Abstract
We develop reduced, stochastic models for high dimensional, dissipative dynamical systems that relax very slowly to equilibrium and can encode long term memory. We present a variety of empirical and first principles approaches for model reduction, and build a mathematical framework for analyzing the reduced models. We introduce the notions of universal and asymptotic filters to characterize ‘optimal’ model reductions for sloppy linear models. We illustrate our methods by applying them to the practically important problem of modeling evaporation in oil spills.
Similar content being viewed by others
References
Amir, A., Oreg, Y., Imry, Y.: On relaxations and aging of various glasses. Proc. Natl Acad. Sci. U.S.A. 109, 1850–1855 (2012)
Arnold, H.M., Moroz, I.M., Palmer, T.N.: Stochastic parametrizations and model uncertainty in the Lorenz ’96 system. Philos. Trans. R. Soc. Lond. A 371, 20120510 (2013)
Baladi, V.: Positive Transfer Operators and Decay of Correlations, vol. 16. Advanced Series in Nonlinear Dynamics. World Scientific, Singapore (2000)
Berkenbusch, M.K., Claus, I., Dunn, C., Kadanoff, L.P., Nicewicz, M., Venkataramani, S.C.: Discrete charges on a two dimensional conductor. J. Stat. Phys. 116, 1301–1358 (2004)
Berry, T., Harlim, J.: Forecasting turbulent modes with nonparametric diffusion models: learning from Noisy data. Physica D 320, 57–76 (2016)
Bouchaud, J.-P., Cugliandolo, L.F., Kurchan, J., Mezard, M.: Out of equilibrium dynamics in spin-glasses and other glassy systems. In: Spin Glasses and Random Fields, pp. 161–223. World Scientific, Singapore (1998)
Bouchaud, J.-P.: Aging in glassy systems: new experiments, simple models, and open questions. In: Cates, M.E., Evans, M. (eds.) Soft and Fragile Matter: Nonequilibrium Dynamics, Metastability and Flow, pp. 285–304. Institute of Physics, Bristol (2000)
Box, G.E.P., Jenkins, G.M., Reinsel, G.C., Ljung, G.M.: Time Series Analysis: Forecasting and Control. Wiley Series in Probability and Statistics. Wiley, Hoboken (2015)
Brown, K.S., Sethna, J.P.: Statistical mechanical approaches to models with many poorly known parameters. Phys. Rev. E 68, 021904 (2003)
Budisic, M., Mohr, R., Mezic, I.: Applied Koopmanism. Chaos 22(4), 047510 (2012)
Chekroun, M.D., Kondrashov, D., Ghil, M.: Predicting stochastic systems by noise sampling, and application to the El Niño-southern oscillation. Proc. Natl Acad. Sci. U.S.A. 108, 11766–11771 (2011)
Chorin, A.J., Hald, O.H.: Stochastic Tools in Mathematics and Science, vol. 58. Texts in Applied Mathematics. Springer, New York (2014)
Chorin, A.J., Hald, O.H., Kupferman, R.: Optimal prediction and the Mori–Zwanzig representation of irreversible processes. Proc. Natl Acad. Sci. U.S.A. 97, 2968–2973 (2000)
Chorin, A., Hald, O., Kupferman, R.: Optimal prediction with memory. Physica D 166, 239–257 (2002)
Chorin, A.J., Lu, F.: Discrete approach to stochastic parametrization and dimension reduction in nonlinear dynamics. Proc. Natl Acad. Sci. U.S.A. 112, 9804–9809 (2015)
Chorin, A., Stinis, P.: Problem reduction, renormalization, and memory. Commun. Appl. Math. Comput. Sci. 1, 1–27 (2007)
Coifman, R.R., Lafon, S.: Diffusion maps. Appl. Comput. Harmonic Anal. 21, 5–30 (2006)
Comeau, D., Zhao, Z., Giannakis, D., Majda, A.J.: Data-driven prediction strategies for low-frequency patterns of north pacific climate variability. Clim. Dyn. 48(5), 1855–1872 (2015)
Crisanti, A., Ritort, F.: Violation of the fluctuation–dissipation theorem in glassy systems: basic notions and the numerical evidence. J. Phys. A 36, R181 (2003)
Darve, E., Solomon, J., Kia, A.: Computing generalized langevin equations and generalized Fokker–Planck equations. Proc. Natl Acad. Sci. U.S.A. 106, 10884–10889 (2009)
Dixon, P.K., Wu, L., Nagel, S.R., Williams, B.D., Carini, J.P.: Scaling in the relaxation of supercooled liquids. Phys. Rev. Lett. 65, 1108–1111 (1990)
Fingas, M.F.: A literature review of the physics and predictive modelling of oil spill evaporation. J. Hazard. Mater. 42, 157–175 (1995)
Fingas, M.: Modeling evaporation using models that are not boundary-layer regulated. J. Hazard Mater. 107, 27–36 (2004)
Fingas, M.: Modeling oil and petroleum evaporation. J. Pet. Sci. Res. 2(3), 104–115 (2013)
Flajolet, P., Odlyzko, A.: Singularity analysis of generating functions. SIAM J. Discret. Math. 3, 216–240 (1990)
Giannakis, D., Majda, A.J.: Nonlinear Laplacian spectral analysis for time series with intermittency and low-frequency variability. Proc. Natl Acad. Sci. U.S.A. 109, 2222–2227 (2012)
Givon, D., Kupferman, R., Stuart, A.: Extracting macroscopic dynamics: model problems and algorithms. Nonlinearity 17, R55 (2004)
Givon, D., Kupferman, R., Hald, O.H.: Existence proof for orthogonal dynamics and the Mori–Zwanzig formalism. Isr. J. Math. 145, 221–241 (2005)
Harlim, J., Kang, E.L., Majda, A.J.: Regression models with memory for the linear response of turbulent dynamical systems. Commun. Math. Sci. 11(2), 481–498 (2013)
Hoult, D.P. (ed.): Oil on the Sea: Proceedings of a Symposium on the Scientific and Engineering Aspects of Oil Pollution of the Sea. Springer, New York (1969)
Jazwinski, A.H.: Stochastic Processes and Filter Theory. Academic Press, New York (1970)
Kampen, N.V.: Stochastic Processes in Physics and Chemistry, 3rd edn. North-Holland Personal Library. North Holland, Amsterdam (2007)
Kawasaki, K.: Simple derivations of generalized linear and nonlinear Langevin equations. J. Phys. A 6, 1289 (1973)
Kawasaki, K.: Theoretical methods dealing with slow dynamics. J. Phys. 12, 6343 (2000)
Kondrashov, D., Chekroun, M., Ghil, M.: Data-driven non-Markovian closure models. Physica D 297, 33–55 (2015)
Kubo, R.: The fluctuation–dissipation theorem. Rep. Prog. Phys. 29, 255 (1966)
Kutner, M., Nachtsheim, C., Neter, J., Li, W.: Applied Linear Statistical Models. McGraw-Hill/Irwin, Chicago (2004)
Lin, K., Lu, F.: Stochastic parametrization, filtering, and the Mori–Zwanzig formalism. Preprint (2017)
Lu, F., Lin, K.K., Chorin, A.J.: Data-based stochastic model reduction for the Kuramoto–Sivashinsky equation. Physica D 340, 46–57 (2017)
Mackay, D., Matsugu, R.S.: Evaporation rates of liquid hydrocarbon spills on land and water. Can. J. Chem. Eng. 51, 434–439 (1973)
Majda, A.J., Harlim, J.: Physics constrained nonlinear regression models for time series. Nonlinearity 26, 201 (2013)
Matan, K., Williams, R.B., Witten, T.A., Nagel, S.R.: Crumpling a thin sheet. Phys. Rev. Lett. 88, 076101 (2002)
Moghimi, S., Ramírez, J.M., Restrepo, J.M., Venkataramani, S.C.: Mass exchange dynamics of surface and subsurface oil in shallow-water transport. Preprint (2017)
Mori, H.: Transport, collective motion, and Brownian motion. Prog. Theor. Phys. 33, 423–455 (1965)
Oded, B., Rubinstein, S.M., Fineberg, J.: Slip-stick and the evolution of frictional strength. Nature 463, 76–9 (2010)
Oppenheim, A.V., Schafer, R.W., Buck, J.R.: Discrete-Time Signal Processing, 2nd edn. Prentice-Hall Signal Processing Series. Prentice Hall, Englewood Cliffs (1999)
Ott, E.: Chaos in Dynamical Systems. Cambridge University Press, New York (1993)
Polya, G., Szegö, G.: Problems and Theorems in Analysis II: Theory of Functions, Zeros, Polynomials, Determinants, Number Theory, Geometry, Classics in Mathematics. Springer, New York (1998)
Restrepo, J.M., Venkataramani, S.C., Dawson, C.: Nearshore sticky waters. Ocean Model. 80, 49–58 (2014)
Restrepo, J.M., Ramírez, J.M., Venkataramani, S.C.: An oil fate model for shallow waters. J. Marine Sci. Eng. 3, 1504–1543 (2015)
Spaulding, M.L.: A state-of-the-art review of oil spill trajectory and fate modeling. Oil Chem. Pollut. 4, 39–55 (1988)
Stinis, P.: Renormalized Mori–Zwanzig-reduced models for systems without scale separation. Proc. R. Soc. Lond. Ser. A 471, 20140446 (2015)
Stiver, W., Mackay, D.: Evaporation rate of spills of hydrocarbons and petroleum mixtures. Environ. Sci. Technol. 18(11), 834–840 (1984)
Sutton, O.G.: Wind structure and evaporation in a turbulent atmosphere. Proc. R. Soc. Lond. Ser. A 146, 701–722 (1934)
Takens, E.: Detecting strange attractors in turbulence. In: Rand, D., Young, L.-S. (eds.) Dynamical Systems and Turbulence, pp. 366–381. Springer, Berlin (1981)
Transtrum, M.K., Machta, B.B., Sethna, J.P.: Geometry of nonlinear least squares with applications to sloppy models and optimization. Phys. Rev. E 83, 036701 (2011)
Vautard, R., Yiou, P., Ghil, M.: Singular-spectrum analysis: a toolkit for short, Noisy chaotic signals. Physica D 58, 95–126 (1992)
Vautard, R., Ghil, M.: Singular spectrum analysis in nonlinear dynamics, with applications to paleoclimatic time series. Physica D 35, 395–424 (1989)
Venturi, D., Cho, H., Karniadakis, G.: Mori–Zwanzig approach to uncertainty quantification. In: Ghanem, R., Higdon, D., Owhadi, H. (eds.) Handbook of Uncertainty Quantification. Springer, Heidelberg (2016)
Venturi, D., Karniadakis, G.: Convolutionless Nakajima–Zwanzig equations for stochastic analysis in nonlinear dynamical systems. Proc. R. Soc. Math. Phys. Eng. Sci. 470, 20130754–20130754 (2014)
Walker, G.: On periodicity in series of related terms. Proc. R. Soc. Lond. Ser. A 131, 518–532 (1931)
Waterfall, J.J., Casey, F.P., Gutenkunst, R.N., Brown, K.S., Myers, C.R., Brouwer, P.W., Elser, V., Sethna, J.P.: Sloppy-model universality class and the Vandermonde matrix. Phys. Rev. Lett. 97, 150601 (2006)
Yule, G.U.: On a method of investigating periodicities in disturbed series, with special reference to Wolfer’s sunspot numbers. Philos. Trans. R. Soc. Lond. Ser. A 226, 267–298 (1927)
Zwanzig, R.: Problems in nonlinear transport theory. In: Garrido, L. (ed.) Systems Far from Equilibrium, vol. 132. Lecture Notes in Physics. Springer, Berlin (1980)
Zwanzig, R.: Nonlinear generalized Langevin equations. J. Stat. Phys. 9, 215–220 (1973)
Zwanzig, R.: Nonequilibrium Statistical Mechanics. Oxford University Press, New York (2001)
Acknowledgements
S.V. would like to acknowledge the many, very illuminating discussions with Kevin Lin who was very generous with his time and his ideas. We are grateful to an anonymous referee for pointing out the potential connections between our work and the sloppy models universality class. This viewpoint turns out to be particularly fruitful. This work was funded in part by a Grant from GoMRI. We also received support from NSF-DMS-1109856 and NSF-OCE-1434198.
Author information
Authors and Affiliations
Corresponding author
Additional information
It is with immense gratitude that we dedicate this article to Leo Kadanoff. Two of the authors (SV and JR) first met Leo as postdocs. Our lives would have been very different if not for the outsize role that Leo played in our professional development and also in our personal growth. His door, and his mind, were always open. He reminded us to ask questions, about how humility gave the courage to know what we knew and we did not. He taught us to be fearless about pursuing a wide range of interests. The fun he had with science was infectious, and the skill he had to ask the right questions is impossible to match. Our experiences were by no means unique. There are hundreds of people whose lives Leo touched in the same way. So many of his informal seminar or lunch questions turned into full research enterprises. It is no exaggeration to say that at some point people needed only to know that Leo had been the one to ask the question in order to assure themselves that their scientific investigations were worthwhile. The last time that one of us saw Leo was in May of 2015. Coincidentally, it was at a talk on the subject of oil spill modeling, and included some of the rudimentary ideas that grew into this paper. Leo came down to the (new) James Franck Institute for the talk. He was just as sharp as ever, and he made sure that the graduate students in the audience got all the physical intuition that the speaker elided over, by interjecting appropriately. It was classic Leo. How we miss him!
Appendices
Appendix 1: The Memory Kernel for Multiple Observables
One other comment is that we can indeed compute the memory kernel explicitly for the evaporation process (4), not just for the case with one observable, the mass \(M_n\), but also more generally if we have a vector-valued linear observable \(\varPhi \), i.e l scalar-valued observables \(\varPhi = \{\phi ^1,\phi ^2,\ldots ,\phi ^l\}^T\). Each scalar linear observable \(\phi ^i\) is given by an element of \(\mathcal {H}^*\), and we will denote the corresponding bra-vector by \(\langle {\phi ^i}|\). Using the Gram–Schmidt procedure if necessary, we can assume that the vectors \(\langle {\phi ^i}|\) given an orthonormal basis for their span, a l-dimensional subspace of \(\mathcal {H}^*\). The orthogonal projection \(P^* : \mathcal {H}^* \rightarrow \mathcal {H}^*\) onto this subspace is given by
It follows that \(\langle {\phi ^i}| P^* = \langle {\phi ^i}| P = \langle {\phi ^i}|\) and (11) gives
The quantities \( \langle {\phi ^i}|{\xi _{n+1}}\rangle \) are the entries of the “vector” observable \(\varPhi _{n+1}\). Defining the matrices \(H_k\) by \((H_k)_{ij} = \langle {\phi ^i}| \varLambda (Q \varLambda )^k\big \vert {\phi ^j}\big \rangle \) for \(k= 0,1,2,\ldots \) and the (column) vectors \(\beta _n\) by the entries \(\beta ^i_n = \langle {\phi ^i}| (\varLambda Q)^{n+1} \big \vert {\rho _0}\big \rangle \), we have the Mori–Zwanzig decomposition
If \(\big \vert {\rho _0}\big \rangle \) is in the span of \(\big \vert {\phi ^i}\big \rangle \), then \(Q \big \vert {\rho _0}\big \rangle = 0\) so that the noise \(\beta _n\) is identically zero. Taking \(\big \vert {\rho _0}\big \rangle = \big \vert {\phi ^1}\big \rangle ,\big \vert {\phi ^2}\big \rangle ,\ldots ,\big \vert {\phi ^l}\big \rangle \) in turn, and collecting the corresponding column vectors \(\varPhi _n\) into a \(l \times l\) matrix \(\varXi _n\), we have
is a symmetric matrix for each n, and
As before, we can determine the memory kernel \(H_k\) using the \(\mathcal {Z}\)-transform. Defining the matrices
we get
The matrix \(\varXi _n\) is symmetric for all n, so that \(H_n\) is also symmetric for all n. We expect that the norm \(\Vert \varXi _n\Vert \) typically decays no faster than 1/n. This is true for instance if the constant functions are in the range of P, or more generally if there are continuous functions \(\psi \) with \(\psi (0) > 0\) in the range of P. In this case, we expect that the norm of \(H_n\) decays no faster that \(1/(n \log ^2(n))\) indicating again that, generically, one expects fat tails in the memory kernel for the system (4) if we use the Mori–Zwanzig decomposition based on any finite set of linear observables.
Appendix 2: Orthogonal Dynamics
We will now compute the statistics of the noise process \(\beta _n\) in the Mori–Zwanzig decomposition (12) with the usual approach through the study of the projection equation (9) and the orthogonal dynamics (10). Since the orthogonal dynamics are linear, it suffices to solve the system
where \(x \in [0,1]\) is fixed. A calculation reveals that, for any continuous function \(\phi \),
We will thus associate \(\langle {F_0}|\) with the “function” \(F_0(w) = \delta (w-x) - 1\). We can follow this computation to solve the orthogonal dynamics equations recursively. For example,
so that \(\langle {F_1}|\) corresponds to the function \(F_1(w) = e^{-x\tau } \delta (w-x)- e^{-w \tau } - e^{-x \tau } + \frac{1-e^{-\tau }}{\tau }\). Using the fact that Q and \(\varLambda \) are self-adjoint operators on \(\mathcal {H}\), and further \(\langle {\psi }| \varLambda \big \vert {\phi }\big \rangle = \int \psi (w) e^{-w \tau } \phi (w) dw\) so that \(\varLambda \) is diagonal on the “basis” \(\{\delta (w-x)\}_{\{x \in [0,1]\}}\), an inductive argument shows that \(F_n(w) = e^{-nx\tau } \delta (w-x) + \varPsi _n(w;x)\) where \(\varPsi _n\) is a smooth, symmetric function \(\varPsi _n(w;x) = \varPsi _n(x;w)\). We will use these conclusions to verify the full solution for \(\langle {F_n}|\) that we obtain below by independent means.
Consider the \(\mathcal {Z}\)-transform \(\hat{\langle {F}|} = \sum z^{-n} \langle {F_n}|\). The orthogonal dynamics imply
Using the ansatz \(\hat{F}(z,x,w) = \hat{A}(z,x) \delta (w-x) + \hat{\varPsi }(z,x,w)\) corresponding to a decomposition of \(F_n\) into its singular and regular parts, we get the pair of equations
where we have suppressed the arguments (z, x, w) for \(\hat{A}\) and \(\hat{\varPsi }\) for clarity. We can solve the first equation to obtain
Using this in the second equation, we obtain
where \(C(z,w) = \int e^{-x \tau } \hat{\varPsi } dx\) is determined in terms of the required solution \(\hat{\varPsi }\) self-consistently. Multiplying by \(e^{-x \tau }\) and integrating in x, and solving the resulting equation for C(z, w), we obtain
Using this result in the computation for \(\hat{\varPsi }\) gives
This gives a complete solution of orthogonal dynamics equation by
The singular part of \(F_n\) is therefore \(e^{-n \tau x} \delta (w-x)\) as we noted above. Further, the regular part \(\hat{\varPsi }\) is symmetric in w and x, implying this property for each of the functions \(\varPsi _n\). Finally, for an observable given by a continuous function g, the solution to the orthogonal dynamics is given by
For the observable \(M_{n}\), the prediction for the total mass at the next time step, we have \(\langle {g}| = \langle {1}| \varLambda \). The \(\mathcal {Z}\)-transform of the memory kernel is given by
The \(\mathcal {Z}\)-transform of the expected values of the noise sequence \(\beta _n\) is given by
and the correlations between the noise \(\beta _n\) and the mass \(M_j\) are given by \(\mathbb {E}[\beta _n M_j] = \bar{\sigma }^2 \langle {1}|{(\varLambda Q)^n \varLambda ^j}|{1}\rangle \) (see Sect. 5). Taking the (two index) \(\mathcal {Z}\)-transform, noting that \(\beta _0 = 0\), we have
It is not true that \(\mathbb {E}[\beta _n M_j] = 0\) if \(n > j\), as one would expect in the Mori–Zwanzig decomposition for a system with an invariant measure. In particular,
Appendix 3: Sampling Initial Conditions
For any prescribed value \(0< \bar{\sigma }^2 < \infty \), we can indeed find a family of I-dependent distributions such that
by appropriately truncating and rescaling a distribution that has finite mean but infinite variance. For example, the function
satisfies \(f \ge 0\) on \((0,\infty )\) and \(\int _0^\infty f(x) dx = 1\), so f is indeed a nonmalized density on \((0,\infty )\). Further \(\int _0^\infty x f(x) dx = 1\) and \(\int _0^L x^2 f(x) dx \sim \sqrt{\frac{32}{75} L}\) for \(L \gg 1\). We can therefore define a sequence of I dependent distributions by truncating the support of f and renormalizing to have unit mass, i.e.
where \(L_I\) is any sequence satisfying \(L_I \ge 2/3\) for all I, \(L_I \nearrow \infty \) and \({\sqrt{\frac{32}{75 I^2} L_I} \rightarrow \bar{\sigma }^2}\) as \(I \rightarrow \infty \). Given such a sequence \(L_I\), the normalization \(c_I\) is determined by \(\int _0^I f_I(x) dx = 1\) so that \(c_I \rightarrow 1\).
Appendix 4: Asymptotic Solutions of the Yule–Walker Equations
We seek a solution to (30) as an asymptotic series in n, i.e. solutions of the form
The difficulty in solving this system is evident if we expand the coefficient matrix A as a power series in n:
Assuming \(L \ge 3\), the two matrices displayed in the expansion of A are singular. The first matrix has rank 1, the second has rank 2. Indeed the first \(L-1\) matrices in the expansion of A are all singular and their (row) nullspaces are nested
and so on. The determinant of A is thus very close to zero (\(\det \left\{ A \right\} \sim O(n^{-L^2})\) as we see below) so it is not clear that we have solutions for \(h^{(n)}\) where the leading order behavior stays O(1) instead of diverging with n. Proving the boundedness of \(h^{(n)}\) and determining the O(1) solution thus requires consideration of L solvability conditions given by the vectors that span the common (row)-nullspaces of the initial j terms in the expansion of A for \(j=1,2,\ldots ,L-1\). Higher order terms will require even longer expansion of the matrices and more solvability conditions.
In the general case of a process with slowly decaying correlations, it is still true that the matrix of coefficients in the Yule–Walker equation is nearly singular, and one does have to go through the process described above to find optimal, reduced dimensional, models for such systems. For the evaporation process (4) however, the coefficient matrix has a special structure, that we exploit to find the solutions for the optimal filter \(h^{(n)}\). The matrix A is a Cauchy matrix [48] i.e its entries are of the form \(A_{ij} = 1/(x_i-y_j)\). In particular, we can choose \(x_i = 2n-i\) and \(y_j = j\). The determinant of a Cauchy matrix \(A_{ij} = 1/(x_i-y_j)\) is given by [48]
For the particular matrix A from above, the terms in the numerator are all bounded by L and the terms in the denominator are all \(\approx 2n\) if \(n \gg L\). Consequently, \(\det {A} \sim O(n^{-L^2})\). The matrix \(\hat{A}_m\) obtained by replacing the mth column of A by the vector \(v_i = \frac{1}{2n -i}\) is also a Cauchy matrix \(\hat{A}_{ij} = 1/(x_i - \hat{y}_j)\), with the same choice \(x_i = 2n-i\) and
Cramer’s rule now yields,
Rights and permissions
About this article
Cite this article
Venkataramani, S.C., Venkataramani, R.C. & Restrepo, J.M. Dimension Reduction for Systems with Slow Relaxation. J Stat Phys 167, 892–933 (2017). https://doi.org/10.1007/s10955-017-1761-7
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10955-017-1761-7