BZPEERReading, trust and collaboration for research
Menu

Discover

GraphFollow connections around a paper, topic or saved view.

Workspaces

HomePick up where your research flow left off.InboxHandle requests, replies and notifications from one place.CollectionsCurate reading lists, evidence sets and shareable dossiers.ResearchSee your private provenance graph, trail and imports.CollaborationMove from discovery into focused discussion rooms.PublishDraft, preview and publish full-text research articles.

Network

ExpertsFind specialists worth following or asking for help.CommunitiesJoin research circles organized around shared work.PartnersSee external organizations active around the same topics.

Opportunities

ListingsBrowse roles, calls and collaborations with clearer fit signals.

Account

Log inReturn to your reading and collaboration workspace.Create accountStart saving work, following people and joining teams.
Log inCreate account

Researcher profile

Andrea Morandi

Andrea Morandi contributes to research discovery and scholarly infrastructure.

ResearcherAffiliation not importedOpen to collaborate
astro-ph.COComputation and LanguageMachine LearningArtificial IntelligenceComputational Engineering, Finance, and Sciencehep-ph

Trust snapshot

Quick read

Trust 21 - EmergingVerification L1Unclaimed author
8works
0followers
6topics
4close collaborators

Actions

Decide how to stay connected

Follow researcher0

Identity and collaboration

How to connect with this researcher

Claiming links this public author record to a researcher profile and unlocks direct collaboration workflows.

Log in to claim

Direct collaboration

Open a focused conversation when the fit is right

Claim this author entity first to unlock direct invitations.

Research graph

See the researcher in context

Open full explorer

Inspect adjacent work, topics, institutions and collaborators without jumping out to a separate graph page.

Building this graph slice

BZPEER is loading the nearby papers, people, topics and institutions for this page.

Published work

8 published item(s)

preprint2026arXiv

Correcting Selection Bias in Sparse User Feedback for Large Language Model Quality Estimation: A Multi-Agent Hierarchical Bayesian Approach

[Abridged] Production LLM deployments receive feedback from a non-random fraction of users: thumbs sit mostly in the tails of the satisfaction distribution, and a naive average over them can land 40-50 percentage points away from true system quality. We treat this as a topic- and sentiment- stratified selection-bias problem and propose a three-agent hierarchical Bayesian pipeline that does not require ground-truth labels on individual interactions. A Topic Clustering Agent partitions the stream via UMAP + HDBSCAN over text embeddings; a Bias Modeling Agent fits a two-stage hierarchical Beta-Binomial under NUTS, inferring per-topic selection rates $s_c$ and quality $q_c$ with partial pooling; a Synthesis Agent reweights $q_c$ by true topic prevalence $\hatπ_c = n_c/N$ to report a bias-corrected aggregate posterior $\bar Q = \sum_c \hatπ_c q_c$ with credible interval, plus drift signals for online recalibration. Validation uses UltraFeedback (N=10,232 retained interactions, $C=18$ clusters, $Q^\star=0.6249$) with simulated topic- and sentiment-dependent selection biases. We compare five Bayesian variants against Naive and IPW baselines. A mild prior on the feedback channel (typical positive-feedback rate and negative-to-positive ratio, both readable from any production dashboard without labels) keeps Hierarchical-Informed within 4-13 pp of $Q^\star$ as the bias ratio sweeps from 1:1 to 30:1, with 95% credible intervals covering $Q^\star$ in 50/50 random-seed replicates at $κ_{\max}=10$. Without channel-side priors, every weak-prior variant misses $Q^\star$ by 22-33 pp: the per-cluster sufficient statistics admit a one-parameter family of equally good fits, and the prior on the bias channel (not on latent quality) is what breaks the degeneracy.

0 citations0 reviewsNo code linkedOpen access
preprint2026arXiv

RTLC -- Research, Teach-to-Learn, Critique: A three-stage prompting paradigm inspired by the Feynman Learning Technique that lifts LLM-as-judge accuracy on JudgeBench with no fine-tuning

LLM-as-a-judge is now the default measurement instrument for open-ended generation, but on the public JudgeBench benchmark even strong instruction-tuned judges barely scrape past random on objective-correctness pairwise items. We introduce RTLC, a three-stage prompting recipe -- Research, Teach-to-Learn, Critique -- that promotes a single black-box LLM into an ensemble-of-thought judge with no fine-tuning, retrieval, or external tools. Stage 1 wraps the input in a fixed pedagogical scaffold porting the Feynman Learning Technique (study $\to$ teach $\to$ find gaps $\to$ simplify) into LLM prompting. Stage 2 draws N=10 independent candidate verdicts at temperature 0.4. Stage 3 acts as its own critic, cross-comparing the candidate set against the original question to emit one critiqued verdict at temperature 0. On JudgeBench-GPT (350 hard pairwise items), Claude 3.7 Sonnet's pairwise accuracy climbs from 64.6% (single-shot vanilla prompt) to 78.6% (RTLC critique-of-10) -- an absolute 14.0-percentage-point gain. RTLC also beats N=10 self-consistency majority voting (77.7%) and a zero-shot first candidate (74.0%). A clean three-step ablation attributes +9.4 pp to the Teach-to-Learn scaffold, +3.7 pp to N=10 marginalisation, and +0.9 pp to explicit critique. We discuss the cost-accuracy frontier (RTLC sits above self-consistency at every working point), the error-budget breakdown across the four JudgeBench categories (knowledge, reasoning, math, coding), and how RTLC composes orthogonally with post-hoc judge-score calibration, with the two interventions compounding multiplicatively in practice.

0 citations0 reviewsNo code linkedOpen access
preprint2026arXiv

Scale-Equivariant Generative Forecasting: Weight-Tied Dilated Convolutions, Wavelet Scattering Inputs, and Spectral-Consistency Training for Self-Similar Time Series

Many natural and engineered time series -- equity returns, climate anomalies, turbulent velocities, neural recordings, packet-level network traffic -- are approximately self-similar: their horizon-$T$ distribution is tied to the horizon-$1$ distribution by one scaling exponent $H$. Standard deep generative sequence models (transformers, dilated TCNs, the WaveNet family) ignore this. Their receptive fields are wide, but kernel parameters live independently at every dilation level, yielding a multi-scale architecture, not a scale-equivariant one. We make three contributions. First, we give a precise definition of discrete scale equivariance for 1D causal networks and prove that dyadic dilation commutes (up to boundary effects) with any dilated-convolution stack whose kernel weights are shared across levels. Tying the kernel shrinks the convolutional parameter budget by an $L$-fold factor (where $L$ is depth) and hard-wires self-similarity in as an inductive bias. Second, we wrap this Scale-Equivariant WaveNet (SE-WaveNet) backbone in three components that carry the same prior: a one-level Daubechies-4 wavelet input, a Hurst-FiLM block exposing the local scaling exponent, and a spectral-consistency training term targeting the $|f|^{-(2H+1)}$ power-law spectrum. The head is a conditional normalising flow, chosen to preserve equivariance. Third, on 30 years of S&P 500 daily log-returns, SE-WaveNet samples reproduce the empirical scaling-collapse diagnostic on the Allan-Variance top-25 universe (median $\mathcal{C}^\star = 0.020$), while a vanilla WaveNet at matched capacity does not ($\geq 0.06$). NLL, KS-calibration, and tail energy distance tie or beat the baseline, with $L\times$ fewer convolutional parameters.

0 citations0 reviewsNo code linkedOpen access
preprint2026arXiv

Sequential Consensus for Multi-Agent LLM Debates: A Wald-SPRT compute governor with calibration-based failure detection

Multi-agent LLM debate improves factuality and reasoning, but most recipes pick a fixed round count, over-spending on easy items and under-spending on hard ones. We adapt Wald's Sequential Probability Ratio Test (SPRT) as a plug-in compute governor for LLM debates. After each round, an LLM judge emits a [0,1] consensus score on the latest agent positions; a Wald monitor accumulates the log-likelihood ratio of "useful convergence" vs "not yet useful" under a Beta likelihood family, and stops when either boundary is crossed or returns a capped best-effort outcome at R_max. Under i.i.d. assumptions the rule inherits SPRT type-I/type-II error guarantees; in deployment the calibration itself is the more important object, since it estimates whether the judge score actually separates useful from unhelpful convergence in a given domain. We evaluate two tracks: (i) a Monte-Carlo study under calibrated Beta models characterising working curves, error rates, capping behaviour, and sensitivity; and (ii) a real-LLM evaluation on 200 attempted MMLU and 200 attempted GSM8K items with three heterogeneous agents (gpt-5, claude-opus-4-6, gemini-2.5-pro) and a claude-opus-4-6 judge, using disjoint 40-item calibration subsets. On GSM8K the rule stops in 1.01 average rounds (4.06 LLM calls) at 97.0% accuracy vs 99.0% for fixed-5 debate at 15 calls: a 3.7x call reduction at -2pp accuracy. On MMLU the calibrated KL collapses to about 0 and the rule caps on 99.5% of items at 2.1x cost. The takeaway is not that SPRT makes debate more accurate, but that a classical sequential test serves as a cheap compute-control and failure-detection layer for multi-agent LLM systems.

0 citations0 reviewsNo code linkedOpen access
preprint2026arXiv

Two Ways to De-Bias an LLM-as-a-Judge: A Continuous-Score Comparison of Hierarchical Bayesian Calibration and Neural-ODE Score Transport

[Abridged] Using a Large Language Model (LLM) as an automatic rater (LLM-as-a-judge) is cheap but potentially biased: some judges run lenient, others strict, the middle of the scale gets compressed, and verbose answers may be over-rewarded. A common remedy is post-hoc calibration: leave the cheap judge in place and, on a modest set of paired anchors, fit a transformation from raw judge scores to an estimate of the human rating. We compare two correctors that take opposing views on how this mapping should be modeled: a parametric, small-anchor hierarchical Bayesian linear correction with per-score uncertainty, and a non-parametric Neural-ODE (FFJORD) score-transport flow. Both are run head-to-head on UltraFeedback fine-grained_score (1700 paired examples, 200 held out), with calibration split into three operational sub-questions: population-mean recovery, per-item accuracy, and distributional-shape match. The headline result is that the choice between methods is primarily a data-budget question. Both correctors close the raw $+0.71$-point mean offset to within $\pm 0.08$ of the GPT-4 reference, at 100 and at 1500 anchors. Past that, the methods swap roles. With 100 anchors, the linear corrector reconstructs the human-score distribution roughly twice as well by KL divergence (0.031 vs. 0.058) and ties the flow on MAE. With 1500 anchors the flow wins on every metric (MAE 0.320 vs. 0.359, Pearson 0.922 vs. 0.896, KL 0.026 vs. 0.037). The Bayesian linear corrector saturates well below 1500 anchors: residual $\tanh$-shaped non-linearity is, by construction, structure a linear correction cannot fit. The flow keeps improving as labels grow. We translate these findings into an explicit decision rule for production deployments.

0 citations0 reviewsNo code linkedOpen access
preprint2012arXiv

Mass Reconstruction using Particle Based Lensing II: Quantifying substructure with Strong+Weak lensing and X-rays

We report a mass reconstruction of A1689 using Particle Based Lensing (PBL), a new technique for Strong+Weak lensing that allows a variable resolution depending on the data density and the signal-to-noise. Using PBL we also calculate the covariance matrix for the resulting mass map. The reconstruction of A1689 shows a secondary mass peak in the north-east direction confirming previous optical observations. This indicates that the central region of the cluster is still undergoing a weak merger. We have used this mass map to measure power ratios of the dark matter distribution and compared it to the X-ray distribution. We find that the power in the X-ray distribution is lower suggesting a smoother and rounder gas distribution compared to the dark matter distribution. We fitted an NFW profile to the profile derived from the mass map and we find that the lensing mass within 1 Mpc is $1.5\pm0.33\times 10^{15}M_\odot$. This is higher than the X-ray mass.

0 citations0 reviewsNo code linkedOpen access
preprint2011arXiv

Cluster-Cluster Lensing and the Case of Abell 383

Extensive surveys of galaxy clusters motivate us to assess the likelihood of cluster-cluster lensing (CCL), namely, gravitational-lensing of a background cluster by a foreground cluster. We briefly describe the characteristics of CCLs in optical, X-ray and SZ measurements, and calculate their predicted numbers for $Λ$CDM parameters and a viable range of cluster mass functions and their uncertainties. The predicted number of CCLs in the strong-lensing regime varies from several ($<10$) to as high as a few dozen, depending mainly on whether lensing triaxiality bias is accounted for, through the c-M relation. A much larger number is predicted when taking into account also CCL in the weak-lensing regime. In addition to few previously suggested CCLs, we report a detection of a possible CCL in A383, where background candidate high-$z$ structures are magnified, as seen in deep Subaru observations.

0 citations0 reviewsNo code linkedOpen access
preprint2011arXiv

Prediction for the neutrino mass in the KATRIN experiment from lensing by the galaxy cluster A1689

The KATRIN experiment in Karlsruhe Germany will monitor the decay of tritium, which produces an electron-antineutrino. While the present upper bound for its mass is 2 eV/$c^2$, KATRIN will search down to 0.2 eV$/c^2$. If the dark matter of the galaxy cluster Abell 1689 is modeled as degenerate isothermal fermions, the strong and weak lensing data may be explained by degenerate neutrinos with mass of 1.5 eV$/c^2$. Strong lensing data beyond 275 kpc put tension on the standard cold dark matter interpretation. In the most natural scenario, the electron antineutrino will have a mass of 1.5 eV/$c^2$, a value that will be tested in KATRIN.

0 citations0 reviewsNo code linkedOpen access