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.
Upper-truncated Poisson distribution Created 2026-10-07 Updated 2026-11-05
A Poisson distribution of parameter conditioned to lie in . It is the occupancy law of a unit-resource loss network. Its mean is the carried load of an Erlang loss resource, . Differentiating the mean with respect to gives its variance, which is positive for and ; this proves strict monotonicity of the carried load.
Vacuum fluctuation Created 2026-10-25 Updated 2026-11-05
A vacuum fluctuation is a nonzero quantum variance in a vacuum state. For a single electromagnetic normal mode, dimensionless optical quadratures and have zero means and unit variances in the vacuum state, despite zero photon number. The mode's zero-point energy is .
Varentropy Created 2026-09-24 Updated 2026-11-05
The varentropy of a discrete random variable is the variance of its information content:
It measures variation in surprise across outcomes, while Shannon entropy is the expected value of that surprise. Assuming finite Shannon entropy, the variance may still be infinite; finite varentropy requires an integrable squared logarithm.
For a Bernoulli random variable with success probability ,
Indeed the information content takes values and , and the variance of a two-valued random variable is times the squared difference. At , every outcome supplies one bit, so Shannon entropy is maximal but varentropy is zero. For independent and identically distributed random variables, the information content of a length- word is a sum, giving variance when finite.
Variance Created 2026-09-24 Updated 2026-10-14
The variance measures the mean squared distance of a random variable from its expected value. For a real square-integrable random variable ,
The square prevents positive and negative deviations from cancelling. It also gives the variance squared units: if measures a length in metres, its variance has units of square metres. Its nonnegative square root is the standard deviation, which has the same units as .
For example, let take , and with equal probabilities. Its expected value is zero, but its variance is , and its standard deviation is . The constant random variable zero has the same expected value and zero variance: a mean alone does not describe the spread. The variance is zero exactly when almost surely.
The variance is useful because squared deviations interact neatly with linear combinations. The covariance supplies the cross terms when combining different random variables, and a covariance matrix organizes those terms for many variables. The characterization of variance as the minimum mean squared error of a constant also explains why the expected value is the best constant prediction under squared error.
Variance bound for a weakly self-bounding function Created 2026-09-28 Updated 2026-11-05
For a weakly self-bounding function of independent random variables, the coordinatewise variance inequality gives
where .
Variance comparison for compound portfolio approximations Created 2026-10-14 Updated 2026-11-05
Put , , and . A compound Poisson distribution with count mean and severity law has variance . A compound binomial distribution with trials, probability , and that same severity law has variance . The individual-policy variance is . The Cauchy-Schwarz inequality gives , proving the ordering. Equality of the first two variances requires constant ; it need not imply equality of the distributions. If all probabilities and severity laws are identical, the binomial compound approximation is exactly the individual model.
Variance-component likelihood-ratio test at a boundary Created 2026-10-05 Updated 2026-11-05
Testing a single nonnegative random-effect variance against zero puts the null on the boundary of the parameter space. In the usual regular increasing-independent-groups limit, the likelihood-ratio test statistic has limit , rather than . A finite-sample simulation calibrated for the actual design avoids relying on this asymptotic approximation.
For independent identically distributed random vectors with finite second absolute moments, expand . Their independence makes , proving the identity. The right side is the sum of the coordinate variances. Applied to the displacement in an optimal transport plan, it separates motion of the barycenter of a measure from relative particle motion.
Variance-minimizing exponential retention Created 2026-10-07 Updated 2026-11-05
For exponential distribution claims of mean and compound Poisson distribution count parameter , the total party variance under excess of loss reinsurance is . Its derivative is , so is the unique global minimum, with value .
For any aggregate with positive finite variance , quota share reinsurance with retained fraction has total party variance . Equal sharing uniquely minimizes this objective. The two party totals remain dependent; adding their individual variances is a different objective from the fixed variance of their sum.
For a Pareto distribution of shape three and minimum with count rate , the sum of insurer and reinsurer aggregate variances is for , and for . The first branch has its unique minimum at ; the second is increasing. The branches agree in value and slope at , giving the displayed global minimum. This criterion minimizes the sum of party variances; the total portfolio variance includes their covariance and remains .
Variance normalization Created 2026-09-24 Updated 2026-11-05
Variance normalization rescales a centered sum to have variance one. Let and assume . Then
For independent variables with finite variances, . With dependence it also contains . For example, if every equals the same random variable of variance , the sum's variance is , not .
This is the natural scale for central limit theorems, but variance one alone does not imply a Normal distribution or convergence to one. Appropriate assumptions, such as a Lindeberg condition for a suitable independent triangular array, control the shape of the normalized sum. Centering, choosing the correct variance and proving a distributional limit are distinct steps.
The variance of a corrected heterodyne overlap estimate quantifies uncertainty in , where is corrected peak beat current and are separately measured arm currents. The consistency bound for separately measured heterodyne arm powers requires the physical overlap not to exceed one, but measured estimates fluctuate.
For small errors, let have covariance , and put . Linearization gives
Thus common influences can cancel rather than contribute three independent uncertainty terms. This is the covariance form of the primary measurement-propagation rule.[ref]
For an original model, all three current readings share a multiplicative calibration error, with additional independent errors on each. The common contribution to is zero. The relative standard uncertainty is only , or at overlap . Treating the common calibration error as independent in all three readings would instead give about relative uncertainty.
The common-gain cancellation in heterodyne overlap uncertainty is exact for a single positive gain : replacing by leaves the ratio unchanged. A separate radio-frequency gain uncertainty affecting only does not cancel, nor does an additive dark-current offset generally have this property.
NIST's heterodyne receiver procedure corrected electrical loss and impedance effects separately from matching optical powers and modes.[ref] Build the covariance from those actual shared and separate calibration paths. This first-order variance model also needs reconsideration when an arm current is close to zero or background subtraction makes its uncertainty comparable to its value.
For a full-column-rank design matrix in a normal linear model, a fitted mean at covariate vector is . The covariance matrix of the least-squares estimator immediately gives the displayed variance. Replacing by the independent unbiased residual variance gives its squared standard error and a Student t confidence interval on statistical degrees of freedom. A future response instead contributes an additional error variance when its error is independent of the fitted data.
Variance of a long-time mean in a harmonic trap Created 2026-10-20 Updated 2026-11-05
For stationary covariance , the mean over duration has variance . At this becomes . Closely sampled positions remain correlated and cannot be counted as independent.
Variance of an estimator Created 2026-09-24 Updated 2026-10-14
The variance of an estimator is the variance of its sampling distribution under the parameter value .
The variance of a pedestal-corrected sample quantifies the random spread remaining after subtracting an estimated baseline from a signal acquisition. Subtracting a mean reset charge-injection pedestal removes an offset; it does not automatically remove reset pedestal variance. For a raw signal sample and pedestal estimate ,
The covariance term determines whether subtraction suppresses shared fluctuations or adds independent noise. The GUM's correlated-input rule gives this identity directly for the linear measurement model .[ref]
Two correction procedures have different statistical consequences. First suppose signal acquisitions have independent pedestals of variance and independent readout noise of variance . A separate mean of independent blank acquisitions, with individual variance , is reused for every signal sample:
After removing any constant means, independence gives
Consequently, averaging corrected samples gives
The shared blank correction creates a common uncertainty floor, as in shared calibration error.
Alternatively, a baseline and signal reading from the same acquisition can share a fluctuating pedestal. In an idealized correlated double sampling model, and , so : the common cancels. Independent readout noises remain and their variances add. More generally, if paired readings each have standard deviation0.20millivolt and correlation coefficient0.9, their difference has standard deviation
Treating that pair as independent would predict0.283millivolt instead.
An experimental assessment records paired zero-input samples at the actual acquisition delays and estimates their covariance, then distinguishes that procedure from reusing a separate blank mean. The ACF2101 datasheet specifies output noise and charge-transfer errors separately; a charge-transfer limit is not itself a pedestal variance or evidence for a particular correlation.[ref]
Variance of a sample mean Created 2026-09-24 Updated 2026-10-14
Let be an IID sample of real-valued random variables with finite expected value and finite variance , where . Its sample mean satisfies
Linearity of expectation gives the first identity. The variance of a sum and vanishing cross covariances give ; scaling by scales the variance by . Pairwise independent random variables with the same expected value and variance suffice for these identities, even without full independence. Grinstead and Snell state the corresponding finite-variance independent-trials result.[ref]
The standard deviation of the sample mean is . Four times as many observations halve this scale; they do not divide it by four. In the fair Bernoulli distribution example, , so sample sizes and give standard deviations and , respectively. The common-coin averaging example shows why merely repeating observations with a common marginal law does not justify this calculation.
Chebyshev's inequality turns the variance identity into a probability bound: for ,
This proves convergence in probability of the sample mean to in this finite-variance setting, a form of the weak law of large numbers. It is an upper bound, so values above one convey no improvement over the universal probability bound one. Cauchy sample mean stability lies outside these hypotheses.
Variance of a uniform distribution Created 2026-09-24 Updated 2026-11-05
A uniform variable on has mean zero and variance .