Paul D. Hale 2026-11-05
Paul D. Hale is a metrology researcher named with Tracy S. Clement and Dylan F. Williams on a 2001 NIST paper about optical receiver frequency response. It compares frequency-domain optical heterodyne detection with time-domain measurements using an oscilloscope, including corrections for the electrical connection and finite optical stimulus.[ref]
An oscilloscope trace need not be the receiver's isolated impulse response. In an original idealized model of linear time-invariant systems, let be the receiver impulse response, the optical pulse envelope and the readout impulse response. Ignoring mismatch for this illustration, the measured waveform is the convolution
A broader measured pulse can therefore reflect the receiver, the stimulus or the readout. If all three envelopes are normalized Gaussian functions, with independent convolution widths, their variances add:
For illustrative widths , and , the inferred receiver width is , rather than . These are chosen model values, not results reported by Hale.
The example motivates characterizing the stimulus and readout as well as the specimen. Dividing by where that product is nearly zero also amplifies measurement noise, so a formal Fourier transform inversion does not guarantee a reliable corrected response. This is why measurement uncertainty belongs alongside the quoted frequency response.
Peak-height calibration from an exponential mass-function cutoff Created 2026-10-06 Updated 2026-11-05
Defining characteristic mass by an exponential factor fixes its peak height to , rather than one. Hence and the variance at other masses follows the same power-law scaling. The precise characteristic-mass convention determines this factor.
The Pearson correlation coefficient is the covariance divided by the product of the standard deviations. For real random variables with finite, strictly positive variances,
It is the expected value of the product of the two centered variables after dividing each by its standard deviation. The Cauchy-Schwarz inequality gives . This removes measurement units, while retaining the direction and strength of the linear relationship. Constant random variables have zero standard deviation, so their Pearson correlation coefficient is undefined.
For the covariance model , with independently uniform on , the calculation gives
The positive correlation is less than one because the independent term introduces extra variation. For the correlation is . Both examples appear in the covariance diagram.
For nonzero real , the correlation of and is . Positive rescaling and added offsets therefore leave the correlation unchanged; reversing one scale reverses its sign. Equality means that one random variable is an affine function of the other almost surely, with a nonzero slope of the corresponding sign. This is the equality condition in the Cauchy-Schwarz inequality, and includes perfect positive correlation.
Zero correlation says that the centered variables are orthogonal in their square-integrable inner product. It permits nonlinear statistical dependence, as the example in uncorrelated random variables demonstrates. More generally, a correlation is a statement about a joint probability distribution; identifying effects of interventions is the separate task of causal inference. This distinction keeps an observed association from being mistaken for a demonstrated mechanism.
An early source for standardized association is Francis Galton's 1888 paper “Co-relations and their Measurement, chiefly from Anthropometric Data”. He compared stature and forearm measurements and changed each measurement to units of its own variability so the two regression directions could be compared. His scale used probable error, so it should not be identified word for word with the modern standard deviation formula above. [ref].
Perfect positive correlation Created 2026-10-07 Updated 2026-11-05
For square-integrable variables of positive variance, correlation one is equivalent to with . The squared norm of the difference between the two standardized centered variables is , proving the equivalence. For uncentered variables the relationship can have a nonzero intercept; it need not be proportionality.
Periodic utility indifference payment for Gaussian income Created 2026-10-07 Updated 2026-10-14
For Gaussian income of mean , standard deviation and tradable correlation , exponential utility makes the maximum acceptable per-period fee equal to the mean minus the correlated-income hedge cost and the residual variance penalty. Compare optimized utilities with and without the contract; a fee above this level makes the contract strictly inferior. At perfect correlation the residual penalty vanishes, leaving only the mean and hedge-cost terms.
Peter Bartlett 2026-11-05
Peter Bartlett studies the foundations of machine learning, statistical learning theory and reinforcement learning. His Berkeley research description emphasizes the joint demands of computational efficiency and statistical performance, including prediction with complex outputs and sequential decisions.[ref]
His Introduction to Time Series Analysis, Lecture 20 by Peter Bartlett gives a concrete lesson about estimation. The raw periodogram can have an approximately correct expected value while its variance remains substantial as the sample grows. For Gaussian white noise, an interior-frequency periodogram ordinate is distributed as a scaled chi-squared distribution with two degrees of freedom. Spectral smoothing is therefore doing statistical work, rather than merely making a graph look smoother.[ref]
Figure 1.
Peter Bartlett
. Portrait identified by the UC Berkeley Department of Statistics profile. Source.
Peter M. Aronow 2026-11-05
Peter M. Aronow develops statistical methods for causal inference, particularly design-based inference. Yale's institutional biography identifies research on complex experimental designs, network settings and incomplete survey data alongside appointments spanning political science and statistics.[ref]
His joint work with Cyrus Samii studies how to estimate sampling variance when a sampling design never observes some pairs together. A missing product cannot simply be deleted from the population variance formula: doing so can produce either sign of bias. Their construction uses Young inequality to bound unobserved products by individually estimable powers and obtain a conservative variance estimator.[ref]
For an original exact illustration, choose one of two fixed values with equal probabilities. As in the single-draw variance identifiability obstruction, the Horvitz-Thompson estimator has true variance . Retaining only diagonal contributions gives , with
Opposite signs therefore produce downward bias. The elementary inequality , obtained from , instead gives
Consequently is a conservative variance estimator for this stipulated sampling design; its excess in expected value is exactly .
With , the true variance is16. The two possible values of are18 and2, averaging10; those of are36 and4, averaging20. Although , a realized value of can be below16. Conservativeness here is an expectation inequality, rather than a bound for every sample. By itself it also does not prove a specified confidence interval coverage probability.
Figure 1.
A conservative variance estimator can still fall below the true variance in an individual sample
. Original exact two-unit example with fixed values 3 and -1, each selected with probability one half. Diagonal-only estimates are 18 and 2; corrected estimates are 36 and 4. Their means are 10 and 20, while the true variance of the Horvitz-Thompson total is16. Details.
For real , put in . Its mean second quadrature is , which is generally nonzero. Subtracting its square from the second moment gives the displayed variance and squeezing precisely when , for nonzero finite amplitude. A useful counterexample to confusing moment and variance is .
Phase-difference variance Created 2026-09-28 Updated 2026-11-05
For the physical free energy
the Boltzmann factor is . At , the Gaussian thermal phase fluctuations therefore obey
The temperature factor is necessary when is the energy stiffness defined by the parent. Defining a dimensionless stiffness instead replaces by .
The difference cancels the spatially uniform phase mode. At any fixed separation, removes the small- singularity. What grows in low dimensions is the variance as the separation tends to infinity.
In one dimension, the integral equals , so the variance is . In two dimensions, with microscopic cutoff and , it grows as
For the Gaussian field, the order parameter correlation is
It decays exponentially in one dimension and algebraically in two. Thus nonzero-temperature Gaussian fluctuations do not leave conventional long-range order in these dimensions. This is the long-wavelength mechanism related to the Mermin-Wagner theorem; vortices and interactions require the fuller theory of the Berezinskii–Kosterlitz–Thouless transition in the two-dimensional case.
Thorlabs' silicon FDS100 experiment sends a collimated, fibre-coupled laser beam through a beam splitter to the photodiode and a reference optical power meter. Varying reverse bias and electrical load tests linearity. The measured readout approaches saturation when its electric potential difference consumes the bias.[ref] A separate low-signal setup uses a laser, polarizers, an optical chopper, optical attenuators and a lock-in amplifier inside a light-tight enclosure. These are actual apparatus choices, not the parameters of the constructed example above.[ref]
For an independent dark current check, Thorlabs enclosed a photodiode in nested metal boxes to exclude light and suppress interference, used a Keithley 6487 ammeter supplying reverse bias, and monitored temperature. Only one specimen per type was tested; the nearby temperature sensor did not directly contact the semiconductor. Such a dark control measures the baseline but cannot certify spectral responsivity.[ref]
To infer spectral responsivity at a selected wavelength, measure the radiant flux actually reaching the test photodiode, rather than just the reference arm. A reference-arm measurement calibration must include the beam splitter ratio and path losses. With signed readouts in one fixed orientation, subtract a beam-blocked baseline and use
for direct electric current readout or a calibrated linear transimpedance amplifier, respectively. Repeating at several incident powers tests whether this ratio is constant. Blocking only the source estimates ambient light plus electronic baseline; covering the detector excludes ambient light too. Their difference distinguishes optical background from dark current and readout offset. For a broad optical spectrum, the ratio is a spectrum-weighted response, not generally at one wavelength.
For independent Poisson distribution collection, the root mean square shot noise in a one-sided rectangular frequency band of signal bandwidth is
Here is reverse dark current; thermal electrical noise and amplifier noise also contribute.[ref]
Subtracting a mean dark current removes its offset, not its fluctuations: subtracting a constant leaves the variance unchanged. Consequently makes dark current shot noise small relative to signal shot noise; it does not by itself establish high signal-to-noise ratio. With an independently measured baseline, the variance of that estimate must also be propagated. Increasing the measurement signal bandwidth raises the admitted noise. Calibration of wavelength, reference radiant flux, readout gain and collection geometry, together with drift and linearity, determines what measurement uncertainty can be claimed.
Photoelectron peak overlap and resolution Created 2026-10-18 Updated 2026-10-20
Choose two hypothetical environments A and B of one chemical element, with line centers and . Model each line by a normalized Gaussian function with common standard deviation , and stipulate integrated count weights , . On the binding energy axis , the normalized count density is
All energies in this expression use electronvolts. Each component's integrated area equals its count weight, and . There is no background, statistical counting fluctuation, or additional emission feature in this stipulated model. These choices deliberately isolate photoelectron peak overlap and resolution.
Figure 1.
Two chemical components can blend into one maximum
. Original arbitrary normalized data, not a measured spectrum. The chosen Gaussian mixture distribution has centers and and integrated count weights and . Changing the common standard deviation from to changes the number of maxima from two to one, while leaving the centroid at . These count weights are not mole fractions; a chemical abundance inference requires cross section and analyzer-response calibration. Details.
Direct integration, or the law of total expectation applied to the component label, gives the centroid
The centroid lies between the chemical environments and is not either component's binding energy. Expanding within each component, the odd cross term integrates to zero. The resulting variance is
For the variance is and the two maxima are separated. For it is and there is one maximum. Each Gaussian's full width at half maximum is , approximately and respectively. The chemical separation remains in both cases. The number of visible maxima therefore does not equal the number of chemical components.
One interpretation of this change is additional independent Gaussian instrument blur. The convolution of probability densities adds the variances, so broadening to requires response standard deviation . A normalized response with zero mean preserves the centroid and integrated counts. Real line shapes and backgrounds can depart from this model, and separating hidden components is an inverse problem; a plausible fit does not alone establish uniqueness.
Finally, count weights describe detection. In a simple response model, let denote the number of emitting sites sampled in environment , its effective photoionization cross section, and the probability that its emitted electron contributes to the selected measured line. Then
The common factor includes exposure. Here includes escape from the sample, analyzer transmission and detection. Count weights equal site fractions only if the relevant response factors are equal or corrected. They are not automatically molecular mole fractions; different molecules can supply different numbers of the selected emitting sites. In an original example, equal site numbers with give the plotted count ratio despite a site ratio.
Photon antibunching Created 2026-10-18 Updated 2026-11-05
Photon antibunching means suppression of close pairs compared with pairs at some later delay: for some . If correlations decay to the independent-count reference, is a sufficient signature. It violates the classical intensity bounds in photon bunching. For the specified single-mode number measurement, second-order optical coherence also shows that is equivalent to . This variance criterion concerns that mode or gate; it must not be applied unchanged to arbitrarily long integrated counting intervals.
The single-mode squeeze operator changes occupation in pairs. Consequently an optical squeezed vacuum has only even photon numbers, with mean and variance . Its large number fluctuations are compatible with quadrature squeezing: they concern different observables.
P. H. Peskun 2026-11-05
P. H. Peskun authored the 1973 paper underlying Peskun ordering, a comparison of the efficiency of Markov chain Monte Carlo algorithms with the same stationary target.[ref]
For finite reversible Markov chains, increasing every off-diagonal transition probability while preserving the stationary law cannot increase the asymptotic variance of time averages. The comparison formalizes why spending less time repeating the current state can help statistical estimation, provided the ordering's hypotheses hold. It compares the variance of estimators, not merely their one-step acceptance rates.[ref]
Pinned capillary-wave correlation Created 2026-10-14 Updated 2026-11-05
Adding positive quadratic confinement to the membrane energy changes its height spectrum to . The continuous height-shift symmetry is explicitly broken. Connected correlation function decay on the finite length and the height-difference variance saturates for every internal dimension, although its saturation value can depend on a microscopic cutoff. A negative confinement coefficient gives an unstable quadratic model rather than pinning.
Pivotal quantity Created 2026-09-24 Updated 2026-11-05
A pivotal quantity is a function of observed data and unknown parameters whose sampling distribution does not depend on those parameters. For independent normal observations, has the Student t-distribution with degrees of freedom. Although the expression contains the unknown , its law does not depend on or the unknown variance. Bounding this pivot by fixed quantiles and rearranging gives a confidence interval. A pivot need not itself be a statistic, since it may contain an unknown parameter.
For free planar turning with constant , no imposed torque and negligible inertia, model an unwrapped angle by the stochastic differential equation
Here is standard one-dimensional Brownian motion, with increment variance , and brackets denote expected values over repeated realizations. The angular variance is conventionally reported in . “Unwrapped” means recording accumulated turns continuously, rather than resetting the angle at each revolution. Directly subtracting wrapped camera angles creates artificial jumps; for an unmarked ellipsoid the wrapping period is , and resolving successive turns additionally requires sufficiently frequent images.
For a planar arrow , the orientational autocorrelation is
For an unoriented planar axis, use an autocorrelation unchanged by replacing either endpoint angle by that angle plus :
The same thus gives arrow and axis correlation times and , respectively, where a correlation time here is the lag at which the relevant curve reaches . A short explanation is enough: has a Gaussian distribution of variance , so its characteristic function gives . The integer selects what the observable can distinguish; it does not change the underlying angular spreading.
For an original numerical example, choose . After , the angular variance is and its root mean square displacement is , although the mean displacement is zero. The planar arrow autocorrelation is , while the planar axis autocorrelation is . Calling both “the orientation correlation” without specifying the observable would hide a factor of four in decay rate.
Figure 1.
Planar rotational diffusion: arrows, unoriented axes and loss of angular memory
. Original schematics, three simulated Brownian motion angle paths and exact ensemble autocorrelation curves for the chosen . The paths use independent Gaussian distribution increments of variance with . The curves are model predictions, not fits to these three paths or experimental measurements. Both curves concern planar turning; neither is the three-dimensional unit-vector law. Details.
Plug-in bandwidth selection Created 2026-10-14 Updated 2026-11-05
Estimate the unknown curvature in the AMISE minimizer and substitute it into that formula. A consistent positive curvature estimate makes the selected bandwidth ratio-consistent. Pilot smoothing is needed because differentiating an undersmoothed density estimator magnifies variance. A normal-reference pilot can initialize an estimate of a higher derivative functional; the final density-curvature estimate is computed from the data rather than assuming the density is normal.
Pointwise minimax rate for Hölder density estimation Created 2026-09-28 Updated 2026-11-05
Over a one-dimensional Hölder density class of smoothness , the optimized pointwise mean-squared error has order . A kernel of sufficiently high order balances squared bias against variance .
Pointwise minimax rate for Lipschitz regression Created 2026-10-05 Updated 2026-11-05
At an interior target point in regular fixed-design nonparametric regression, mean functions with Lipschitz bound have pointwise minimax risk of order for fixed positive as . A smoothing bandwidth of order balances squared bias of an estimator and variance; the two-point lower bound with a triangular bump gives the matching lower rate.