Latent Laplace Diffusion for Irregular Multivariate Time Series
Zinuo You, Jin Zheng, John Cartlidge · 2026
Read Latent Laplace Diffusion for Irregular Multivariate Time Series by Zinuo You, Jin Zheng, John Cartlidge (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-2605-19805.txt.
Introduction
You, Zheng and Cartlidge present Latent Laplace Diffusion for uneven many sided time series and long range outlooks. Discrete methods distort time structure by new time grids, while smooth models drift through one by one solvers. Their model treats the target as a latent path, made across the horizon without step by step summing. A modal setup from random port rules, with mean change in the Laplace domain, handles uneven time marks. Tests beat rivals on long range outlooks, and it fills missing values at time marks.
Catalog record
| CREATOR | Zinuo You, Jin Zheng, John Cartlidge |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 116039 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2605.19805
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Latent Laplace Diffusion for Irregular Multivariate Time Series Zinuo You, Jin Zheng, John Cartlidge Abstract Irregular multivariate time series impose a trade-off for long-horizon forecasting: discrete methods can distort temporal structure via re-gridding, while continuous-time models often require sequential solvers prone to drift. To bridge this gap, we present Latent Laplace Diffusion (LLapDiff), a generative framework that models the target as a low-dimensional latent trajectory, enabling horizon-wide generation without step-by-step integration over physical time. We guide the reverse process utilizing a stable modal parameterization motivated by stochastic port-Hamiltonian dynamics, and parameterize its mean evolution in the Laplace domain via learnable complex-conjugate poles, enabling direct evaluation over irregular timestamps. We also link continuous dynamics to irregular observations through renewal-averaging analysis, which maps sampling gaps to effective event-domain poles and motivates a gap-aware history summarizer. Extensive experiments show that LLapDiff improves over baselines in long-horizon forecasting, and its continuous-time generative nature supports missing-value imputation by querying the same model at historical timestamps. Code is available at https://github.com/pixelhero98/LLapDiffusion . 1 Introduction Real-world multivariate time series spanning healthcare, climate science, and finance are rarely observed on a clean and synchronous grid. Instead, measurements arrive at nonuniform timestamps and exhibit variable-wise missingness that is often informative ( 44 ; 34 ) . Modeling in this regime requires respecting physical time and sampling gaps without reducing irregularity to a mere preprocessing artifact, while producing stable long-horizon trajectories. Conversely, unconstrained rollouts easily drift or diverge, particularly as gaps become larger and available observations become sparse. Most existing methods for irregular multivariate time series (IMTS) commonly fall into three families. First, discrete-time pipelines interpolate or impute onto a regular grid ( 6 ; 37 ) , or discretize events into tokens/patches to apply strong sequence models ( 48 ; 50 ) . While efficient, aggressive re-gridding beyond native-bin indexing can distort timing information under severe irregularity and blur the semantics of missingness ( 28 ) . Second, continuous-time models, such as Neural ODE/CDE families ( 9 ; 31 ; 22 ) , continuous-time RNNs ( 34 ) , and continuous-time Transformers ( 10 ) , incorporate timestamps naturally. But, these methods typically require sequential numerical integration, increasing cost and accumulating error over long horizons ( 5 ; 45 ) . Third, diffusion/score-based generative models ( 20 ; 40 ) provide strong uncertainty quantification and sample quality, and have been adapted to forecasting and imputation by conditioning denoising on historical context ( 41 ; 29 ) . However, most time-series diffusion methods denoise directly in the observation space and handle irregularity mainly through masks and time embeddings. Consequently, the denoisers lack explicit dynamical structure or stability control, making long-horizon generation fragile under irregular sampling ( 1 ; 23 ) . In the face of these challenges, we present Latent Laplace Diffusion (LLapDiff), a conditional generative model that combines continuous-time inductive bias without ODE/SDE solver calls. Here, solver-free refers to avoiding numerical integration over the physical time; the reverse process is iterative only over diffusion noise levels. First, LLapDiff represents irregular target horizons as low-dimensional latent trajectories and performs diffusion in this latent space, avoiding direct denoising over sparse masked observations. Then, to guide the reverse process toward stable dynamics without numerical solvers, we draw inspiration from stochastic port-Hamiltonian systems ( 33 ; 43 ) and parameterize the latent mean evolution in the Laplace domain ( 2 ) via stable complex-conjugate poles ( 19 ) , enabling horizon-wide generation (sampling still relies on reverse steps) over irregular grids. Finally, we analyze irregular sampling via a renewal-averaging perspective ( 12 ; 3 ) , showing how random gaps map continuous-time poles to effective event-domain poles. This further motivates a gap-aware history summarizer that aggregates observations and time gaps to adapt learned dynamics to the sampling pattern. More broadly, LLapDiff formulates forecasting as continuous latent trajectory generation, which offers missing-value imputation as a byproduct by simply generating the trajectory at historical query timestamps. Our contributions are threefold. First, we present LLapDiff, a latent diffusion framework for irregular multivariate time series that models its evolution as low-dimensional latent trajectories while preserving timestamp fidelity. Second, we derive a Laplace-domain modal parameterization grounded on stochastic port-Hamiltonian dynamics, allowing horizon-wide generation with explicit stability control through constrained pole damping. Third, we establish a theoretical link between continuous-time poles and random sampling gaps, which motivates a gap-aware history conditioning that improves long-horizon modeling under varying irregularities. 2 Related Work Irregular Time Series Modeling . Discrete-time pipelines often align observations to a regular grid via interpolation or recurrence ( 6 ; 8 ; 13 ) and apply strong sequence backbones on tokenized representations ( 47 ; 7 ) . Although efficient, aggressive re-gridding can distort temporal structure under severe irregularity and entangle missingness with preprocessing artifacts. Meanwhile, other discrete approaches avoid gridding by incorporating time gaps via attention ( 38 ; 24 ) or irregularity-aware encoding ( 42 ; 11 ) . In parallel, graph-based forecasters ( 50 ; 46 ) construct temporal graphs but become fragile under severe sparsity due to weak connectivity. While continuous-time approaches, including Neural ODE/CDE ( 31 ; 22 ; 26 ) and continuous Transformers ( 10 ) , handle timestamps naturally but rely on costly sequential integration. Structured dynamical parameterizations, like SSMs ( 18 ; 17 ) , offer efficient long-context inference and have been adapted in time-series ( 1 ) . Diffusion for Time Series . Diffusion-based methods learn conditional generation by denoising with historical context. Forecasting models such as TimeGrad ( 29 ) and TSDiff ( 23 ) have been improved via non-autoregressive ( 36 ) and multi-resolution architectures ( 35 ) , while imputation methods such as CSDI ( 41 ) perform well under structured missingness. However, many existing approaches denoise directly in observation space, treating irregularity primarily as a masking problem. This leaves dynamical structure implicit ( 32 ) , which can lead to unstable long-horizon trajectories as models must relearn consistency without physical constraints. Physics-informed Priors for Deep Learning . Physics-inspired architectures encode conservation and dissipation structure (e.g., Hamiltonian formulations ( 16 ) ) to provide stability-oriented inductive biases, with recent variants enforcing global Lyapunov stability by construction ( 30 ) . Complementary works connect generative modeling to PDEs via first-principles loss functions that encourage samples to satisfy physical constraints ( 4 ) . Furthermore, neural Laplace models ( 21 ) enable solver-free evaluation for differential equations. LLapDiff unifies these threads by applying diffusion to latent trajectories and utilizing a stable modal parameterization to achieve horizon-wide generation over irregular grids, enabling forecasting while also supporting imputation via trajectory queries at queried timestamps. 3 Problem Formulation Preliminary . Consider N entities observed at some physical timestamps \{t_{j}\}_{j=1}^{\mathcal{T}} . At each observation time t_{j} , we have inputs \bm{X}_{t_{j}}\in\mathbb{R}^{N\times d_{x}} , targets (often a subset of inputs) \bm{Y}_{t_{j}}\in\mathbb{R}^{N\times d_{y}} , and their masks \bm{M}^{x}_{t_{j}}\in\{0,1\}^{N\times d_{x}},\bm{M}^{y}_{t_{j}}\in\{0,1\}^{N\times d_{y}} . For a timestamp t_{i} , we define the observed history \mathcal{H}_{t_{i}}=\{(t_{j},\bm{X}_{t_{j}},\bm{M}^{x}_{t_{j}})\}_{j=i-\ell+1}^{i} and targets \mathcal{Y}_{t_{i}}=\{(t^{q}_{r},\bm{Y}_{t^{q}_{r}},\bm{M}^{y}_{t^{q}_{r}})\}_{r=1}^{h} corresponding to h queries \{t^{q}_{r}\}_{r=1}^{h} . These queries may be future points (forecasting) or historical timestamps with missing entries (imputation). Latent target representation . We represent the target as a low-dimensional latent trajectory \bm{z}\in\mathbb{R}^{h\times d_{z}} on a latent manifold. During training, we encode ground-truth targets \bm{z}=\mathrm{VAE}_{\mathrm{enc}}(\mathcal{Y}_{t_{i}}) with a