Whittle Likelihood Methods

• Whittle likelihood is a frequency-domain quasi-likelihood that approximates the Gaussian likelihood by replacing heavy covariance operations with independent spectral evaluations.
• It enables efficient inference for stationary time series and random fields by reducing computational complexity from O(n³) to O(n log n).
• Extensions such as debiasing and modulation adjust for edge effects and missing data, improving parameter estimation in mixed and spatial models.

The Whittle likelihood is a frequency-domain quasi-likelihood that serves as an efficient substitute for the exact Gaussian likelihood in the analysis of second-order stationary time series and random fields. Its fundamental innovation is the replacement of computationally intensive time-domain likelihoods, which require operations on covariance matrices, with a product of independent spectral likelihoods at the discrete Fourier frequencies. The approach is valid under asymptotic regimes and for processes (not necessarily Gaussian) that satisfy fourth-order stationarity. The Whittle likelihood and its refinements are central to a range of modern inference methods for time series, spatial data, and mixed models, especially where computational scalability and robustness to non-Gaussianity or missing data are required (Pypkowski et al., 23 Dec 2025, Guillaumin et al., 2019, Vu et al., 2024).

1. Mathematical Foundations and Derivation

Let $\{X_t\}_{t=1}^n$ denote a zero-mean, second-order stationary time series with autocovariance $c(\tau; \alpha)$ and spectral density

$$S(\omega; \alpha) = \sum_{\tau = -\infty}^\infty c(\tau; \alpha) e^{-i \omega \tau}.$$

The periodogram at discrete frequencies $\omega_j = 2\pi j / n$ ($$j = -\lfloor n/2 \rfloor, \ldots, \lfloor (n-1)/2 \rfloor$$) is defined as

$$I(\omega_j) = \frac{1}{n}\biggl|\sum_{t=1}^n X_t e^{-i \omega_j t} \biggr|^2.$$

For Gaussian $X$, the exact time-domain log-likelihood is

$$l(\alpha) = -\frac{1}{2} \log|C(\alpha)| - \frac{1}{2} X^T C(\alpha)^{-1} X,$$

where $C(\alpha)$ is the $n \times n$ covariance matrix. Whittle (1953) showed that, under mild regularity, this can be approximated (up to an additive constant) by the frequency-domain Whittle log-likelihood:

$c(\tau; \alpha)$0

The “Whittle quasi-likelihood” is therefore

$c(\tau; \alpha)$1

The derivation leverages the asymptotic diagonalization of $c(\tau; \alpha)$2 by the discrete Fourier transform: $c(\tau; \alpha)$3 and $c(\tau; \alpha)$4.

The estimation extends beyond Gaussian processes: for any fourth-order stationary process with finite moments, maximizing $c(\tau; \alpha)$5 yields estimators that are consistent and asymptotically normal (Pypkowski et al., 23 Dec 2025).

2. Extensions, Refinements, and Bias Correction

Mixed Models and Missing Data

For mixed models $c(\tau; \alpha)$6, with fixed mean $c(\tau; \alpha)$7 and stationary error $c(\tau; \alpha)$8, the joint Whittle objective for parameters $c(\tau; \alpha)$9 is

$$S(\omega; \alpha) = \sum_{\tau = -\infty}^\infty c(\tau; \alpha) e^{-i \omega \tau}.$$0

where $$S(\omega; \alpha) = \sum_{\tau = -\infty}^\infty c(\tau; \alpha) e^{-i \omega \tau}.$$1 is the periodogram of the residual process $$S(\omega; \alpha) = \sum_{\tau = -\infty}^\infty c(\tau; \alpha) e^{-i \omega \tau}.$$2. Both fixed and random-effect parameters are estimated jointly, without closed-form expressions for the regression parameters.

Debiasing and Modulation

Finite-sample bias, due to spectral blurring and boundary effects, is substantial, especially in spatial or short time series. Bias correction is achieved by substituting $$S(\omega; \alpha) = \sum_{\tau = -\infty}^\infty c(\tau; \alpha) e^{-i \omega \tau}.$$3 by the expected periodogram:

$$S(\omega; \alpha) = \sum_{\tau = -\infty}^\infty c(\tau; \alpha) e^{-i \omega \tau}.$$4

For missing data, define $$S(\omega; \alpha) = \sum_{\tau = -\infty}^\infty c(\tau; \alpha) e^{-i \omega \tau}.$$5 as a binary mask and consider modulated residuals $$S(\omega; \alpha) = \sum_{\tau = -\infty}^\infty c(\tau; \alpha) e^{-i \omega \tau}.$$6, with adjusted autocovariance

$$S(\omega; \alpha) = \sum_{\tau = -\infty}^\infty c(\tau; \alpha) e^{-i \omega \tau}.$$7

The corresponding debiased and modulated likelihood becomes

$$S(\omega; \alpha) = \sum_{\tau = -\infty}^\infty c(\tau; \alpha) e^{-i \omega \tau}.$$8

These refinements are systematically developed in the context of spatial random fields in (Guillaumin et al., 2019, Goodwin et al., 29 May 2025, Sykulski et al., 2016).

3. Algorithmic and Computational Properties

The main computational advantage is the $$S(\omega; \alpha) = \sum_{\tau = -\infty}^\infty c(\tau; \alpha) e^{-i \omega \tau}.$$9 complexity for all core steps, achieved via FFTs:

• Fourier transforms and periodogram computation are performed in $\omega_j = 2\pi j / n$0.
• For missing values, the modulation simply zeros out unobserved entries.
• Debiasing and modulation require $\omega_j = 2\pi j / n$1 pre-computations and at most two FFTs per evaluation.
• Joint optimization is performed over all parameters with gradient-based routines; the Whittle objective is jointly non-linear in regression and covariance parameters without analytic conditional solutions, distinguishing it from time-domain Gaussian ML (Pypkowski et al., 23 Dec 2025).

For large spatial domains, the multiparametric framework and FFT-based evaluation enable inference for $\omega_j = 2\pi j / n$2 up to $\omega_j = 2\pi j / n$3 (Guillaumin et al., 2019).

4. Theory: Statistical Properties and Regimes of Validity

The Whittle estimator is consistent and asymptotically normal under broad conditions. For data exhibiting only fourth-order stationarity, large-sample arguments ensure that estimators converge with asymptotic variance matching the time-domain ML estimator under Gaussianity (Pypkowski et al., 23 Dec 2025, Sykulski et al., 2016). In spatial settings ($\omega_j = 2\pi j / n$4), edge effects yield bias of order $\omega_j = 2\pi j / n$5 (smallest grid length), which can be addressed by the debiased spatial Whittle likelihood using expected periodograms (Guillaumin et al., 2019, Goodwin et al., 29 May 2025).

The method is robust to non-Gaussianity, in the sense that moment conditions are sufficient, and it directly accommodates complex patterns of missingness when the modulation correction is applied.

5. Empirical Performance and Applications

Extensive simulations and real-world analyses demonstrate the practical utility.

• For mixed models under Gaussian and non-Gaussian (AEP) errors, the Whittle likelihood yields estimates comparable to ML, with reduced computational burden and robustness under model misspecification (Pypkowski et al., 23 Dec 2025).
• In spatial statistics, the debiased spatial Whittle dramatically reduces bias, with RMSEs up to 50% lower than the classical approximation, even for grids with large proportions of missing data (Guillaumin et al., 2019).
• Real-world groundwater datasets ($\omega_j = 2\pi j / n$6) showed Whittle-based methods matched long-range and periodic autocovariances more flexibly than exponential ML fits, and outperformed on out-of-sample prediction in gap-filling tasks (Pypkowski et al., 23 Dec 2025).
• In simulation studies, the debiased approach reduced parameter bias by one to two orders of magnitude relative to the standard Whittle estimator and performed comparably to exact ML at a fraction of the computational cost (Sykulski et al., 2016).

6. Comparison with Other Likelihoods and Boundary Corrections

The Whittle likelihood is an asymptotic approximation to the Gaussian likelihood. Explicit matrix decompositions reveal that its departure from the exact time-domain likelihood is due to omitting best linear predictors outside the observed domain—i.e., boundary leakage effects. This insight underpins new pseudo-likelihoods that further reduce finite-sample bias by explicitly incorporating these predictors or plug-in AR model corrections (Rao et al., 2020).

Key differences between standard and debiased (or boundary-corrected) Whittle estimators can be summarized as:

Method	Accounts for Blurring/Aliasing	Finite-Sample Bias	Computational Complexity
Standard Whittle	No	High	$\omega_j = 2\pi j / n$7
Debiased/Boundary-corrected	Yes	Low	$\omega_j = 2\pi j / n$8
Exact Gaussian ML	Yes	Minimal	$\omega_j = 2\pi j / n$9

Debiasing is essential for accurate inference in spatial and small-$$j = -\lfloor n/2 \rfloor, \ldots, \lfloor (n-1)/2 \rfloor$$0 time series and in applications where parameter bias translates directly to poor prediction or scientific misinterpretation (Sykulski et al., 2016, Guillaumin et al., 2019, Rao et al., 2020).

7. Summary and Best Practices

The Whittle likelihood provides a computationally optimal and statistically robust framework for parameter estimation in stationary time series, spatial fields, and mixed models, particularly when handling large datasets, non-Gaussian errors, and missing data. Debiased and modulated extensions should be employed to correct for edge effects and missingness. In practical inference, the Whittle family of estimators offers a tradeoff between asymptotic efficiency, finite-sample performance, and computational cost that can be tuned via bias correction and model-based (e.g., AR) boundary augmentation (Pypkowski et al., 23 Dec 2025, Sykulski et al., 2016, Guillaumin et al., 2019).