A Christov Integrating Factor Method for Approximating Rational PDE Solutions

arXiv:2611.07154 · nlin.PS, cs.NA, math.NA · Submitted 2026-11-04 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: Nonlinear Sciences. The day in nonlinear science: chaos, patterns, networks and adaptation.

Kai: We're looking at the paper: "A Christov Integrating Factor Method for Approximating Rational PDE Solutions".

Mira: Here's the gist of the paper. Rational solutions of partial differential equations (PDEs) are notoriously difficult to approximate via spectral Fourier methods due to their algebraically slow decay rate This work discusses approximating rational PDE solutions in a basis of orthogonal functions known as the Christov functions, allowing for the computation of its spectrum via the fast Fourier transform An explicit fourth-order integrating factor method is derived Basis and Properties of Christov Functions The Christov functions are a family of rational functions that are orthogonal on the real line These functions are defined as ϕn(x) = r(2πi/n) / (1 + 2ix / 1 - 2ix n / 1 - 2ix n) for n ∈ Z Key properties include: Poles: The nth Ch function ϕn(x) has a degree n + 1 pole at x = −i/2 when n is positive, and a degree n pole at x = i/2 and a simple pole at x = −i/2 when n is nonpositive.

Kai: Oscillations: The Ch modes increase in frequency near the origin as n → ∞.

Mira: Orthogonality: The Ch functions are mutually orthonormal, meaning ⟨ϕm, ϕn⟩ = δmn.

Kai: Derivative Recurrence Relation: The Ch functions satisfy the skew-symmetric recurrence relation d/dxϕn(x) = −nϕn−1(x) + i(2n + 1)ϕn(x) + (n + 1)ϕn+1(x).

Title and authors: Kai: So what this means is that for rational functions that are square integrable, they develop a method that is spectrally accurate in space and fourth-order accurate in time, which is pretty strong for a numerical scheme <ref:2611.07154#pg2>.

Mira: They also mention they can compute the coefficients using modified fast Fourier transforms when you use a coordinate transformation, specifically x equals one over two times the tangent of theta squared <ref:2611.07154#pg2>.

Lev: That coordinate transformation is what lets them rapidly compute the modal coefficients, and they say that for benchmark simulations, they consistently achieve seven to eight decimal places of accuracy in only a few seconds on a standard personal laptop <ref:2611.07154#pg2>.

Kai: Seven to eight decimal places in just a few seconds is quite fast for these kinds of calculations, even with the overhead of using Christov functions instead of simpler methods <ref:2611.07154#pg2>.

Mira: They also mention that they improved on a previous work by introducing a split-step Christov method, but this new one is more versatile for approximating rational partial differential equations <ref:2611.07154#pg2>.

Lev: The paper also points out that the convergence rate is linked to the width of an analytic strip in the complex plane of a Fourier-like problem, which tells you how quickly those coefficients fall off <ref:2611.07154#pg2>.

The paper's summary: Kai: The authors suggest several improvements to this Christov Integrating Factor Method, like using an adaptive basis selection for the Christov functions improvement one <ref:2611.07154#pg1>.

Mira: That makes sense. If you can dynamically adjust the number of Christov modes based on how much round-off error you get in individual coefficients, that should help keep things accurate without wasting computation improvement one <ref:2611.07154#pg1>.

Lev: From an engineering standpoint, I think that adaptive selection would be a big win for real hardware where you have finite precision and limited memory improvement one <ref:2611.07154#pg1>.

Kai: And then there's the explicit fourth-order Runge Kutta integrating factor method implementation itself, which they really stress about because it leads to a larger region of stability than direct pseudo spectral methods improvement two <ref:2611.07154#pg3>.

Mira: That's interesting because direct methods often have very strict time step restrictions that limit how long you can simulate something, but the IF method seems to bypass those constraints improvement two <ref:2611.07154#pg3>.

Lev: If the integrating factor method gives you a larger region of stability, it means you could potentially use much larger time steps, which would significantly reduce the overall CPU runtimes for simulations improvement five.

Kai: And they also mentioned using pre-computed banded matrices like D1, D2, and D3 for derivatives and Hilbert transforms to compute spectral operators more efficiently using fast modified Fourier transforms improvement three <ref:2611.07154#pg3>.

Mira: That sounds like a practical step. If you can use pre-computed structures for those operators, it streamlines the computation of the spectral differentiation matrices improvement three <ref:2611.07154#pg3>.

The paper's improvements: Kai: So to wrap up, this paper on "A Christov Integrating Factor Method for Approximating Rational PDE Solutions" shows a way to approximate rational solutions using Christov functions that are spectrally accurate in space and fourth-order accurate in time <ref:2611.07154#pg2>.

Mira: The main implication is that this integrating factor method offers a more efficient scheme compared to straightforward pseudo spectral approximations because it supports larger regions of stability <ref:2611.07154#pg2>.

Lev: It seems like the convergence rate is determined by how far the nearest pole in the complex plane is from the real axis, which gives you a way to predict how many Christov modes you actually need for a target accuracy improvement six.

Kai: And they show that for problems like nonlinear Schrödinger equations, they can achieve accuracy of ten to six decimal places in just a few seconds on a standard personal laptop <ref:2611.07154#pg2>.

Mira: It's exciting because it applies this to physical equations like the nonlinear Schrödinger equation, modified Korteweg-de Vries, and Benjamin-Ono equations <ref:2611.07154#pg3>.

Lev: For things like momentum and Hamiltonian—the conserved quantities—they manage to approximate them by constructing those integrands as products of Christov expansions, and the error in those quantities stays bounded over time intervals from zero to ten <ref:2611.07154#pg3>.

Kai: It’s a lot of practical stuff here—from the method's stability region to how fast they can compute results on real hardware improvement five.

Mira: And it lays out the idea that applying an integrating-factor approach with Christov functions is a novel way to handle these rational solutions in nonlinear systems <ref:2611.07154#pg3>.

Lev: Overall, this work develops a method that is comparable to split step methods but more accurate than Yoshida, though not quite as accurate as PRK or RKN Appendix Details.

Kai: It's a solid development for anyone trying to simulate these complex nonlinear systems, and we're looking forward to seeing how they push these ideas further <ref:2611.07154#pg3>.

Conclusion: Kai: So we've been talking about this paper, "A Christov Integrating Factor Method for Approximating Rational PDE Solutions," which uses Christov functions to tackle rational partial differential equations and achieves high accuracy in both space and time.

Mira: It really hinges on using these orthogonal functions, the Christov functions, to build a basis that decays exponentially fast when you transform the spatial coordinates <ref:2611.07154#pg2>.

Lev: And the real payoff is their fourth-order Runge Kutta integrating factor method for time evolution; they show it has a larger region of stability than direct pseudo spectral methods <improvement five>.

Kai: That means that for something like the nonlinear Schrödinger equation, they got accuracy of ten to six decimal places in just a few seconds on a laptop <ref:2611.07154#pg2>.

Mira: The assumptions underpinning this are that the functions we're solving are analytic and bounded rational functions, which means we can predict the convergence rate based on how far a pole is from the real axis in our transformed coordinate system <ref:2611.07154#pg2>.

Lev: And they also showed that they can compute those spectral derivatives and Hilbert transform using pre-computed banded matrices, which makes the actual computation of those operators much faster <improvement three>.

Kai: It’s interesting how they manage to approximate conserved quantities like power and momentum just by setting up the right products of Christov expansions <ref:2611.07154#pg3>.

Mira: I think what this means for the world is that it gives us a tool specifically tailored for solving these rational equations in physical systems, especially when you're dealing with nonlinear waves like rogue waves <ref:2611.07154#pg3>.

Lev: For real hardware, it’s important that the overhead of this method lands predictably so we can actually use it to run simulations instead of just sitting there <improvement five>.

Kai: So, to wrap up, "A Christov Integrating Factor Method for Approximating Rational PDE Solutions" gives us a way to get high accuracy for rational equations using Christov functions in space and a fourth-order method in time.

Mira: It's a clever way to handle these problems by linking the spectral properties of the functions directly to the time integration scheme <ref:2611.07154#pg3>.

Lev: It’s comparable to other methods like split step but more accurate than Yoshida, which is a solid result for explicit integrators <Appendix Details>.

Kai: So, it shows that even though the basis functions are complicated rational functions, we can get good results if we use the right orthogonal system and a smart time-stepping approach.

Mira: It’s definitely a useful piece of machinery for those who need to solve these specific types of nonlinear equations <ref:2611.07154#pg3>.

Lev: Next up, we're looking at how higher-order interactions enhance synchronization in coupled oscillators, the case of the Kuramoto model.

TROY I. JOHNSON, JUSTIN T. COLE

University of Colorado Colorado Springs

nlin.PS, cs.NA, math.NA

Submitted: 2026-11-04

Updated: 2026-11-04

License: http://creativecommons.org/licenses/by/4.0/

Importance score: 83/100

The gist: Rational solutions of partial differential equations (PDEs) are notoriously difficult to approximate via spectral Fourier methods due to their algebraically slow decay rate This work discusses…

Key concepts

Christov Functions
These are a specific family of orthogonal rational functions defined on the real line. They form a basis set that is particularly well-suited for approximating rational solutions to PDEs, offering advantages over standard Fourier methods due to their specific algebraic properties.
Integrating Factor Method (IF)
This is a novel time-stepping technique used to evolve the PDE. It transforms the equation into an integrating factor form, which provides a larger region of stability for explicit solvers and allows for high accuracy in time evolution when combined with spectral approximations.
Spectral Differentiation Matrices
Derivatives and Hilbert transforms are approximated using matrices constructed from Christov functions. These matrices allow for efficient computation of spectral operators by utilizing fast modified Fourier transforms, speeding up the calculation of higher-order derivatives.
Convergence Rate Prediction
The exponential decay rate of the Christov function coefficients is determined by the distance from a function's pole to the real axis in a transformed coordinate. This allows researchers to predict how many basis functions are needed for a desired level of accuracy.

Terminology

Summary

Rational solutions of partial differential equations (PDEs) are notoriously difficult to approximate via spectral Fourier methods due to their algebraically slow decay rate This work discusses approximating rational PDE solutions in a basis of orthogonal functions known as the Christov functions, allowing for the computation of its spectrum via the fast Fourier transform An explicit fourth-order integrating factor method is derived

Basis and Properties of Christov Functions

The Christov functions are a family of rational functions that are orthogonal on the real line These functions are defined as ϕn(x) = r(2πi/n) / (1 + 2ix / 1 - 2ix n / 1 - 2ix n) for n ∈ Z Key properties include:

** Poles:**

The nth Ch function ϕn(x) has a degree n + 1 pole at x = −i/2 when n is positive, and a degree n pole at x = i/2 and a simple pole at x = −i/2 when n is nonpositive.

** Oscillations:**

The Ch modes increase in frequency near the origin as n → ∞.

** Orthogonality:**

The Ch functions are mutually orthonormal, meaning ⟨ϕm, ϕn⟩ = δmn.

** Derivative Recurrence Relation:**

The Ch functions satisfy the skew-symmetric recurrence relation d/dxϕn(x) = −nϕn−1(x) + i(2n + 1)ϕn(x) + (n + 1)ϕn+1(x).

** Hilbert Transform Eigenfunction:**

The Ch functions are eigenfunctions of the Hilbert transform, satisfying Hϕn = i sgn(n)ϕn(x).

Transformation to Fourier Series and Convergence Rate

A transformation relating the Ch expansion to the Fourier series of a periodic function is established using a coordinate transformation x = 1/2 tan θ squared. The Ch coefficients are computed via modified Fourier transforms, and for analytic and bounded rational functions, these coefficients decay exponentially fast in n, specifically ˇfn ≈ O(e −nκ). p. 4

Where the paper says it

Consider complexifying θ within some analytic strip |Re θ| ≤ π and |Im θ| < κ, such that the function f 1 2 tan θ 2 1 − itan θ 2 is uniformly bounded within the strip. Then the Ch coefficients decay like | ˇfn| ≈ O(e −|nκ| ), that is, exponentially fast [11].

The convergence rate is determined by the distance from the nearest pole to the real axis in θ, denoted as κ. p. 4
Where the paper says it

Hence, the value of κ in Eq. 1) that corresponds to the pole of f(z) closest to the real axis fixes the asymptotic convergence rate, | ˇfn| ∼ e −|nκ| .

For a function with a pole at z = ξ + iη, this distance is given by κ = Im 2 tan−1(2(ξ + iη)). p. 4
Where the paper says it

8) implies a pole at θ = 2 tan−1 (2(ξ + iη)). We then note that the complex form of the inverse tangent is given by, tan−1 (z) = − i 2 log 1+iz 1−iz , and the complex logarithm is, log(z) = log |z| + i arg(z), and so the distance from this pole in θ to the real axis is given by (3.

Spectral Differentiation and Hilbert Transform Matrices

Derivatives and the Hilbert transform can be approximated using matrices constructed from Ch functions.

** First Derivative:**

The first derivative coefficient is related to the original function coefficients by ˇf(1)n = n ˇfn−1 + i(2n + 1) ˇfn − (n + 1) ˇfn+1.

** Second and Third Derivatives:**

Higher-order derivatives can be obtained recursively, and the second derivative coefficients are related to the function coefficients via a complex recurrence relation. The third derivative coefficients are given by a relation involving seven nearest Ch modes. p. 6

Where the paper says it

A Ch derivative coefficient involves itself {n} and nearest two {n − 1, n + 1} modes. Alternatively, the derivative coefficients can be computed via matrix multiplication (4.

Integrating Factor Method for Time Evolution

A novel time-stepping approach is implemented using an explicit fourth-order Runge-Kutta integrator and the differentiation and Hilbert transform matrices. The method involves transforming the equation into an integrating factor form, leading to a system of differential equations vˇt = e −LtMFhN [w + U] − N[U]i. p. 9

Where the paper says it

Taking the integrating-factor (IF) transformation wˇ = e Ltvˇ removes the Lwˇ term in Eq. 3) and yields the system of differential equations (5.

This IF method supports larger regions of stability compared to direct pseudo-spectral methods. p. 0
Where the paper says it

Relative to a straightforward pseudo-spectral approximation, the integrating factor method supports larger regions of stability resulting in a more efficient scheme. Christov functions, spectral methods, partial differential equations, integrating factor, rational functions AMS subject classifications.

Application to Nonlinear PDEs

The Ch-IF method is tested on several physically important nonlinear PDEs, including the nonlinear Schrödinger (NLS), modified Korteweg-de Vries (mKdV), and Benjamin-Ono (BO) equations. For the NLS equation, the method achieves accuracy of O(10−6) or better in a few seconds on a standard personal laptop. p. 1

Where the paper says it

For benchmark simulations considered in this work, we are consistently able to achieve seven to eight decimal places of accuracy in only a few seconds on a standard personal laptop. In a previous work, we introduced a split-step Christov method [12].

For the mKdV and BO equations, fourth-order convergence is observed as time-step size decreases. p. 1
Where the paper says it

For rational functions that are square-integrable, we develop a method that is spectrally accurate in space and fourth-order accurate in time; we integrate the IF equations with fourth-order RungeKutta method. The exponential convergence rate is related to the width of an analytic strip in the complex plane of a Fourier-like problem [11].

Conserved Quantities and Stability

The method accurately approximates conserved quantities such as power, momentum, and Hamiltonian by constructing integrands as products of Ch expansions. For localized solutions of the NLS equation, the error in these quantities remains bounded over time intervals [0, 10]. p. 1

Where the paper says it

These equations are relevant in the field of nonlinear waves and support soliton-like rational solutions. One notable example is the Peregrine soliton solution of the NLS equation (which we approximate below), a wellknown rogue wave model for deep water waves [13, 14] and fiber optics [15].

The method's effectiveness is highlighted by its enhanced stability properties for explicit integrators.

Conclusion and Limitations

The integrating-factor time-stepping method for evolving rational solutions of nonlinear partial differential equations of physical interest was developed The novelty lies in the application of an integrating-factor approach with Christov functions, a family of orthogonal rational functions defined on the real line. The biggest limitation is the clustering of spatial points near the origin, which necessitates the concentration of important dynamics near the origin. p. 1

Where the paper says it

thermore, via a coordinate transformation, it is possible to rapidly compute the modal coefficients via a modified fast Fourier transform. The main limitation of this basis is that it clusters grid points near the origin.

Declarations

The authors declare that they have no conflict of interest. No funding was received or utilized for the research or authorship of this paper. Both Troy I. p. 0

Where the paper says it

A CHRISTOV INTEGRATING FACTOR METHOD FOR APPROXIMATING RATIONAL PDE SOLUTIONS CORRESPONDING AUTHOR: TROY I. JOHNSON ∗ AND JUSTIN T.

Johnson and Justin T. p. 0
Where the paper says it

A CHRISTOV INTEGRATING FACTOR METHOD FOR APPROXIMATING RATIONAL PDE SOLUTIONS CORRESPONDING AUTHOR: TROY I. JOHNSON ∗ AND JUSTIN T.

p. 0
Where the paper says it

A CHRISTOV INTEGRATING FACTOR METHOD FOR APPROXIMATING RATIONAL PDE SOLUTIONS CORRESPONDING AUTHOR: TROY I. JOHNSON ∗ AND JUSTIN T.

Cole contributed to the conceptualization of this manuscript The figures were made by Troy I. p. 0
Where the paper says it

A CHRISTOV INTEGRATING FACTOR METHOD FOR APPROXIMATING RATIONAL PDE SOLUTIONS CORRESPONDING AUTHOR: TROY I. JOHNSON ∗ AND JUSTIN T.

Johnson Both Troy I. p. 0
Where the paper says it

A CHRISTOV INTEGRATING FACTOR METHOD FOR APPROXIMATING RATIONAL PDE SOLUTIONS CORRESPONDING AUTHOR: TROY I. JOHNSON ∗ AND JUSTIN T.

Johnson and Justin T. Cole worked on every draft of the manuscript and read and approved the final manuscript. p. 0
Where the paper says it

JOHNSON ∗ AND JUSTIN T. COLE† Abstract.

The authors acknowledge motivating and insightful discussions with Denis Silantyev.

Appendix Details

The decay rate of Ch coefficients is influenced by the multiplicity of a pole, as seen in the case of repeated roots. The Christov integrating-factor method uses a spectral Ch approximation in space and a fourth-order Runge-Kutta IF method in time. p. 8

Where the paper says it

An explicit fourth-order Runge-Kutta integrator is used to advance the time step. The effectiveness of the method is highlighted in the next section for several physically important nonlinear PDEs.

p. 8
Where the paper says it

An explicit fourth-order Runge-Kutta integrator is used to advance the time step. The effectiveness of the method is highlighted in the next section for several physically important nonlinear PDEs.

The pseudo-code for the one Ch-IF time-step is provided. p. 9
Where the paper says it

5 seconds, whereas using similar parameters in a pseudo-spectral 4th order Runge-Kutta method took over 2 minutes. Pseudo-code is given in Appendix C.

The comparison to split-step methods shows that the Ch-IF method is comparable to them, more accurate than Yoshida, but not quite as accurate as PRK and RKN. p. 1
Where the paper says it

The advantage of the IF method is that it is easier to accommodate generic nonlinearities relative to the splitstep method. An advantage that both the split-step and integrating-factor methods have over a direct pseudo-spectral method is a significantly larger region of stability in the selection of timestep.

p. 1
Where the paper says it

The advantage of the IF method is that it is easier to accommodate generic nonlinearities relative to the splitstep method. An advantage that both the split-step and integrating-factor methods have over a direct pseudo-spectral method is a significantly larger region of stability in the selection of timestep.

References

[1] B. Fornberg, A practical guide to pseudospectral methods, Cambridge University Press (1996)

[2] C. Canuto, M. Y. Hussaini, A. Quarteroni, and T. A. Zang, Spectral Methods: Fundamentals in Single Domains, Springer-Verlag, Berlin (2006)

[3] J. Yang, Nonlinear Waves in Integrable and Nonintegrable Systems, SIAM, Philadelphia (2010)

[4] C. E. Grosch and S. A. Orszag, Numerical solution of problems in unbounded regions: coordinate transforms, J. Comp. Phys. 25 pp. 273-296 (1977)

[5] J. P. Boyd, Spectral methods using rational basis functions on an infinite interval, J. p. 0

Where the paper says it

Rational solutions of partial differential equations (PDEs) are notoriously difficult to approximate via spectral Fourier methods due to their algebraically slow decay rate. In this work we discuss approximating rational PDE solutions in a basis of orthogonal functions known as the Christov functions, allowing for the computation of its spectrum via the fast Fourier transform.

Comp. Phys. 69, pp. 112-142 (1987)

[6] J. P. Boyd, The rate of convergence of Hermite function series, Math. p. 0

Where the paper says it

Intuitively, this approach makes sense because the Hermite functions form an orthogonal basis on the real line. However, the truncated series approximation only converges spectrally fast if the function decays exponentially fast at ±∞ [2].

of Comp. 35 pp. 1309- 1316 (1980)

[7] C. Christov, A complete orthonormal system of functions in l2(−∞,∞) space, SIAM J. p. 2

Where the paper says it

2) ⟨f, g⟩ = Z ∞ −∞ f ∗ (x)g(x)dx, where ∗ denotes complex conjugation. The Ch functions are mutually orthonormal [9], that is (2.

Appl. Math. 42 pp. 1337-1344 (1982)

[8] J. P. Boyd, The orthogonal rational functions of Higgins and Christov and algebraically mapped Chebyshev polynomials, J. p. 0

Where the paper says it

Rational solutions of partial differential equations (PDEs) are notoriously difficult to approximate via spectral Fourier methods due to their algebraically slow decay rate. In this work we discuss approximating rational PDE solutions in a basis of orthogonal functions known as the Christov functions, allowing for the computation of its spectrum via the fast Fourier transform.

Approx. Theory 61 pp. 98-105 (1990)

[9] A. Iserles and M. Webb, A family of orthogonal rational functions and other orthogonal systems with a skew-Hermitian differentiation matrix, J. p. 0

Where the paper says it

An alternative approach we look to continue developing in this work is the application of the Christov (Ch) functions [7–9] as an approximation basis. The Ch functions are a family of rational functions that are orthogonal on the real line.

Fourier Anal. p. 0
Where the paper says it

COLE† Abstract. Rational solutions of partial differential equations (PDEs) are notoriously difficult to approximate via spectral Fourier methods due to their algebraically slow decay rate.

and Appl. 26, 19 (2020)

[10] J. P. Boyd, Chebyshev and Fourier Spectral methods 2nd ed. p. 0

Where the paper says it

COLE† Abstract. Rational solutions of partial differential equations (PDEs) are notoriously difficult to approximate via spectral Fourier methods due to their algebraically slow decay rate.

, Dover, New York, (2001)

[11] R. E. A. C. Paley and N. Wiener, Fourier transforms in the complex domain, Amer. p. 0

Where the paper says it

For sufficiently smooth solutions, this approach yields exponentially decaying Fourier modes which can be rapidly computed via fast Fourier transform [1,2]. The key assumption here is that the function reaches its boundary conditions within the computational domain.

Math. Soc. Colloq. Publ., 19 Providence, (1934)

[12] J. T. Cole and T. p. 0

Where the paper says it

JOHNSON ∗ AND JUSTIN T. COLE† Abstract.

p. 0
Where the paper says it

JOHNSON ∗ AND JUSTIN T. COLE† Abstract.

I. Johnson, A split-step Christov method for approximating rational PDE solutions, J. p. 0
Where the paper says it

A CHRISTOV INTEGRATING FACTOR METHOD FOR APPROXIMATING RATIONAL PDE SOLUTIONS CORRESPONDING AUTHOR: TROY I. JOHNSON ∗ AND JUSTIN T.

p. 0
Where the paper says it

A CHRISTOV INTEGRATING FACTOR METHOD FOR APPROXIMATING RATIONAL PDE SOLUTIONS CORRESPONDING AUTHOR: TROY I. JOHNSON ∗ AND JUSTIN T.

Comp. Phys. (2025)

[13] C. Kharif and E. Pelinovsky, Physical mechanisms of the rogue wave phenomenon, European J. p. 1

Where the paper says it

These equations are relevant in the field of nonlinear waves and support soliton-like rational solutions. One notable example is the Peregrine soliton solution of the NLS equation (which we approximate below), a wellknown rogue wave model for deep water waves [13, 14] and fiber optics [15].

Mechanics-B/Fluids 22 pp. 603-634 (2003)

[14] A. Chabchoub, N. P Hoffmann, and N. Akhmediev, Rogue wave observation in a water wave tank, Phys. p. 1

Where the paper says it

These equations are relevant in the field of nonlinear waves and support soliton-like rational solutions. One notable example is the Peregrine soliton solution of the NLS equation (which we approximate below), a wellknown rogue wave model for deep water waves [13, 14] and fiber optics [15].

Rev. Lett. 106 204502 (2011)

[15] G. Agrawal, Nonlinear fiber optics, Academic Press, Elsevier (2013)

[16] J. R. Higgins, Completeness and Basis Properties of Sets of Special Functions, Cambridge University Press (1977)

[17] J. A. C. Weideman, Computing the Hilbert transform on the real line, Math. p. 4

Where the paper says it

12), is included because Eq. 11) is computing the Fourier transform of the function f 1 2 tan θ 2 1 − itan θ 2 = f (x) (1 − 2ix), so one recovers the function f(x) by dividing out the term, 1 − itan θ 2 .

Comp. 64 pp. 745-762 (1995)

[18] M. J. Ablowitz, Nonlinear dispersive waves, Cambridge Press (2011)

[19] A. Calini, C. M. Schober, and M. Strawn Linear instability of the Peregrine breather: Numerical and analytical investigations, Appl. p. 1

Where the paper says it

These equations are relevant in the field of nonlinear waves and support soliton-like rational solutions. One notable example is the Peregrine soliton solution of the NLS equation (which we approximate below), a wellknown rogue wave model for deep water waves [13, 14] and fiber optics [15].

Num. Math. 141 pp. 36-43 (2019)

[20] M. J. Ablowitz and J. T. Cole, Transverse instability of rogue waves, Phys. p. 1

Where the paper says it

These equations are relevant in the field of nonlinear waves and support soliton-like rational solutions. One notable example is the Peregrine soliton solution of the NLS equation (which we approximate below), a wellknown rogue wave model for deep water waves [13, 14] and fiber optics [15].

Rev. Lett. 127 104101 (2021)

[21] Y. Ohta and J. Yang, General high-order rogue waves and their dynamics in the nonlinear Schrödinger equation, Proc. p. 1

Where the paper says it

These equations are relevant in the field of nonlinear waves and support soliton-like rational solutions. One notable example is the Peregrine soliton solution of the NLS equation (which we approximate below), a wellknown rogue wave model for deep water waves [13, 14] and fiber optics [15].

Royal Soc. A 468 pp. 1716-1740 (2012)

[22] G. Biondini and D. Mantzavinos, Universal nature of the nonlinear stage of modulational instability, Phys. p. 0

Where the paper says it

An explicit fourth-order integrating factor method is derived. The effectiveness of the method is tested on several nonlinear PDEs.

Rev. Lett. 116 043902 (2016)

[23] H. C. Yuen and W. E. Ferguson, Relationship between Benjamin-Feir instability and recurrence in the nonlinear Schrödinger equation, Phys. p. 1

Where the paper says it

An advantage that both the split-step and integrating-factor methods have over a direct pseudo-spectral method is a significantly larger region of stability in the selection of timestep. To highlight the versatility of this method, we approximate rational solutions of the nonlinear Schr¨odinger (NLS), modified Korteweg-de Vries (mKdV), and Benjamin-Ono (BO) equations.

Fluids 21 pp. 1275-1278 (1978)

[24] S. Baqer, T.P. Horikis and D.J. Frantzeskakis, Physical vs mathematical origin of the extended KdV and mKdV equations, AIMS Mathematics 10 pp. p. 1

Where the paper says it

An advantage that both the split-step and integrating-factor methods have over a direct pseudo-spectral method is a significantly larger region of stability in the selection of timestep. To highlight the versatility of this method, we approximate rational solutions of the nonlinear Schr¨odinger (NLS), modified Korteweg-de Vries (mKdV), and Benjamin-Ono (BO) equations.

9295-9309 (2025)

[25] R.M. Miura, C.S. Gardner and M.D. Kruskal, Kortweg-de Vries equation and generalizations, II: existence of conservation laws and constants of motion, Journal of Mathematical Physics 9 pp. p. 9

Where the paper says it

Below we test the performance of this method on several nonlinear dispersive PDEs with well-known rational solutions. When we do not have a known solution we monitor the accuracy of the method by conservation laws (see below) and localization of the Ch coefficients.

1204-1209 (1968)

[26] M.A. Helal, Soliton solution of some nonlinear partial differential equations and its applications in fluid mechanics, Chaos, Solitons & Fractals 13 pp. p. 1

Where the paper says it

These equations are relevant in the field of nonlinear waves and support soliton-like rational solutions. One notable example is the Peregrine soliton solution of the NLS equation (which we approximate below), a wellknown rogue wave model for deep water waves [13, 14] and fiber optics [15].

1917-1929 (2002)

[27] F. Verheest, C.P. Oliver and W.A. Hereman, Modified kortweg-de vries solitons at supercritical densities in two-electron temperature plasmas, Journal of Plasma Physics 82 (2016)

[28] F. Verheest and W.A. Hereman, Collissions of acoustic solitons and their electric fields in plasmas at critical compositions, Journal of Plasma Physics 85 (2019)

[29] H. Leblond and D. Mihalache, Few-optical-cycle soltions: modified Kortweg-de Vries sineGordon equation versus other non-slowly-varying envelope-approximation models, Physical Review A 79 (2009)

[30] H. Leblond and D. Mihalache, Optical solitons in the fewcycle regime: recent theoretical results, Romanian Reports in Physics 63 pp. p. 1

Where the paper says it

We conclude in Section 7. Several appendices with extra technical notes, figures and results are also included.

1254-1266 (2011)

[31] M. J. Ablowitz and H. Segur, Solitons and the inverse scattering transform, SIAM (1981)

[32] A. Ankiewicz and N. Akhmediev, Rogue wave-type solutions of the mKdV equation and their relation to known NLSE rogue wave solutions, Nonlinear Dyn 91 pp. p. 1

Where the paper says it

These equations are relevant in the field of nonlinear waves and support soliton-like rational solutions. One notable example is the Peregrine soliton solution of the NLS equation (which we approximate below), a wellknown rogue wave model for deep water waves [13, 14] and fiber optics [15].

1931-1938 (2018)

[33] B. T. Benjamin and J. p. 1

Where the paper says it

An advantage that both the split-step and integrating-factor methods have over a direct pseudo-spectral method is a significantly larger region of stability in the selection of timestep. To highlight the versatility of this method, we approximate rational solutions of the nonlinear Schr¨odinger (NLS), modified Korteweg-de Vries (mKdV), and Benjamin-Ono (BO) equations.

E. Feir, The disintegration of wave trains on deep water, Part 1. p. 1
Where the paper says it

These equations are relevant in the field of nonlinear waves and support soliton-like rational solutions. One notable example is the Peregrine soliton solution of the NLS equation (which we approximate below), a wellknown rogue wave model for deep water waves [13, 14] and fiber optics [15].

Theory, J. Fluid Mech. 27 pp. 417-430 (1967)

[34] H. Ono, Algebraic solitary waves in stratified fluids, J. p. 1

Where the paper says it

To highlight the versatility of this method, we approximate rational solutions of the nonlinear Schr¨odinger (NLS), modified Korteweg-de Vries (mKdV), and Benjamin-Ono (BO) equations. These equations are relevant in the field of nonlinear waves and support soliton-like rational solutions.

Phys. Soc. Japan 39 pp. 1082–1091 (1975)

[35] R. L. James and J. A. C. Weideman, Pseudospectral Methods for the Benjamin-Ono Equation, Advances in Computer Methods for Partial Differential Equations 7 (1992)

[36] J. P. Boyd and Z. Xu, Comparison of three spectral methods for the Benjamin-Ono equation: Fourier psuedospectral, rational Christov functions and Gaussian radial basis functions, Wave Motion 48 pp. p. 0

Where the paper says it

Rational solutions of partial differential equations (PDEs) are notoriously difficult to approximate via spectral Fourier methods due to their algebraically slow decay rate. In this work we discuss approximating rational PDE solutions in a basis of orthogonal functions known as the Christov functions, allowing for the computation of its spectrum via the fast Fourier transform.

702-706 (2011)

[37] J. T. Cole and T. I. Johnson, A split-step Christov method for approximating rational PDE solutions, J. Comp. Phys. (2025)

The decay rate of Ch coefficients is influenced by the multiplicity of a pole. The Christov integrating-factor method uses a spectral Ch approximation in space and a fourth-order Runge-Kutta IF method in time. The comparison to split-step methods shows that the Ch-IF method is comparable to them, more accurate than Yoshida, but not quite as accurate as PRK and RKN.

Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.

Improvements for AI systems

  1. Adaptive basis selection for Christov functions: The system can dynamically adjust the number of Ch modes based on a threshold for round-off error in individual coefficients to maintain accuracy while minimizing computational cost.

  2. Fourth-order Runge-Kutta integrating factor method implementation: The core solver can utilize an explicit fourth-order Runge-Kutta method integrated with the Christov integrating factor equations to achieve high accuracy in time, as shown by the O(10-6) convergence for NLS solutions.

  3. Spectral differentiation and Hilbert transform matrix construction: The system should employ pre-computed banded matrices like D1, D2, and D3 for derivatives and Hilbert transforms to efficiently compute spectral operators using fast modified Fourier transforms.

  4. Nonlinear operator decomposition: The system can break down the nonlinear operator into real space applications NR[u] and spectral space applications NCh to handle different types of nonlinearities effectively within the IF framework.

  5. Stability region assessment: The system can utilize the larger region of stability inherent in the integrating-factor method compared to direct pseudo-spectral methods for time step selection, potentially leading to significantly lower CPU runtimes.

  6. Asymptotic convergence rate prediction: By analyzing the pole location in the complex plane using Eq (3.1), the system can predict the exponential decay rate of Ch coefficients and thus estimate the required number of modes N for a target accuracy level based on function location 'a'.

Abstract

Rational solutions of partial differential equations (PDEs) are notoriously difficult to approximate via spectral Fourier methods due to their algebraically slow decay rate. In this work we discuss approximating rational PDE solutions in a basis of orthogonal functions known as the Christov functions, allowing for the computation of its spectrum via the fast Fourier transform. An explicit fourth-order integrating factor method is derived. The effectiveness of the method is tested on several nonlinear PDEs. Relative to a straightforward pseudo-spectral approximation, the integrating factor method supports larger regions of stability resulting in a more efficient scheme.

Related papers