搜索结果：fast

We present fast-vollib, an open-source Python library that provides high-performance European option pricing, implied volatility (IV) computation, and Greeks under the Black-76, Black-Scholes, and Black-Scholes-Merton models. The library is designed as a drop-in alternative to the de-facto-standard py_vollib and py_vollib_vectorized packages, with pluggable PyTorch and JAX execution backends, a CUDA fused-kernel Triton contribution for batched IV workloads, and a compatibility-first public API. In addition to a vectorized Halley-method IV solver, fast-vollib ships an experimental, fully-vectorized implementation of Jäckel's "Let's Be Rational" (LBR) algorithm with NumPy/Numba, torch.compile, JAX, and Triton single-pass GPU kernels for batched option chains. This note announces the library and describes its public API surface, with source, documentation, and packaging artifacts available at: GitHub (https://github.com/raeidsaqur/fast-vollib), Docs (https://raeidsaqur.github.io/fast-vollib/), PyPI (https://pypi.org/project/fast-vollib/).

Large Chunk Test-Time Training (LaCT) has shown strong performance on long-context 3D reconstruction, but its fully plastic inference-time updates remain vulnerable to catastrophic forgetting and overfitting. As a result, LaCT is typically instantiated with a single large chunk spanning the full input sequence, falling short of the broader goal of handling arbitrarily long sequences in a single pass. We propose Elastic Test-Time Training inspired by elastic weight consolidation, that stabilizes LaCT fast-weight updates with a Fisher-weighted elastic prior around a maintained anchor state. The anchor evolves as an exponential moving average of past fast weights to balance stability and plasticity. Based on this updated architecture, we introduce Fast Spatial Memory (FSM), an efficient and scalable model for 4D reconstruction that learns spatiotemporal representations from long observation sequences and renders novel view-time combinations. We pre-trained FSM on large-scale curated 3D/4D data to capture the dynamics and semantics of complex spatial environments. Extensive experiments show that FSM supports fast adaptation over long sequences and delivers high-quality 3D/4D reconstruc

Large language models (LLMs) are trained for downstream tasks by updating their parameters (e.g., via RL). However, updating parameters forces them to absorb task-specific information, which can result in catastrophic forgetting and loss of plasticity. In contrast, in-context learning with fixed LLM parameters can cheaply and rapidly adapt to task-specific requirements (e.g., prompt optimization), but cannot by itself typically match the performance gains available through updating LLM parameters. There is no good reason for restricting learning to being in-context or in-weights. Moreover, humans also likely learn at different time scales (e.g., System 1 vs 2). To this end, we introduce a fast-slow learning framework for LLMs, with model parameters as "slow" weights and optimized context as "fast" weights. These fast "weights" can learn from textual feedback to absorb the task-specific information, while allowing slow weights to stay closer to the base model and persist general reasoning behaviors. Fast-Slow Training (FST) is up to 3x more sample-efficient than only slow learning (RL) across reasoning tasks, while consistently reaching a higher performance asymptote. Moreover, FST-

Autoregressive video diffusion models enable streaming generation, opening the door to long-form synthesis, video world models, and interactive neural game engines. However, their core attention layers become a major bottleneck at inference time: as generation progresses, the KV cache grows, causing both increasing latency and escalating GPU memory, which in turn restricts usable temporal context and harms long-range consistency. In this work, we study redundancy in autoregressive video diffusion and identify three persistent sources: near-duplicate cached keys across frames, slowly evolving (largely semantic) queries/keys that make many attention computations redundant, and cross-attention over long prompts where only a small subset of tokens matters per frame. Building on these observations, we propose a unified, training-free attention framework (FAST-AR) for FAST-AutoRegressive diffusion, consisting of three components: TempCache compresses the KV cache via temporal correspondence to bound cache growth; AnnCA accelerates cross-attention by selecting frame-relevant prompt tokens using fast approximate nearest neighbor (ANN) matching; and AnnSA sparsifies self-attention by restri

Von Economo neurons (VENs) are large bipolar projection neurons found exclusively in the anterior cingulate cortex (ACC) and frontal insula of species with complex social cognition, including humans, great apes, and cetaceans. Their selective depletion in frontotemporal dementia (FTD) and altered development in autism implicate them in rapid social decision-making, yet no computational model of VEN function has previously existed. We introduce the Fast Lane Hypothesis: VENs implement a biological speed-accuracy tradeoff (SAT) by providing a sparse, fast projection pathway that enables rapid social decisions at the cost of deliberate processing accuracy. We model VENs as fast leaky integrate-and-fire (LIF) neurons with membrane time constant 5 ms and sparse dendritic fan-in of eight afferents, compared to 20 ms and eighty afferents for standard pyramidal neurons, within a spiking cortical circuit of 2,000 neurons trained on a social discrimination task. Networks are evaluated under three clinically motivated conditions across 10 independent random seeds: typical (2% VENs), autism-like (0.4% VENs), and FTD-like (post-training VEN ablation). All configurations achieve equivalent asymp

Vision-Language-Action (VLA) tasks require reasoning over complex visual scenes and executing adaptive actions in dynamic environments. While recent studies on reasoning VLAs show that explicit chain-of-thought (CoT) can improve generalization, they suffer from high inference latency due to lengthy reasoning traces. We propose Fast-ThinkAct, an efficient reasoning framework that achieves compact yet performant planning through verbalizable latent reasoning. Fast-ThinkAct learns to reason efficiently with latent CoTs by distilling from a teacher, driven by a preference-guided objective to align manipulation trajectories that transfers both linguistic and visual planning capabilities for embodied control. This enables reasoning-enhanced policy learning that effectively connects compact reasoning to action execution. Extensive experiments across diverse embodied manipulation and reasoning benchmarks demonstrate that Fast-ThinkAct achieves strong performance with up to 89.3% reduced inference latency over state-of-the-art reasoning VLAs, while maintaining effective long-horizon planning, few-shot adaptation, and failure recovery.

We introduce a deep generative framework for high-dimensional Bayesian inference that enables efficient posterior sampling. As telescopes and simulations rapidly expand the volume and resolution of astrophysical data, fast simulation-based inference methods are increasingly needed to extract scientific insights. While diffusion-based approaches offer high-quality generative capabilities, they are hindered by slow sampling speeds. Our method performs posterior sampling an order of magnitude faster than a diffusion baseline. Applied to the problem of CMB delensing, it successfully recovers the unlensed CMB power spectrum from simulated observations. The model also remains robust to shifts in cosmological parameters, demonstrating its potential for out-of-distribution generalization and application to observational cosmological data.

Constraint-based causal discovery relies on repeated conditional independence tests, but fast nonparametric tests often sacrifice calibration, especially when variables depend on the conditioning set through nonlinear relationships. We introduce BLITZ (Broad-to-Local Independence Testing via residualiZation), a nonparametric conditional independence test designed to run well under a second while maintaining the accuracy needed for the thousands of queries performed by constraint-based causal discovery algorithms. BLITZ first removes broad smooth dependence on the conditioning set using low-order polynomial regression, then applies a small nonlinear feature map and residualizes those features with shallow tree regressions. The resulting statistic tests residual cross-covariance, with a moment-matched chi-square approximation to the null distribution. We show theoretically that the two-stage design reduces the effective complexity faced by the tree residualizers, allowing shallow trees to control residual conditional-mean bias while avoiding excessive overfitting. In simulations, BLITZ provides better null calibration than fast kernel, random-feature, and regression-based competitors

We rigorously prove the existence and uniqueness of fast traveling pulse solutions to the singularly perturbed neural field system with linear feedback and Heaviside nonlinearity structure within a spatial convolution. Although a long-standing open problem, the pulse is well-accepted to often exist based on its original singular construction, closed form when it exists, and follow-ups, but prior to this study, there has not been a proof that overcomes the difficulties of (i) solving for the fast speed and width functions using the implicit function theorem at $ε=0$ and (ii) tracking the resultant formal homoclinic orbit near its singular orbit during fast, slow, and mixed time scales. First, we provide new, first-order approximations of the pulse speed and width. We then show that the formal pulse is close to the front and back at threshold crossing points and that the Hausdorff distance between the formal pulse and the singular homoclinic orbit converges to zero, demonstrating that the formal pulse is a true solution to the original system. Broadly, our methods provide insight into nonlocal geometric singular perturbation theory.

Autoregressive sequence models, such as Transformer-based vision-language action (VLA) policies, can be tremendously effective for capturing complex and generalizable robotic behaviors. However, such models require us to choose a tokenization of our continuous action signals, which determines how the discrete symbols predicted by the model map to continuous robot actions. We find that current approaches for robot action tokenization, based on simple per-dimension, per-timestep binning schemes, typically perform poorly when learning dexterous skills from high-frequency robot data. To address this challenge, we propose a new compression-based tokenization scheme for robot actions, based on the discrete cosine transform. Our tokenization approach, Frequency-space Action Sequence Tokenization (FAST), enables us to train autoregressive VLAs for highly dexterous and high-frequency tasks where standard discretization methods fail completely. Based on FAST, we release FAST+, a universal robot action tokenizer, trained on 1M real robot action trajectories. It can be used as a black-box tokenizer for a wide range of robot action sequences, with diverse action spaces and control frequencies.

We detail the emission behaviors of three long-period pulsars detected using the Five-hundred-meter Aperture Spherical radio Telescope (FAST) during the CRAFTS survey. Their rotational periods range from 1.83 s to 4.75 s, and the null fractions measure between 28% and 53%. PSR J1945+1211 and PSR J2323+1214 exhibited quasi-periodic nulls, with duration of around 57 seconds. The longest null was observed in PSR J1945+1211, lasting 76 seconds. PSR J2323+1214 displayed varying null fractions between its leading and trailing components. For the first time in PSR J2323+1214, we detected five dwarf pulses, which are much weaker and narrower pulses than typical burst pulses. In addition, we investigate the microstructure of PSR J1900-0134 for the first time, revealing intricate pulses of up to 2.05 milliseconds and noting its complex emission characteristics. Bright pulses occur in all of these sources at different rates. These observations suggest complex magnetospheric processes, potentially related to magnetic reconnections, and provide insights into the origins of bright and microstructure pulses, as well as their distinctions from ordinary pulses.

Diffusion models have recently advanced Combinatorial Optimization (CO) as a powerful backbone for neural solvers. However, their iterative sampling process requiring denoising across multiple noise levels incurs substantial overhead. We propose to learn direct mappings from different noise levels to the optimal solution for a given instance, facilitating high-quality generation with minimal shots. This is achieved through an optimization consistency training protocol, which, for a given instance, minimizes the difference among samples originating from varying generative trajectories and time steps relative to the optimal solution. The proposed model enables fast single-step solution generation while retaining the option of multi-step sampling to trade for sampling quality, which offers a more effective and efficient alternative backbone for neural solvers. In addition, within the training-to-testing (T2T) framework, to bridge the gap between training on historical instances and solving new instances, we introduce a novel consistency-based gradient search scheme during the test stage, enabling more effective exploration of the solution space learned during training. It is achieved

In this work we address the analysis of discrete-time models of structured metapopulations subject to environmental stochasticity. Previous works on these models made use of the fact that migrations between the patches can be considered fast with respect to demography (maturation, survival, reproduction) in the population. It was assumed that, within each time step of the model, there are many fast migration steps followed by one slow demographic event. This assumption allowed one to apply approximate reduction techniques that eased the model analysis. It is however a questionable issue in some cases since, in particular, individuals can die at any moment of the time step. We propose new non-equivalent models in which we re-scale survival to consider its effect on the fast scale. We propose a more general formulation of the approximate reduction techniques so that they also apply to the proposed new models. We prove that the main asymptotic elements in this kind of stochastic models, the Stochastic Growth Rate (SGR) and the Scaled Logarithmic Variance (SLV), can be related between the original and the reduced systems, so that the analysis of the latter allows us to ascertain the po

We present a detailed single-pulse study of the long-period pulsar PSR J2129+4119 using high-sensitivity FAST observations. Despite locating well below the traditional death line, the pulsar exhibits sustained and multi-modal emission behavior, including nulls, weak pulses, regular emission, and occasional bright pulses. The nulling fraction is measured to be $8.13\% \pm 0.51\%$, with null durations typically under four pulse periods. Fluctuation spectral analysis reveals both phase-modulated subpulse drifting and intermittent beat-like modulation. At the same time, polarization profiles show high linear polarization and stable polarization position angle (PPA) swings consistent with a near-tangential sightline geometry. Quasi-periodic microstructures are detected in 11.54\% of regular pulses, with a mean periodicity and width of 4.57 ms and 4.30 ms, respectively. A well-defined scintillation arc in the secondary spectrum confirms the presence of a localized scattering screen. These results indicate that PSR J2129+4119 remains magnetospherically active and coherently emitting despite its low energy loss rate, offering key insights into pulsar emission physics near the death line.

Neutral hydrogen (HI) is the fundamental component of the interstellar medium. The Galactic Plane Pulsar Snapshot (GPPS) survey is designed for hunting pulsars by using the Five-hundred-meter Aperture Spherical radio Telescope (FAST) from the visible Galactic plane within $|b| \leq 10^{\circ}$. The survey observations are conducted with the L-band 19-beam receiver in the frequency range of 1.0 $-$ 1.5 GHz, and each pointing has an integration time of 5 minutes. The piggyback spectral data simultaneously recorded during the FAST GPPS survey are great resources for studies on the Galactic HI distribution and ionized gas. We process the piggyback HI data of the FAST GPPS survey in the region of $33^{\circ} \leq l \leq 55^{\circ}$ and $|b| \leq 2^{\circ}$. The rms of the data cube is found to be approximately 40 mK at a velocity resolution of $0.1$ km s$^{-1}$, placing it the most sensitive observations of the Galactic HI by far. The high velocity resolution and high sensitivity of the FAST GPPS HI data enable us to detect weak exquisite HI structures in the interstellar medium. HI absorption line with great details can be obtained against bright continuum sources. The FAST GPPS survey

The FAST All Sky HI survey (FASHI) was designed to cover the entire sky observable by the Five-hundred-meter Aperture Spherical radio Telescope (FAST), spanning approximately 22000 square degrees of declination between -14 deg and +66 deg, and in the frequency range of 1050-1450 MHz, with the expectation of eventually detecting more than 100000 HI sources. Between August 2020 and June 2023, FASHI had covered more than 7600 square degrees, which is approximately 35% of the total sky observable by FAST. It has a median detection sensitivity of around 0.76 mJy/beam and a spectral line velocity resolution of ~6.4 km/s at a frequency of ~1.4 GHz. As of now, a total of 41741 extragalactic HI sources have been detected in the frequency range 1305.5-1419.5 MHz, corresponding to a redshift limit of z<0.09. By cross-matching FASHI sources with the Siena Galaxy Atlas (SGA) and the Sloan Digital Sky Survey (SDSS) catalogs, we found that 16972 (40.7%) sources have spectroscopic redshifts and 10975 (26.3%) sources have only photometric redshifts. Most of the remaining 13794 (33.0%) HI sources are located in the direction of the Galactic plane, making their optical counterparts difficult to id

The Five-hundred-meter Aperture Spherical radio Telescope (FAST) has been running for several years. A new Ultra-Wide Bandwidth (UWB) receiver, simultaneously covering 500-3300 MHz, has been mounted in the FAST feed cabin and passed a series of observational tests. The whole UWB band is separated into four independent bands. Each band has 1048576 channels in total, resulted in a spectral resolution of 1 kHz. At 500-3300 MHz, the antenna gain is around 14.3-7.7 K/Jy, the aperture efficiency is around 0.56-0.30, the system temperature is around 88-130 K, and the HPBW is around 7.6-1.6 arcmin. The measured standard deviation of pointing accuracy is better than ~7.9 arcsec, when zenith angle (ZA) is within 26.4deg. The sensitivity and stability of the UWB receiver are confirmed to satisfy expectation by spectral observations, e.g., HI and OH. The FAST UWB receiver already has a good performance for taking sensitive observations in various scientific goals.

Neural fields have revolutionized the area of 3D reconstruction and novel view synthesis of rigid scenes. A key challenge in making such methods applicable to articulated objects, such as the human body, is to model the deformation of 3D locations between the rest pose (a canonical space) and the deformed space. We propose a new articulation module for neural fields, Fast-SNARF, which finds accurate correspondences between canonical space and posed space via iterative root finding. Fast-SNARF is a drop-in replacement in functionality to our previous work, SNARF, while significantly improving its computational efficiency. We contribute several algorithmic and implementation improvements over SNARF, yielding a speed-up of $150\times$. These improvements include voxel-based correspondence search, pre-computing the linear blend skinning function, and an efficient software implementation with CUDA kernels. Fast-SNARF enables efficient and simultaneous optimization of shape and skinning weights given deformed observations without correspondences (e.g. 3D meshes). Because learning of deformation maps is a crucial component in many 3D human avatar methods and since Fast-SNARF provides a co

Pulsar polarization profiles are very basic database for understanding the emission processes in pulsar magnetosphere. After careful polarization calibration of the 19-beam L-band receiver and verification of beam-offset observation results, we obtain polarization profiles of 682 pulsars from observations by the Five-hundred-meter Aperture Spherical radio Telescope (FAST) during the survey tests for the Galactic Plan Pulsar Snapshot (GPPS) survey and other normal FAST projects. Among them, polarization profiles of about 460 pulsars are observed for the first time. The profiles exhibit diverse features. Some pulsars have a polarization position angle curve with a good S-shaped swing, and some with orthogonal modes; some have components with highly linearly components or strong circularly polarized components; some have a very wide profile, coming from an aligned rotator, and some have an interpulse from a perpendicular rotator; some wide profiles are caused by interstellar scattering. We derive geometry parameters for 190 pulsars from the S-shaped position angle curves or with orthogonal modes. We find that the linear and circular polarization or the widths of pulse profiles have va

We report two phenomena detected in PSR J0344$-$0901 from two observations conducted at frequency centered at 1.25 GHz using the Five-hundred-meter Aperture Spherical radio Telescope (FAST). The first phenomenon manifests as shifting in the pulse emission to later longitudinal phases and then gradually returns to its original location. The event lasts for about 216 pulse periods, with an average shift of about $0.7^\circ$ measured at the peak of the integrated profile. Changes in the polarization position angle (PPA) are detected around the trailing edge of the profile, together with an increase in the profile width. The second phenomenon is characterized by the apparent movement of subpulses, which results in different subpulse track patterns across the profile window. For the first time in this pulsar, we identify four emission modes, each with unique subpulse movement, and determine the pattern periods for three of the emission modes. Pulse nulling was not detected. Modeling of the changes in the PPA using the rotating vector model gives an inclination angle of $75.12^\circ \pm 3.80^\circ$ and an impact parameter of $-3.17^\circ \pm 5.32^\circ$ for this pulsar. We speculate that