Central moment Created 2026-09-28 Updated 2026-11-05
For a random variable with finite mean , its th central moment is when defined. The first is zero; the second is the variance; the standardized third gives skewness. Translating by a constant leaves central moments unchanged, because the mean translates by the same amount. For equally likely values one and three, the second central moment is one, whereas the second raw moment is five. Odd central moments can cancel between positive and negative deviations, so an absolute moment provides a different integrability check.
This turbulence closure assumes the velocity increment skewness is the same throughout the equilibrium range, with its value fixed by the Kolmogorov two-thirds law and Kolmogorov four-fifths law. It yields a simple crossover between differentiable and inertial-scale increments, but suppresses the scale dependence of skewness and has no independent spectral-transfer dynamics.
Cornish-Fisher expansion Created 2026-10-14 Updated 2026-11-05
The Cornish-Fisher expansion inverts an Edgeworth expansion to approximate quantiles. For a standardized iid sum, let be the skewness of one summand and the standard normal quantile. The first distribution function correction is . Substituting and expanding gives . This is a fixed-quantile asymptotic, with fixed in .
Cubic intensity-noise bias in pulse correlation is the difference between the average of a third-order signal and the signal predicted from the average intensity when a fluctuating common intensity multiplier is cubed. Assume every shot has the same pulse shape, , with nonnegative scalar . The ideal third-order integral gives
The three intensity factors follow from the second-harmonic generation and sum-frequency generation gate.[ref]
Let , with , variance and third central moment . Expanding the cube gives
For , write the coefficient of variation as and, when , skewness as . The relative multiplier is . For zero variance the multiplier is one without needing to define skewness.
Equally likely and have mean one and give mean cubic multiplier . Rare bright shots can matter much more: stipulated probabilities 0.9 and 0.1 at and also give mean one, but
The extra signal need not mean that the mean input intensity increased. These are original probability examples, not measurements reported in the correlator paper.
For unchanged pulse shape, common fluctuations multiply every delay by the same expected factor. Normalizing the averaged trace by its peak cancels that factor. Shape changes correlated with brightness destroy this simplification. For example, two equally frequent shot classes with central signals 1 and 8 and side/center ratios 0.01 and 0.10 give ratio of averaged signals , whereas averaging the two ratios gives 0.055. They answer different questions.
A controlled measurement records a synchronized input-intensity monitor and detector background alongside each trace. Compare intensity-binned signals in an unsaturated range and test whether normalized shapes agree. Common attenuation in third-order pulse correlation predicts the cubic scaling only while the nonlinear gate, pulse shape and readout retain the assumed response. A mean power reading alone does not determine the required third moment.
Edgeworth series Created 2026-10-05 Updated 2026-11-05
An Edgeworth expansion refines the central limit theorem by adding corrections determined by higher cumulants. Let be independent and identically distributed with mean zero, variance and skewness . For the standardized sum , the first density correction is
Here is the standard normal density. This is an asymptotic statement for a sequence, not a universal density formula for one arbitrary random variable. Sufficient smoothness and moment conditions are needed; one strong sufficient setting is that the common probability density function is a non-degenerate Schwartz function. In that setting the displayed remainder is valid for each fixed . Lattice-valued variables need additional corrections rather than this density formula.[ref]
To identify the coefficient, expand the log characteristic function of the standardized sum:
for fixed . Exponentiation and Fourier inversion turn into the Hermite polynomial times . The smoothness assumptions justify inversion beyond this formal calculation. Higher-order terms use further cumulants and Hermite polynomials.
The approximation improves central asymptotics, but it need not be a nonnegative normalized density after truncation. In particular the cubic correction can become negative in a far tail, where a fixed-point expansion does not provide a small relative error. Inverting a distribution-function Edgeworth expansion produces the Cornish-Fisher expansion for fixed quantiles.
Gaussian mixture distribution Created 2026-10-14 Updated 2026-11-05
A Gaussian mixture distribution is a weighted sum of normal distribution densities, with nonnegative weights summing to one. It need not itself be a normal distribution: components with different means and variances can produce skewness or several modes. Its raw moments are weighted sums of the component moments.
Histogram 2026-10-14
A histogram groups numerical observations into adjacent intervals called bins and draws one rectangle per bin. Its attraction is that a long list of values becomes a visible pattern of concentration, gaps and skewness. Specify the bin edges, the treatment of values on those edges, and what the vertical axis means: count, proportion, or density. Those choices are part of the graphic. [ref]
Let observations be covered by disjoint bins with positive widths and counts , so . For example, use left-closed, right-open bins, closing the final right edge too. The three common height choices have different meanings:
For the density convention, a histogram is a piecewise constant probability density function estimate:
Its normalization follows by adding rectangular areas:
If the observations have units, density has reciprocal units; a density height is not a probability and can exceed . With equal bin widths, count heights and density heights differ only by a common scale factor. With unequal widths, count-height areas give wider bins extra visual weight unrelated to their counts. Dividing by width removes that distortion. R's histogram documentation explicitly describes area-one normalization for unequal bins. [ref]
Reuse the twelve invented values under box plot, with bins . The counts are and the widths are . Hence the density heights and areas are exactly
The second and third bars have equal density heights even though the third contains twice as many observations: it is also twice as wide. The last bar contains only two observations; its count-height area would be , larger than the six-observation bar's . Its correct density area is only .
Figure 1.
Unequal toy bins: counts determine heights, density makes area represent proportion
. Both panels use the same twelve invented values and bin edges . Left: count heights , whose areas are not proportional to counts. Right: density heights , with total area . Equal density heights for the second and third bins express equal concentration per unit width. No bars represent measured data. Details.
Bin choice trades detail against stability. Broad bins merge distinctions: the toy bin hides the empty stretch between and . Narrow bins expose gaps but can make each individual observation dominate a bar. With one-unit bins starting at , the bins each contain two values and appear equally prominent; that apparent pattern depends on grouping the repeated integers together. Moving the bin origin also changes membership without changing the observations. For density estimation, this is a form of the bias-variance tradeoff: averaging over broad bins loses local detail, while narrow bins give more variable density estimates across repeated statistical samples. [ref] A useful practice is to compare a few plausible widths and offsets, keeping common edges when comparing groups. Automatic bin selectors are starting points, rather than guarantees of a unique correct picture. [ref]
A histogram retains only bin membership. Its density estimate spreads each bin's fraction uniformly across that bin, so our last bar assigns density throughout even though the only toy values there are and . It does not recover within-bin positions or observation order, and a visually convincing shape does not establish a population probability distribution. A box plot provides a complementary compact summary of quartiles; neither graphic replaces the original observations.
For isotropic incompressible flow, directional averaging gives . Positive mean vortex stretching therefore implies negative longitudinal derivative skewness. Statistical homogeneity alone does not determine the sign in a fixed direction.
Normal Q-Q plot 2026-10-14
A normal Q-Q plot compares order statistics with standard normal quantiles . For a normal distribution with , the population quantile function is , so the diagnostic sample relation is
The appropriate reference is a line with an intercept and positive slope, not necessarily the identity line. A line can be fitted through two central sample quantiles; R's qqline uses the quantile levels and by default. Neither that line nor its fitting rule is part of the definition of a normal Q-Q plot. [ref]
To isolate curvature, retain the same 19 plotting positions and construct two more ordered sequences:
The are continuous uniform distribution quantiles on , while the apply the hyperbolic sine to the reference quantiles. Both examples are symmetric about zero. At ranks , is approximately and is approximately . Again, these are constructed values rather than random draws or real data.
Figure 1.
Normal Q-Q curvature distinguishes compressed and stretched outer quantiles
. Left: the constructed uniform distribution quantiles bend inward relative to the central tangent . Right: the constructed values bend outward relative to the central tangent . Both panels use 19 standard normal quantiles at . Dashed lines are analytic central tangents, not fitted lines or uncertainty bands; only the chosen finite range is illustrated. Details.
Read these shapes relative to their central slope. The uniform distribution example has , with derivative ; at the upper endpoint the tangent predicts , while . Its lower end lies above the tangent and its upper end below it: the outer quantiles are compressed. The hyperbolic sine example has ; its tangent predicts at the upper endpoint, while . Its lower end lies below the tangent and its upper end above it: the outer quantiles are stretched. Unequal departures at the two ends can suggest skewness. The direction of these descriptions assumes that reference quantiles are horizontal and observed values vertical. The graphs illustrate finite-range shape differences; a sample pattern cannot establish an asymptotic tail probability law or membership in a heavy-tailed distribution class.
A nearly straight normal Q-Q plot is a diagnostic, not a proof of a normal distribution or a formal hypothesis test. Random statistical samples fluctuate, particularly at extreme order statistics, and apparent curvature depends on sample size, plotting positions and the reference line. A calibrated hypothesis test or uncertainty band requires its own assumptions and must account for parameters estimated from the same observations. Sorting also discards the original observation order: rearranging a time series leaves its Q-Q plot unchanged, so the plot cannot establish independence.
For a normal linear model, R's diagnostic normal Q-Q plot uses standardized regression residuals, also called internally studentized residuals. [ref] Even with independent errors, the regression residuals have covariance matrix , where is the hat matrix; fitting parameters generally creates dependencies. Dividing by a shared estimated standard deviation creates further dependence, and the distribution of an internally studentized Gaussian residual is not exactly the standard normal distribution. The normal Q-Q plot therefore helps assess error shape alongside other diagnostics rather than certifying independent normal random variables.
Let the summands have mean mu, variance , and cumulants . For the standardized sum , expansion of its log-characteristic function gives a normal leading term and successively smaller cumulant terms. Fourier inversion turns these terms into Probabilists' Hermite polynomials. Under sufficient moments and a smooth nonlattice condition, the Edgeworth expansion begins
for bounded z. Higher orders include the fourth cumulant and products of lower cumulants. This is a central expansion in powers of ; it need not remain positive, normalized, or accurate in far tails after truncation.
For a saddlepoint density approximation, require the summand cumulant-generating function K to be finite on an interval about zero. At a sum value , find the interior saddle satisfying . Exponential tilting makes the tilted mean y, so the tilted sum is approximated locally by its normal probability density at its mean. Undoing the tilt gives
Equivalently this follows by deforming the transform-inversion contour through its stationary point. Suitable analytic and nonlattice regularity gives relative error at fixed interior y. This approximation incorporates a nonquadratic tail exponent and is positive, while an Edgeworth truncation uses a normal probability density times a polynomial. It uses more information from the distribution and stronger exponential-moment assumptions. Its integral may differ from one by a higher-order amount, so positivity does not imply exact normalization.
For the centered Laplace distribution, , hence and . Its skewness vanishes, so the order- Edgeworth correction is zero. With , the requested expansion, omitting terms of order in the standardized correction, is
For bounded z the exact probability density function is this expression times . The prefactor is because this is the probability density function of the unstandardized sum; that change of units should not be confused with the orders inside the Edgeworth expansion.
For the saddlepoint, . Write and . The root in is
These formulas include without a division by y. Substitution yields
This is the leading saddlepoint approximation, without its order- relative correction. Expanding d near one recovers the same central normal approximation, while retaining d captures the Laplace-sum large-deviation exponent away from the center.
Figure 1.
Leading Edgeworth and saddlepoint approximations compared with the exact Laplace-sum density
. Details.
Write , with , and differentiate :
Using and gives
Thus the first three cumulants are the mean, variance and third central moment. A finite moment-generating function near zero justifies these differentiations. For positive claims with only three finite ordinary moments, the same calculations are valid as left derivatives at zero of the logarithm of the Laplace transform of a nonnegative random variable; positive exponential moments are not needed for the resulting moment identities.
For the independent Poisson distribution count, the random-sum transform identity gives
Consequently the compound Poisson distribution has
These are raw severity moments, not severity cumulants.
The printed gamma density uses rate , not scale. Direct integration against gives
Therefore the required matching equations for the shifted gamma distribution are
For and nonzero positive claims with finite , these equations give positive and the unique solution
This is the three-cumulant shifted gamma approximation. In particular, its skewness equals , the skewness of .
The competing normal approximation to a compound Poisson aggregate is
To justify its large-count limit, let be the claim characteristic function. For fixed severity law with , its expansion is . The characteristic function of is
Thus it converges in distribution to a standard normal distribution as .
The normal approximation is simple, uses only , and works well in the centre when the count is large and standardized skewness is small. It imposes symmetry, whereas positive compound-Poisson claims have positive third cumulant. The shifted-gamma approximation reproduces that third cumulant and can better describe moderate right skewness. It requires a finite and reliably known third moment, and matching three cumulants alone does not guarantee accurate extreme tails.
Neither continuous approximation reproduces the genuine atom , which matters at small . The normal distribution also assigns probability to negative losses. The gamma approximation is bounded below by , but its fitted shift need not be nonnegative: exponential claims of mean give , so it too can assign mass to negative losses. If , it instead excludes some genuinely possible small losses. These support and atom discrepancies prevent either approximation from being a universal replacement for the exact aggregate law.
Let . Near the interior minimum, Taylor expansion gives
On the scale , the leading exponent is and . If the integral is localized at this minimum, the rescaled integration endpoints tend to infinity in both directions, so the Gaussian integral gives Laplace's method:
Formally expanding the cubic term gives an order- correction odd in , whose Gaussian integral is zero. Quartic terms and the square of the cubic term produce the order- correction. Sufficient smoothness near the minimum and control of the exterior integral justify this formal calculation.
The printed local assumptions alone do not guarantee localization. A useful localization condition for Laplace's method is that outside each sufficiently small fixed neighbourhood of , for some , and that for some . For , the exterior contribution is at most
which is exponentially smaller than the displayed local term. To see why something of this kind is needed, take on . Its unique minimum is at zero and , yet on we have . This boundary interval alone contributes at least , dominating the proposed local approximation. The Laplace formula therefore has its usual implicit tail assumption, rather than following from uniqueness of the interior minimum alone.
For the gamma function, set to obtain
Here has its unique minimum at , with and . It tends to infinity at both endpoints, and its integral is finite for positive exponents, so localization holds. Laplace's method yields Stirling's formula:
For a prior density and independent observations, Bayes' theorem writes the desired posterior expectation as a ratio of integrals:
Suppose the log-posterior kernel has a unique concentrating interior mode , with curvature of order , and the smoothness and tail conditions hold for both integrals. Expanding quadratically gives a common Gaussian factor . Replacing the smooth amplitude by gives
This approximation does not require to be positive: it treats as an amplitude, without taking its logarithm. For more accuracy, expand the amplitude and the cubic term in the log-posterior kernel. With , the Gaussian moments are and . Terms multiplying cancel between numerator and denominator; the two remaining first corrections are
under the stronger smoothness, derivative-scaling and integrability assumptions needed for this expansion. The first term accounts for the curvature of the amplitude and the second for posterior skewness. This is posterior expectation by Laplace approximation. If there are several modes, boundary modes, or nonconcentrating tails, the appropriate separate contributions must be included; a single interior Gaussian approximation cannot then be assumed.
Combining the Kolmogorov two-thirds law with the Kolmogorov four-fifths law gives the velocity increment skewness in the inertial range:
The constant-skewness turbulence closure assumes this same value throughout the universal equilibrium range, including the dissipative crossover. This is an extra modeling assumption; neither similarity law establishes it near .
Write , , and . Since and , substituting into the Kolmogorov equation for structure functions gives
Dividing by yields the normalized constant-skewness structure-function equation:
At small , , recovering for differentiable velocity increments. At large , the dominant balance gives , recovering the assumed Kolmogorov two-thirds law.
This turbulence closure is inexpensive: a single ordinary differential equation interpolates between the two regimes while respecting the local third-order balance. Its prescribed skewness cannot independently predict scale-dependent derivative statistics, transient spectral evolution or internal intermittency.
EDQNM closure instead evolves the turbulent energy spectrum through modeled spectral triad interactions. Its quasi-normal fourth-order approximation is supplemented by eddy damping and a Markovian treatment of correlation memory. It retains more information about scale-dependent transfer and evolving spectra, at the cost of spectral integrals and modeled damping times. It is still approximate and does not automatically describe coherent structures or internal intermittency. The simple constant-skewness model is therefore useful as an equilibrium interpolation; EDQNM closure addresses a broader spectral evolution problem.
Put , and , where . The small- Taylor series give
In the Kármán-Howarth equation, the coefficient of on the left is . On the right, the triple-correlation term contributes and the viscous term contributes . Hence
Comparing with the mean enstrophy balance identifies the mean vortex stretching:
Also , whereas differentiability gives . Thus . Homogeneity makes , so the longitudinal velocity-gradient skewness is . Substitution gives
Positive mean vortex stretching therefore requires negative skewness. A centered Gaussian distribution has zero third moment, so the derivative statistics cannot be Gaussian. This conclusion uses the physically positive mean stretching, not homogeneity alone.
The probability density function has mean zero but a longer or stronger negative tail, representing intermittent compression. The illustration uses a standardized Gaussian mixture distribution solely to sketch this sign of skewness; it is not turbulence simulation data or a fitted universal distribution.
Figure 1.
Illustrative negatively skewed longitudinal velocity-gradient density
. Details.
Write and . The derivatives of the cumulant-generating function are
Since and , the first three cumulants are
The last equality follows by expanding the cube and taking expectations. We use the paper's unstandardized third central moment throughout. The usual skewness coefficient divides this quantity by ; its sign is the same whenever the variance is positive.
There is a minor domain qualification: finite ordinary moments do not guarantee a finite moment-generating function for positive arguments. The differentiations are valid in a neighborhood of zero when a positive exponential moment exists. For positive variables with the stated finite moments, all formulas also follow from left derivatives at zero, because for and dominated convergence applies. Thus no extra exponential-moment assumption is needed for the moment identities themselves.
For the random sum of independent claims, condition on . Independence of the claims and independence from the count give
Taking expectations and then logarithms yields the random-sum transform identity and the cumulant composition for a random sum:
Here is the probability generating function. These identities hold wherever the transforms are finite, in particular for with positive claims.
For reuse in the three cases, write and . Applying the chain rule three times gives
or equivalently
This expresses the requested third central moment in raw claim moments, while keeping the contribution of claim-count variability explicit.
The central limit theorem gives , so, with the cumulative distribution function of the standard normal distribution,
The error has two tails. Using the supplied quantile, whose two-sided coverage is approximately , gives
The supplied quantile would be appropriate for a one-sided calculation; it would give only approximately two-sided coverage. Using the exact quantile instead of would give draws; we consistently use the paper's rounded value below.
For routine Monte Carlo integration, the central limit theorem calculation is preferable as a practical budget: it uses about one fifteenth as many draws as the Chebyshev's inequality bound. However, its coverage is approximate. Finite variance alone supplies no universal rate of convergence to the normal distribution; an integrand with large skewness or a heavy-tailed distribution can make the approximation poor even at a large sample count. If an unconditional finite-sample guarantee is required under only the stated assumptions, use the Chebyshev's inequality budget.
Shifted gamma distribution Created 2026-10-14 Updated 2026-10-20
A gamma distribution translated by a real constant . With shape and rate , its support is and its first three cumulants are , , and . Its skewness is . A negative shift permits negative values; introducing a shift does not automatically respect the nonnegative support of an insurance loss.
Skewness Created 2026-10-05 Updated 2026-11-05
For a real random variable with finite third absolute moment and positive variance , its skewness is . It is dimensionless and unchanged by positive rescaling or translation; a reflection reverses its sign. A symmetric distribution has zero skewness, but the converse fails: values with probabilities have mean zero and third moment zero without symmetry. Constant variables have no defined standardized skewness because . The cosmological reduced skewness uses a different normalization.
Three-cumulant shifted gamma approximation Created 2026-10-14 Updated 2026-11-05
Approximate a compound Poisson distribution of positive claims by a shifted gamma distribution whose first three cumulants agree. Here are raw severity moments, while the compound cumulants are . Solve , , and to obtain the displayed parameters. The approximation captures positive skewness, but not the atom at zero. Its shift can be negative; with exponential claims of mean it is . Matching three cumulants alone supplies no general tail-error bound.
For Gaussian initial conditions and the Einstein-de Sitter second-order standard perturbation theory density kernel, the unsmoothed density contrast obeys
There are three choices of the second-order field and two connected Wick contractions; the angular average of is . Thus the reduced skewness is . Smoothing changes this result, and the unsmoothed calculation requires a finite variance.