CAST: Causal Anchored Simplex Transport for Distribution-Valued Time Series

Jiecheng Lu, Jieqi Di, Runhua Wu, Yuwei Zhou · 2026

Read CAST: Causal Anchored Simplex Transport for Distribution-Valued Time Series by Jiecheng Lu, Jieqi Di, Runhua Wu, Yuwei Zhou (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-16919.txt.

Introduction

Lu's team studies forecasting for distribution-valued time series, where observations are shares or mixtures that live on the probability simplex. Their method, CAST, retrieves past successors from causal context, steadies them with a persistence anchor, and applies bounded local transport on ordered supports, keeping every stage on the simplex. They expose a failure mode, latent transition-kernel aliasing, where similar distributions evolve differently by regime, and prove summary-only forecasters carry an irreducible error floor. On eleven benchmarks, CAST ranks best on average for one-step and rollout scores.

Catalog record

CREATORJiecheng Lu, Jieqi Di, Runhua Wu, Yuwei Zhou
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE101412
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2605.16919

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

CAST: Causal Anchored Simplex Transport for Distribution-Valued Time Series Jiecheng Lu, Jieqi Di, Runhua Wu, Yuwei Zhou Abstract Many decision-facing stochastic systems are observed through aggregate distributions rather than scalar trajectories: queue occupancies, mobility shares, public-health mixtures, generation-source shares, ecological compositions, and air-quality severity profiles all live on the probability simplex and evolve over time. We study causal (time-respecting online) forecasting for these distribution-valued time series and argue that the transition operator itself should be structured around the simplex. We introduce CAST (Causal Anchored Simplex Transport), a successor-local operator that (i) retrieves empirical successors from causal context, (ii) stabilizes them with a persistence anchor, and (iii) applies a bounded local stochastic transport on ordered supports; every stage preserves the simplex by construction. We identify a structural failure mode, latent transition-kernel aliasing, where similar observed distributions evolve differently under different contextual regimes, and prove that any forecaster depending only on an aliased summary incurs an irreducible weighted Jensen–Shannon excess-risk lower bound, while the CAST hypothesis class contains the regime-aware Bayes successor; for ordered supports an additional Pinsker separation holds whenever the transported successor lies outside the no-transport anchor hull. On a suite of eleven public and simulated benchmarks spanning ecology, energy, diet, mortality, employment, air quality, severe weather, mobility, and G / G / 1 G/G/1 , G t / G / 1 G_{t}/G/1 queue occupancy, CAST achieves the best average rank on both one-step KL (1.27) and autoregressive rollout JSD (1.91), winning 8/11 sections on each metric against a broad statistical, compositional, recurrent, convolutional, Transformer, and modern time-series baseline set, and top-2 on all 11 sections for offline KL. Component ablations and a controlled synthetic aliasing experiment corroborate the theory. The code release is available at this link . 1 Introduction Modern stochastic systems are increasingly observed, controlled, and planned through aggregate distributions rather than isolated events or scalar trajectories. The rapid deployment of large language model services has in particular turned cloud inference into a high-throughput stochastic system with heterogeneous request sizes, variable service times, memory and cache constraints, and batching effects, placing queueing and scheduling at the center of high-performance inference [ 1 ] . For capacity planning, rebalancing, and risk-sensitive control, what matters is how the distribution of system states—queue length, workload, congestion regime—evolves over the next several intervals. The same pattern recurs across mobility platforms [ 2 ] , public-health panels, power systems, ecological studies, and air-quality severity reports. Distribution-valued forecasting. We formalize this by letting p_{t}\in\Delta^{D-1}:=\Bigl\{p\in\mathbb{R}_{\geq 0}^{D}:\textstyle\sum_{j=1}^{D}p(j)=1\Bigr\} denote a probability vector observed at time t , and studying the causal one-step map \hat{p}_{t+1}=f(p_{1:t}) that uses only present and past distributions. Here causal means time-respecting online prediction, not interventional causal inference. This is a compositional time series problem: observations are not arbitrary Euclidean vectors but proportions of a whole, and the simplex is the natural sample space [ 3 , 4 ] . Classical compositional forecasters apply log-ratio transformations, state-space recursions, exponential smoothing, or VAR dynamics in unconstrained coordinates and then map back to the simplex [ 5 , 6 , 7 ] ; Wasserstein autoregressive and functional distributional models treat probability distributions as serially dependent objects [ 8 , 9 ] . These provide important statistical baselines but leave a modeling question open: can the transition operator itself be structured as a causal map on the simplex? Why the full distribution. Forecasting the whole distribution provides a signal that a single trajectory forecast cannot. A future window may place substantial mass on both a light-load and a severe-congestion regime, or split across compositions with very different operational consequences. Distribution-valued forecasts are therefore the natural primitive for early warning, resource allocation, risk-sensitive planning, and distributionally robust optimization [ 10 , 11 , 12 , 13 ] . Queueing makes this concrete: a discrete-time single-server queue obeys L_{t+1}=\max\{0,\,L_{t}+A_{t}-C_{t}\} [ 14 , 15 , 16 ] , and recent work casts next-step queue-length distribution prediction as its own forecasting target on simulated G/G/1 and G_{t}/G/1 systems [ 17 ] . Setting p_{t}(k)\approx\mathbb{P}(L_{t}=k) turns the problem into one of forecasting how probability mass redistributes across an ordered support—directly exposing the probability of idleness, congestion, or overflow. The structural challenge: latent-kernel aliasing. In distribution-valued dynamics the same observed composition can arise from different latent transition mechanisms: a queue occupancy histogram may look identical while the system is entering high load or draining after a burst; a similar air-quality histogram may appear under different seasonal or regional regimes; similar ecological compositions may have different successors because the sites have different histories. In such cases the current distribution alone does not determine the next distribution, and a forecaster depending only on a fixed visible summary must collapse multiple latent kernels into one averaged successor. Existing deep time-series architectures—recurrent, convolutional, attention-based, residual-MLP, patching, inverted, multiscale, frequency-decomposed, cross-dimension, channel-mixing, exogenous-aware, foundation-model [ 18 , 19 , 20 , 21 , 22 , 23 , 24 , 25 , 26 , 27 , 28 , 29 , 30 , 31 , 32 , 33 , 34 , 35 , 36 , 37 , 38 , 39 , 40 ] —can be cross-applied to probability vectors, but their output heads enforce validity through a final softmax and do not encode empirical successor locality, latent-regime disambiguation, or ordered mass motion. Figure 1: CAST overview. (1) Real-world systems evolve as distributions on a simplex rather than scalar trajectories; (2) the same p_{t} can arise from histories with different successors, so (3) current-only forecasters collapse to a mixture average; (4) CAST encodes history, retrieves empirical successors r_{t} , forms a persistence–retrieval anchor a_{t}=\lambda_{t}p_{t}+(1-\lambda_{t})r_{t} , and applies bounded local transport T_{t} on ordered supports, \hat{p}_{t+1}=(1-\rho_{t})a_{t}+\rho_{t}T_{t}a_{t} , simplex-valid by construction; (5) this resolves regime aliasing and stabilizes rollout. CAST. We introduce CAST, a causal successor-local operator on the simplex (Figure 1 ). Given the causal history, CAST retrieves empirical successors from past context-successor pairs, forms a persistence-successor anchor, and, when the support is ordered, applies a bounded local stochastic transport operator: \hat{p}_{t+1}=(1-\rho_{t})\,a_{t}+\rho_{t}\,T_{t}a_{t},\qquad a_{t}=\lambda_{t}p_{t}+(1-\lambda_{t})\,r_{t}, where r_{t} is a retrieved causal successor, a_{t} is a persistence–retrieval anchor, T_{t} is a radius-bounded row-stochastic transport, and \rho_{t} is a bounded transport strength. Every stage preserves the simplex by construction. This structure encodes a simple inductive bias: future distributions should be close to empirical successors observed in similar causal contexts, stabilized by persistence and corrected only by small local mass motion when the support has a natural order. Retrieval addresses aliasing (two visible distributions with different histories have different successors); transport addresses ordered mass motion (on queue occupancies, AQI levels, ages, severity bins, probability mass should move locally, not through an unconstrained global correction). Contributions. (i) Model: we propose CAST, a causal successor-local transition model for distribution-valued time series, whose prediction is a convex combination of an empirical-successor anchor and a bounded local stochastic transport of that anchor on ordered supports. (ii) Theory: we prove a weighted Jensen–Shannon aliasing lower bound that any fixed-summary forecaster must pay when multiple latent regimes share the same observed summary, show that the CAST hypothesis class contains the regime-aware Bayes successor under mild memory assumptions, and give an ordered-support Pinsker separation for transports that leave the anchor hull; approximation bounds under imperfect retrieval/transport and simplex-closure/rollout-drift control complete the picture. (iii)

Read the complete text