{"total":19,"items":[{"citing_arxiv_id":"2606.30773","ref_index":2,"ref_count":1,"confidence":0.88,"is_internal_anchor":false,"paper_title":"Diffusion-warm sampling of the XY model enables fast thermalization at scale","primary_cat":"quant-ph","submitted_at":"2026-06-29T18:06:09+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":7.0,"formal_verification":"none","one_line_summary":"A temperature-conditioned diffusion model trained on small XY lattices produces accurate larger-lattice samples and cuts MCMC thermalization time by roughly 10x.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2606.30267","ref_index":7,"ref_count":1,"confidence":0.88,"is_internal_anchor":false,"paper_title":"Pathway variability, coat stiffening and mechanical adaptation during clathrin-mediated endocytosis","primary_cat":"q-bio.SC","submitted_at":"2026-06-29T13:15:19+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":7.0,"formal_verification":"none","one_line_summary":"Hybrid simulation and non-Euclidean elasticity theory demonstrate that clathrin coats develop adaptive rigidity and memory during growth, producing flat, stalled, or closed outcomes through two energy-landscape gates and matching experiments without fitted parameters.","context_count":1,"top_context_role":"background","top_context_polarity":"unclear","context_text":"[5] M. J. Bowick and L. Giomi. Two-Dimensional Matter: Order, Curvature and Defects.Adv. Phys., 58(5):449-563, 2009. [6] J. Bradbury, R. Frostig, P. Hawkins, M. J. John- son, Y. Katariya, C. Leary, D. Maclaurin, G. Nec- ula, A. Paszke, J. VanderPlas, S. Wanderman-Milne, and Q. Zhang. JAX: composable transformations of Python+NumPy programs, 2018. [7] F. M. Brodsky. Diversity of Clathrin Function: New Tricks for an Old Protein.Annu. Rev. Cell Dev. Biol., 28(1):309-336, 2012. [8] C. P. Broedersz and F. C. MacKintosh. Modeling semiflexible polymer networks.Rev. Mod. Phys., 86(3):995-1036, 2014. [9] D. Bucher, F. Frey, K. A. Sochacki, S. Kummer, J.-P. Bergeest, W. J. Godinez, H.-G. Kr ¨ausslich, K."},{"citing_arxiv_id":"2606.28489","ref_index":145,"ref_count":1,"confidence":0.88,"is_internal_anchor":false,"paper_title":"pop-cosmos: Galaxy size evolution across structural and star-formation classifications in COSMOS-Web","primary_cat":"astro-ph.GA","submitted_at":"2026-06-26T18:00:02+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":5.0,"formal_verification":"none","one_line_summary":"Galaxy size-mass relations exhibit double power-law breaks at different pivot masses for quiescent versus bulge-dominated samples, coinciding with AGN activity scales.","context_count":1,"top_context_role":"background","top_context_polarity":"unclear","context_text":"Dashes are used for parameters where we do not have galaxies to constrain them. To compare with the scale radius𝑅0 in the single-power law fits, instead of reporting𝛾, we report log10 (𝑅 ∗/kpc), which is the value of the scale radius𝑅50 at𝑀 ∗ =5×10 10 M⊙ in Equation 3. sample redshift log10 (𝑅 ∗/kpc)𝛼 𝛽 𝛿 𝜎int 𝑁gal 𝑁gal bin dex( ≤10.3 M ⊙ ) (>10.3 M ⊙ ) quiescent:[0.0,0.5)0.39±0.02 0.15±0.02>1.52(4.13)>11.56(12.08)0.15±0.01633 304 log10 (sSFR/yr −1 ) [0.5,1.0)0.19±0.01−0.15±0.05 1.07±0.21 10.76±0.18 0.14±0.01665 1315 ≤ −11[1.0,1.5)0.07±0.02−0.45±0.19 0.98±0.21 10.45±0.25 0.15±0.01221 898 [1.5,2.0) −0.01±0.02-0.94±0.34-0.15±0.0152 529 [2.0,3.0) −0.16±0.04->0.71(0.99)-0.14±0.0211 166 bulge-dominated:[0.0,0.5)0.40±0.02 0.24±0.10 † <1.47(7.05)>12.52(12.94) † 0.20±0.01454 205 B/T≥0.6[0.5,1.0)0.24±0.02−0.01±0.03 1.70±1.16 11.35±0.21 0.18±0.011298 1099 [1.0,1.5)0.19±0.01−0.10±0.05 0.97±0.30 10.83±0.26 0.18±0.011379 1519 [1.5,2.0)0.11±0.02−0.11±0.05>1.10(1.66)>11.04(11.31)0.19±0.011014 1016 [2.0,3.0)0.03±0.03−0.11±0.04>1.33(4.10)>11.42(11.96)0.21±0.011348 545 early-type:[0.0,0.5)0.42±0.03 0.28±0.03>0.74(2.89)>11.66(12.37)0.21±0.01396 350 𝑛 >2.5[0.5,1.0)0.27±0.01 0.08±0.03>1.15(1.82)>11.24(11.54)0.18±0.011189 1660 [1.0,1.5)0.18±0.02−0.01±0.03>1.52(2.99)>11.40(11.77)0.20±0.011450 1290 [1.5,2.0)0.13±0.02−0.06±0.02>1.38(3.58)>11.38(11.88)0.21±0.011247 981 [2.0,3.0)0.08±0.03−0.04±0.05>0.68(3.07)>11.34(12.08)0.22±0.011948 636 † Bimodal posterior, only the main mode reported. Table 7.Posterior median and 68 per cent credible interval for the parameters describing the size-redshift relation. samplelog 10 (𝐵 𝑧 /kpc)𝛽 𝑧 full0.64±0.02 0.71±0.07 star-forming:log 10 (sSFR/yr −1 )>−11 0.74±0.03 0.82±0.04 quiescent, linear:log10 (sSFR/yr −1 ) ≤ −11 0.52±0.06 1.12±0.09 quiescent, double-power law:log10 (sSFR/yr −1 ) ≤ −11 0.48±0.07 1.16±0.22 disc-dominated:B/T≤0.2 0.77±0.02 0.66±0.04 intermediate:0.2<B/T<0.6 0.59±0.03 0.81±0.07 bulge-dominated, linear:B/T≥0.6 0.49±0.06 0.74±0.09 "},{"citing_arxiv_id":"2606.25169","ref_index":2,"ref_count":2,"confidence":0.88,"is_internal_anchor":false,"paper_title":"Laplace-Fisher Gate Identities for Optimal Matrix-Gated Blended Score Estimation","primary_cat":"math.ST","submitted_at":"2026-06-23T21:00:32+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":7.0,"formal_verification":"none","one_line_summary":"The Laplace-Fisher Gate Identity supplies the variance-optimal matrix blending coefficients for Tweedie and target-score estimators under an OU diffusion, enabling improved finite-reference score estimation and posterior density surrogates.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2606.23326","ref_index":1,"ref_count":1,"confidence":0.88,"is_internal_anchor":false,"paper_title":"Online forecast reconciliation using linear models","primary_cat":"stat.ME","submitted_at":"2026-06-22T13:35:43+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":6.0,"formal_verification":"none","one_line_summary":"A framework for online forecast reconciliation is developed via multivariate linear models on graph hierarchies, ridge regression, and recursive least squares, with a demonstration on district heating load data.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2606.19148","ref_index":124,"ref_count":1,"confidence":0.88,"is_internal_anchor":false,"paper_title":"Fast Computation of Free-Support Wasserstein Medians","primary_cat":"stat.CO","submitted_at":"2026-06-17T14:50:29+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":7.0,"formal_verification":"none","one_line_summary":"Direct fixed-weight solver for free-support Wasserstein medians relocates atoms using OT barycentric projections and inverse-distance weights, achieving monotone descent on smoothed objectives with fewer subproblems than nested Weiszfeld baselines.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2606.11308","ref_index":143,"ref_count":1,"confidence":0.88,"is_internal_anchor":false,"paper_title":"pop-cosmos: Disentangling galaxy properties from observables using data-driven approaches","primary_cat":"astro-ph.GA","submitted_at":"2026-06-09T18:00:20+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":6.0,"formal_verification":"none","one_line_summary":"A beta-VAE analysis of pop-cosmos models finds that five latent dimensions capture the rest-frame optical SED, corresponding to stellar mass, recent star formation, dust, and two gas ionization states.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2606.00233","ref_index":80,"ref_count":1,"confidence":0.88,"is_internal_anchor":false,"paper_title":"Density Evolution: A Multiscale View of Density Estimation","primary_cat":"math.ST","submitted_at":"2026-05-29T18:08:31+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":5.0,"formal_verification":"none","one_line_summary":"A review reframing density estimation as 'density evolution' across scales, linking kernel smoothing to heat flow, mixtures to compression, and topology to level sets, while stating three structural results on modes, Gaussian semigroups, and log-concavity.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2606.00219","ref_index":4,"ref_count":1,"confidence":0.88,"is_internal_anchor":false,"paper_title":"21cmEMUv3: a hybrid diffusion-LSTM emulator of 21cmFAST summary observables","primary_cat":"astro-ph.CO","submitted_at":"2026-05-29T18:00:05+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":6.0,"formal_verification":"none","one_line_summary":"21cmEMUv3 emulates the cylindrical 21cm power spectrum via score-based diffusion and six other 21cmFAST observables via LSTM networks at sub-percent accuracy, then uses the emulator to infer a lower limit on soft-band X-ray luminosity from HERA data.","context_count":1,"top_context_role":"background","top_context_polarity":"unclear","context_text":"Parameter Distribution and limits Description log10 𝑓∗,10 N (−1.2,0.2) ∈ [−2,−0.5]SHMR normalisation for ACGs 𝛼∗ N (0.5,0.15) ∈ [0,1]power-law index of the SHMR 𝑡∗ N (0.55,0.3) ∈ [0.01,1]characteristic SF timescale log10 𝑓esc,10 N (−1.3,0.4) ∈ [−3,0]normalisation of the𝑓 esc −M h relation for ACGs 𝛼esc N (0,0.5) ∈ [−1,1]power-law index of the𝑓 esc −M h relation log10 𝑓∗,7 N (−2.5,0.8) ∈ [−4,−1]SHMR normalisation for MCGs log10 𝑓esc,7 N (−1.5,0.8) ∈ [−3,−1]normalisation of the𝑓 esc −M h relation for MCGs log10 𝐿acg X<2keV/SFRN (40.5,1) ∈ [38,43] erg s−1 𝑀⊙ yr−1 soft-band X-ray luminosity per unit SFR for ACGs log10 𝐿mcg X<2keV/SFRN (41.5,1) ∈ [39,44] erg s−1 𝑀⊙ yr−1 soft-band X-ray luminosity per unit SFR for MCGs 𝐸0 N (500,300) ∈ [100,1500]keV minimum energy of X-ray photons that can escape their host galaxy"},{"citing_arxiv_id":"2605.30330","ref_index":31,"ref_count":1,"confidence":0.88,"is_internal_anchor":false,"paper_title":"When, why, and how do diffusion posterior samplers fail? A finite-sample lens","primary_cat":"cs.LG","submitted_at":"2026-05-28T17:57:46+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":7.0,"formal_verification":"none","one_line_summary":"A finite-sample perspective reveals that inexact likelihood approximations cause under- or over-estimation of posterior spread at intermediate timesteps, leading to early-stopping sensitivity, mode weighting errors, and hallucinations even from multimodal priors alone.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2605.08976","ref_index":1,"ref_count":1,"confidence":0.88,"is_internal_anchor":false,"paper_title":"Score-Based Generative Modeling through Anisotropic Stochastic Partial Differential Equations","primary_cat":"cs.CE","submitted_at":"2026-05-09T14:36:05+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":6.0,"formal_verification":"none","one_line_summary":"Anisotropic SPDEs preserve geometric data structure over longer timescales in score-based generative modeling, yielding better image quality than standard SDE baselines and flow matching in unconditional and conditional tasks.","context_count":1,"top_context_role":"background","top_context_polarity":"background","context_text":"3 Score-based generative modeling Generative modeling operates as a two-pass procedure: In the first (forward) pass, the information in the data is systematically destroyed. The destruction process is described by a transformation process (Ut)t∈I evolving over a time domain I, which is either discrete, I={0, . . . , T} , or continuous, I= [0, T] , for someterminal time T∈[0,∞) . During this pass, the score (i.e. the gradient of the log-density) of the transformation process (Ut)t∈I is learned by a neural network. In thebackwardpass, the destroyed data is stochastically reconstructed, resulting in new, previously unseen samples resembling the original data. Because of this dual perspective, the transformation(U t)t∈I is called theforward process, while its time reversal U t :=U T−t fort∈I(1) is referred to as thebackward process. 3.1 Forward pass (data destruction) Generally, in generative modeling, the goal is to learn a data distribution µ and generate new samples that closely resemble the data. Conceptually, the data distribution µ is a probability measure on RD, where D is a finite index set. In practice, µ 3 SBGMTHROUGHANISOTROPICSPDES-PREPRINT- MAY12, 2026 is unknown and only implicitly represented through a dataset, which we treat as an independent and identically µ-distributed sequence. Specifically, inscore-basedgenerative modeling, the data is transformed stochastically into a progressively simpler representation, while thescore(i.e., the gradient of the log-density) of the resulting forward process is learned. Conceptually, the information contained in the data is gradually destroyed until reaching a sufficiently simple terminal representation, after which the objective is to generate new data resembling the original distribution. From an analytical perspective, the data is smoothed over time, thereby progressively simplifying the learning task - a principle commonly referred to asregularization by noise. 3.2 Learning the score function The score"},{"citing_arxiv_id":"2605.06538","ref_index":1,"ref_count":1,"confidence":0.88,"is_internal_anchor":false,"paper_title":"Diffusion-Based Posterior Sampling: A Feynman-Kac Analysis of Bias and Stability","primary_cat":"cs.LG","submitted_at":"2026-05-07T16:37:29+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":8.0,"formal_verification":"none","one_line_summary":"Diffusion posterior samplers produce biased outputs that can be expressed as an Ornstein-Uhlenbeck path expectation via a surrogate Gaussian path and Feynman-Kac representation, with STSL flattening the spatially varying bias term.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2604.25289","ref_index":34,"ref_count":1,"confidence":0.88,"is_internal_anchor":false,"paper_title":"Exploring Time Conditioning in Diffusion Generative Models from Disjoint Noisy Data Manifolds","primary_cat":"cs.LG","submitted_at":"2026-04-28T06:53:57+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":5.0,"formal_verification":"none","one_line_summary":"Aligning the DDIM forward diffusion process with flow-matching manifold evolution enables high-quality generation without time conditioning, and class-conditional synthesis is possible with an unconditional denoiser by using separate time spaces per class.","context_count":1,"top_context_role":"method","top_context_polarity":"use_method","context_text":"choice is deliberate: compared with stochastic samplers, DDIM is more sensitive to local geometric ambiguity between neighboring noisy manifolds, so it provides a particularly sharp test of the claims in Section 2 [ 65, 66, 69, 67]. If xi Exploring Time Conditioning in Diffusion Generative Models from Disjoint Noisy Data Manifolds 𝑁𝑜𝑛𝑒 𝒕𝑒𝑚𝑏𝒕𝑜𝑟𝑡ℎ𝑜 𝑈𝑛𝑖𝑓𝑜𝑟𝑚𝑟𝑎𝑑𝑖𝑎𝑙 𝐿𝑎𝑡𝑒 𝑒𝑥𝑝𝑎𝑛𝑠𝑖𝑜𝑛 Figure 8: Qualitative ImageNet result with DiT-B/2 [34] at 256×256 , DDIM sampling [15], and classifier-free guidance scale 5.5 [70, 71]. Columns from left to right are None, Uniform-radial, Late-expansion, tortho, and temb. Each row shows representative ImageNet classes. The None baseline collapses, while schedule-level manifold disjoint and orthogonal time-space separation recover clean samples without explicitly feeding the timestep into the denoiser."},{"citing_arxiv_id":"2604.21210","ref_index":2,"ref_count":1,"confidence":0.88,"is_internal_anchor":false,"paper_title":"The Feedback Hamiltonian is the Score Function: A Diffusion-Model Framework for Quantum Trajectory Reversal","primary_cat":"quant-ph","submitted_at":"2026-04-23T02:02:13+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":8.0,"formal_verification":"none","one_line_summary":"The García-Pintos feedback Hamiltonian equals the score function of the quantum trajectory distribution, linking quantum feedback to diffusion-model reversal.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2604.21180","ref_index":2,"ref_count":1,"confidence":0.88,"is_internal_anchor":false,"paper_title":"Uncertainty-Aware Spatiotemporal Super-Resolution Data Assimilation with Diffusion Models","primary_cat":"physics.flu-dyn","submitted_at":"2026-04-23T00:56:25+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":6.0,"formal_verification":"none","one_line_summary":"DiffSRDA uses denoising diffusion models to perform uncertainty-aware spatiotemporal super-resolution data assimilation, achieving EnKF-like quality from low-resolution forecasts on an ocean jet testbed.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2604.10465","ref_index":1,"ref_count":1,"confidence":0.88,"is_internal_anchor":false,"paper_title":"Rethinking the Diffusion Model from a Langevin Perspective","primary_cat":"cs.LG","submitted_at":"2026-04-12T05:18:07+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":5.0,"formal_verification":"none","one_line_summary":"Diffusion models are reorganized under a Langevin perspective that unifies ODE and SDE formulations and shows flow matching is equivalent to denoising under maximum likelihood.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2511.07571","ref_index":1,"ref_count":1,"confidence":0.88,"is_internal_anchor":false,"paper_title":"Forecasting implied volatility surface with generative diffusion models","primary_cat":"q-fin.CP","submitted_at":"2025-11-10T19:31:04+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":6.0,"formal_verification":"none","one_line_summary":"A conditioned diffusion model with SNR-weighted arbitrage penalty generates one-day-ahead arbitrage-free implied volatility surfaces and outperforms baselines on market data.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2506.02371","ref_index":3,"ref_count":1,"confidence":0.88,"is_internal_anchor":false,"paper_title":"SFBD Flow: A Continuous-Optimization Framework for Training Diffusion Models with Noisy Samples","primary_cat":"cs.LG","submitted_at":"2025-06-03T02:16:02+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":5.0,"formal_verification":"none","one_line_summary":"SFBD Flow converts the iterative SFBD approach into a continuous optimization framework for diffusion models on noisy samples, with its Online SFBD instantiation outperforming baselines.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2502.04575","ref_index":4,"ref_count":1,"confidence":0.88,"is_internal_anchor":false,"paper_title":"Complexity Analysis of Normalizing Constant Estimation: from Jarzynski Equality to Annealed Importance Sampling and beyond","primary_cat":"stat.ML","submitted_at":"2025-02-07T00:05:28+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":7.0,"formal_verification":"none","one_line_summary":"Derives Õ(d β² A² / ε⁴) oracle complexity for AIS estimating normalizing constant Z to relative error ε and introduces reverse diffusion sampler for geometric paths with large action.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null}],"limit":50,"offset":0}