If , a stationary reversible Markov chain satisfiesConsequently the variance of a sum of consecutive values is at most times the stationary variance of one value. This bounds the Markov chain Monte Carlo asymptotic variance uniformly over square-integrable functions.
Stationary increments Created 2026-09-24 Updated 2026-11-05
A stochastic process has Stationary increments when the probability distribution of depends only on the duration , whenever both times belong to the index set. The clock origin then does not change the increment law. For Brownian motion, that law is ; for a rate- Poisson process, it is a Poisson distribution with expected value . Neither process itself is a stationary process: their variances grow with . Stationary increments and a stationary process describe different properties.
Stationary process Created 2026-09-28 Updated 2026-10-14
A stationary process, in the strict sense, has the same joint probability distribution after every common shift of its time indices. More precisely, for every positive integer , every choice of times , and every integer shift ,Here equality means equality of probability distributions, not equality of realized observations. A sample path can fluctuate, stay positive for a while, or show accidental runs while its generating law remains stationary. Strong white noise is a basic example: shifting time preserves every joint law of its independent and identically distributed random variables.
The distinction between strict stationarity and weak stationarity is useful. The former preserves all finite joint laws; the latter preserves only the finite mean and covariance structure. Strict stationarity with finite second moments implies weak stationarity, but strict stationarity itself does not require finite variance. A Gaussian process with weak stationarity is also strictly stationary, because each of its finite-dimensional distributions, a joint normal distribution, is determined by its mean vector and covariance matrix.
Stationary trapped Brownian displacement Created 2026-10-20 Updated 2026-11-05
For overdamped Brownian motion in potential , the relaxation time is and the equilibrium covariance function is . Expanding the squared displacement between two equilibrium samples gives the title expression. Initializing the particle exactly at the centre instead gives variance . Both are valid but concern different initial ensembles; their saturation values differ by a factor of two. These formulas neglect inertia and require times long compared with momentum relaxation.
Statistical decision theory Created 2026-09-24 Updated 2026-11-05
Statistical decision theory compares procedures by the losses their decisions incur under a statistical model. Specify observation laws , an action space and a loss function . A decision rule maps data to an action, and its risk function isFor estimating a normal mean from observations of known variance , the sample mean has squared-error loss risk . Under zero-one loss, the same framework describes classification in statistical learning and statistical hypothesis testing.
Different criteria compare the whole risk function differently. A Bayes decision rule minimizes risk averaged over a prior distribution. A minimax decision rule minimizes the worst-case risk. An admissible decision rule has no competitor with no greater risk everywhere and smaller risk somewhere. These criteria require the model, allowed actions and loss to be specified; accuracy alone does not define what mistakes cost.
Statistical fluctuation Created 2026-10-25 Updated 2026-11-05
A Statistical fluctuation is variation between repeated random outcomes or samples. Its scale can be quantified by variance or standard deviation. For independent signal counts with a Poisson mean proportional to deposited energy, the absolute fluctuation grows as the square root of energy and its relative size decreases as the inverse square root.
Statistical physics Created 2026-09-24 Updated 2026-10-13
Statistical physics explains collective behavior using probability distributions over microscopic possibilities. Its central surprise is that incomplete microscopic knowledge can still produce precise predictions for internal energy, entropy and heat capacity. The bridge is statistical mechanics: specify the available microstates, assign their probabilities, then compute the expected values of measured quantities.
A two-level system already displays the idea. Warming it makes its higher energy more likely; its heat capacity first rises, then falls as the two probabilities approach equality. With two distinguishable units, merely counting microstates explains why the middle energy has twice the state degeneracy. These small examples expose the machinery before the much larger sums needed for a photon gas or an Ising model.
The appeal is economical: the same microscopic energies and probabilities explain both an average reading and its variance. A measurement's spread therefore becomes useful physical information rather than an inconvenience to discard.
For an equally populated ideal pair of angular momentum eigenstates, the magnetic moment component takes two values, , and the displacements at fixed velocity are . Compare a specified classical alternative: every chemical atom carries a permanent magnetic moment of fixed magnitude , with independently isotropic initial orientations and no dissipative alignment. If , the sphere's area element is , so has a uniform distribution on . With precession preserving this component in the ideal geometry, the displacement fills the interval continuously. A pair of bands therefore tests allowed components rather than merely proving that silver is magnetic. Other classical preparation assumptions would be different models and must be stated.
To include finite instrumental width, choose a normalized Gaussian distributionwith standard deviation , representing a combined field-off spatial width. At fixed velocity the model probability density functions areEach has unit integral and zero mean; the different shapes, rather than a different average displacement, distinguish them. The discrete model has variance , whereas the classical model has variance . Their common mean also explains why the raw average landing position is insufficient to identify the internal quantum state.
Now choose a uniform distribution of forward velocities from to , explicitly as an example rather than the oven's actual distribution. Average either conditional probability density function over that interval. The positive-channel displacement ranges from to , since slower chemical atoms spend longer accelerating and drifting. The two components broaden into bands even though their internal magnetic moment values remain discrete. The corresponding classical profile continues to populate the center. The displayed strip darkness is proportional to predicted deposited number density, with equal total arrivals; it is not a photograph or a calibrated optical response of a developed film.
Discrete magnetic components and continuous classical deflections
. Original model predictions with the chosen flight parameters, comparing one forward speed with a finite speed interval. Both alternatives include the same Gaussian spatial width. Synthetic deposition strips use a common density scale; none of the curves or strips is measured data. Details.Increasing the spatial width until it is comparable with can hide the two maxima. At fixed velocity, differentiating the two-state probability density function twice gives : the central minimum becomes a central maximum when . Conversely, a broad distribution of velocities need not create a new internal component: it spreads each existing component through the dependence. Further instrumental limits include collisions, aperture clipping of strongly deflected trajectories, asymmetric pole geometry, and nonlinear film development. Field-off widths and a mapped magnetic field gradient are needed to separate such effects. For a modern counting experiment, channel calibration and efficiency corrections belong to Stern-Gerlach expectation estimation; spatial separation alone does not supply those statistics.
Steven L. Heston 2026-11-05
Steven L. Heston introduced the Heston model, a stochastic volatility model in which the asset price and its instantaneous variance evolve together. The variance follows a square-root diffusion, while correlation between the price and variance innovations allows asymmetric option-price patterns.[ref]
The model is useful because richer uncertainty does not remove all analytical structure. An affine expression for the characteristic function of the log price allows Fourier inversion to recover the probabilities used in European option valuation. The resulting pricing formula includes numerical integration; “closed-form” here does not mean that every price is an elementary expression without a numerical step.[ref]
Institutional profile: www.rhsmith.umd.edu/directory/steve-heston
Portrait of Steven L. Heston
. Portrait selected from the matching University of Maryland institutional profile. Source. Stochastic convolution 2026-11-05
A stochastic convolution propagates random forcing through a linear evolution. Let be a strongly continuous semigroup on a Hilbert space , let be a Cylindrical Wiener process on a Hilbert space , and let be a bounded linear operator. The expressionis an Itô integral whenever its integrand is square integrable in the Hilbert-Schmidt operator norm. In particular,This Itô isometry tests whether the accumulated forcing is actually -valued. The factor evolves the contribution arriving at time through the remaining time . For with deterministic initial value , the mild solution is .[ref]
An Introduction to Stochastic PDEs by Martin Hairer. §6, Definition 6.5, equation 6.3 and Remark 6.6; §4.4, Hilbert-space stochastic integration
For , and real Brownian motion , with ,This is the zero-initial-value Ornstein-Uhlenbeck process: damping limits the variance despite continued forcing. With , the limit is instead. These formulas follow by integrating in the Itô isometry.
The word convolution describes the dependence on the time difference . It does not mean an ordinary integral against a differentiable noise function; Brownian motion requires a stochastic integral.
Stochastic cooling variance model 2026-10-18
Let be a centered scalar beam-deviation coordinate in a deliberately controlled toy model. Let the readout be , where the centered measurement noise has variance and is independent of and of earlier noise. Stipulate a correctly timed correction of strength : . This scalar observation is a teaching surrogate, not a claim that real cooling measures each particle. From variance additivity for independent random variables,For fixed current and , differentiate the next-step variance with respect to feedback gain:The unique one-step minimizing feedback gain and resulting variance areThe noisier the readout relative to the current spread, the smaller the useful correction. The exact one-step cooling condition follows by subtraction:Thus excessive positive feedback gain heats this model. The threshold depends on the current variance: a gain that initially cools can later heat if the beam has become narrower or the noise has grown.
For chosen dimensionless and , gives . Taking instead gives , larger than the initial spread. Holding fixed givesso the variance approaches the nonzero noise floor . More generally, for the fixed-gain limit is . This long-run stability condition differs from the condition for cooling the current beam on the very next step.
If the toy feedback gain is recomputed after every step and , the reciprocal recurrence is even simpler:With the chosen inputs, . This ideal model has no additional diffusion, finite observation time or lost correlation; its indefinitely falling spread is not a prediction for an operating storage ring.
To expose the missing physics mathematically, replace by an arbitrary centered sample signal . Defining and givesUnwanted contributions from other particles or electronics can enlarge without increasing useful ; poor timing or lost correlation can reduce or even reverse its sign. This is why real stochastic cooling needs measured beam fluctuations, appropriate beam mixing and correctly phased feedback control, rather than an imagined catalogue of individual particle positions.
Stochastic gradient descent Created 2026-09-24 Updated 2026-11-05
Stochastic gradient descent updates an optimization parameter using a random estimate of the gradient:For a finite-sum objective , selecting an index uniformly and taking gives an unbiased stochastic gradient. Averaging several selected gradients gives a mini-batch gradient. The saving is fewer component evaluations per update, while the randomness adds variance. Unbiasedness or a controlled bias is an assumption, rather than an automatic property of any approximate gradient.[ref]
https://leon.bottou.org/publications/pdf/sirev-2018.pdf. Bottou, Curtis and Nocedal, Optimization Methods for Large-Scale Machine Learning, §4, stochastic-gradient formulation and noise/stepsize trade-offs
A simple noise example is and , where independent centered noises have variance . With a constant step size ,Its mean tends to zero, but its limiting variance is : solve . Thus constant steps can leave a stochastic gradient noise floor even for this strictly convex quadratic. Smaller or diminishing steps change the balance between progress and noise; a nonconvex stationarity estimate must not be read as guaranteed convergence to a global minimum.
Stochastic gradient noise floor 2026-10-25
A residual accuracy level in a bound for stochastic gradient descent with a fixed step size and nonvanishing gradient noise. For a convex function with an -Lipschitz gradient, step size and conditional total variance at most , typical bounds have floors for averaged squared gradient norms, for averaged gradient norms, and for the best expected objective function gap. Such floors are upper bounds depending on the analysis and assumptions, rather than claims that these exact values are attained or that every last iterate converges.
Target: gradient-descent.bigb, under Stochastic gradient descent. Stochastic gradient second-moment identity 2026-10-25
For an unbiased stochastic gradient at with finite conditional second moment, expand in the squared Euclidean norm. The cross term has zero conditional expectation, which proves the displayed identity. It separates the squared true gradient from the total conditional variance of its random estimate.
Target: gradient-descent.bigb, under Stochastic gradient descent. Stochastic heating Created 2026-10-24 Updated 2026-11-05
Stochastic heating increases random particle kinetic energy through decorrelating electromagnetic kicks. A simple one-dimensional model makes the normalization explicit. Let be a velocity distribution with number density , and suppose constant-coefficient velocity-space diffusion givesThe coefficient has dimensions of squared velocity per time, and a freely diffusing component has variance . For sufficiently decaying tails, two applications of integration by parts yieldThe first velocity moment is unchanged, so this is growth of random motion rather than acceleration of the mean flow. In three independent isotropic components with coefficient in each, the heating rate per volume is instead . More generally each diagonal component contributes its own diffusion coefficient.
This idealization describes externally supplied, rapidly decorrelating forcing; its energy budget is not that of an isolated particle population. Screening, resonant self-consistent fields, collisions and velocity-dependent diffusion can change the result. In other plasma settings, stochastic heating also describes irregular orbits driven by spatial structure or overlapping resonances; that mechanism must be specified separately.
Stochastic process Created 2026-09-24 Updated 2026-10-14
A stochastic process describes uncertainty about a whole history: not just the probability distribution of a measurement now, but how measurements at different times fit together. It is a family of random variables on one probability space, with index set . Time is the usual index: for successive observations or for continuous time. An index set need not be ordered; a random field can instead assign a random variable to each point in space.
There are two complementary views. Fix a time and is a random variable; fix a sample space outcome and is one sample path. A model specifies the probability distribution of these histories. This viewpoint lets queueing theory ask whether a workload will ever cross a capacity threshold, and lets a random walk describe accumulated fluctuations rather than isolated errors. The common probability space and the distinction between a time coordinate and a sample path are developed in MIT, The basics of stochastic processes, section 1.
A concrete construction starts with a Bernoulli process of independent fair bits . Keeping records the next trial; adding the signs gives a simple random walk on the integer line, and . Every increment of this random walk is , yet the positions are dependent: contains all the previous increments. Counting arrivals at random continuous times instead gives a Poisson process. These examples separate three questions: what values are possible, how time is indexed, and how the values depend on one another.
For the fair simple random walk on the integer line, the expected value is and the variance is . At time four, returning to zero requires exactly two positive steps, so . A rate-two Poisson process instead has . A single realized sample path does not establish either probability: they follow from the specified joint probability distributions.
The most revealing feature is often dependence. Two stochastic processes can have identical marginal distributions at every time while one forgets everything at the next step and the other repeats its first value forever. Their finite-dimensional distributions and covariance functions distinguish them. That extra structure is what makes a stochastic process useful for prediction.
Stochastic volatility model Created 2026-10-03 Updated 2026-11-05
A stochastic volatility model gives an asset a separate random volatility or variance process. For example, a stock can have pricing dynamics , with random , instead of the constant variance of the Black-Scholes model. The Heston model uses a CIR model for ; an Ornstein-Uhlenbeck process can instead model a signed volatility whose square is the variance.
The extra state lets current uncertainty affect future uncertainty, and correlated asset and variance shocks change the distribution of future returns and option values. This differs from a local volatility model, where the coefficient is a deterministic function of time and current spot alone.
With two independent sources of uncertainty and only a stock and bank account, the variance risk is generally not fully hedgeable. A physical variance model then does not uniquely determine its pricing dynamics: a volatility-risk premium or additional traded instruments must be specified. Heston's original model makes the joint asset/variance system tractable through an affine characteristic-function calculation.[ref]
A delivery has value , where are the mean and variance of under the stock-numeraire measure in the Black-Scholes model, and . Split the normal integral at and complete the square in times its normal probability density.
Stop loss variance minimization principle 2026-10-06
Among retained payouts with and the same expected value as , aggregate stop loss reinsurance minimizes the variance. Pointwise , and subtracting the identical squared distance of their common expected value from proves the claim. Equality requires equal payouts almost surely.
Stratified log-rank statistic 2026-10-07
Compute observed-minus-expected event contributions within each stratum's risk set, then add them across independent strata. Under the null, the corresponding conditional variances add. Standardizing gives an approximate normal distribution test, or its square gives a one-degree-of-freedom chi-squared distribution test. The comparison requires the usual exchangeability and independent censoring assumptions within strata.


