pith. sign in

arxiv: 1907.09162 · v1 · pith:V7Y3RUCNnew · submitted 2019-07-22 · 💰 econ.EM · math.PR· stat.CO· stat.ME

On the simulation of the Hawkes process via Lambert-W functions

Pith reviewed 2026-05-24 17:57 UTC · model grok-4.3

classification 💰 econ.EM math.PRstat.COstat.ME
keywords Hawkes processsimulationLambert-W functioninverse transform samplingpoint processeseconometricscomputational efficiency
0
0 comments X

The pith

The inverse transform sampling method for Hawkes processes, when expressed using Lambert-W functions, yields a faster implementation than recent alternatives.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper revives the 1979 inverse sampling transform for generating realizations of the Hawkes process by showing it can be solved through Lambert-W functions. This mathematical reformulation supports an optimized algorithm whose author reports superior runtime performance over more recent simulation techniques. A sympathetic reader cares because Hawkes processes appear in econometrics and other fields where repeated simulation is needed, and faster methods reduce the cost of such work. The contribution lies in making the oldest approach practically competitive again rather than replacing it with entirely new machinery.

Core claim

The inverse sampling transform approach to simulating the Hawkes process can be conveniently discussed in terms of Lambert-W functions, and an optimized implementation of this approach is computationally more performing than more recent alternatives available for the simulation of the Hawkes process.

What carries the argument

Lambert-W function reformulation of the inverse sampling transform applied to the cumulative intensity of the Hawkes process.

If this is right

  • The Lambert-W formulation supplies a concrete, implementable algorithm for Hawkes process simulation that revives the inverse transform method.
  • Under equivalent coding conditions this algorithm runs faster than more recent published alternatives.
  • Faster simulation directly reduces the wall-clock time required for Monte Carlo studies that rely on many realizations of the process.
  • The same Lambert-W inversion step applies to any Hawkes process whose cumulative intensity admits an explicit inverse expressible via the Lambert-W function.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • If the speed advantage generalizes, large-scale econometric simulations that currently use newer methods could switch to this older route without loss of accuracy and with lower compute cost.
  • The approach invites similar Lambert-W treatments for other self-exciting point processes whose intensity integrals are not analytically invertible by elementary functions.
  • Independent re-implementations in languages other than the original would test whether the reported advantage survives differences in library support for the Lambert-W function.

Load-bearing premise

The performance advantage rests on the assumption that the Lambert-W implementation and the benchmarked alternatives are coded and timed under equivalent conditions without hidden optimizations or hardware-specific advantages.

What would settle it

A controlled runtime benchmark on the same machine and in the same programming language, with both the Lambert-W method and a recent alternative implemented at comparable optimization levels, in which the Lambert-W version does not complete faster.

read the original abstract

Several methods have been developed for the simulation of the Hawkes process. The oldest approach is the inverse sampling transform (ITS) suggested in \citep{ozaki1979maximum}, but rapidly abandoned in favor of more efficient alternatives. This manuscript shows that the ITS approach can be conveniently discussed in terms of Lambert-W functions. An optimized and efficient implementation suggests that this approach is computationally more performing than more recent alternatives available for the simulation of the Hawkes process.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 0 minor

Summary. The manuscript reformulates the inverse transform sampling (ITS) method for simulating Hawkes processes in terms of Lambert-W functions and asserts that an optimized implementation of this approach is computationally more efficient than more recent alternatives.

Significance. If the performance claim is substantiated with reproducible benchmarks, the work could revive a simple closed-form simulation technique for Hawkes processes, offering practical value in applications such as financial modeling and seismology where fast, accurate simulation of self-exciting point processes is needed.

major comments (2)
  1. [Abstract] Abstract: the central claim that the Lambert-W ITS implementation 'is computationally more performing than more recent alternatives' is presented without any benchmark details, timing protocol, error metrics, hardware specification, or list of compared methods, so the empirical superiority cannot be evaluated.
  2. [Abstract] The performance advantage rests on the unstated assumption that all compared implementations (Lambert-W ITS and the 'more recent alternatives') were coded, optimized, and timed under equivalent conditions; no evidence or protocol is supplied to support this equivalence.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the careful review and constructive feedback on the abstract. We agree that the performance claims would benefit from additional context and will revise the manuscript to address the points raised.

read point-by-point responses
  1. Referee: [Abstract] Abstract: the central claim that the Lambert-W ITS implementation 'is computationally more performing than more recent alternatives' is presented without any benchmark details, timing protocol, error metrics, hardware specification, or list of compared methods, so the empirical superiority cannot be evaluated.

    Authors: The abstract summarizes the key finding; the supporting benchmarks, including timing protocol, error metrics, hardware specifications, and compared methods (e.g., thinning and other referenced alternatives), appear in Section 4. We will revise the abstract to briefly reference the benchmark setup and reproducibility details from the numerical experiments. revision: yes

  2. Referee: [Abstract] The performance advantage rests on the unstated assumption that all compared implementations (Lambert-W ITS and the 'more recent alternatives') were coded, optimized, and timed under equivalent conditions; no evidence or protocol is supplied to support this equivalence.

    Authors: All methods were implemented and timed by the authors in the same computational environment to ensure equivalence. We will revise the manuscript to explicitly state this and refer readers to the detailed protocol in the numerical experiments section. revision: yes

Circularity Check

0 steps flagged

No circularity: reformulation and empirical timing claim are independent of inputs

full rationale

The paper reformulates the inverse transform sampling method for Hawkes simulation in terms of Lambert-W functions and reports runtime comparisons for an optimized implementation. No derivation step equates a claimed result to its own fitted parameters or prior self-citation; the Lambert-W equivalence is a direct algebraic rewriting of the existing ITS integral, and the performance claim is an external benchmark rather than a quantity defined in terms of itself. The central content remains self-contained against external timing measurements.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract-only review supplies no explicit free parameters, axioms, or invented entities.

pith-pipeline@v0.9.0 · 5589 in / 917 out tokens · 43189 ms · 2026-05-24T17:57:36.958247+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.