Pith. sign in

REVIEW 3 major objections 5 minor 58 references

Operator theoretic causality analysis of fluid flows using linearized dynamics

T0 review · 3 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read The paper proposes that for linearized systems, the matrix exponential of the dynamics operator encodes which flow modes cause which at any future time.

desk verdict A sound operator-theoretic core wrapped in some overclaimed equivalence and a missing example; the Couette flow analysis is the strongest part. read the letter →

arxiv 2506.08118 v3 pith:SDCAYHSQ submitted 2025-06-09 nlin.CD math-phmath.MP

Ankit Srivastava , Victor Jimenez-Fernandez , Louis Cattafesta , Scott Dawson This is my paper · ORCID
classification nlin.CDmath-phmath.MP MSC 37N1062M1093B0593B0776E05
keywords causalitylinearizeddynamicsmatrixexponentialfluidflowsGrangertransferentropydynamicmodedecompositionbalancedtruncation
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

This paper proposes reading causality in fluid flows directly from the linearized governing equations rather than inferring it from time-series data. Its central claim is that the matrix exponential $e^{A\tau}$ of the linearized dynamics encodes how a perturbation in one flow mode affects another at any later time, making causality a property of the dynamics operator itself. The paper defines immediate, delayed, and global causality, and shows that the matrix $M=(\alpha I-|A|)^{-1}$ provides a global indicator of whether any indirect causal path exists across all time horizons. Under stated assumptions, Granger causality and transfer entropy reduce to the squared entries of the matrix exponential, so LOCA agrees with data-driven methods when variables are uncorrelated. The payoff would be a causality analysis that avoids correlation-induced false inferences and can identify which physical modes truly drive a flow.

What carries the argument

The central object is the matrix exponential $e^{A\tau}$ of the linearized operator $A$, together with the global matrix $M=(\alpha I-|A|)^{-1}$ built from the elementwise absolute value of $A$. In the power series $e^{A\tau}=I+A\tau+\tfrac12 A^2\tau^2+\cdots$, the $k$-th term accounts for indirect causal pathways of length $k$; this is why a zero direct coupling can still yield delayed or global causality. $M$ is a Neumann-series upper bound on $\sum_k |A^k|$, so its zero entries are rigorous statements of causal disconnection across all times.

What would settle it

Take a two-variable linear system with known $A$ and deliberately correlated state variables, generate noise-forced data, and compare the Granger $F$-statistic with $([e^{A\Delta t}]_{ij})^2 \sigma_j^2/\sigma_{\epsilon,i}^2$. If $RSS_r-RSS_u$ differs from $([e^{A\Delta t}]_{ij})^2\sum_n c_j(t_{n-1})^2$ by an amount that grows with the correlation between $c_j$ and the other variable, then Eq. (3.6) is violated and LOCA and Granger will disagree exactly where the paper says they converge.

Watch

Extended reading notes

Core claim

For a system in canonical form $\dot{\mathbf{c}}=A\mathbf{c}$, the paper's central claim is that the matrix exponential $e^{A\tau}$ 'encapsulates the essential causal relations among the modes at any future time': entry $(i,j)$ is the counterfactual response of mode $i$ at time $\tau$ to an initial perturbation of mode $j$. Immediate causality is the direct coupling $A_{ij}\neq 0$; delayed causality is the corresponding entry of $e^{A\tau}$; and global causality is measured by $M=(\alpha I-|A|)^{-1}$, chosen so that a zero entry definitively rules out any causal path over all times. A modified $F$-statistic $F_{j\to i}=[e^{A\tau}]_{ij}^2 \sigma_j^2/\sigma_{\epsilon,i}^2$ weights these entries by state and error variances. The paper further shows that LOCA and Granger causality agree when state variables are uncorrelated, that transfer entropy agrees in the linear Gaussian case, and that a DMD-style estimate of the matrix exponential recovers the same causal structure from data.

Load-bearing premise

The load-bearing premise is that when a variable is omitted from the restricted regression, the remaining coefficients stay identical to those in the unrestricted model, so the omitted variable's contribution appears directly in the residual; in correlated systems least-squares refits the remaining coefficients and that identity no longer holds.

Editorial extensions

If this is right

  • In linearized fluid systems, the cause-effect structure can be computed from the operator alone, without long time-series records, and interpreted physically through the underlying mechanisms.
  • A zero entry in $M$ identifies modes that are causally disconnected, and this directly implies loss of controllability if the input acts on the cause mode and loss of observability if the measurement is of the effect mode.
  • Granger causality and transfer entropy, in the linear Gaussian setting, measure essentially the squared matrix-exponential entries, so LOCA supplies the same information as data-driven causality analyses when variables are uncorrelated.
  • Truncation of modes by energy (POD) can discard low-energy modes that carry dominant causal influence, whereas balanced truncation preserves the leading causal interactions.
  • The DMD-based variant LOCA-DMD estimates the matrix exponential from data and reproduces the operator-based causality metric with short data records.

Reading between the lines

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

  • Editorial extension: the equivalence with Granger causality rests on the restricted regression keeping its coefficients fixed when a variable is dropped; when variables are correlated the remaining coefficients re-fit, so the difference between LOCA and Granger should be quantifiable as a correlation-dependent correction term.
  • Editorial extension: because $M$ bounds rather than equals $\sum_k |A^k|$, it certifies absence of causality but can overstate the strength of present causality; an exact global measure would require the generally unavailable closed form of the absolute-power series.
  • Editorial extension: for nonlinear chaotic flows, LOCA describes infinitesimal perturbations, so its causal rankings may differ from finite-amplitude causal pathways; comparing LOCA with data-driven causality on the same chaotic attractor would test how far the linearized notion extends.
  • Editorial extension: a practical diagnostic follows—when applying Granger or transfer entropy to fluid data, decorrelating variables first (e.g., with POD modes) should make the statistical estimate converge to the operator-based result, and the residual discrepancy measures correlation contamination.
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 / 5 minor

Summary. This paper introduces Linear Operator Causality Analysis (LOCA), a framework for quantifying causal interactions among modes of linearized dynamical systems. Causality is defined through the matrix exponential of the system matrix A, with immediate causality given by nonzero off-diagonal entries of A, delayed causality by entries of e^{Aτ}, and global causality by the Neumann-series bound M=(αI−|A|)^{-1}. The paper establishes formal connections between LOCA and Granger causality/transfer entropy, controllability/observability, and transitive closure in graph theory, and proposes a DMD-based data-driven estimator (LOCA-DMD). The methods are illustrated on a nilpotent toy problem and on linearized Couette flow, where LOCA is shown to capture direct and indirect causal pathways, to be robust to correlated variables when applied in POD coordinates, and to motivate balanced truncation for model reduction that preserves causal information.

Significance. The counterfactual definition of delayed causality via [e^{Aτ}]_{ij} is exact and well motivated for linear time-invariant systems, and the paper's distinction between immediate, delayed, and global causality is conceptually useful for fluid mechanics. The identification of the global causality metric with a transitive-closure-like bound, the connection to controllability/observability, and the demonstration that low-energy modes can be causally relevant (Figures 6 and 9) are valuable contributions. The paper also shows that balanced truncation outperforms POD truncation in preserving causal interactions, which is a practically important insight. However, the claim that LOCA is equivalent to Granger causality rests on a restrictive assumption (Eq. 3.6) that is not satisfied generically, and the global metric is misdescribed as 'integrated over all time.' These issues temper the generality of the results, but they do not invalidate the core operator-theoretic picture.

major comments (3)
  1. [§3.1, Eq. (3.6)] The derivation of the Granger equivalence assumes that the restricted-model regression coefficients are identical to the unrestricted coefficients except for the omitted variable, so that the omitted term appears directly in the restricted residual. Under least-squares fitting this is generically false: by the Frisch–Waugh–Lovell theorem, the retained coefficients in the restricted regression shift to partially absorb the omitted variable unless the omitted regressor is uncorrelated with all retained regressors. Thus Eq. (3.6), and consequently the scaling result (3.11), holds only under an orthogonality condition that is not stated in the derivation. The paper's own results (Figures 2 and 5) confirm that LOCA and Granger agree only after transforming to POD coordinates, whose temporal coefficients are uncorrelated by construction. The equivalence claim in Section 3.1 should therefore be presented as conditional on such an orthogonality condition, or the derivation should be revised to account for the re-fitting of retained coefficients.
  2. [§2.3 and §6] The global causality metric M=(αI−|A|)^{-1} is described in the conclusions as providing 'a measure of global causality integrated over all time.' This is not mathematically accurate: M is an elementwise upper bound of the weighted Neumann series Σ |A|^k (and hence of Σ |A^k|), not the time integral of |e^{At}|. Indeed, for the nilpotent example in Eq. (2.16), |e^{At}|_{31}=t^2/2, whose integral over t diverges, while the corresponding entry of M is finite. M is better characterized as a transitive-closure-like indicator of whether any indirect causal path of any length exists, not as a time-integrated causal effect. The terminology in §2.3 and §6 should be corrected to avoid overstating the temporal interpretation.
  3. [Abstract and §5] The abstract states that the method is demonstrated on two fluid flow examples, 'linearized Couette flow, and a nonlinear wake flow featuring chaotic dynamics,' and that 'in both cases' the framework captures direct and indirect causal interactions. However, the manuscript body contains no section or results for a nonlinear wake flow; the only fluid application is linearized Couette flow (Section 5). This is a substantive discrepancy between the claimed and actual scope of the demonstrations. The abstract (and any related statements in the introduction) should be revised to match the content, or the missing nonlinear example should be added.
minor comments (5)
  1. [§2.2, Eq. (2.8)] The definition of F_{j→i} appears after the discussion of the error term in (2.7) and is then used to show agreement with Granger causality. It would be clearer to state explicitly that (2.8) is chosen as an operator-based analogue of the Granger F-statistic (as Eq. (3.11) demonstrates), so that the direction of the logical dependency is transparent.
  2. [§2.3, after Eq. (2.20)] The statement 'e^{At} exhibits no causality if the time lag is very small or close to 2' should be more precise: it is the (3,1) element that vanishes at t=2; other entries of e^{At} may be nonzero.
  3. [§3.1, Eq. (3.12)] The transfer entropy expression is introduced with a brief justification; it would be helpful to state explicitly that the same coefficient-constancy assumption underlying (3.6) applies here, and to provide a direct reference for the log-likelihood form of the Granger statistic.
  4. [§3.2, Eq. (3.16)] The choice of α in the global causality metric is left unspecified in the definition; the later choice α=1.05ρ(|A|) (used in Figure 10) should be mentioned in §3.2 for completeness, with a note that the zero pattern of M is independent of α.
  5. [Throughout] The paper alternates between 'linearised' and 'linearized'; consistency is recommended.

Circularity Check

1 steps flagged · score 2.0 of 10

Mild definitional circularity: the LOCA F-statistic in Eq. (2.8) is defined to match the Granger expression derived under Eq. (3.6), so the claimed equivalence is partly by construction; the core matrix-exponential causality measure is otherwise self-contained.

  1. self definitional [Section 2, Eq. (2.8); Section 3.1, Eqs. (3.6)-(3.11)]
    "While our defined notion of causality is proportional to G^τ_{j→i}, we will ultimately modify this definition to account for the statistics of the state variables ... we define a modified F-statistic ... F_{j→i} = [e^{Aτ}]^2_{ij} σ^2_j / σ^2_{ε,i}. ... We can thus conclude that subject to the assumptions of this analysis (most notably that the restricted and unrestricted models share the same coefficients except for those involving the excluded variable), our notion of causality defined in (2.8) agrees with Granger analysis."

    Equation (2.8) is introduced as a definition, not a prediction from the linearized equations: the ratio [e^{Aτ}]^2 σ^2_j / σ^2_{ε,i} is exactly what the Granger derivation (3.6)-(3.11) produces after dropping the sample-size factor. The claimed equivalence 'LOCA agrees with Granger' therefore holds by construction once the coefficient-invariance assumption in (3.6) is granted. The remaining content is the validity of that assumption, which is not established and is contradicted by the authors' own Couette results in the original (v,η) coordinates; agreement appears only in POD coordinates where temporal coefficients are uncorrelated by construction.

full rationale

The central causal object G^τ_{j→i} = [e^{Aτ}]_{ij} is a genuine counterfactual derivative for linear systems and is not fitted to data. The global metric M=(αI-|A|)^{-1} is derived from the Neumann series of |A| and is an independent mathematical bound, not a re-labeled empirical pattern. The DMD-based estimator in Section 4 is a standard least-squares companion to the operator and is tested against synthetic data, not used as an input to the theory. The only definitional circularity is the 'modified F-statistic' in Eq. (2.8), which is deliberately chosen to replicate the Granger scaling derived under the explicit but generally false assumption (3.6) that omitting a regressor leaves the other least-squares coefficients unchanged. Because the paper states this as an assumption and because the core operator-theoretic measure has independent content, the overall circularity is mild. The two self-citations (Lopez-Doriga et al. 2024, Martinez-Sanchez et al. 2023) appear in the literature review and are not load-bearing.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The main free choices are the error covariance Q (which sets the F amplitude), the time horizon τ, and the cutoff α for the global metric. The strongest ad hoc assumption is Eq (3.6) in the Granger equivalence. No new physical entities are introduced.

free parameters (3)
  • Error covariance matrix Q (σ_ε^2)
    Used in the Lyapunov equation (2.9) to compute state variances needed in the F-statistic (2.8). In §5 the authors assume equal noise amplitude for all states, so the overall scale of Q is arbitrary and affects the absolute F amplitude.
  • Time horizon τ = Δt = 1, 5, 20
    Delayed causality is evaluated at user-chosen horizons; the paper shows results are highly sensitive to τ (Fig 3).
  • α in global metric M = 1.05 ρ(|A|) or 1.05|λ_1|
    Chosen to ensure convergence of the Neumann series (αI-|A|)^{-1}; the 1.05 factor is arbitrary and the metric is a bound rather than an exact integrated measure.
assumptions (4)
  • domain assumption The error term ε in (2.7) is additive with known covariance Q, and state statistics are governed by the discrete Lyapunov equation (2.9).
    Introduced in §2.2 to convert the raw matrix exponential into an F-statistic; without data, Q must be assumed.
  • ad hoc to paper Restricted and unrestricted Granger models share all coefficients except the omitted variable's (Eq 3.6).
    This is the key step in the Granger equivalence; it is generally false under least-squares fitting with correlated variables, as the paper's own POD results later confirm.
  • standard math The error term ε_i is uncorrelated with the past of c_j, so the cross-term in Eq (3.9) vanishes.
    Standard Gauss-Markov assumption used to simplify the RSS difference; the paper uses this to drop the cross term.
  • standard math Spectral radius condition α > ρ(|A|) ensures convergence of the series defining M.
    Needed for the Neumann series expansion; the authors choose α accordingly, but the result is an upper bound, not an equality with the causal sum.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Operator theoretic causality analysis of fluid flows using linearized dynamics." pith.science (2026). https://pith.science/paper/SDCAYHSQ

@misc{pith2026250608118,
  author       = {Pith},
  title        = {Pith review of: Operator theoretic causality analysis of fluid flows using linearized dynamics},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/SDCAYHSQ}},
  note         = {Machine review of arXiv:2506.08118}
}
read the original abstract

This paper presents an operator-theoretic framework, Linear Operator Causality Analysis (LOCA), for analyzing causality in linearized dynamical systems, focusing here on fluid flows. Our proposed approach, which can be characterized as a special case of Dynamic Causal Effect (DCE) analysis, utilizes the matrix exponential of linearized differential equations to determine causal relationships between system modes at any future time. We further develop an upper bound that quantifies the presence and extent of global causality across all time horizons. This approach provides a physics-based alternative to data-driven statistical and information-theoretic causality measures such as Granger causality and transfer entropy. Unlike these data-driven techniques that infer causality from time-series data, LOCA leverages the linearized governing equations, yielding a physically-motivated and interpretable measure of causal interactions. We identify the conditions under which LOCA gives equivalent results to data-driven causality analysis methods, and further discuss connections to key system properties such as controllability, observability, and graph-theoretic transitive closure. To complement this operator-based approach, we introduce a data-driven methodology akin to Dynamic Mode Decomposition (DMD) that estimates causal connections directly from time series data by approximating the matrix exponential. We argue that LOCA also mitigates common issues in data-driven causality analyses, such as misleading inferences due to correlated variables or state truncation. We demonstrate our method on two fluid flow examples: linearized Couette flow, and a nonlinear wake flow featuring chaotic dynamics. In both cases, we demonstrate how our framework captures both direct and indirect causal interactions among flow structures.

Figures

Figures reproduced from arXiv: 2506.08118 by the authors.

Figure 1
Figure 1. Global causality estimates for the two systems given in (4.4). (a) The system matrix (A (1) ) is upper triangular and the corresponding M is consequently also upper triangular. (b) Immediate causality in A (2) from 1 → 2 and 2 → 3 is zero, but causal connections emerge in M along these directions through indirect channels. to estimate the system matrix, which can then be used to infer causal relationships between th… view at source ↗
Figure 2
Figure 2. Causality metrics (amplitude of 𝐹-statistic) obtained using LOCA (a,d,g), LOCA-DMD (b,e,h) and Granger analysis (c,f,i) for Couette flow with Re = 2000, and 𝑘 𝑥 = 𝑘 𝑧 = 1, over time horizons 𝛥𝑡 = 1 (a-c), 5 (d-f) and 20 (g-i). Each pixel corresponds to 𝑣 or 𝜂 at a given location in the domain. Pixels in the centre of the domain are larger, corresponding to the larger distance between collocation points in this regio… view at source ↗
Figure 3
Figure 3. Amplitude of 𝐹-statistic for all methods as a function of the timestep, for linearised Couette flow with the same parameters as figure 2. Cause and effect variables are taken at the same location in the centre of the channel for 𝑣 → 𝜂 (a) and adjacent locations near the channel centre for 𝜂 → 𝜂 (b). 0 X0-16 [PITH_FULL_IMAGE:figures/full_fig_p016_3.png] view at source ↗
Figures from the paper (7 more)
Figure 4
Figure 4. Figure 4: (a-c) Log amplitude of entries of discrete time operators log10  [A𝑑] 2 𝑖 𝑗 , (d-f) log of state variance log10 (𝜎 2 𝑗 ), and (g-i) log of the product of these terms log10  [A𝑑] 2 𝑖 𝑗 𝜎 2 𝑗  , which is proportional to 𝐹 assuming uniform noise in each state. Quantit…
Figure 5
Figure 5. Figure 5: Causality (𝐹-statistic) between POD coefficients for timesteps 𝛥𝑡 = 1 (a-c), 5 (d-f), and 20 (g-i), computed using LOCA (a,d,g), LOCA-DMD (b,e,h) and Granger analysis (c,f,i). agreement between all methods in figure 5. This suggests that LOCA and Granger analyses agree…
Figure 6
Figure 6. Figure 6: Causality (𝐹-statistic) between POD coefficients computed using LOCA for a timestep 𝛥𝑡 = 20, where the dynamics are projected onto the first 𝑟 = (a) 5, (b) 10, (c) 15 and (d) 20 POD modes. Dotted lines indicate the truncation levels used in other subplots, highlighting…
Figure 7
Figure 7. Figure 7: As for figure 4, but using POD coefficients: (a-c) Log amplitude of entries of discrete time operators log10  [PITH_FULL_IMAGE:figures/full_fig_p019_7.png]
Figure 8
Figure 8. Figure 8: Causality (𝐹-statistic) identified between balancing coordinate states using LOCA for timesteps 𝛥𝑡 = (a) 1, (b) 5, and (c) 20. Other methods are not shown but yield similar results [PITH_FULL_IMAGE:figures/full_fig_p020_8.png]
Figure 9
Figure 9. Figure 9: Causality (𝐹-statistic) identified between balancing coordinate states using LOCA, where the dynamics are projected onto the first 𝑟 = (a) 5, (b) 10, (c) 15 and (d) 20 state variables. such a subspace using balanced truncation of the linear system, which is designed to…
Figure 10
Figure 10. Figure 10: Logarithm of the magnitude of the global causality indicator M = (𝛼I − |A|)−1 for the linear Couette flow system. 𝛼 is chosen to be 1.05|𝜆1 |, where 𝜆1 is the eigenvalue of |A| with largest magnitude. approaches, such as Granger causality and Transfer Entropy, which d…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

58 extracted references · 56 canonical work pages

  1. [1]

    , " * write output.state after.block = add.period write newline

    ENTRY address author booktitle chapter edition editor howpublished institution journal key month note number organization pages publisher school series title type volume year eprint label extra.label sort.label short.list INTEGERS output.state before.all mid.sentence after.sentence after.block FUNCTION init.state.consts #0 'before.all := #1 'mid.sentence ...

  2. [2]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in bbl.in capitalize " " * FUNCT...

  3. [3]

    , Cavalieri, A.V.G

    Abreu, L.I. , Cavalieri, A.V.G. , Schlatter, P. , Vinuesa, R. & Henningson, D.S. 2020 Spectral proper orthogonal decomposition and resolvent analysis of near-wall coherent structures in turbulent pipe flows . J. Fluid Mech. 900 , A11

    work page 2020

  4. [4]

    & Sameshima, K

    Baccal \'a , L.A. & Sameshima, K. 2001 Partial directed coherence: a new concept in neural structure determination . Biological Cybernetics 84 (6), 463--474

    work page 2001

  5. [5]

    , Encinar, M.P

    Bae, H.J. , Encinar, M.P. & Lozano-Dur \'a n, A. 2019 Causal analysis of self-sustaining processes in the logarithmic layer of wall-bounded turbulence . Journal of physics. Conference series 1001

    work page 2019

  6. [6]

    & Bossomaier, T.R.J

    Barnett, L.C. & Bossomaier, T.R.J. 2012 Transfer entropy as a log-likelihood ratio . Phys. Rev. Lett. 109 13 , 138105

    work page 2012

  7. [7]

    & Seth, A.K

    Barnett, L. & Seth, A.K. 2014 The MVGC multivariate G ranger causality toolbox: a new approach to G ranger-causal inference . Journal of Neuroscience Methods 223 , 50--68

    work page 2014

  8. [8]

    & Seth, A.K

    Barnett, L. & Seth, A.K. 2015 Granger causality for state-space models . Phys. Rev. E 91 (4), 040101

    work page 2015

Show all 58 references
  1. [9]

    2003 A numerical algorithm to solve A^TX A-X = Q

    Barraud, A.Y. 2003 A numerical algorithm to solve A^TX A-X = Q . IEEE Transactions on Automatic Control 22 (5), 883--885

    2003

  2. [10]

    & Piersol, A.G

    Bendat, J.S. & Piersol, A.G. 2011 Random data: analysis and measurement procedures\/ . John Wiley & Sons, Hoboken, New Jersey

    2011

  3. [11]

    , Holmes, P

    Berkooz, G. , Holmes, P. & Lumley, J. L. 1993 The proper orthogonal decomposition in the analysis of turbulent flows . Annual Review of Fluid Mechanics 25 (1), 539--575

    1993

  4. [12]

    John Wiley & Sons, New York

    Bollen, Kenneth A 1989 Structural equations with latent variables\/ . John Wiley & Sons, New York

    1989

  5. [13]

    , Gallusser, F

    Brodersen, K.H. , Gallusser, F. , Koehler, J. , Remy, N. & Scott, S.L. 2015 Inferring causal impact using B ayesian structural time-series models . The Annals of Applied Statistics pp. 247--274

    2015

  6. [14]

    & Goldburg, W.I

    Cerbus, R.T. & Goldburg, W.I. 2013 Information content of turbulence . Phys. Rev. E 88 (5), 053012

    2013

  7. [15]

    & Jimenez, J

    Del Alamo, J.C. & Jimenez, J. 2006 Linear energy amplification in turbulent channels . J. Fluid Mech. 559 , 205--213

    2006

  8. [16]

    , Gaviglio, J.J

    Favre, A.J. , Gaviglio, J.J. & Dumas, R. 1957 Space-time double correlations and spectra in a turbulent boundary layer . J. Fluid Mech. 2 (4), 313--342

    1957

  9. [17]

    1969 Investigating causal relations by econometric models and cross-spectral methods

    Granger, C.W.J. 1969 Investigating causal relations by econometric models and cross-spectral methods . Econometrica: Journal of the Econometric Society pp. 424--438

    1969

  10. [18]

    1986 Excitation of direct resonances in plane P oiseuille flow

    Gustavsson, L.H. 1986 Excitation of direct resonances in plane P oiseuille flow . Studies in Applied Mathematics 75 (3), 227--248

    1986

  11. [19]

    1965 The environment and disease: association or causation?

    Hill, A.B. 1965 The environment and disease: association or causation?

    1965

  12. [20]

    , Lumley, J

    Holmes, P. , Lumley, J. L. , Berkooz, G. & Rowley, C. W. 2012 Turbulence, coherent structures, dynamical systems and symmetry\/ . Cambridge University Press

    2012

  13. [21]

    & Rowley, C.W

    Ilak, M. & Rowley, C.W. 2008 Modeling of transitional channel flow using balanced proper orthogonal decomposition . Phys. Fluids. 20 , 034103

    2008

  14. [22]

    & Rubin, D.B

    Imbens, G.W. & Rubin, D.B. 2015 Causal inference in statistics, social, and biomedical sciences\/ . Cambridge University Press

    2015

  15. [23]

    1927 La diffusion de la lumi\'ere par les atomes

    Kramers, H.A. 1927 La diffusion de la lumi\'ere par les atomes . Atti del Congresso Internazionale dei Fisici 2 , 545

    1927

  16. [24]

    1926 On the theory of dispersion of x-rays

    de Kronig, R. 1926 On the theory of dispersion of x-rays . Journal of the Optical Society of America 12 (6), 547--556

    1926

  17. [25]

    1975 Wave breakdown and turbulence

    Landahl, M.T. 1975 Wave breakdown and turbulence . SIAM Journal on Applied Mathematics 28 (4), 735--756

    1975

  18. [26]

    1973 Causation

    Lewis, D. 1973 Causation . The Journal of Philosophy 70 (17), 556--567

    1973

  19. [27]

    , Atzori, M

    Lopez-Doriga, B. , Atzori, M. , Vinuesa, R. , Bae, H.J. , Srivastava, A. & Dawson, S.T.M. 2024 Linear and nonlinear G ranger causality analysis of turbulent duct flows. In Journal of Physics: Conference Series\/ , , vol. 2753 , p. 012017 . IOP Publishing

    2024

  20. [28]

    & Arranz, G

    Lozano-Dur \'a n, A. & Arranz, G. 2022 Information-theoretic formulation of dynamical systems: causality, modeling, and control . Phys. Rev. Research 4 (2), 023195

    2022

  21. [29]

    , Bae, H.J

    Lozano-Dur \'a n, A. , Bae, H.J. & Encinar, M.P. 2020 a\/ Causality of energy-containing eddies in wall turbulence . J. Fluid Mech. 882 , A2

    2020

  22. [30]

    , Constantinou, N.C

    Lozano-Dur \'a n, A. , Constantinou, N.C. , Nikolaidis, M.-A. & Karp, M. 2020 b\/ Cause-and-effect of linear mechanisms sustaining wall turbulence . J. Fluid Mech. 914

    2020

  23. [31]

    , Darden, L

    Machamer, P. , Darden, L. & Craver, C.F. 2000 Thinking about mechanisms . Philosophy of Science 67 (1), 1--25

    2000

  24. [32]

    , Arranz, G

    Mart \'i nez-S \'a nchez, \'A . , Arranz, G. & Lozano-Dur \'a n, A. 2024 Decomposing causality into its synergistic, unique, and redundant components . Nat. Commun. 15

    2024

  25. [33]

    , L \'o pez, E

    Mart \' nez-S \'a nchez, \'A . , L \'o pez, E. , Le Clainche, S. , Lozano-Dur \'a n, A. , Srivastava, A. & Vinuesa, R. 2023 Causality analysis of large-scale structures in the flow around a wall-mounted square cylinder . J. Fluid Mech. 967 , A1

    2023

  26. [34]

    , Rezaeiravesh, S

    Massaro, D. , Rezaeiravesh, S. & Schlatter, P. 2023 On the potential of transfer entropy in turbulent dynamical systems . Scientific Reports 13

    2023

  27. [35]

    1990 Causality, feedback and directed information

    Massey, J. 1990 Causality, feedback and directed information. In Proc. Int. Symp. Inf. Theory Applic.(ISITA-90)\/ , , vol. 2

    1990

  28. [36]

    & Sharma, A.S

    McKeon, B.J. & Sharma, A.S. 2010 A critical-layer framework for turbulent pipe flow . J. Fluid Mech. 658 , 336--382

    2010

  29. [37]

    1981 Principal component analysis in linear systems: Controllability, observability, and model reduction

    Moore, B.C. 1981 Principal component analysis in linear systems: Controllability, observability, and model reduction . IEEE transactions on automatic control 26 (1), 17--32

    1981

  30. [38]

    Cambridge University Press

    Pearl, J 2000 Causality: Models, reasoning and inference\/ . Cambridge University Press

    2000

  31. [39]

    2014 Probabilistic reasoning in intelligent systems: networks of plausible inference\/

    Pearl, J. 2014 Probabilistic reasoning in intelligent systems: networks of plausible inference\/ . Elsevier

    2014

  32. [40]

    2005 Model reduction for fluids using balanced proper orthogonal decomposition

    Rowley, C.W. 2005 Model reduction for fluids using balanced proper orthogonal decomposition. International Journal of Bifurcation and Chaos 15 (3), 997--1013

    2005

  33. [41]

    & Dawson, S.T.M

    Rowley, C.W. & Dawson, S.T.M. 2017 Model reduction for flow analysis and control . Annu. Rev. Fluid Mech. 49 (1), 387--417

    2017

  34. [42]

    1974 Estimating causal effects of treatments in randomized and nonrandomized studies

    Rubin, D.B. 1974 Estimating causal effects of treatments in randomized and nonrandomized studies. Journal of Educational Psychology 66 (5), 688

    1974

  35. [43]

    2016 Information flow and causality as rigorous notions ab initio

    San Liang , X. 2016 Information flow and causality as rigorous notions ab initio. Phys. Rev. E 94 , 052201

    2016

  36. [44]

    2005 Information transfer between dynamical system components

    San Liang , X & Kleeman, R. 2005 Information transfer between dynamical system components . Phys. Rev. Lett. 95 (24), 244101

    2005

  37. [45]

    & Lozano-Dur \'a n, A

    San Liang , X. & Lozano-Dur \'a n, A. 2016 A preliminary study of the causal structure in fully developed near-wall turbulence . CTR Proc. of the Summer Prog p. 233

    2016

  38. [46]

    2010 Dynamic mode decomposition of numerical and experimental data

    Schmid, P.J. 2010 Dynamic mode decomposition of numerical and experimental data . J. Fluid Mech. 656 , 5 -- 28

    2010

  39. [47]

    & Hussain, F

    Schoppa, W. & Hussain, F. 2002 Coherent structure generation in near-wall turbulence . J. Fluid Mech. 453 , 57--108

    2002

  40. [48]

    2000 Measuring information transfer

    Schreiber, T. 2000 Measuring information transfer . Phys. Rev. Lett. 85 (2), 461

    2000

  41. [49]

    2020 Causality and passivity: From electromagnetism and network theory to metamaterials

    Srivastava, A. 2020 Causality and passivity: From electromagnetism and network theory to metamaterials . Mechanics of Materials p. 103710

    2020

  42. [50]

    , May, R

    Sugihara, G. , May, R. , Ye, H. , Hsieh, C.-H. , Deyle, E. , Fogarty, M. & Munch, S. 2012 Detecting causality in complex ecosystems . Science 338 (6106), 496--500

    2012

  43. [51]

    1973 A probabilistic theory of causality

    Suppes, P. 1973 A probabilistic theory of causality . British Journal for the Philosophy of Science 24 (4)

    1973

  44. [52]

    , Brunton, S

    Taira, K. , Brunton, S. L. , Dawson, S. T. M. , Rowley, C. W. , Colonius, T. , McKeon, B. J. , Schmidt, O. T. , Gordeyev, S. , Theofilis, V. & Ukeiley, L. S. 2017 Modal analysis of fluid flows: An overview . arXiv preprint arXiv:1702.01453

    2017 arXiv

  45. [53]

    1941 On the calculus of relations

    Tarski, A. 1941 On the calculus of relations . The Journal of Symbolic Logic 6 (3), 73--89

    1941

  46. [54]

    , Lozano-Dur \'a n, A

    Tissot, G. , Lozano-Dur \'a n, A. , Cordier, L. , Jim \'e nez, J. & Noack, B.R. 2014 Granger causality in wall-bounded turbulence . Journal of Physics: Conference Series 506 , 012006

    2014

  47. [55]

    Journal of Computational Dynamics 1 (2), 391--421

    Tu, Jonathan H , Rowley, Clarence W , Luchtenburg, Dirk M , Brunton, Steven L & Kutz, J Nathan 2014 On dynamic mode decomposition: Theory and applications . Journal of Computational Dynamics 1 (2), 391--421

    2014

  48. [56]

    1962 A theorem on B oolean matrices

    Warshall, S. 1962 A theorem on B oolean matrices . J. ACM 9 , 11--12

    1962

  49. [57]

    2001 Introduction to graph theory

    West, D.B. 2001 Introduction to graph theory

    2001

  50. [58]

    & Peraire, J

    Willcox, K. & Peraire, J. 2002 Balanced model reduction via the proper orthogonal decomposition . AIAA J. 40 (11), 2323--2330

    2002

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.