Pith. sign in

REVIEW 4 major objections 4 minor 32 references

Solving high-dimensional nonlinear filtering problems using a tensor train decomposition method

T0 review · 4 major / 4 minor · reviewed 2026-08-14 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read Quantized tensor trains solve 3D nonlinear filtering in seconds

desk verdict A promising QTT-based filter for 3D problems whose central complexity claim does not survive its own Table 5. read the letter →

arxiv 1908.04010 v1 pith:W2HQHOWG submitted 2019-08-12 math.NA cs.CEcs.NAcs.SYeess.SYmath.OC

Sijing Li , Zhongjian Wang , Stephen S.T. Yau , Zhiwen Zhang This is my paper · ORCID
classification math.NAcs.CEcs.NAcs.SYeess.SYmath.OC MSC 15A6935R6065M1260G3565M99
keywords nonlinearfilteringforwardKolmogorovequationsDuncan-Mortensen-Zakaiequationtensortraindecompositionquantizedlow-rankapproximationconvergenceanalysisreal-timealgorithm
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 sets out to establish that nonlinear filtering — estimating a hidden state from noisy observations — can be solved in real time even in dimension three, by storing the conditional density in the quantized tensor train (QTT) format. The key observation is that the solution of the forward Kolmogorov equation, which evolves the density between observations, has low-dimensional structure that can be extracted once and reused. After an offline phase builds a low-rank QTT approximation of the evolution operator, each new observation is assimilated by fast QTT operations. The paper proves convergence to the pathwise robust DMZ solution with an $L^1$ error of $O(N_t^{-\alpha}+\tau+h^2+\text{QTT rounding terms})$ and demonstrates on two 3D examples that filtering updates take seconds instead of thousands of seconds.

What carries the argument

The central object is the quantized tensor train (QTT) decomposition: a $d$-dimensional tensor is written as a product of small cores $G_1(i_1)\cdots G_d(i_d)$ with TT-ranks $r_k$, and each spatial dimension is reshaped into virtual binary dimensions so storage drops to $O(d\log_2(N) r^2)$. The method's engine is the exponential transformation $\sigma(x,t)=\exp(h^T(x,t)S^{-1}(t)y_t)u(x,t)$, which converts the stochastic DMZ equation into a deterministic FKE, plus the offline assembly of the operator $A=\frac12\Delta_d-C_d-\frac12 Q_d$ in QTT format and its power $(\tau A+I)^{\Delta T/\tau}$. The QTT rank of the discrete Laplacian is bounded by 4, the convection operator's rank by $5dr$ when the drift functions have rank $r$, and TT-SVD rounding controls the ranks of everything else; these bounds are what make the online operations — Hadamard products and matrix-vector products — fast.

What would settle it

Run a filtering problem whose conditional density becomes more complex over time — for example, a bimodal drift that splits the density into well-separated peaks, or a cubic sensor in four dimensions — and record the effective QTT rank at each observation time. If the rank exceeds the assumed constant bound and keeps growing, the stated online complexity and convergence guarantee will not hold.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central discovery is that the forward Kolmogorov equation (FKE) — a parabolic PDE for the unnormalized conditional density — admits low-rank quantized tensor train representations over the filtering horizon, so the whole filtering loop can be run in compressed form. Starting from the pathwise robust DMZ equation (the stochastic PDE for the unnormalized conditional density), the paper freezes each observation interval, solves the FKE with a finite-difference operator assembled in QTT format, and assimilates new observations through the exponential transformation. Theorem 5.7 bounds the $L^1$ error between the QTT-recovered density and the DMZ solution by three contributions: the time-frozen approximation of the observation path, the finite-difference discretization error $O(\tau+h^2)$, and the QTT rounding errors. Numerically, the method matches finite-difference estimates on two 3D examples (mean-square errors $0.007$ and $0.023$) while cutting the per-run CPU time from $2052$ s and $4079$ s to about $15$ s and $17$ s.

Load-bearing premise

The method works only if the compressed representation of the density stays small over the entire filtering window; the paper assumes exactly this (Proposition 4.3) and checks it numerically on only two three-dimensional examples. If the compression size grows with time or dimension, the promised speed-up and the error bound both collapse.

Editorial extensions

If this is right

  • If the ranks stay bounded, the online per-interval cost is $O(d\log_2(N) r^6)$, so refining the spatial grid makes the solver only logarithmically more expensive.
  • The offline/online split means the same low-rank propagator can be reused for every observation path of a given system, making repeated or long-horizon filtering much cheaper after a one-time setup.
  • Theorem 5.7 gives explicit error control: reduce $N_t$ for the path-frozen error, $\tau$ and $h$ for the finite-difference error, and the TT-rounding precisions for the QTT error.
  • In the two 3D examples, the method matches the finite-difference density estimates while reducing CPU time from 2052 s and 4079 s to about 15 s and 17 s.

Reading between the lines

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

  • A natural stress test, not performed in the paper, is a time-dependent drift or observation function: the paper itself notes the low-rank structure may shift over time, so an adaptive or dynamic low-rank update would be needed.
  • The offline phase must build low-rank QTT tensors from full discretized operators; the paper reports this cost is modest for the 3D tests, but for much larger $d$ or $N$ the offline construction could become the bottleneck unless cross-approximation is used.
  • Assumption 4 requires the observation term $|2h^T y_t|$ to be bounded by a constant, but real observations driven by Brownian motion are not bounded; a fully rigorous guarantee would need a truncation of observations or a probabilistic bound on the exceptional paths.
  • If the small-rank conjecture survives in $d>3$, the $O(N d r^2)$ cost of assimilating the exponential observation factor becomes the main bottleneck, so the scaling in very high dimension deserves direct measurement.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 4 minor

Summary. The paper proposes a tensor-train (QTT) method for solving the forward Kolmogorov equation (FKE) that arises in nonlinear filtering, with an offline stage that precomputes low-rank QTT representations of the discretized PDE operators and propagator, and an online stage that assimilates observations through an exponential (Hadamard) update and propagates the density in QTT format. A convergence theorem (Theorem 5.7) is stated that combines the observation-time splitting error, the finite-difference discretization error, and the QTT rounding errors. Numerical experiments on two 3D filtering problems compare the method against finite differences and particle filters, reporting large speedups at moderate resolution and small mean-square errors relative to the FD reference.

Significance. If the assumed low-rank structure persists, the method is a promising route toward real-time nonlinear filtering: the paper gives explicit convergence analysis, identifies the three sources of error separately, and demonstrates on two 3D examples that the FKE density has low effective QTT ranks, yielding speedups of two orders of magnitude over finite differences. The numerical comparison is honest in reporting averaged MSEs over 100 paths and in showing convergence with respect to the TT-rounding precision. However, the high-dimensional and real-time claims go beyond what is demonstrated: only d=3 is tested, and the complexity analysis contains a load-bearing gap that the paper's own timing data contradicts.

major comments (4)
  1. [Section 4.3, Proposition 4.3 and Table 5] The claimed O(N d r^2) cost for constructing exp[h^T(x,t_j)(y_{t_j}-y_{t_{j-1}})] in QTT format is contradicted by the paper's own timing data. The proof says this is obtained 'from a full multidimensional array' via Theorem 2.1 of [25], but a full array has N^d entries; the cited theorem does not justify a construction that avoids materializing the full array for a general nonlinear h(x)\cdot y_t. In Table 5, doubling N from 64 to 128 increases tEXP from 5.20 s to 46.28 s (a factor of 8.9), while the averaged QTT rank grows only from 8.78 to 8.89; this scaling is close to O(N^d), not O(N d r^2). Since this step is repeated at every observation time, the online complexity stated in Proposition 4.3 and the associated real-time claim are not supported.
  2. [Section 5.1, Assumption (Asm.4) and Lemma 5.4] The convergence theorem requires |2h^T y_t| \le c2 on the observation times, but for the SDE model (2) the observation process y_t is not bounded. Even with h bounded on the numerical domain, h^T y_t is unbounded along a Brownian path; for a fixed path the maximum is finite, but the resulting constant c2 is path-dependent and typically grows with the number of observation times, and the paper does not quantify it. The constant c2 controls the factor e^{c2} in Eq. (42) and enters the final bounds (36) and (53). The numerical experiments do not indicate that the simulated observation paths are truncated or that Asm.4 is verified, so the convergence statement as formulated does not cover the stochastic model actually simulated in Section 6.
  3. [Section 5.3, Lemma 5.5 and Theorem 5.7] The QTT contribution to the error is controlled by powers c7^{Nt}, with c7 = (1+ǫ1)(1+ǫ2)c6 e^{c2} defined after Eq. (49). The paper does not show that c7 < 1, and for a stable explicit scheme with c6 potentially larger than 1 this will generally not be the case. Consequently, the term ǫ1 c7^{Nt} in Eq. (46) and the analogous terms in (53) grow exponentially in the number of observations, so the displayed bound does not establish an error of order O(Nt^{-α} + τ + h^2 + ǫ1 + ǫ2) uniformly in Nt. As stated, the convergence theorem is therefore not proven by the given estimates unless the authors add a mechanism (such as explicit conditions guaranteeing c7 < 1 or a choice of ǫ1, ǫ2 that compensates the exponential growth).
  4. [Abstract, Section 7, and Section 6] The paper claims in the abstract and conclusions that it solves 'high-dimensional' NLF problems in a real-time manner, but all numerical experiments are for d=3 (Examples 1 and 2 in Section 6.2). The central assumption of bounded QTT ranks (Proposition 4.3) is verified only for these 3D problems, as shown in Tables 1-4; no evidence is provided for d>3, and Section 7 itself defers d>3 to future work. The high-dimensional claim in the title and abstract is therefore an overstatement of the demonstrated results.
minor comments (4)
  1. [Abstract] The word 'roblem' in 'NLF roblem' is a typo and should be corrected to 'problem'.
  2. [Section 6.3, Table 5] The table should specify whether tFKE and tEXP are total costs over the whole simulation or per observation interval; the text says the FD method at N=2^7 would cost about 13 hours, but without a per-step breakdown the real-time claim cannot be checked against the observation interval ΔT=0.05.
  3. [Section 6.1, Eq. (60)] The definition of the effective QTT-rank r_eff is nonstandard and not derived; the authors should either explain its motivation or report the maximum QTT-rank, which is the quantity that enters the complexity estimates.
  4. [References] Reference [14] has a typo in the author name: 'Vlanimir A. Kazeev' should be 'Vladimir A. Kazeev'.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the QTT solution is compared against independent finite-difference and particle-filter references, and the cited self-results are parameter-free theorems used as background.

full rationale

The paper's derivation chain is standard and non-circular: the DMZ equation is transformed into the pathwise robust DMZ equation using the exponential change of variables from Yau and Yau [30], then into a forward Kolmogorov equation, which is discretized by finite differences and solved in QTT format. The convergence analysis in Theorem 5.7 is a triangle-inequality splitting into three independent error terms: the time-freezing error (Proposition 5.1, cited from [30]), the finite-difference discretization error (Lemma 5.4), and the QTT-rounding error (Lemma 5.5). Each term is proven separately and the cited propositions are external, parameter-free mathematical theorems with stated assumptions that do not include the present paper's numerical results; the fact that [30] shares an author is therefore not load-bearing circularity. The central numerical claim is validated against finite-difference and particle-filter benchmarks, not against a fitted quantity, so there is no 'fitted input called prediction' pattern. The bounded-rank assumption in Proposition 4.3 is an explicit assumption, and the paper verifies slow rank growth numerically in Tables 1-4; this is an empirical check, not a circular definition. Potential concerns about the complexity statement in Proposition 4.3, such as the claim that a full multidimensional array can be converted to QTT format in O(N d r^2) time when the array has N^d entries, or the scaling of tEXP in Table 5 being inconsistent with linear-in-N growth, are correctness and rigor issues rather than circularity: they do not make the derived result equivalent to its input by construction. Overall, no step in the paper's derivation reduces to its own assumptions or to a self-citation chain.

Assumptions & free parameters 0 free parameters · 4 assumptions · 0 invented entities

No new physical entities. The main burden is the low-rank assumption on QTT ranks, which is empirical, and the bounded-observation assumption, which is restrictive for true Brownian observation paths.

assumptions (4)
  • domain assumption Pathwise robust DMZ equation is well-posed and equivalent to the FKE under growth conditions on f and h with at most linear growth.
    Relied on via Proposition 5.2 and Proposition 5.1 from [30], the third author's earlier work, in Section 5.1.
  • domain assumption Observation paths satisfy |2 h^T y_t| <= c2 (Asm.4), so the exponential transformation is bounded and the domain truncation is valid.
    Asm.4 in Section 5.1 is used in Lemma 5.4 and Theorem 5.7; it is not guaranteed by the SDE model in Eq. (2).
  • ad hoc to paper The QTT-ranks of operators and of the solution remain bounded by a small r throughout the evolution.
    Assumed in Proposition 4.3 and verified numerically only for two 3D examples in Tables 1-4; the core efficiency claim depends on it.
  • standard math Explicit finite-difference stability conditions h < 1/Cf and tau < (d/h^2 + dLf + d/2 Ch^2)^{-1} hold.
    Lemma 5.4 derives these as sufficient conditions for the error estimate; they are standard CFL-type restrictions for explicit schemes.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Solving high-dimensional nonlinear filtering problems using a tensor train decomposition method." pith.science (2026). https://pith.science/paper/W2HQHOWG

@misc{pith2026190804010,
  author       = {Pith},
  title        = {Pith review of: Solving high-dimensional nonlinear filtering problems using a tensor train decomposition method},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/W2HQHOWG}},
  note         = {Machine review of arXiv:1908.04010}
}
read the original abstract

In this paper, we propose an efficient numerical method to solve high-dimensional nonlinear filtering (NLF) problems. Specifically, we use the tensor train decomposition method to solve the forward Kolmogorov equation (FKE) arising from the NLF problem. Our method consists of offline and online stages. In the offline stage, we use the finite difference method to discretize the partial differential operators involved in the FKE and extract low-dimensional structures in the solution space using the tensor train decomposition method. In addition, we approximate the evolution of the FKE operator using the tensor train decomposition method. In the online stage using the pre-computed low-rank approximation tensors, we can quickly solve the FKE given new observation data. Therefore, we can solve the NLF roblem in a real-time manner. Under some mild assumptions, we provide convergence analysis for the proposed method. Finally, we present numerical results to show the efficiency and accuracy of the proposed method in solving high-dimensional NLF problems.

Figures

Figures reproduced from arXiv: 1908.04010 by the authors.

Figure 1
Figure 1. Comparison of a trajectory of the almost linear problem ob [PITH_FULL_IMAGE:figures/full_fig_p022_1.png] view at source ↗
Figure 2
Figure 2. The estimations of density function at t = 10s in the almost linear problem. the first example, in order to satisfy the CFL stability condition. We fix a higher TT-rounding precision ǫ = 5 × 10−5 due to the higher nonlinearity in this example. In [PITH_FULL_IMAGE:figures/full_fig_p023_2.png] view at source ↗
Figure 3
Figure 3. Comparison of a trajectory of the cubic sensor problem o [PITH_FULL_IMAGE:figures/full_fig_p023_3.png] view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: The estimations of density function at t = 12s in the cubic sensor problem Remark 6.1. The PF method is a very popular method in solving NLF problem, which is a Monte Carlo method and it requires a certain amount of sample to compute statistical quantities. Since the P…
Figure 5
Figure 5. Figure 5: Error of estimated states between the QTT method and r [PITH_FULL_IMAGE:figures/full_fig_p025_5.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

32 extracted references · 32 canonical work pages

  1. [25]

    I. V. Oseledets. Approximation of 2 d × 2d matrices using tensor decomposition. SIAM Journal on Matrix Analysis and Applicatoins , 31(4):2130–2145, 2010

    work page 2010

  2. [30]

    Real time solution of the n onlinear filtering problem without memory II

    Shing-Tung Yau and Stephen S-T Yau. Real time solution of the n onlinear filtering problem without memory II. SIAM Journal on Control and Optimization , 47(1):163–195, 2008

    work page 2008

  3. [1]

    M. S. Arulampalam, S. Maskell, N. Gordon, and T. Clapp. A tutorial on particle filters for online nonlinear/non-gaussian Bayesian tracking. IEEE Transactions on signal processing, 50(2):174–188, 2002

    work page 2002

  4. [2]

    Bain and D

    A. Bain and D. Crisan. Fundamentals of stochastic filtering , volume 3. Springer, 2009

    work page 2009

  5. [3]

    Baras, G

    J. Baras, G. Blankenship, and W. Hopkins. Existence, uniquenes s, and asymptotic be- havior of solutions to a class of Zakai equations with unbounded coe fficients. IEEE Transactions on Automatic Control , 28(2):203–214, 1983

    work page 1983

  6. [4]

    Bensoussan, R

    A. Bensoussan, R. Glowinski, and A. Rascanu. Approximation of t he Zakai equation by the splitting up method. SIAM Journal on Control and Optimization , 28(6):1420–1431, 1990. 26

    work page 1990

  7. [5]

    Cheng, T

    M. Cheng, T. Y. Hou, and Z. Zhang. A dynamically bi-orthogonal m ethod for stochastic partial differential equations I: derivation and algorithms. J. Comput. Phys. , 242:843–868, 2013

    work page 2013

  8. [6]

    Cheng, T

    M. Cheng, T. Y. Hou, and Z. Zhang. A dynamically bi-orthogonal m ethod for stochas- tic partial differential equations II: adaptivity and generalizations . J. Comput. Phys. , 242:753–776, 2013

    work page 2013

Show all 32 references
  1. [7]

    S. V. Dolgov, B. N. Khoromskij, and I. V. Oseledets. Fast solutio n of parabolic problems in the tensor train/quantized tensor train format with initial applica tion to the Fokker- Planck equation. SIAM Journal on Scientific Computing , 34(6):A3016–A3038, 2012

    2012

  2. [8]

    T. Duncan. Probability densities for diffusion processes with applic ations to nonlinear filtering theory and detection theory. Technical report, Stanford Univ CA Stanford Elec- tronics Labs, 1967

    1967

  3. [9]

    Fleming and S

    W. Fleming and S. Mitter. Optimal control and nonlinear filtering fo r nondegenerate diffusion processes. Stochastics, 8(1):63–77, 1982

    1982

  4. [10]

    Gustafsson, F

    F. Gustafsson, F. Gunnarsson, N. Bergman, U. Forssell, J. J ansson, R. Karlsson, and P. J. Nordlund. Particle filters for positioning, navigation, and tracking. IEEE Transactions on signal processing , 50(2):425–437, 2002

    2002

  5. [11]

    Gy¨ ongy and N

    I. Gy¨ ongy and N. Krylov. On the splitting-up method and stoch astic partial differential equations. The Annals of Probability , 31(2):564–591, 2003

    2003

  6. [12]

    K. Ito. Approximation of the Zakai equation for nonlinear filter ing. SIAM Journal on Control and Optimization , 34(2):620–634, 1996

    1996

  7. [13]

    Kallianpur

    G. Kallianpur. Stochastic filtering theory , volume 13. Springer Science & Business Media, 2013

    2013

  8. [14]

    Kazeev, and Boris N

    Vlanimir A. Kazeev, and Boris N. Khoromskij. Low-rank explicit QT T representation of the laplace operator and its inverse. SIAM Journal on Matrix Analysis and Applications , 33(3):724–758, 2012

    2012

  9. [15]

    Khoromskij

    Boris N. Khoromskij. O(d log N)-Quantics approximation of N − d tensors in high- dimensional numerical modeling. Constructive Approximation , 34:257–280, 2011

    2011

  10. [16]

    P. E. Kloeden and E. Platen. Numerical solution of Stochastic Differential Equations . Springer-Verlag Berlin, Heidelberg, 1992

    1992

  11. [17]

    Kurtz and J

    T. Kurtz and J. Xiong. Particle representations for a class of nonlinear SPDEs . Stochastic Processes and their Applications , 83(1):103–126, 1999

    1999

  12. [18]

    H. Kushner. Approximations to optimal nonlinear filters . IEEE Transactions on Auto- matic Control , 12(5):546–556, 1967. 27

    1967

  13. [19]

    Lototsky, R

    S. Lototsky, R. Mikulevicius, and B. L. Rozovskii. Nonlinear filter ing revisited: a spectral approach. SIAM Journal on Control and Optimization , 35(2):435–461, 1997

    1997

  14. [20]

    Luo and Stephen S

    X. Luo and Stephen S. T. Yau. Complete real time solution of the general nonlinear fil- tering problem without memory. IEEE Transactions on Automatic Control , 58(10):2563– 2578, 2013

    2013

  15. [21]

    Luo and Stephen S

    X. Luo and Stephen S. T. Yau. Hermite spectral method to 1-D forward Kolmogorov equation and its application to nonlinear filtering problems. IEEE Transactions on Au- tomatic Control, 58(10):2495–2507, 2013

    2013

  16. [22]

    Mortensen

    R. Mortensen. Optimal control of continuous-time stochast ic systems. Technical report, California Univ Berkeley Electronics Research Lab , 1966

    1966

  17. [23]

    Morton and D

    K. Morton and D. Mayers. Numerical solution of partial differen tial equations: an intro- duction. Cambridge University Press , 2005

    2005

  18. [24]

    N. Nagase. Remarks on nonlinear stochastic partial differentia l equations: an application of the splitting-up method. SIAM Journal on Control and Optimization , 33(6):1716–1730, 1995

    1995

  19. [26]

    I. V. Oseledets. Tensor-train decomposition. SIAM Journal on Scientific Computing , 33(5):2295-2317, 2011

    2011

  20. [27]

    I. V. Oseledets and E. E. Tyrtyshnikov. TT-cross approximat ion for multidimensional arrays. Linear Algebra and its Applications , 432:70–88, 2010

    2010

  21. [28]

    Pardouxt

    E. Pardouxt. Stochastic partial differential equations and filt ering of diffusion processes. Stochastics, 3:127–167, 1980

    1980

  22. [29]

    Z. Wang, X. Luo, Stephen S.-T. Yau, and Z. Zhang. Proper ort hogonal decomposition method to nonlinear filtering problems in medium-high dimension. to appear in IEEE Transactions on Automatic Control , 2020

    2020

  23. [31]

    M. Zakai. On the optimal filtering of diffusion processes. Probability Theory and Related Fields, 11(3):230–243, 1969

    1969

  24. [32]

    Q. Zhang. Nonlinear filtering and control of a switching diffusion w ith small observation noise. SIAM Journal on Control and Optimization , 36(5):1638–1668, 1998. 28

    1998

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.