Memory-Aware Validated N-Body Integration: Centering, Structured Taylor Recurrences and Workspace Organization

This preprint presents an independent computational investigation of representation, interval enclosure quality, and memory efficiency in validated Newtonian N-body integration. The study documents a sequence of mathematical and computational experiments (MF01–MF26), comparing residual-based certification, algebraic force representations, coordinate centering, structured Taylor recurrences, precision selection, and workspace organization within a CAPD-based rigorous integration framework. The initial investigation shows that the independently developed residual-certification approach does not outperform an established CAPD reference implementation. Subsequent experiments demonstrate that suitable coordinate centering can substantially reduce the widths of rigorously enclosed physical endpoint intervals. Further investigations examine moving coordinate frames, relative-pair force representations, analytic Jacobian evaluation, complete first-variational Taylor recurrences, and precision-selection strategies. Several approaches improve isolated computations but do not yield corresponding benefits in complete validated integrations. The strongest implementation-level results arise from combining coordinate centering, structured pairwise Taylor recurrences, persistent workspaces, relative-variation caching, and a reduced auxiliary expression graph. In the MF26 scaling experiments, the resulting memory-aware implementation reduces median peak process memory by approximately 71.7% for a 48-body test, from 2,582.59 MiB to 731.12 MiB, compared with the prepared centered CAPD reference. Median process runtime decreases by approximately 19.2%, from 16.984 s to 13.724 s. A longer 16-body test spanning approximately 18.3 days shows a memory reduction of approximately 68.2%, accompanied by a 3.0% increase in median runtime. All 108 complete MF26 integration processes satisfy their prescribed accuracy targets. Independent algebraic and MPFR-based checks support the numerical and implementation audits within their explicitly documented scope. A three-factor ablation study identifies the reduction of the auxiliary expression graph as the primary source of memory savings. Persistent workspaces and relative-variation caching exhibit distinct time–memory trade-offs. The implementation retains the full first variational matrix, the required Taylor orders, and CAPD's rigorous enclosure machinery. Its storage complexity remains cubic in the number of bodies at fixed Taylor order; the observed improvement concerns the memory factor rather than the asymptotic storage exponent. The results constitute a reproducible implementation and benchmarking study for selected synthetic Newtonian N-body problems. They do not establish a new gravitational theory, a generally superior numerical integrator, or a CAPD-independent rigorous solver. The scientific novelty of the particular implementation combination remains subject to further literature review and independent assessment. Keywords: Validated Numerics; N-Body Integration; Interval Arithmetic; CAPD; Rigorous ODE Integration; Taylor Methods; Memory Optimization; Automatic Differentiation; Scientific Computing; Computational Reproducibility. Publication status: Independent research preprint; not peer-reviewed.

Authors

Publication Details

Journal
Zenodo (CERN European Organization for Nuclear Research)
Published
2026-11-15
DOI
https://doi.org/10.5281/zenodo.24260147
Primary Topic
Numerical Methods and Algorithms
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

Memory-Aware Validated N-Body Integration: Centering, Structured Taylor Recurrences and Workspace Organization

Stefan Beuchert
Zenodo (CERN European Organization for Nuclear Research)
Numerical Methods and Algorithms
preprint

Memory-Aware Validated N-Body Integration: Centering, Structured Taylor Recurrences and Workspace Organization

Stefan Beuchert
preprint en

Abstract

This preprint presents an independent computational investigation of representation, interval enclosure quality, and memory efficiency in validated Newtonian N-body integration. The study documents a sequence of mathematical and computational experiments (MF01–MF26), comparing residual-based certification, algebraic force representations, coordinate centering, structured Taylor recurrences, precision selection, and workspace organization within a CAPD-based rigorous integration framework. The initial investigation shows that the independently developed residual-certification approach does not outperform an established CAPD reference implementation. Subsequent experiments demonstrate that suitable coordinate centering can substantially reduce the widths of rigorously enclosed physical endpoint intervals. Further investigations examine moving coordinate frames, relative-pair force representations, analytic Jacobian evaluation, complete first-variational Taylor recurrences, and precision-selection strategies. Several approaches improve isolated computations but do not yield corresponding benefits in complete validated integrations. The strongest implementation-level results arise from combining coordinate centering, structured pairwise Taylor recurrences, persistent workspaces, relative-variation caching, and a reduced auxiliary expression graph. In the MF26 scaling experiments, the resulting memory-aware implementation reduces median peak process memory by approximately 71.7% for a 48-body test, from 2,582.59 MiB to 731.12 MiB, compared with the prepared centered CAPD reference. Median process runtime decreases by approximately 19.2%, from 16.984 s to 13.724 s. A longer 16-body test spanning approximately 18.3 days shows a memory reduction of approximately 68.2%, accompanied by a 3.0% increase in median runtime. All 108 complete MF26 integration processes satisfy their prescribed accuracy targets. Independent algebraic and MPFR-based checks support the numerical and implementation audits within their explicitly documented scope. A three-factor ablation study identifies the reduction of the auxiliary expression graph as the primary source of memory savings. Persistent workspaces and relative-variation caching exhibit distinct time–memory trade-offs. The implementation retains the full first variational matrix, the required Taylor orders, and CAPD's rigorous enclosure machinery. Its storage complexity remains cubic in the number of bodies at fixed Taylor order; the observed improvement concerns the memory factor rather than the asymptotic storage exponent. The results constitute a reproducible implementation and benchmarking study for selected synthetic Newtonian N-body problems. They do not establish a new gravitational theory, a generally superior numerical integrator, or a CAPD-independent rigorous solver. The scientific novelty of the particular implementation combination remains subject to further literature review and independent assessment. Keywords: Validated Numerics; N-Body Integration; Interval Arithmetic; CAPD; Rigorous ODE Integration; Taylor Methods; Memory Optimization; Automatic Differentiation; Scientific Computing; Computational Reproducibility. Publication status: Independent research preprint; not peer-reviewed.

Zenodo (CERN European Organization for Nuclear Research)
Numerical Methods and Algorithms
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon