arXiv subjects

Hongwei Xiong

Publications and source records attributed to Hongwei Xiong.

At least 19 recordsLinked to original sources

arXiv2026

A sign-blocking method for mitigating the fermion sign problem

The fermion sign problem remains the primary obstacle in simulating the thermodynamic properties of various fermionic systems. In this work, we present a sign-blocking method to mitigate the numerical instability inherent in the sign problem. In the sign-blocking method, the Monte Carlo importance sampling remains identical to traditional methods; instead, the sign-blocking method is applied during the post-processing of signed samples. Given the significant progress in simulating the 2D Fermi-Hubbard model over the past decade, a wealth of energy benchmarks is available for comparison. Consequently, we use the 2D Fermi-Hubbard model as a benchmark to validate the sign-blocking method. Surprisingly, our results align exceptionally well with existing state-of-the-art benchmarks, even in regimes previously considered challenging. The physical mechanism of the sign-blocking method lies in uncovering the correlation between energy and sign factors through data blocking, thereby successfully inferring the fermionic system's energy. Our findings suggest that the sign-blocking method holds promise for complex quantum systems, particularly when combined with appropriate simulation techniques such as auxiliary-field formalisms that trace out the fermionic degrees of freedom.

physics.comp-ph↗

arXiv2026

Simulation of strongly quantum-degenerate uniform electron gas using the pseudo-fermion method

For strongly quantum-degenerate systems at finite temperatures, the fermion sign problem remains the major obstacle to first-principles simulations. In this work, we apply the recently proposed pseudo-fermion method - designed to overcome the sign problem - to strongly quantum-degenerate uniform electron gases. We find that the pseudo-fermion method can efficiently and highly accurately infer the energy of the uniform electron gas while being free from the fermion sign problem. For example, in the strongly quantum-degenerate regime where RPIMC fails (33 spin-polarized electrons at the density parameter $r_s = 0.5$), the relative deviation between the pseudo-fermion method and the exact CPIMC result is only 0.6%. In particular, the pseudo-fermion method bridges the gap where neither CPIMC nor RPIMC can accurately simulate the regime $1 \le r_s \le 2$ at the reduced temperature $\theta = 0.0625$. This work demonstrates that the pseudo-fermion method opens a new pathway for studying strongly quantum-degenerate systems in a sign-problem-free manner.

physics.comp-ph↗

arXiv2026

BFA++: Hierarchical Best-Feature-Aware Token Prune for Multi-View Vision Language Action Model

Vision-Language-Action (VLA) models have achieved significant breakthroughs by leveraging Large Vision Language Models (VLMs) to jointly interpret instructions and visual inputs. However, the substantial increase in visual tokens, particularly from multi-view inputs, poses serious challenges to real-time robotic manipulation. Existing acceleration techniques for VLMs, such as token pruning, often result in degraded performance when directly applied to VLA models, as they overlook the relationships between different views and fail to account for the dynamic and task-specific characteristics of robotic operation. To address this, we propose BFA++, a dynamic token pruning framework designed specifically for VLA models. BFA++ introduces a hierarchical pruning strategy guided by two-level importance predictors: an intra-view predictor highlights task-relevant regions within each image to suppress spatial noise, while an inter-view predictor identifies critical camera views throughout different manipulation phases to reduce cross-view redundancy. This design enables efficient token selection while preserving essential visual cues, resulting in improved computational efficiency and higher manipulation success rates. Evaluations on the RoboTwin benchmark and real-world robotic tasks demonstrate that BFA++ consistently outperforms existing methods. BFA++ improves the success rate by about 10% on both the {\pi}0 and RDT models, achieving speedup of 1.8X and 1.5X, respectively. Our results highlight that context-sensitive and task-aware token pruning serves as a more effective strategy than full visual processing, enabling faster inference and improved manipulation accuracy in real-world robotic systems.

cs.RO↗

arXiv2025

A Pseudo-Fermion Propagator Approach to the Fermion Sign Problem

In this work, within the framework of path integral Monte Carlo, we construct a pseudo-fermion propagator by replacing the original fermionic determinant with its absolute value. This modified propagator defines an auxiliary system free from the fermion sign problem, enabling efficient simulations of fermionic systems. We found that by shifting the pseudo-fermion energy based on the energy of a non-interacting fermion system, we can efficiently and reliably infer the energy of fermionic systems in various situations, from strong quantum degeneracy to weak quantum degeneracy. We have performed first-principles simulations of quantum dots confined in a two-dimensional harmonic potential and found excellent agreement with benchmark results provided by other established methods. We believe that this pseudo-fermion propagator framework opens up new possibilities for first-principles simulations of fermionic systems.

physics.comp-ph↗

arXiv2025

Study of the uniform electron gas through parametrized partition functions

We investigate the energy per particle, static structure factor, and momentum distribution of the uniform electron gas for different conditions defined by the dimensionless temperature $\Theta = 0.25 - 1.0$ and average interparticle distance $r_s = 0.5 - 80.0$ using path-integral Monte Carlo (PIMC) simulations. For small $r_\text{s}$ ($r_\text{s}\leq10$) where the sign problem is particularly challenging, we employ a recent approach based on an analytic continuation of the partition function using a real parameter $\xi$, which allows a generalization from bosons ($\xi=1$) to fermions ($\xi=-1$). We show that the results are in good agreement with other state-of-the-art methods while requiring low computational resources. For large $r_\text{s}$ ($r_\text{s}=80$), we use direct PIMC exploiting the good behaviour of the thermodynamic properties for negative $\xi$. In this framework we demonstrate that, for large $r_s$, the small negative region of $\xi$ can be utilized to extract information about the true fermionic limit, where $\xi = -1$.

cond-mat.mtrl-sci↗

arXiv2024

Ab initio simulation of the universal properties of unitary Fermi gas in a harmonic trap

Chang and Bertsch [Phys. Rev. A 76, 021603(R) (2007)] proposed a simple formula for the ground state energy of a unitary Fermi gas in a harmonic trap, based on their Green's function Monte Carlo simulations of up to 22 fermions, combined with general assumptions about the universal thermodynamic behavior of the unitary Fermi gas. In this work, we perform the ab initio simulations of the ground state energy of up to one hundred fermions using the fictitious identical particle method to overcome the Fermion sign problem, and we find that the formula proposed by Chang and Bertsch remains highly accurate. Since the number of fermions we simulate is much larger than that simulated by Chang and Bertsch when they proposed the formula, our work provides strong evidence for the universal validity of the formula. Our work demonstrates that fictitious identical particles provide a valuable tool for the ab initio simulations of ultracold Fermi gases.

cond-mat.quant-gas↗

arXiv2024

Ab initio simulations of the thermodynamic properties and phase transition of Fermi systems based on fictitious identical particles and physics-informed neural networks

Fictitious identical particle thermodynamics has emerged as a powerful tool to overcome the fermion sign problem, enabling highly accurate simulations of one thousand fermions in warm dense matter (T. Dornheim et al., J. Phys. Chem. Lett. 15, 1305 (2024)). However, inferring the thermodynamic properties of Fermi systems from a large number of exact numerical simulations of the bosonic sector still poses subtle challenges, especially in the regime of high quantum degeneracy and in the presence of phase transitions. In this work, we demonstrate that physics-informed neural networks (PINNs), trained on data from extensive and sign-problem-free numerical simulations of the bosonic sector, offer a valuable means to infer the thermodynamic properties of Fermi systems. PINNs can play a particularly crucial role in capturing phase transitions. To illustrate the methodology of fictitious identical particles combined with PINNs for simulating the thermodynamics of Fermi systems, we explore its application in realistic scenarios, including ultracold Fermi gases in periodic potentials, and phase transitions of pair condensation formed in the unitary limit in a three-dimensional harmonic trap. For the spatially continuous Fermi-Hubbard model, we efficiently and reliably simulated hundreds of fermions here. For the Fermi gas in the unitary limit, based on the fictitious identical particle combined with PINNs, our approach confirms the universal result of the critical temperature with the increasing of the number of fermions, and is consistent with the experimental observations.

cond-mat.quant-gas↗

arXiv2023

Quadratic scaling path integral molecular dynamics for fictitious identical particles and its application to fermion systems

Recently, fictitious identical particles have provided a promising way to overcome the fermion sign problem and have been used in path integral Monte Carlo (PIMC) to accurately simulate warm dense matter with up to 1000 electrons (T. Dornheim et al., arXiv:2311.08098 (2023)). The inclusion of fictitious identical particles in path integral molecular dynamics (PIMD) can provide another way to simulate fermion systems. In a recent paper (J. Chem. Phys. 159, 154107 (2023)), Feldman and Hirshberg improved the recursive formula for PIMD of N identical bosons, significantly reducing the computational complexity from $O(PN^3)$ to $O(N^2+PN)$. In this paper, we extend this latest recursive formula for bosons to PIMD of fictitious identical particles to improve the efficiency of simulating fermion systems. We also provide the virial estimator for calculating energy by using the recursive technique. As an example, we use the quadratic scaling PIMD for fictitious identical particles to study the simulation of hundreds of fermions in a two-dimensional periodic potential, in the hope of providing a simulation tool for two-dimensional Fermi-Hubbard model and other strongly correlated fermion systems, such as the simulation of ultracold fermionic gases in optical lattices.

cond-mat.quant-gas↗

arXiv2023

Robust and Efficient Interference Neural Networks for Defending Against Adversarial Attacks in ImageNet

The existence of adversarial images has seriously affected the task of image recognition and practical application of deep learning, it is also a key scientific problem that deep learning urgently needs to solve. By far the most effective approach is to train the neural network with a large number of adversarial examples. However, this adversarial training method requires a huge amount of computing resources when applied to ImageNet, and has not yet achieved satisfactory results for high-intensity adversarial attacks. In this paper, we construct an interference neural network by applying additional background images and corresponding labels, and use pre-trained ResNet-152 to efficiently complete the training. Compared with the state-of-the-art results under the PGD attack, it has a better defense effect with much smaller computing resources. This work provides new ideas for academic research and practical applications of effective defense against adversarial attacks.

cs.CV↗

arXiv2023

On the thermodynamics of fermions at any temperature based on parametrized partition function

In this work we study the recently developed parametrized partition function formulation and show how we can infer the thermodynamic properties of fermions based on numerical simulation of bosons and distinguishable particles at various temperatures. In particular, we show that in the three dimensional space defined by energy, temperature and the parameter characterizing parametrized partition function, we can map the energies of bosons and distinguishable particles to fermionic energies through constant-energy contours. We apply this idea to both noninteracting and interacting Fermi systems and show it is possible to infer the fermionic energies at all temperatures, thus providing a practical and efficient approach to obtain thermodynamic properties of Fermi systems with numerical simulation. As an example, we present energies and heat capacities for 10 noninteracting fermions and 10 interacting fermions (more fermions are provided in the appendix) and show good agreement with the analytical result for noninteracting case.

cond-mat.quant-gas↗

arXiv2022

Numerical simulation of two-component attractive Fermi gases based on parametrized partition function

The zero-temperature and finite-temperature thermodynamics of two-component Fermi gases with finite-range attractive interaction suffer from fermion sign problem, which seems like an insurmountable problem in exact numerical simulations. In a recent work, we find a reliable method to simulate the thermodynamic properties of single-component Fermi gases for both noninteracting and repulsively interacting cases based on the method of parametrized partition function and the $ξ_E$ curve of constant energy. In the present work, this method is generalized to two-component Fermi gases with finite-range attractive interaction, which shows clearly that our method has good chance to apply to various Fermi systems. From the simulated heat capacity, we find a peak at the temperature below the Fermi temperature which implies the pairing of fermions with different spin. At high temperature, the simulated heat capacity approaches the classical value. The reasonable result in this work validates the application of our method to attractive cases, which implies a wide range of applications, from nuclear physics, BCS-BEC crossover, superconductivity, to neutron star, etc..

cond-mat.quant-gas↗

arXiv2022

On the thermodynamic properties of fictitious identical particles and the application to fermion sign problem

By generalizing the recently developed path integral molecular dynamics for identical bosons and fermions, we consider the finite-temperature thermodynamic properties of fictitious identical particles with a real parameter $ξ$ interpolating continuously between bosons ($ξ=1$) and fermions ($ξ=-1$). Through general analysis and numerical experiments we find that the average energy may have good analytical property as a function of this real parameter $ξ$, which provides the chance to calculate the thermodynamical properties of identical fermions by an extrapolation with a simple polynomial function after accurately calculating the thermodynamic properties of the fictitious particles for $ξ\geq 0$. Using several examples, it is shown that our method can efficiently give accurate energy values for finite-temperature fermionic systems. Our work provides a chance to circumvent the fermion sign problem for some quantum systems.

cond-mat.stat-mech↗

arXiv2021

An Analytical Approach to the Universal Wave Function and its Gravitational Effect

Based on quantum origin of the universe, in this article we find that the universal wave function can be far richer than the superposition of many classical worlds studied by Everett. By analyzing the more general universal wave function and its unitary evolutions, we find that on small scale we can obtain Newton's law of universal gravity, while on the scale of galaxies we naturally derive gravitational effects corresponding to dark matter, without modifying any physical principles or hypothesizing the existence of new elementary particles. We find that an auxiliary function having formal symmetry is very useful to predict the evolution of the classical information in the universal wave function.

physics.gen-ph↗

arXiv2018

Abnormal Quantum Gravity Effect: Experimental Scheme with Superfluid Helium Sphere and Applications to Accelerating Universe

From the general assumption that gravity originates from the coupling and thermal equilibrium between matter and vacuum, after a derivation of Newton's law of gravitation and an interpretation of the attractive gravity force between two classical objects, we consider the macroscopic quantum gravity effect for particles whose wave packets are delocalized at macroscopic scale. We predict an abnormal repulsive gravity effect in this work. For a sphere full of superfluid helium, it is shown that with a gravimeter placed in this sphere, the sensitivities of the gravity acceleration $Δg/g$ below $10^{-8}$ could be used to test the abnormal quantum gravity effect, which satisfies the present experimental technique of atom interferometer, free-fall absolute gravimeters and superconducting gravimeters. We further propose a self-consistent field equation including the quantum effect of gravity. As an application of this field equation, we give a simple interpretation of the accelerating universe due to dark energy. Based on the idea that the dark energy originates from the quantum gravity effect of vacuum excitations due to the coupling between matter and vacuum, without any fitting parameter, the ratio between dark energy density and matter density (including dark matter) is calculated as 2.2, which agrees quantitatively with the result 7/3 obtained from various astronomical observations.

gr-qc↗

arXiv2018

Notes on the Origin of Gravity and the Laws of Newton

Following Verlinde's recent work on the origin of gravity and the laws of Newton, we consider further the origin of the starting point of Verlinde's work and Unruh effect. Simple derivations are given for this starting point and Unruh effect with the standard theory about quantum mechanics, special relativity and 'vacuum' background based on the gauge field theory. Our studies give the physical origin of the Planck length. The present work supports further the idea that gravity is not a fundamental force. In the present work, the concept of holographic screen is not absolutely necessary to derive the universal gravity and the laws of Newton.

hep-th↗

arXiv2018

New quantum gravity effect, dark energy, accelerating universe, black hole and experimental scheme using superfluid Helium and atom interferometer

Considerable attention has been focused on Verlinde's recent work, claiming that Newton's gravity is not a fundamental force. In a recent work (arXiv:1012.5858), we give further the logic basis and basic clues to derive the Newton's gravity, inertia law and Einstein's weak equivalence principle. In this work, we show that if the gravity is not a fundamental force, in special case, it could be repulsive when quantum wavepacket effect is considered. This quantum gravity effect leads to several physical effects: (1) It is consistent with the universe with accelerating expansion, if the gravity and quantum effect of the fluctuating 'vacuum' (dark energy) is considered. The role of the cosmological constant is naturally interpreted when the gravity and quantum effect of the whole 'vacuum' background is considered. (2) It leads to new idea about black hole information paradox, no-hair theorem and Hawking radiation. (3) With a sphere full of superfluid Helium, we propose a feasible experimental scheme to test our idea with an atom interferometer placed in the sphere. Our calculations show that the accuracy Delta g/g below 10^(-8) could be used to test our idea, which satisfies the present experimental technique of atom interferometer.

hep-th↗

arXiv2018

Repulsive gravitational effect of a quantum wave packet and experimental scheme with superfluid helium

We consider the gravitational effect of quantum wave packets when quantum mechanics, gravity, and thermodynamics are simultaneously considered. Under the assumption of a thermodynamic origin of gravity, we propose a general equation to describe the gravitational effect of quantum wave packets. In the classical limit, this equation agrees with Newton's law of gravitation. For quantum wave packets, however, it predicts a repulsive gravitational effect. We propose an experimental scheme using superfluid helium to test this repulsive gravitational effect. Our studies show that, with present technology such as superconducting gravimetry and cold atom interferometry, tests of the repulsive gravitational effect for superfluid helium are within experimental reach.

hep-th↗

arXiv2018

On the quantitative calculation of the cosmological constant of the quantum vacuum

It is widely believed that as one of the candidates for dark energy, the cosmological constant should relate directly with the quantum vacuum. Despite decades of theoretical effects, however, there is still no quantitative interpretation of the observed cosmological constant. In this work, we consider the quantum state of the whole universe including the quantum vacuum. Everett's relative-state formulation, vacuum quantum fluctuations and the validity of Einstein's field equation at macroscopic scales imply that our universe wave function might be a superposition of states with different cosmological constants. In the density matrix formulation of this quantum universe, the quasi-thermal equilibrium state is described by a specific cosmological constant with the maximum probability. Without any fitting parameter, the ratio between the vacuum energy density due to the cosmological constant (dark energy) and the critical density of the universe is 68.85% based on simple equations in our theoretic model, which agrees very well with the best current astronomical observations of 68.5%.

gr-qc↗