Paper: arXiv 2307.12843
Authors: Gero Junike, Hauke Stier
Abstract
We provide a unified framework to obtain numerically certain quantities, such as the distribution function, absolute moments and prices of financial options, from the characteristic function of some (unknown) probability density function using the Fourier-cosine expansion (COS) method. The classical COS method is numerically very efficient in one-dimension, but it cannot deal very well with certain integrands in general dimensions. Therefore, we introduce the damped COS method, which can handle a large class of integrands very efficiently. We prove the convergence of the (damped) COS method and study its order of convergence. The method converges exponentially if the characteristic function decays exponentially. To apply the (damped) COS method, one has to specify two parameters: a truncation range for the multivariate density and the number of terms to approximate the truncated density by a cosine series. We provide an explicit formula for the truncation range and an implicit formula for the number of terms. Numerical experiments up to five dimensions confirm the theoretical results.
Complexity vs Empirical Score
- Math Complexity: 8.5/10
- Empirical Rigor: 5.0/10
- Quadrant: Holy Grail — high math complexity, high empirical rigor
Why this score: The paper employs dense mathematical analysis including multivariate characteristic functions, convergence proofs, and the damped COS method, scoring high on math complexity. It includes numerical experiments up to five dimensions and provides explicit/implicit formulas for parameters, indicating strong empirical validation, warranting a high rigor score.
Research Flowchart
flowchart TD
A["Research Goal:<br>Develop unified framework for<br>multivariate density & option pricing<br>from characteristic function"] --> B{"Key Methodology<br>Damped COS Method"}
B --> C["Inputs:<br>Characteristic Function<br>Truncation Range L<br>Number of Terms N"]
C --> D["Computation:<br>Fourier-Cosine Expansion<br>& Numerical Integration"]
D --> E{"Convergence Analysis"}
E -->|"Exponential Decay"| F["Outcome:<br>Exponential Convergence"]
E -->|"General Integrands"| G["Outcome:<br>Efficient Multivariate<br>Density Estimation<br>up to 5D"]
F --> H["Final Outcomes:<br>European Option Prices<br>Absolute Moments<br>Distribution Functions"]
G --> H
Evidence Card
What the authors claim, on what data, and who has checked it. Read from the paper's own text by a language model: every quote is verbatim, nothing was re-run. How to read this card
Proposes
The paper proposes the damped COS method for multivariate integrals and provides convergence proofs and formulas for truncation range and number of terms.
Source sentence
We introduce the damped COS method, which is able to avoid the expensive application of the discrete cosine transform if the Fourier transform of the function of interest is given in closed form.
Assumes 2
- Assumption A1: there exists a damping factor α such that δg ∩ δ∞w is non‑empty and α belongs to this intersection. model
Source sentence
Assumption A1. We assume that δg ∩δ∞ w̸ = ∅and α ∈δg ∩δ∞ w .
- Assumption A2: there exists a damping factor α such that δg ∩ δw is non‑empty and α belongs to this intersection. model
Source sentence
Assumption A2. We assume that δg ∩δw̸ = ∅and α ∈δg ∩δw.
Data
- Multivariate normal distribution (synthetic) synthetic unknown
Source sentence
Table 2: Truncation ranges L = (L, ..., L) for error tolerance ε = 10−4 for a multivariate normal distribution evaluated at y = (4.60517, ..., 4.60517) ...
- Variance Gamma (VG) distribution (synthetic) synthetic unknown
Source sentence
We approximate the CDF for the VG(a, s, η, θ, σ) distribution in d = 3 dimensions with parameters a = 10, s = 0.1, η = 0, θ = (−0.03, −0.03, −0.03) and σ = (0.2, 0.2, 0.2) at 1000 different y ∈Rd by the classical…
Costs
Not applicable
Finds 3
- Both the classical and damped COS methods converge exponentially when the Fourier transform of the (damped) density decays exponentially. theoretical
Source sentence
damped COS methods converge exponentially if the Fourier transform of the (damped) density decays exponentially.
- In two dimensions the classical COS‑i and damped COS‑iv methods are 500–5000 times faster than crude Monte Carlo for an absolute error tolerance of about 10⁻². performance
Source sentence
We observe that the classical COS-i and the damped COS-iv methods in d = 2 dimensions are significantly faster than MC, approximately by a factor between 500 and 5000.
- For dimensions d ≤ 4 and error tolerance ε ≤ 10⁻³ the damped COS‑iv method outperforms Monte Carlo; for d = 5 it only outperforms MC when ε ≤ 10⁻⁵. performance
Source sentence
We observe that the damped COS-iv method is faster than MC for d ≤4 and ε ≤10−3. For d = 5, the damped COS-iv method outperforms MC for ε ≤10−5; otherwise, a MC simulation is faster.
Checked by others
No other paper in the archive has been found to support or contradict these findings (compared with 16 related papers). Coverage is partial, so this is neither replication nor refutation.
4 papers in the archive : Quasi-Monte Carlo with Domain Transformation for Efficient … ; Fast calculation of Counterparty Credit exposures and …
Extracted 2026-11-14 · same card as JSON · A language model chose these items: check the source sentence before relying on one.