Paper: arXiv 2507.08482
Authors: Rihito Sakurai, Koichi Miyamoto, Tsuyoshi Okubo
Abstract
Efficient computation of Greeks for multi-asset options remains a key challenge in quantitative finance. While Monte Carlo (MC) simulation is widely used, it suffers from the large sample complexity for high accuracy. We propose a framework to compute Greeks in a single evaluation of a tensor train (TT), which is obtained by compressing the Fourier transform (FT)-based pricing function via TT learning using tensor cross interpolation. Based on this TT representation, we introduce two approaches to compute Greeks: a numerical differentiation (ND) approach that applies a numerical differential operator to one tensor core and an analytical (AN) approach that constructs the TT of closed-form differentiation expressions of FT-based pricing. Numerical experiments on a five-asset min-call option in the Black-Sholes model show significant speed-ups of up to about \(10^{“5”} \times\) over MC while maintaining comparable accuracy. The ND approach matches or exceeds the accuracy of the AN approach and requires lower computational complexity for constructing the TT representation, making it the preferred choice.
Complexity vs Empirical Score
- Math Complexity: 8.5/10
- Empirical Rigor: 7.0/10
- Quadrant: Holy Grail — high math complexity, high empirical rigor
Why this score: The paper utilizes advanced tensor network mathematics (TT decompositions, TCI algorithms) and Fourier transform theory, indicative of high math complexity. It presents concrete empirical results with specific speed-up metrics (10^5x) and comparative accuracy assessments on a defined 5-asset option model, demonstrating implementation-heavy validation.
Research Flowchart
flowchart TD A["Research Goal"] --> B["Methodology"] B --> C["Computation"] C --> D["Outcomes"] A --> A1["Efficient Greeks for<br>Multi-Asset Options"] B --> B1["FT-Based Pricing"] B --> B2["TT Compression"] B1 --> B3["Data: Black-Scholes<br>5-Asset Min-Call"] B2 --> B3 C --> C1["ND: Numerical Differentiation"] C --> C2["AN: Analytical Differentiation"] C1 --> C3["Single TT Evaluation"] C2 --> C3 D --> D1["Speed-up: ~10<sup>5</sup>x<br>vs Monte Carlo"] D --> D2["Accuracy: MC Comparable"] D --> D3["Preferred: ND Approach"]
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 a framework that computes Greeks for multi-asset options via a single evaluation of a tensor train (TT) representation obtained by compressing the Fourier‑transform based pricing function.
Source sentence
We propose a framework to compute Greeks in a single evaluation of a tensor train (TT), which is obtained by compressing the Fourier transform (FT)-based pricing function via TT learning using tensor cross interpolation.
Assumes 4
- Underlying asset dynamics follow the Black‑Scholes model with constant risk‑free rate and volatilities. model
Source sentence
In this article, we consider the underlying asset prices⃗S(t) = (S1(t), · · · , Sd(t)) ∈Rd in the Black‑Scholes (BS) model described by the following stochastic differential equation dSm(t) = rSm(t)dt + σmSm(t)dWm(t)
- Correlations between Brownian motions are given by a time‑independent symmetric matrix. model
Source sentence
Here, W1(t), · · · , Wd(t) are Brownian motions whose pairwise correlations are determined by a time‑independent symmetric matrix ρ = (ρmn), namely dWm(t)dWn(t) = ρmndt.
- Tensor‑train low‑rank structure exists because correlations between distant variables are weak and nearby variables are strongly correlated. other
Source sentence
Assuming that correlations between distant variables are sufficiently weak while correlations between nearby variables are strong, tensor train, a kind of tensor network, can be applied to exploit this structure as follows.
- No closed‑form expressions are available for the price or Greeks of the min‑call option, so high‑precision Monte‑Carlo is used as ground truth. data
Source sentence
Because no closed‑form expressions are available for either the price or the Greeks of the min‑call option, we treat results obtained from high‑precision MC simulations as the ground truth.
Data
- Five‑asset min‑call option under Black‑Scholes synthetic unknown
Source sentence
We demonstrate both approaches through pricing a five‑asset min‑call option under the Black–Scholes (BS) model with different correlation settings dubbed constant, constant with noise, and random.
Costs
Not applicable
Finds 4
- Both TT‑based methods achieve speed‑ups of about 10^5 over Monte‑Carlo while maintaining comparable accuracy. performance
Source sentence
Numerical experiments on a five‑asset min‑call option in the Black‑Sholes model show significant speed‑ups of up to about $10^{5} \times$ over MC while maintaining comparable accuracy.
- The ND approach matches or exceeds the accuracy of the AN approach and has lower offline computational complexity. performance
Source sentence
The ND approach matches or exceeds the accuracy of the AN approach and requires lower computational complexity for constructing the TT representation, making it the preferred choice.
- TT‑based methods reduce computational complexity by approximately 10^3‑10^4 times and execution‑time by 10^4‑10^5 times compared to MC in the online phase. performance
Source sentence
These TT‑based methods reduce computational complexity by approximately 103 −104 times and achieve execution‑time speed‑ups on the order of 104−105 compared to MC.
- Accuracy declines in the random correlation case, especially for Vega, due to higher required bond dimensions. negative
Source sentence
Accuracy declines in the random correlation case, especially for Vega. This highlights two distinct factors influencing the accuracy of TT representations.
Checked by others
No other paper in the archive has been found to support or contradict these findings (compared with 64 related papers). Coverage is partial, so this is neither replication nor refutation.
1 paper in the archive
Code
No repository by the authors found in the paper.
Extracted 2026-11-14 · same card as JSON · A language model chose these items: check the source sentence before relying on one.
Cited References (69)
- A Highly Efficient Tensor Network Algorithm for Multi-Asset Fourier Options Pricing (2022) — link
- $O(d\log N)$ -Quantics Approximation of N-d Tensors in High-Dimensional Numerical Modeling (2011) — link
- Tensor Numerical Methods in Scientific Computing (2018) — link
- The ITensor Software Library for Tensor Network Calculations (2022) — link
- tf-quant-finance (2023)
- Options futures and other derivatives (2003)
- Stochastic calculus for finance I & II (2004)
- Quantum computing for finance (2023) — link
and 61 more
- Quantum speedup of Monte Carlo methods (2015) — link
- Option valuation using the fast Fourier transform (1999)
- A simple option formula for general jump-diffusion and other exponential L 'e vy processes (2001)
- A practical introduction to tensor networks: Matrix product states and projected entangled pair states (2014) — link
- Developments in the Tensor Network — from Statistical Mechanics to Quantum Entanglement (2022) — link
- Supervised Learning with Tensor Networks (2016)
- Exponential machines (2016) — link
- Tensor-Train Decomposition (2011) — link
- TT-cross approximation for multidimensional arrays (2010) — link
- Analysis of Fourier Transform Valuation Formulas and Applications (2010) — link
- Monte Carlo methods in financial engineering (2004)
- Tensor-Train Decomposition (2011) — link
- Efficient classical simulation of the approximate quantum Fourier transform (2007) — link
- Multiscale Space-Time Ansatz for Correlation Functions of Quantum Systems Based on Quantics Tensor Trains (2023) — link
- Parallel cross interpolation for high-precision calculation of high-dimensional integrals (2020) — link
- TT-cross approximation for multidimensional arrays (2010) — link
- Learning Feynman Diagrams with Tensor Trains (2022) — link
- Quantum Fourier Transform Has Small Entanglement (2023) — link
- Superfast Fourier Transform Using QTT Approximation (2012) — link
- Quantics Tensor Cross Interpolation for High-Resolution Parsimonious Representations of Multivariate Functions (2024) — link
- The density-matrix renormalization group in the age of matrix product states (2011) — link
- Quantum-Inspired Tensor Neural Networks for Partial Differential Equations (2022) — link
- Low-Rank Tensor Approximation for Chebyshev Interpolation in Parametric Option Pricing (2020) — link
- Solving high-dimensional partial differential equations using deep learning (2018) — link
- Pricing High-Dimensional Bermudan Options with Hierarchical Tensor Formats (2023) — link
- Quantum annealing in the transverse Ising model (1998) — link
- A quantum-inspired approach to exploit turbulence structures (2022)
- Quantum-inspired method for solving the Vlasov-Poisson equations (2022) — link
- Multiscale Space-Time Ansatz for Correlation Functions of Quantum Systems Based on Quantics Tensor Trains (2023)
- Linear image coding for regression and classification using the tensor-rank principle (2001) — link
- Multilinear Analysis of Image Ensembles: TensorFaces (2002)
- Tensor-train approximation of the chemical master equation and its application for parameter inference (2021) — link
- Learning tensor networks with tensor cross interpolation: New algorithms and libraries (2025)
- Numerical solution of the incompressible Navier-Stokes equations for chemical mixers via quantum-inspired Tensor Train Finite Element Method (2023) — link
- Quantum-inspired framework for computational fluid dynamics (2024)
- Hierarchical Tensor Approximation of Output Quantities of Parameter-Dependent PDEs (2015) — link
- Pricing with a smile (1994)
- A closed-form solution for options with stochastic volatility with applications to bond and currency options (1993) — link
- Managing Smile Risk (2002)
- Optimal damping with a hierarchical adaptive quadrature for efficient Fourier pricing of multi-asset options in Lévy models (2023)
- Quasi-Monte Carlo for Efficient Fourier Pricing of Multi-Asset Options — → 2403.02832
- The Variance Gamma (V.G.) Model for Share Market Returns (1990)
- Processes of normal inverse Gaussian type (1997)
- Deeply Learning Derivatives (2018) — link
- Accelerating Python for Exotic Option Pricing (2018)
- Failure of Fourier pricing techniques to approximate the Greeks — → 2306.08421
- Computing Greeks for L 'e vy Models: The Fourier Transform Approach (2016) — link
- Option Pricing Formulae using Fourier Transform : Theory and Application (2010) — link
- Efficient simulation of Greeks of multiasset European and Asian style options by Malliavin calculus and quasi-Monte Carlo methods (2014) — link
- Gauss-Kronrod quadrature formulae - A survey of fifty years of research (2016)
- Chebyshev and Fourier Spectral Methods (2001)
- Permutation of Tensor-Train Cores for Computing Moments on Stochastic Differential Equations (2025) — link
- Compactness of quantics tensor train representations of local imaginary-time propagators (2025) — link
- Adaptive sampling-based optimization of quantics tensor trains for noisy functions: applications to quantum simulations (2025) — link
- Theory of rational option pricing (1973)
- Levy processes and infinitely divisible distributions, by Ken-iti Sato. Pp. 486. £50. 1999. ISBN 0 521 55302 4 (Cambridge University Press). (2001)
- Analysis of Fourier Transform Valuation Formulas and Applications (2010) — link
- Spectral Methods in MATLAB (2000) — link
- Finitely correlated states on quantum spin chains (1992)
- Matrix Product Ground States for One-Dimensional Spin-1 Quantum Antiferromagnets (1993) — link
- Malliavin calculus in finance: Theory and practice (2021)
2 of 69 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2596 linked papers.