82%. Furthermore, vEcho proactively identified 37 additional known vulnerabilities beyond the 120 documented in the dataset, and has discovered 51 novel 0-day vulnerabilities in open-source projects.

cs.CR↗

arXiv2024

Offline Inverse Constrained Reinforcement Learning for Safe-Critical Decision Making in Healthcare

Reinforcement Learning (RL) applied in healthcare can lead to unsafe medical decisions and treatment, such as excessive dosages or abrupt changes, often due to agents overlooking common-sense constraints. Consequently, Constrained Reinforcement Learning (CRL) is a natural choice for safe decisions. However, specifying the exact cost function is inherently difficult in healthcare. Recent Inverse Constrained Reinforcement Learning (ICRL) is a promising approach that infers constraints from expert demonstrations. ICRL algorithms model Markovian decisions in an interactive environment. These settings do not align with the practical requirement of a decision-making system in healthcare, where decisions rely on historical treatment recorded in an offline dataset. To tackle these issues, we propose the Constraint Transformer (CT). Specifically, 1) we utilize a causal attention mechanism to incorporate historical decisions and observations into the constraint modeling, while employing a Non-Markovian layer for weighted constraints to capture critical states. 2) A generative world model is used to perform exploratory data augmentation, enabling offline RL methods to simulate unsafe decision sequences. In multiple medical scenarios, empirical results demonstrate that CT can capture unsafe states and achieve strategies that approximate lower mortality rates, reducing the occurrence probability of unsafe behaviors.

cs.LG↗

arXiv2024

Hybrid silicon all-optical switching devices integrated with two-dimensional material

We propose and demonstrate hybrid all-optical switching devices that combine silicon nanocavities and two-dimensional semiconductor material. By exploiting the refractive index modulation caused by photo-induced carriers in the two-dimensional material instead of the silicon substrate, we overcome the switching performance limitation imposed by the substrate material. Air-mode photonic crystal nanobeam cavities capable of efficient interaction with two-dimensional materials are fabricated, and molybdenum ditelluride, a two-dimensional material with rapid carrier recombination, is transferred onto the cavities. The molybdenum ditelluride flake is excited by an optical pump pulse to shift the resonant wavelength of the cavity for switching operation. We have successfully achieved all-optical switching operations on the time scale of tens of picoseconds while requiring low switching energies of a few hundred femtojoules.

physics.optics↗

arXiv2023

Van der Waals functionalization of ultrahigh-Q silica microcavities for $χ^{(2)}$-$χ^{(3)}$ hybrid nonlinear photonics

Optical nonlinear processes are indispensable in a wide range of applications including ultrafast laser sources, microscopy, metrology, and quantum information technologies. Combinations of the diverse nonlinear processes should further lead to the development of unique functionalities, but simultaneous use of second- and third-order nonlinear processes is generally difficult. Second-order effects usually overwhelm the higher-order ones, except in centrosymmetric systems where the second-order susceptibility vanishes to allow the use of the third-order nonlinearity. Here we demonstrate a hybrid photonic platform whereby the balance between second- and third-order susceptibilities can be tuned flexibly. Ultrahigh-Q silica microcavities capable of generating third-order effects are functionalized by atomically thin tungsten diselenide, and we observe cavity-enhanced second-harmonic generation and sum-frequency generation with continuous-wave excitation at a power level of only a few hundred microwatts. Pump power dependence exhibits drastic increase and saturation of the second-harmonic light, originating from the dynamic phase-matching process. We show that the coexistence of second- and third-order nonlinearities in a single device can be achieved by carefully choosing the size and the location of the two-dimensional material. Our approach can be generalized to other types of cavities, unlocking the potential of hybrid systems with controlled nonlinear susceptibilities for novel applications.

physics.optics↗

arXiv2020

Granularity of wagers in games and the possibility of savings

In a casino where arbitrarily small bets are admissible, any betting strategy M can be modified into a savings strategy that, not only is successful on each casino sequence where M is (thus accumulating unbounded wealth inside the casino) but also saves an unbounded capital, by permanently and gradually withdrawing it from the game. Teutsch showed that this is no longer the case when a fixed minimum wager is imposed by the casino, thus exemplifying a savings paradox where a player can win unbounded wealth inside the casino, but upon withdrawing a sufficiently large amount out of the game, he is forced into bankruptcy. We study the potential for saving under a shrinking minimum wager rule (granularity) and its dependence on the rate of decrease (inflation) as well as timid versus bold play.

cs.GT↗

arXiv2019

Quantum-mechanical effect in atomically thin MoS2 FET

Two-dimensional (2D) layered materials-based field-effect transistors (FETs) are promising for ultimate scaled electron device applications because of the improved electrostatics to atomically thin body thickness. However, compared with the typical thickness of ~5-nm for Si-on-insulator (SOI), the advantage of the ultimate thickness limit of monolayer for the device performance has not been fully proved yet, especially for the on-state at the accumulation region. Here, we present much stronger quantum-mechanical effect at the accumulation region based on the C-V analysis for top-gate MoS2 FETs. The self-consistent calculation elucidated that the electrons are confined in the monolayer thickness, unlike in the triangle potential formed by the electric field for SOI, the gate-channel capacitance is ideally maximized to the gate insulator capacitance since the capacitive contribution of the channel can be neglected due to the negligible channel thickness. This quantum-mechanical effect agreed well with the experimental results. Therefore, monolayer 2D channels are suggested to be used to enhance the on-current as well as the gate modulation ability.

physics.app-ph↗

arXiv2019

Integrating high-quality dielectrics with one-nanometer equivalent oxide thickness on two-dimensional electronic devices

Two-dimensional (2D) semiconductors are widely recognized as attractive channel materials for low-power electronics. However, an unresolved challenge is the integration of high-quality, ultrathin high-\k{appa} dielectrics that fully meet the roadmap requirements for low-power applications. With a dangling-bond free surface, the deposition of dielectrics by atomic layer deposition (ALD) on 2D materials is usually characterized with non-uniform nucleation and island formation, producing a highly porous dielectric layer with serious leakage particularly at the small equivalent oxide thickness (EOT) limit. Here, we report the robust ALD of highly uniform high-\k{appa} dielectric on 2D semiconductors by using ~0.3 nm-thick exclusively monolayer molecular crystal as seeding layer. Ultrathin dielectrics down to 1 nm EOT is realized on graphene, MoS2 and WSe2, with considerably reduced roughness, density of interface states, leakage current and improved breakdown field compared to prior methods. Taking advantage of the reduced EOT, we demonstrate graphene RF transistors operating at 60 GHz, as well as MoS2 and WSe2 complementary metal-oxide-semiconductor (CMOS) transistors with Vdd =0.8 V and ideal subthreshold swing (SS) of 60 mV/dec, 20 nm-channel-length MoS2 transistors with on/off ratio over 10^7. These studies highlight that our dielectric integration method is generally applicable for different 2D materials, and compatible with top-down fabrication process on large-area chemical vapor deposited films.

physics.app-ph↗

arXiv2019

Monotonous betting strategies in warped casinos

Suppose that the outcomes of a roulette table are not entirely random, in the sense that there exists a successful betting strategy. Is there a successful `separable' strategy, in the sense that it does not use the winnings from betting on red in order to bet on black, and vice-versa? We study this question from an algorithmic point of view and observe that every strategy $M$ can be replaced by a separable strategy which is computable from $M$ and successful on any outcome-sequence where $M$ is successful. We then consider the case of mixtures and show: (a) there exists an effective mixture of separable strategies which succeeds on every casino sequence with effective Hausdorff dimension less than 1/2; (b) there exists a casino sequence of effective Hausdorff dimension 1/2 on which no effective mixture of separable strategies succeeds. Finally we extend (b) to a more general class of strategies.

math.LO↗

arXiv2018

Accumulation-mode two-dimensional field-effect transistor: Operation mechanism and thickness scaling rule

Understanding the operation mode of a two-dimensional (2D) material-based field-effect transistor (FET) is one of the most essential issues in the study of electronics and physics. The existing Schottky barrier-FET model for devices with global back gate and metallic contacts overemphasizes the metal-2D contact effect, and the widely observed residual conductance cannot be explained by this model. Here, an accumulation-mode FET model, which directly reveals 2D channel transport properties, is developed based on a partial top-gate MoS2 FET with metallic contacts and a channel thickness of 0.65~118 nm. The operation mechanism of an accumulation-mode FET is validated and clarified by carefully performed capacitance measurements. A depletion capacitance-quantum capacitance transition is observed. After the analysis of the MoS2 accumulation-mode FET, we have confirmed that most 2D-FETs show accumulation-mode behavior. The universal thickness scaling rule of 2D-FETs is then proposed, which provides guidance for future research on 2D materials.

cond-mat.mtrl-sci↗

arXiv2018

Direct observation of electron capture & emission processes by the time domain charge pumping measurement of MoS2 FET

Understanding interface properties in MoS2 field effect transistors with a high-k gate insulator is critical for improving the performance of the device. Here, by applying the time domain charge pumping method, the elementary process for capture and emission of electrons to the interface states is monitored directly using a fast acquisition system. The main outcome is the detection of the clear difference in the capture and emission process of electrons to the interface states. In addition to the transient current response for gate capacitance, the current peak is observed during electron capture, while the broad tail is detected during electron emission. This different behavior is associated with the fact that the time constant for electron capture is much shorter than that for electron emission. Moreover, Dit is evaluated to be in the range of 1012 - 1013 cm-2 eV-1, which is comparable with that estimated from subthreshold swing.

physics.app-ph↗

arXiv2018

2D Tunnel Field Effect Transistors (FETs) with a Stable Charge-Transfer-Type p$^+$-WSe$_2$ Source

Two-dimensional (2D) materials are highly promising for tunnel field effect transistors (TFETs) with low subthreshold swing and high drive current because the shorter tunnel distance and strong gate controllability can be expected from the van der Waals gap distance and the atomically sharp heterointerface formed independently of lattice matching. However, the common problem for 2D-2D TFETs is the lack of highly doped 2D materials with the high process stability as the sources. In this study, we have found that p+-WSe2 doped by charge transfer from a WOx surface oxide layer can be stabilized by transferring it onto a h-BN substrate. Using this p$^+$-WSe$_2$ as a source, we fabricate all-solid-state 2D-2D heterostructure TFETs with an Al2O3 top gate insulator, i.e., type-II p$^+$-WSe$_2$ /MoS$_2$ and type-III p$^+$-WSe$_2$ /WSe$_2$. The band-to-band tunneling and negative differential resistance trends are clearly demonstrated at low temperatures. This work suggests that high doped 2D crystal of the charge transfer type is an excellent choice as sources for TFETs.

physics.app-ph↗

arXiv2018

Equivalences between learning of data and probability distributions, and their applications

Algorithmic learning theory traditionally studies the learnability of effective infinite binary sequences (reals), while recent work by [Vitanyi and Chater, 2017] and [Bienvenu et al., 2014] has adapted this framework to the study of learnability of effective probability distributions from random data. We prove that for certain families of probability measures that are parametrized by reals, learnability of a subclass of probability measures is equivalent to learnability of the class of the corresponding real parameters. This equivalence allows to transfer results from classical algorithmic theory to learning theory of probability measures. We present a number of such applications, providing many new results regarding EX and BC learnability of classes of measures, thus drawing parallels between the two learning theories.

math.LO↗

arXiv2018

Band tail interface states and quantum capacitance in a monolayer molybdenum disulfide field-effect-transistor

Although MoS2 field-effect transistors (FETs) with high-k dielectrics are promising for electron device applications, the underlying physical origin of interface degradation remains largely unexplored. Here, we present a systematic analysis of the energy distribution of the interface state density (Dit) and the quantum capacitance (CQ) in a dual-gate monolayer exfoliated MoS2 FET. The CQ analysis enabled us to construct a Dit extraction method as a function of EF. A band tail distribution of Dit with the lowest value of 8*1011 cm-2eV-1 suggests that Dit is not directly related to the sharp peak energy distribution of the S vacancy. Therefore, the Mo-S bond bending related to the strain at the interface or the surface roughness of the SiO2/Si substrate might be the origin. It is also shown that ultra-thin 2D materials are more sensitive to interface disorder due to the reduced density of states. Since all the constituents for the measured capacitance are well understood, I-V characteristics can be reproduced by utilizing the drift current model. As a result, one of the physical origins of the metal/insulator transition is suggested to be the external outcome of interface traps and quantum capacitance.

cond-mat.mtrl-sci↗

arXiv2016

Experimental detection of active defects in few layers MoS2 through random telegraphic signals analysis observed in its FET characteristics

Transition-metal dichalcogenides (TMDs), such as molybdenum disulfide (MoS2), are expected to be promising for next generation device applications. The existence of sulfur vacancies formed in MoS2, however, will potentially make devices unstable and problematic. Random telegraphic signals (RTSs) have often been studied in small area Si metal-oxide-semiconductor field-effect transistors (MOSFETs) to identify the carrier capture and emission processes at defects. In this paper, we have systemically analyzed RTSs observed in atomically thin layer MoS2 FETs. Several types of RTSs have been analyzed. One is the simple on/off type of telegraphic signals, the second is multilevel telegraphic signals with a superposition of the simple signals, and the third is multilevel telegraphic signals that are correlated with each other. The last one is discussed from the viewpoint of the defect-defect interaction in MoS2 FETs with a weak screening in atomically confined two-dimensional electron-gas systems. Furthermore, the position of defects causing RTSs has also been investigated by preparing MoS2 FETs with multi-probes. The electron beam was locally irradiated to intentionally generate defects in the MoS2 channel. It is clearly demonstrated that the MoS2 channel is one of the RTS origins. RTS analysis enables us to analyze the defect dynamics of TMD devices.

cond-mat.mtrl-sci↗

arXiv2016

Optimal asymptotic bounds on the oracle use in computations from Chaitin's Omega

Chaitin's number Omega is the halting probability of a universal prefix-free machine, and although it depends on the underlying enumeration of prefix-free machines, it is always Turing-complete. It can be observed, in fact, that for every computably enumerable (c.e.) real, there exists a Turing functional via which Omega computes it, and such that the number of bits of omega that are needed for the computation of the first n bits of the given number (i.e. the use on argument n) is bounded above by a computable function h(n) = n+o(n). We characterise the asymptotic upper bounds on the use of Chaitin's omega in oracle computations of halting probabilities (i.e. c.e. reals). We show that the following two conditions are equivalent for any computable function h such that h(n)-n is non-decreasing: (1) h(n)-n is an information content measure, (2) for every c.e. real there exists a Turing functional via which omega computes the real with use bounded by h. We also give a similar characterisation with respect to computations of c.e. sets from Omega, by showing that the following are equivalent for any computable non-decreasing function g: (1) g is an information-content measure, (2) for every c.e. set A, Omega computes A with use bounded by g. Further results and some connections with Solovay functions are given.

math.LO↗