Pith. sign in

REVIEW 3 major objections 3 minor 1 cited by

Tensor train representations of Greeks for Fourier-based pricing of multi-asset options

T0 review · 3 major / 3 minor · reviewed 2026-08-06 · deepseek-v4-flash

Pith's one-line read After an offline tensor-train build, Vega, Delta, and Gamma each cost one tensor contraction — the same as the price — with online speed-ups of about 10^4 to 10^5 over Monte Carlo.

desk verdict Solid but overstated: TT-based Greeks are a useful extension, yet the 'comparable accuracy' claim fails for Vega under random correlation, and the small-Vega caveat is untested. read the letter →

arxiv 2507.08482 v1 pith:6URJHN5L submitted 2025-07-11 q-fin.CP quant-ph

classification q-fin.CPquant-ph MSC 91G6065D0515A6965K05
keywords tensortraincrossinterpolationGreeksFourier-basedoptionpricingmulti-assetoptionsBlack-Scholesnumericaldifferentiationmin-call
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper proposes a method to compute option Greeks—derivatives of the price with respect to volatility or initial asset price—for multi-asset options almost for free after a one-time tensor-train compression of the Fourier-based pricing function. The central claim is that applying a numerical differentiation operator to one tensor core leaves bond dimensions unchanged, so evaluating any of the tested Greeks costs no more than evaluating the price itself. An alternative analytical approach compresses closed-form derivative expressions into separate tensor trains, but the numerical approach is simpler, cheaper to build offline, and generally at least as accurate. For a five-asset min-call option under the Black–Scholes model, the authors report online complexity reductions of roughly $10^{3}$ to $10^{4}$ and runtime speed-ups of about $10^{4}$ to $10^{5}$ over Monte Carlo while keeping errors comparable. The practical consequence is fast large-scale risk calculations when parameters move frequently.

What carries the argument

The central object is the tensor train (TT), a product of small three-way cores whose contraction approximates a high-dimensional tensor with small bond dimensions. The tensor cross interpolation (TCI) algorithm builds it by adaptively sampling a tiny fraction of entries. The authors put the local indices in the interleaved order $(z_i, \sigma_i, S^0_i)_{i=1}^d$, bake the parameter dependence into the cores of the characteristic function and the payoff tensor, sum over the Fourier variables, and contract adjacent cores to get a TT whose output is the price on a parameter grid. The numerical-differentiation mechanism is the matrix $D$ applied only to the $\kappa$-th core, yielding the Greek TT while preserving the bond dimensions, so the online cost stays $O(d\chi^2)$.

What would settle it

Take a ten-asset min-call option with a correlation matrix whose off-diagonal entries are all large, use the paper's interleaved TT ordering, and measure the largest bond dimension of the characteristic function at tolerance $10^{-6}$. If the bond dimension grows exponentially in the asset count or exhausts memory, the single-contraction online-cost claim fails for that setting. A simpler check is to compare TT ranks under the paper's ordering against those under a random asset ordering on a small exact tensor.

Watch

Extended reading notes

Core claim

The paper establishes that once a tensor-train representation of the option price is built with the volatility and initial-price grids as local indices, Eqs. (4.10)–(4.12) compute the first- and second-order Greeks by contracting only one tensor core with a numerical differentiation matrix. The bond dimensions remain exactly those of the price tensor train, so each Greek evaluation has the same asymptotic cost as a single price evaluation. In contrast, the analytical approach multiplies the price tensor-train by tensor-train representations of the factors that appear in the closed-form Greek formulas, which increases bond dimensions, especially for Vega. Numerical results for the five-asset min-call with constant, noisy, and random correlation matrices show that both approaches match or improve on a $10^{6}$-path Monte Carlo benchmark in accuracy for price, $\Delta$, and Gamma, while Vega is slightly less accurate only in the random-correlation case and only where Vega is small.

Load-bearing premise

The method's speed-up rests on the assumption that the characteristic function and the payoff tensor, after the interleaved asset ordering, have low tensor-train rank that tensor cross interpolation can recover at tolerance $10^{-6}$; in the random-correlation case this assumption is visibly strained, with characteristic-function bond dimensions growing to about 400.

Editorial extensions

If this is right

  • Once the offline TT is built, evaluating a Greek on the parameter grid costs the same as evaluating the price, which enables fast batch risk computations for many parameter sets.
  • The numerical-differentiation approach avoids building a separate TT for each Greek, so it is cheaper in the offline phase while matching or exceeding the analytical approach's accuracy.
  • The reported online speed-ups over a 10^6-path Monte Carlo benchmark range from about 10^4 to 10^5 in runtime and 10^3 to 10^4 in arithmetic complexity for a five-asset min-call.
  • Accuracy is generally comparable to the benchmark, with the notable exception of Vega under a random correlation matrix, where errors are slightly larger but mainly in the small-Vega region.
  • The method currently evaluates only on the fixed Chebyshev–Lobatto parameter grid; continuous parameter interpolation is left for future work.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Editorial inference: the method's advantage is not intrinsic; it depends on the TT rank of the characteristic function, so correlation structures without short-range decay would likely require different asset orderings or hierarchical tensor networks.
  • Editorial inference: extending the one-core differentiation trick to maturity or interest-rate parameters would be viable only if those parameters enter the TT cores without inflating bond dimensions, a question the paper does not address.
  • Editorial inference: the numerical differentiation matrix is a spectral Chebyshev differentiation operator, so the ND approach effectively differentiates the TT-interpolated function; comparing against a higher-order finite-difference grid could separate discretization error from TT approximation error.
  • Editorial inference: testing a long-memory or block-correlation model would clarify when asset reordering is sufficient and when the low-rank assumption breaks.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 3 minor

Summary. The paper extends a previously proposed tensor-train (TT) pricing framework [33] to the computation of Greeks for multi-asset options. It builds a TT representation of the Fourier-transform pricing integrand that includes the volatility vector and initial asset prices as explicit parameters, and then computes Greeks in two ways: the ND approach applies a numerical differentiation operator to a single TT core, leaving bond dimensions unchanged, while the AN approach constructs separate TTs for analytically differentiated characteristic-function integrands. Numerical experiments for a five-asset min-call option under Black-Scholes with three correlation matrices report online-phase speedups of up to about 1e5 over MC with 1e6 paths, and the authors conclude that ND is generally preferable. The paper explicitly excludes offline TCI/SVD construction time from the reported timings.

Significance. If the accuracy claims are confirmed, the paper offers a clean and useful construction: because differentiation acts locally on one tensor core, all first- and second-order Greeks share the bond dimensions of the price TT, and online evaluation of a Greek costs no more than a single price evaluation. The derivations are parameter-free and the numerical checks against independent MC/MV benchmarks are appropriate; the manuscript also honestly discloses that reported timings exclude offline construction and shows how bond dimensions grow with correlation complexity. The unresolved issue is the accuracy of Vega under random correlations, where the reported RMSE exceeds the MC benchmark error by factors of about 2-5 under the paper's own criterion; this is a correctness-risk concern for the central 'comparable accuracy' claim rather than a circularity or derivation error.

major comments (3)
  1. [5.4, Table 3(c)] Under the paper's own accuracy criterion in Section 5.4 ('checking whether its error is lower than that of the MC'), the random-correlation Vega row of Table 3(c) fails: eTT,ND=1.78e-2 and eTT,AN=4.29e-2 are both larger than eMC,106=8.28e-3. Sections 6.2 and 7 attribute the discrepancy to regions where Vega is small, but no RMSE restricted to the large-Vega region, no error-per-parameter-set breakdown, and no threshold defining 'small Vega' are reported. The abstract's claim of 'comparable accuracy' therefore needs either a qualification or quantitative support in the form of restricted-region errors and error distributions.
  2. [4.2, Eqs. (5.6)-(5.11)] The ND approach's error amplification is left unexamined. The Chebyshev spectral differentiation matrix used in Eqs. (4.10)-(4.12) has entries whose magnitude grows with Np (e.g., O(Np^2) for the corner entries) and a condition number that grows polynomially in Np; with Np=100 and epsilon_TCI=1e-6, amplification of the TCI error to the 1e-2 scale observed in Table 3(c) is plausible. The manuscript reports no sensitivity study in Np or epsilon_TCI and no numerical bound on derivative amplification, so the claim that ND accuracy is acceptable is not yet demonstrated.
  3. [5.3 and 6.1] The reported advantages are conditional on correlation structure and parameter settings in ways that the paper does not fully quantify. For random correlations, Section 5.3 narrows the volatility range to [0.175,0.225], relaxes epsilon_SVD to 1e-8, and Section 5.2 introduces asset reordering heuristics; Section 6.1 shows the phi-bond dimension rising to about 400 in that case. These adjustments mean the TT advantage is not intrinsic to the method but depends on ordering and tolerance choices. The authors should state the verified operating envelope and, since Section 4.4 excludes offline construction from all timings, should either report offline costs for the three correlation cases or clearly limit the speedup claims to the online phase.
minor comments (3)
  1. [4.4] The sentence 'shown in Figure 3 (j)-(l) and Figure 2 (o)' appears to swap the figure references; the ND steps are in Figure 2 (j)-(l) and the AN steps in Figure 3 (m)-(o).
  2. [Throughout] There are several typos: 'F ourier' in the title, 'Black-Sholes' in the abstract, 'reffered' in Section 2.1, 'quantitiy' in Section 3.3, 'tenor train' in Section 2.1, and 'Chebyshve' in Section 7.
  3. [5.3] Section 5.3 states that increasing epsilon_TCI from 1e-6 to 1e-5 significantly reduces Greek accuracy, but no supporting data are shown; the statement should either be accompanied by a small experiment or removed.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: Greeks are produced by applying standard differentiation operators to an independently constructed TT of the Fourier pricing integrand, with external MC/MV benchmarks.

full rationale

The derivation chain is self-contained and contains no circular step. The price TT is built by TCI/SVD from the Fourier integrand phi and v-tilde (Section 4.1, Eqs. 4.1-4.9); it is not fitted to any Greek benchmark. The ND Greeks (Eqs. 4.10-4.12) are obtained by acting on one TT core with the standard Chebyshev spectral differentiation matrix (Eqs. 5.6-5.11); the claimed bond-dimension invariance is a direct structural property of applying the operator to a physical index of a single core, not an imported result. The AN Greeks (Eqs. 4.13-4.19) are TTs of the analytically differentiated Fourier integrals (Eqs. 3.12-3.18), so no quantity being reported was used as an input to a fit. Accuracy is assessed against external MC/MV ground truth (Section 5.4), and the reported errors are computed on 100 randomly chosen parameter sets, so this is not a fitted-input-called-prediction structure. The only overlapping-author reference, [33], provides the prior parameter-dependent TT construction; the present paper reimplements that construction, benchmarks it externally, and its novel ND/AN contribution is independently tested, so the self-citation is not load-bearing in a circular way. The paper itself flags the real limitations (Vega accuracy under random correlation, Table 3(c), Sections 6.2 and 7; narrowed sigma range and loosened SVD tolerances in Section 5.3; unexamined conditioning of the differentiation matrix), but these are empirical accuracy and omitted-analysis concerns, not circularity.

Assumptions & free parameters 7 free parameters · 5 assumptions · 0 invented entities

The method relies on standard Fourier pricing formulas and numerical quadrature, not on fitted physical parameters. The free parameters are numerical hyperparameters chosen by hand to achieve accuracy. The central assumption of low-rank compressibility is empirical and correlation-dependent, and the paper itself documents that random correlations stress it.

free parameters (7)
  • alpha (contour shift) = 5/d (from prior work, stability confirmed)
    Chosen by hand for Fourier inversion; must satisfy alpha_m > 0 and sum alpha_m > 1; results are stated to be stable under slight variations.
  • R_z (integration range half-width) = 25
    Chosen large enough to keep truncation error below MC error; not derived.
  • N_z (Gauss-Kronrod nodes per dimension) = 127
    Grid refinement choice for the Fourier integral.
  • N_p (Chebyshev-Lobatto nodes per parameter) = 100
    Parameter grid resolution; derivative accuracy depends on it.
  • epsilon_TCI = 1e-6
    User-specified tolerance; paper notes accuracy drops sharply at 1e-5.
  • epsilon_SVD = 1e-10 for ND, 1e-8 for AN (phi,v), 1e-10 for V
    Chosen per approach and per correlation case; random case requires looser tolerance due to memory.
  • sigma range for random correlations = [0.175,0.225]
    Narrowed from [0.15,0.25] specifically to maintain accuracy in the random-correlation case.
assumptions (5)
  • domain assumption Interchange of differentiation and integration in Eq. (3.12)
    Assumed to hold for the damped Fourier integrals; not proved for the min-call payoff.
  • ad hoc to paper Low-rank TT compressibility of phi and v over the joint (z, sigma, S0) grid
    Core premise of the method; documented as fragile for random correlations in Section 6.1.
  • ad hoc to paper Interleaved core ordering and asset reordering reduce bond dimensions
    Heuristic stated in Sections 4.1 and 5.2; no guarantee for general correlation matrices.
  • ad hoc to paper TCI converges to the specified tolerance from limited samples
    Paper notes TCI can become trapped in flat regions; accuracy results rely on successful convergence.
  • domain assumption Spectral differentiation matrix on Chebyshev-Lobatto nodes yields the derivative of the true price within desired tolerance
    Standard spectral method; error not independently bounded in the paper.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Tensor train representations of Greeks for Fourier-based pricing of multi-asset options." pith.science (2026). https://pith.science/paper/6URJHN5L

@misc{pith2026250708482,
  author       = {Pith},
  title        = {Pith review of: Tensor train representations of Greeks for Fourier-based pricing of multi-asset options},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/6URJHN5L}},
  note         = {Machine review of arXiv:2507.08482}
}
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.

Figures

Figures reproduced from arXiv: 2507.08482 by the authors.

Figure 1
Figure 1. Diagram of (a) a tensor train and (b) a tensor train operator. For clarity, at [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. The procedure for constructing TT representations of the option price [PITH_FULL_IMAGE:figures/full_fig_p008_2.png] view at source ↗
Figure 3
Figure 3. The procedure for constructing TT representations of Greeks obtained via [PITH_FULL_IMAGE:figures/full_fig_p012_3.png] view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: The bond dimensions are plotted against the bond index [PITH_FULL_IMAGE:figures/full_fig_p020_4.png]
Figure 5
Figure 5. Figure 5: One-dimensional plots of the option price [PITH_FULL_IMAGE:figures/full_fig_p022_5.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Classical Tensor Network and Quantum Fourier Transform Approaches for Large-Scale Carr-Madan Option Pricing

    quant-ph 2026-07 conditional novelty 5.0 of 10

    SFFT tensor-train Carr–Madan pricing matches FFT accuracy on Black–Scholes while using far less memory at large grids, and is compared directly to QFT on simulators and IBM hardware.

Reference graph

Works this paper leans on

40 extracted references · 23 canonical work pages · cited by 1 Pith paper

  1. [33]

    Sakurai, H

    R. Sakurai, H. Takahashi, and K. Miyamoto , Learning parameter dependence for fourier-based option pricing with tensor trains , Mathematics, 13 (2025), https://doi.org/10.3390/math13111828

  2. [1]

    https://indexes.nikkei.co.jp/en/nkave/index/profile?idx=nk225vi

  3. [2]

    Al \`o s and D

    E. Al \`o s and D. G. Lorite , Malliavin calculus in finance: Theory and practice , Chapman and Hall/CRC, 2021

  4. [3]

    Applebaum , Levy processes and infinitely divisible distributions, by ken-iti sato

    D. Applebaum , Levy processes and infinitely divisible distributions, by ken-iti sato. pp. 486. £50. 1999. isbn 0 521 55302 4 (cambridge university press). , The Mathematical Gazette, 85 (2001), p. 568–569, https://doi.org/10.2307/3621820

  5. [4]

    Barndorff-Nielsen , Processes of normal inverse gaussian type , Finance and Stochastics, 2 (1997), pp

    O. Barndorff-Nielsen , Processes of normal inverse gaussian type , Finance and Stochastics, 2 (1997), pp. 41--68

  6. [5]

    Behrens, G

    T. Behrens, G. Junike, and W. Schoutens , Failure of fourier pricing techniques to approximate the greeks , Frontiers of Mathematical Finance, 4 (2025), pp. 102--113, https://doi.org/10.3934/fmf.2024022

  7. [6]

    J. P. Boyd , Chebyshev and Fourier Spectral Methods , Dover Books on Mathematics, Dover Publications, Mineola, NY, second ed., 2001

  8. [7]

    Carr and D

    P. Carr and D. Madan , Option valuation using the fast fourier transform , Journal of computational finance, 2 (1999), pp. 61--73

Show all 40 references
  1. [9]

    K. G. Ernst Eberlein and A. Papapantoleon , Analysis of fourier transform valuation formulas and applications , Applied Mathematical Finance, 17 (2010), pp. 211--240, https://doi.org/10.1080/13504860903326669

  2. [10]

    Fannes, B

    M. Fannes, B. Nachtergaele, and R. F. Werner , Finitely correlated states on quantum spin chains , Communications in Mathematical Physics, 144 (1992), pp. 443 -- 490

  3. [11]

    Glasserman , Monte Carlo methods in financial engineering , vol

    P. Glasserman , Monte Carlo methods in financial engineering , vol. 53, Springer, 2004

  4. [12]

    K. Glau, D. Kressner, and F. Statti , Low-rank tensor approximation for chebyshev interpolation in parametric option pricing , SIAM Journal on Financial Mathematics, 11 (2020), pp. 897--927, https://doi.org/10.1137/19M1244172

  5. [13]

    https://github.com/google/tf-quant-finance

    Google , tf-quant-finance , 2023. https://github.com/google/tf-quant-finance

  6. [14]

    J. C. Hull , Options futures and other derivatives , Pearson, 2003

  7. [15]

    I. G. Ion, C. Wildner, D. Loukrezis, H. Koeppl, and H. De Gersem , Tensor-train approximation of the chemical master equation and its application for parameter inference , The Journal of Chemical Physics, 155 (2021), p. 034102, https://doi.org/10.1063/5.0045521

  8. [16]

    Kastoryano and N

    M. Kastoryano and N. Pancotti , A highly efficient tensor network algorithm for multi-asset fourier options pricing , arXiv:2203.02804, (2022)

  9. [17]

    Kinjo, R

    K. Kinjo, R. Sakurai, T. Kishimoto, and J. Ohkubo , Permutation of tensor-train cores for computing moments on stochastic differential equations , Journal of the Physical Society of Japan, 94 (2025), p. 084001, https://doi.org/10.7566/JPSJ.94.084001

  10. [18]

    Kl\"umper, A

    A. Kl\"umper, A. Schadschneider, and J. Zittartz , Matrix product ground states for one-dimensional spin-1 quantum antiferromagnets , Europhysics Letters, 24 (1993), pp. 293--297, https://doi.org/10.1209/0295-5075/24/4/010

  11. [19]

    Kornev, S

    E. Kornev, S. Dolgov, K. Pinto, M. Pflitsch, M. Perelshtein, and A. Melnikov , Numerical solution of the incompressible navier-stokes equations for chemical mixers via quantum-inspired tensor train finite element method , arXiv:2305.10784, (2023)

  12. [20]

    A. L. Lewis , A simple option formula for general jump-diffusion and other exponential l \'e vy processes , Available at SSRN 282110, (2001)

  13. [21]

    D. B. Madan and E. Seneta , The variance gamma (v.g.) model for share market returns , The Journal of Business, 63 (1990), pp. 511--524

  14. [22]

    R. C. Merton , Theory of rational option pricing , The Bell Journal of economics and management science, (1973), pp. 141--183

  15. [23]

    S. E. Notaris , Gauss-kronrod quadrature formulae - a survey of fifty years of research , Electron. Trans. Numer. Anal., 45 (2016), pp. 371--404

  16. [24]

    Novikov, M

    A. Novikov, M. Trofimov, and I. Oseledets , Exponential machines , arXiv preprint arXiv:1605.03795, (2016)

  17. [25]

    N \'u \ n ez Fern \'a ndez, M

    Y. N \'u \ n ez Fern \'a ndez, M. Jeannin, P. T. Dumitrescu, T. Kloss, J. Kaye, O. Parcollet, and X. Waintal , Learning feynman diagrams with tensor trains , Phys. Rev. X, 12 (2022), p. 041018, https://doi.org/10.1103/PhysRevX.12.041018

  18. [26]

    Núñez Fernández, M

    Y. Núñez Fernández, M. K. Ritter, M. Jeannin, J.-W. Li, T. Kloss, T. Louvet, S. Terasaki, O. Parcollet, J. von Delft, H. Shinaoka, and X. Waintal , Learning tensor networks with tensor cross interpolation: New algorithms and libraries , SciPost Physics, 18 (2025), https://doi....

  19. [27]

    F. D. Oliveira and E. Mordecki , Computing greeks for l \'e vy models: The fourier transform approach , in Trends in Mathematical Economics: Dialogues Between Southern Europe and Latin America, A. A. Pinto, E. A. Gamba, A. N. Yannacopoulos, and C. Herv \'e s-Beloso, eds., Spri...

  20. [28]

    Or\' u s , A practical introduction to tensor networks: Matrix product states and projected entangled pair states , Annals of Physics, 349 (2014), pp

    R. Or\' u s , A practical introduction to tensor networks: Matrix product states and projected entangled pair states , Annals of Physics, 349 (2014), pp. 117--158, https://doi.org/https://doi.org/10.1016/j.aop.2014.06.013

  21. [29]

    Oseledets and E

    I. Oseledets and E. Tyrtyshnikov , Tt-cross approximation for multidimensional arrays , Linear Algebra and its Applications, 432 (2010), pp. 70--88, https://doi.org/https://doi.org/10.1016/j.laa.2009.07.024

  22. [30]

    I. V. Oseledets , Tensor-train decomposition , SIAM Journal on Scientific Computing, 33 (2011), pp. 2295--2317, https://doi.org/10.1137/090752286

  23. [31]

    R. D. Peddinti, S. Pisoni, A. Marini, P. Lott, H. Argentieri, E. Tiunov, and L. Aolita , Quantum-inspired framework for computational fluid dynamics , Communications Physics, 7 (2024), https://doi.org/10.1038/s42005-024-01623-8

  24. [32]

    Sakaue, H

    K. Sakaue, H. Shinaoka, and R. Sakurai , Adaptive sampling-based optimization of quantics tensor trains for noisy functions: applications to quantum simulations , arXiv:2405.12730, (2025)

  25. [34]

    Schmelzle , Option pricing formulae using fourier transform : Theory and application , 2010, https://api.semanticscholar.org/CorpusID:14684726

    M. Schmelzle , Option pricing formulae using fourier transform : Theory and application , 2010, https://api.semanticscholar.org/CorpusID:14684726

  26. [35]

    Schollwock , The density-matrix renormalization group in the age of matrix product states , Ann

    U. Schollwock , The density-matrix renormalization group in the age of matrix product states , Ann. Phys., 326 (2011), pp. 96--192, https://doi.org/10.1016/j.aop.2010.09.012

  27. [36]

    Shinaoka, M

    H. Shinaoka, M. Wallerberger, Y. Murakami, K. Nogaki, R. Sakurai, P. Werner, and A. Kauch , Multiscale Space-Time ansatz for correlation functions of quantum systems based on quantics tensor trains , Phys. Rev. X, 13 (2023), p. 021015, https://doi.org/10.1103/PhysRevX.13.021015

  28. [37]

    Stoudenmire and D

    E. Stoudenmire and D. J. Schwab , Supervised learning with tensor networks , in Advances in Neural Information Processing Systems, D. Lee, M. Sugiyama, U. Luxburg, I. Guyon, and R. Garnett, eds., vol. 29, Curran Associates, Inc., 2016

  29. [38]

    Takahashi, R

    H. Takahashi, R. Sakurai, and H. Shinaoka , Compactness of quantics tensor train representations of local imaginary-time propagators , SciPost Phys., 18 (2025), p. 007, https://doi.org/10.21468/SciPostPhys.18.1.007

  30. [39]

    L. N. Trefethen , Spectral Methods in MATLAB , Society for Industrial and Applied Mathematics, 2000, https://doi.org/10.1137/1.9780898719598

  31. [40]

    Y. Xu, Y. Lai, and H. Yao , Efficient simulation of greeks of multiasset european and asian style options by malliavin calculus and quasi-monte carlo methods , Applied Mathematics and Computation, 236 (2014), pp. 493--511, https://doi.org/https://doi.org/10.1016/j.amc.2014.03.057

  32. [41]

    write newline

    " write newline "" before.all 'output.state := FUNCTION fin.entry add.period write newline FUNCTION new.block output.state before.all = 'skip after.block 'output.state := if FUNCTION not #0 #1 if FUNCTION and 'skip pop #0 if FUNCTION or pop #1 'skip if FUNCTION new.block.check...

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.