Probable error Created 2026-10-14 Updated 2026-11-05
The probable error is the half-width of an interval around a specified central value containing probability . For a normal distribution with , its quantile function giveswhere is the standard normal cumulative distribution function and is the standard deviation. The factor is specific to the normal distribution. The centered Cauchy distribution with parameter has probable error , despite having no finite variance. Thus a central probability width need not require a finite second moment.
Proficiency-test z-score 2026-11-05
A proficiency-test z-score expresses a laboratory result's deviation from the assigned value in proficiency testing in units of the scheme's standard deviation for proficiency assessment:The score is dimensionless because numerator and denominator describe the same kind of quantity in the same units. Its sign gives the direction of the deviation. The denominator is an assessment scale, not automatically the participant's own measurement uncertainty.[ref]
For an original resistance example, take , and a reported value . The score is . Under the guide's commonly used interpretation, is satisfactory, raises a warning, and raises an action signal. Thus this example is questionable; a result of gives and lies exactly at the satisfactory boundary, whereas gives . The scheme must state its interpretation.[ref]
Selection, Use and Interpretation of Proficiency Testing (PT) Schemes by Brian Brookman and Ian Mann. §7.2.4, printed p. 17, score interpretation
The word z does not by itself make this a hypothesis test with a universally fixed error probability. If scores actually followed a standard normal distribution, the stated thresholds would have the corresponding normal-tail probabilities. That distribution is an additional model assumption; the numerical scoring rule alone does not establish it.
Uncertainty in the benchmark can change the comparison. The proficiency-test z-prime score usesWith the same deviation and , an assigned-value standard uncertainty changes the denominator to and the modified score to 2. A smaller modified score may reflect a less certain benchmark, not improved laboratory measurement.[ref]
Selection, Use and Interpretation of Proficiency Testing (PT) Schemes by Brian Brookman and Ian Mann. §7.2.4(b), printed pp. 17–18, modified score and caution
A score should therefore travel with the assigned-value method, assessment scale and relevant uncertainty information. Comparing scores from schemes with different criteria can confuse a change in the benchmark with a change in performance.
Pulsed field gradient NMR 2026-11-05
Pulsed field gradient NMR, or PFG NMR, uses timed magnetic field gradients to encode position in the phase of a nuclear magnetic resonance signal. An appropriately refocused pair of encoding episodes cancels the position-dependent phase for stationary spins. Motion between the episodes can prevent cancellation, making the echo sensitive to displacement.
For an original narrow-pulse derivation in one spatial direction, let each effective encoding pulse have gradient magnitude and short duration , and write , where is the angular gyromagnetic ratio. Neglect displacement during each pulse. For positions separated by an observation interval , a choice of encoding sign gives . The normalized signal is the ensemble average of .
For free diffusion with diffusion coefficient , displacement has a zero-mean normal distribution of variance . Its characteristic function givesFor invented values , and , the attenuation is . Finite pulse duration, restricted motion and other sequence details require their corresponding model rather than automatic use of this narrow-pulse result.
The 1965 experiment of E. O. Stejskal and J. E. Tanner combined a derivation for time-dependent gradients with experimental tests. Reducing gradients during radio-frequency pulses and echo detection relaxed practical limitations of a continuously strong gradient; the authors measured slow diffusion in dry glycerol.[ref]
Spin Diffusion Measurements: Spin Echoes in the Presence of a Time-Dependent Field Gradient by E. O. Stejskal and J. E. Tanner. Page288, Abstract; Theory and experimental tests
PFG NMR can also use a stimulated echo for storage between encoding periods. Distinguishing motion-induced attenuation from other signal losses is part of the measurement, not a consequence of the method's name alone.
Quadratic energy loss from centred Gaussian gate timing error is the leading decrease in collected energy when a symmetric gate is displaced slightly from the centre of a Gaussian pulse. At the optimum the first-order change vanishes, but the second-order change is negative. The Gaussian pulse time-gate energy fraction gives the exact displaced-gate integral; here the local loss and its effect on jitter are derived explicitly.
Let be the centred energy-normalized pulse profile, let , and setThe Leibniz integral rule gives . Since is even and ,The Taylor expansion therefore yieldsThis is an absolute fraction of the full pulse energy. The relative loss compared with the centred gate is instead .
For a gate as wide as the power FWHM , and . An offset gives leading loss , or about percent of the centred gate's collected energy. Neither the full pulse energy nor the gate width has changed.
A zero mean offset does not eliminate this loss. For a sufficiently narrow random offset with mean and variance , the leading expectation isIf has a centred normal distribution, the quadratic model also gives . At , the leading mean loss is and this standard deviation is . These formulas neglect fourth-order effects; they are local approximations, not bounds for large timing excursions.
Exact Gaussian gate capture and its local quadratic loss: a zero first derivative does not remove timing sensitivity.
Details. An apparatus check scans the gate delay about its optimum and compares the measured decrease with the model, while monitoring gate width and the pulse. A gate on the detector's electrical output need not act directly on optical power: the detector impulse response and readout convolution must be included. NIST's impulse-response measurements illustrate that distinction.[ref]
Quantile endpoint at probability one Created 2026-10-14 Updated 2026-11-05
The quantile function is finite at every probability strictly between zero and one, but need not be finite at one. For any unbounded-above normal distribution, for every real , so its generalized inverse at one is . Inverse transform sampling still produces a real random variable almost surely because a continuous uniform distribution puts zero mass at that endpoint. A real-valued implementation may discard endpoint draws or assign an arbitrary finite value there without changing the sampled law.
Quantum variational principle Created 2026-10-03 Updated 2026-10-14
The quantum variational principle estimates the lowest energy by choosing a trial wavefunction and computing its expectation value, without solving the Schrödinger equation exactly. Minimizing over a manageable family of trial wavefunctions gives an upper bound; enlarging that family can improve the estimate.
Let be a self-adjoint operator bounded below, and put . For in its operator domain ,Trial states may also lie in the form domain , where . Then use the quadratic form of a positive quantum Hamiltonian shifted by :This extends the expectation value beyond without requiring to exist. Equality for a normalized state holds exactly in the energy eigenspace at , if that eigenvalue exists. [ref]
https://www.mat.univie.ac.at/~gerald/ftp/book-schroe/schroe.pdf. Gerald Teschl, Mathematical Methods in Quantum Mechanics, Theorem 2.19 and equations (3.50)–(3.51)
.A short proof makes the direction of the bound memorable. If has a complete orthonormal basis of energy eigenstates , writeThe expectation value is a weighted average of allowed energies:Equality requires every occupied energy eigenstate to have energy . [ref]
https://live.ocw.mit.edu/courses/8-05-quantum-physics-ii-fall-2013/61bc31b8d8bf0680c322733910a71aa0_MIT8_05F13_Chap_01.pdf. Barton Zwiebach, MIT 8.05 Wave Mechanics, section 5, Variational Principle
. With continuous spectrum, the spectral theorem replaces the sum by an integral against the scalar spectral measure; the same nonnegative-difference argument works.For an unnormalized trial wavefunction, divide by its squared norm: is the form version of the Rayleigh quotient. The quantum variational principle supplies an upper bound on ; it supplies neither a lower bound on nor, by itself, a guarantee that a bound state exists.
To see the method at work, choose the real Gaussian wave packetThe Gaussian integral gives wavefunction normalization and the two useful position averages:Thus its position probability density is a normal distribution with standard deviation : large means narrow confinement. In units , integration by parts and give the kinetic energyThese smooth, rapidly decaying trial wavefunctions belong to the operator domains of both polynomial Quantum Hamiltonians below. Narrowing the Gaussian wave packet increases its kinetic energy; widening it increases the potential energy in either confining potential.
For the quantum harmonic oscillator, take :The trial values at are respectively . The derivative changes sign at , where kinetic energy and potential energy are both . Here the trial family contains an exact ground state:The zero-point energy remains positive even at the best width. For comparison, removing the potential leaves as : the free Quantum Hamiltonian on the line has spectral bottom zero but no normalizable zero-energy energy eigenstate. A sequence of ever broader trial wavefunctions need not approach an actual bound state.
Gaussian trial widths and the kinetic–potential energy tradeoff
. Normalized Gaussian wave packets with have different position probability densities. The bars show the competing kinetic energy and harmonic potential energy ; their sum is smallest at . Details.For a pure quartic oscillator, keep and set the quartic coefficient to one:The trial values at are respectively . Setting the derivative to zero givesThe sign of this derivative changes from negative to positive at the optimum, giving the smallest energy within this Gaussian wave packet family:At this width, and , so , as expected from the power-law quantum virial theorem for a scale-optimized trial wavefunction. The number is an upper estimate of the ground-state energy, not an asserted exact energy eigenvalue. Indeed,depends on for every , so no member of this trial family is an energy eigenstate. Nevertheless, a single width parameter already turns a difficult Schrödinger equation into a useful, explicit energy estimate.
Raman captured-area fraction 2026-11-05
The Raman captured-area fraction is the part of a specified complete spectral band's area that lies inside a chosen integration interval. For a nonnegative isolated background-corrected band with finite positive complete area,It lies between zero and one. It quantifies one mechanism of Raman integration-window bias: changing the center or width changes the fraction even when the complete area stays fixed. Background errors or overlapping bands can prevent measured intensity from representing the assumed isolated band.
For a Gaussian function centered at with width parameter , substitution of giveswhere is the cumulative distribution function of the standard normal distribution. A centered window with half-width captures . Half-widths of one, two and three capture approximately68.27%,95.45% and99.73% of the complete Gaussian area. These percentages refer to this line shape, not every Raman spectrum.
For an original example, a window extending4cm⁻¹ either side of the fitted center captures about0.95450 when , but about0.81758 when . Even if the height decreases so the complete area is unchanged, the finite-window area falls. If one incorrectly uses the first fraction to correct the broader band, the estimated complete area is only of the true value.
Raman band-area extrapolation uses a justified fraction to estimate complete area as . For example, measured area10 in a centered two-width Gaussian window implies about10.477 in the same area units. The correction depends on the model and its parameters; it is not extra measured signal. At fixed , an absolute area error is multiplied by , while uncertainty in supplies another contribution. Large extrapolations from a narrow interval therefore need especially convincing evidence about the unobserved tails.
Collecting and retaining the full local spectra supports checks of center, width, background and neighboring bands instead of interpreting a single intensity summary alone. Primary manufacturer guidance makes this distinction between simple spectral-intensity images and fitted band parameters explicit.[ref]
What Raman images can reveal by Renishaw. “How do we collect Raman images?”; “How can I analyse Raman images?”, background and fitted spectral parameters
Raman integration-window bias 2026-11-05
Raman integration-window bias is a systematic difference between the spectral-band area one intends to compare and the area captured by a selected finite Raman shift interval. It can create contrast in a Raman band-area map when the complete band area is unchanged: shifting or broadening redistributes intensity across the fixed interval's edges. The target must be stated clearly, because an area over that interval can be correctly measured even when it is a biased proxy for the complete band.
For an original calculation, take an isolated background-corrected Gaussian function bandIts complete area is . Integrating only from to gives the Raman captured-area fractionwhere is the cumulative distribution function of the standard normal distribution. The substitution gives this expression directly. The complete area depends on height and width; the captured fraction also depends on center position relative to the window.
Suppose and the fixed window runs from996 to1004cm⁻¹. A center at1000cm⁻¹ places its edges at two standard widths on each side, giving . Shift the center to1004cm⁻¹ without changing height or width: now . The measured fixed-window area falls to about52.38% of its previous value although the complete band area is identical. This contrast is therefore insufficient evidence for a decrease in the amount of material.
A window that follows the fitted center can keep the captured fraction constant for this ideal fixed-width model. Broadening, neighboring bands, errors in a Raman peak-position map and an imperfect background still matter; a larger window can admit unwanted signal. Manufacturer documentation explicitly distinguishes simple intensity images from fitted-band analysis and warns that broad background can contaminate intensity contrast.[ref]
What Raman images can reveal by Renishaw. “How can I analyse Raman images?”, single-frequency intensity, broad background and curve-fit band parameters
Reporting the actual interval, background treatment and whether the interval moves with the band makes the Raman band-area map scientifically interpretable. Rank-one covariance spike Created 2026-10-06 Updated 2026-11-05
For a unit vector and , a rank-one covariance spike is the covariance matrix . Its eigenvalue is in the direction and one on the orthogonal complement. It arises by adding an independent scalar normal distribution signal to a standard Gaussian random vector.
Rare-outlier three-point distribution 2026-10-14
For , put probability mass at zero and at each of and . This symmetric probability distribution has mean zero. Its second and fourth moments are both , because the nonzero outcomes have square and fourth power equal to one:For the kurtosis is ; for it is . In the latter case of the probability mass is at zero, yet all of the fourth central moment comes from the remaining . The standard score takes values and , with variance one. Each nonzero outcome contributesto kurtosis. At , the two rare outcomes are ten standard deviations from the mean, and their contributions are .
Every member has finite support of a probability distribution, so all its moments are finite. Nevertheless, kurtosis becomes arbitrarily large as approaches zero: high kurtosis does not require an unbounded support of a probability distribution or a heavy-tailed distribution. At the variance vanishes and kurtosis is undefined. At this discrete probability distribution even has the normal distribution value and excess kurtosis zero, so those values do not establish a normal distribution.
Rayleigh radius from Gaussian components 2026-10-18
The joint probability density of isIn polar coordinates, , , and the absolute Jacobian determinant is . Thus the joint density of radius and angle factors asIntegrating over angle gives the Rayleigh distribution. The same factorization shows that the angle has uniform distribution and is independent of the radius. This is a short change of variables argument: an annulus of width has area to first order.
The assumptions are substantive. Unequal component variances give elliptical rather than circular density contours; a nonzero mean displaces the center. Even zero covariance and equal normal distributions do not suffice without independence. For an original counterexample, let and let be an independent Rademacher random variable. Put . Then has the same normal distribution as , and . Nevertheless, all points lie on or , and has densitywhich is positive at zero, unlike the Rayleigh distribution. This pair is not a Gaussian random vector.
R. B. Nottingham 2026-11-05
R. B. Nottingham coauthored The Folded Normal Distribution by F. C. Leone and others with F. C. Leone and L. S. Nelson, published in Technometrics in 1961. Their paper addressed measurements whose algebraic signs were not recorded. Even if a signed quantity follows a normal distribution, its recorded magnitude follows a different probability distribution.[ref]
The Folded Normal Distribution by F. C. Leone and others. Primary byline, abstract and introduction: unsigned measurements and parameter estimation
The practical issue is information loss during observation. If and the instrument reports , then and produce the same folded normal distribution. Thus magnitudes alone cannot distinguish an offset to one side of a reference from an equally large offset to the other side. This conclusion follows from the observation rule, not from the quality of a numerical fitting method.
An illustrative inspection records distances from a reference line but omits which side each point lies on. A pair of signed readings and becomes two identical magnitudes . Retaining the original signs would preserve a distinction that no later analysis of those two magnitudes can reconstruct. The paper's contribution is to treat this altered observation distribution explicitly rather than fit an ordinary signed normal distribution to magnitude data.
The cited work is jointly authored; the primary byline does not assign individual responsibility for particular formulas or tables. The printed initials are retained here without attaching unrelated biographical records to this author.
Recovery of a continuous Brownian integrand from short increments Created 2026-10-05 Updated 2026-10-06
For a bounded continuous adapted process and a Brownian motion , converges in probability to as . The Itô isometry makes the error numerator small relative to ; the normal distribution of the denominator controls its chance of being too close to zero.
Regression toward the mean Created 2026-10-03 Updated 2026-10-14
Regression toward the mean is a tendency for measurements selected because they are extreme to be less extreme on repetition. A large first reading can combine a persistent signal with unusually positive measurement error; a fresh error need not repeat that contribution. This can change the selected group's mean even when the underlying signal stays fixed and no intervention occurs.
The claim concerns a conditional expectation, not a rule that every individual moves toward the population mean. Some second readings are more extreme. Nor does an observed change establish a causal effect: selection alone can produce it. The following original normal distribution model makes both distinctions quantitative.
Reparameterization gradient Created 2026-10-06 Updated 2026-10-25
A reparameterization gradient differentiates an expected value by representing the simulated random variable as a differentiable transformation of parameter-independent noise. For , the identity requires justified differentiation under the integral sign. For a normal distribution, use with drawn from the standard normal distribution.
Repeated-measurement latent signal model 2026-10-14
Imagine an instrument measuring the same fixed quantity twice on each unit. Across units the persistent signal is a latent variable; the two measurement errors are fresh, mutually independent random variables, also independent of . Assume they have mean zero and common finite variance , while has mean and variance . ThenThus the Pearson correlation coefficient is when . Both measurements retain the same marginal mean and variance. Repetition does not make the entire population collapse toward ; it changes what we expect after selecting on .
For , further assuming normal distributions for and both measurement errors gives the conditional expectationSince , this predicted reading is closer to than is. Independence and zero mean errors alone do not justify this linear conditional expectation: without the normal distribution assumption the same expression is the linear least-squares projection, and the actual conditional expectation can be nonlinear. Correlated errors, a changing signal, or different error variances require their own model; a noisy reading by itself supplies no universal shrinkage factor.
Residual constraints from a bounded response 2026-10-25
If observed responses satisfy , their regression residuals necessarily obey the displayed inequalities. Subtracting the corresponding fitted values proves both bounds, without a normal distribution or a fitted-model assumption. On a residual-versus-fitted plot, all points must lie between two parallel lines of slope . This is also a consistency check when are merely the reported observed minimum and maximum, rather than theoretical response limits. A bounded response does not itself prove that a straight conditional mean is inappropriate.
Robert Litterman 2026-11-05
Robert Litterman developed the Black-Litterman model with Fischer Black. Their 1992 paper makes equilibrium expected returns a neutral starting point for portfolio modeling, then incorporates views that can differ in their confidence. The contribution addresses how uncertain inputs enter modern portfolio theory, rather than treating every proposed expected return as equally reliable.[ref]
Global Portfolio Optimization by Fischer Black and Robert Litterman. Publisher abstract: neutral equilibrium reference and views with differing confidence
An elementary Bayesian inference analogy illustrates the role of uncertainty. For an unknown quantity with normal distribution prior and an independent observation error giving , Gaussian conditional expectation gives the posterior meanThe more uncertain input receives less weight. This scalar example explains a modeling principle; the full Black-Litterman model treats related expected returns jointly, with a covariance matrix describing their uncertainty.
Robert Litterman with Fischer Black in 1994
. Historical photograph hosted and identified by Goldman Sachs, linked externally. Source. Samplewise variance upper bound 2026-11-05
A samplewise variance upper bound is a nonnegative statistic satisfying , where is the fixed sampling variance at parameter . It controls each realized value outside a probability-zero exceptional set. This implies the expected-value condition for a conservative variance estimator when the expectation exists; the converse fails, as the parent article's counterexample shows. The primary sampling-design definition only asserts the weaker expectation condition.[ref]
Suppose is an unbiased estimator with finite variance, and let . Then Chebyshev's inequality and the samplewise comparison giveNo independence between and the centre is needed. If , unbiasedness makes the centre equal to almost surely, so coverage still holds. This proves confidence bounds from a samplewise variance bound; it does not justify replacing by a normal distribution critical value.
For an original bounded-data example, take independent and identically distributed measurements known to lie in , and estimate their population mean by the sample mean. With , the inequality givesThe deterministic choice is a samplewise variance upper bound for every such population. Thus covers with probability at least . The radius shows the price of distribution-free reasoning; the estimate may be much wider than a model-specific confidence interval.
Sara van de Geer 2026-11-05
Sara van de Geer develops the mathematical theory of high-dimensional statistics. Her Statistical Theory for High-Dimensional Models by Sara van de Geer studies prediction bounds for the Lasso and Square-root Lasso, and inference after correcting shrinkage bias.[ref]
The distinction is useful: an estimator that predicts well may still bias an individual coefficient toward zero. A de-sparsified estimator adds a correction whose leading error can have an approximately normal distribution. Bounds on the remaining error, together with sparsity and design assumptions, then justify confidence intervals. The same notes connect generic chaining and concentration inequalities with control of random-process suprema.[ref]




