Independent loss alone leaves normalized factorial moments invariant: if every input photon is retained with probability , binomial thinning conditional on a total gives and . Both powers cancel from . Fewer coincidences still mean poorer statistical precision.
Now let each channel contain a signal rate and an independent Poisson process background rate , independent also of the signal. Put . The pair density is . Dividing by proves the displayed formula, for the same bin and timing response. For chosen equal signal fractions , ideal signal values become . Background pulls both dips and peaks toward one.
This calculation assumes additive, mutually independent background counts. Scattered coherent light that interferes with the signal, correlated electronic noise, or recovery effects require a different model. Correcting a dip without independently estimating the backgrounds cannot establish photon antibunching.
In 1998, Jian-Wei Pan, Dik Bouwmeester, Harald Weinfurter and Anton Zeilinger created two photon pairs by two passages of a pulsed pump through a crystal using spontaneous parametric down-conversion. Interfering the inner photons at a nonpolarizing beam splitter implemented a partial Bell-basis measurement. Narrow optical filters and spatial filtering removed distinguishing information; changing the outer polarizer bases checked the conditional correlations. Fourfold detections selected the entanglement swapping events. The reported interference supported conditional quantum entanglement, but the paper reserved a statistically significant Bell inequality violation for improved visibility; it was not a demonstration closing the detection loophole.[ref]
The ideal calculation makes the distinction between conditional and unconditional quantum states explicit. Start with spin singlet states , where and denote orthogonal photon polarizations. Taking the inner product with gives : expanding the four terms leaves precisely the two antisymmetric outer terms. That Bell-basis measurement outcome has probability and leaves the outer pair in a spin singlet state. Ignoring the inner outcome instead leaves the outer density matrix , a maximally mixed state. A conditional preparation flag is useful for later Bell tests only with the timing and setting conditions in setting-independent heralding preserves Bell locality.
An explicit original ideal meter model explains how photon number survives while field quantum coherence changes. Let be the number operator and prepare an atomic qubit in . The dispersive unitary operator
commutes with . For a field quantum superposition , it produces , where . The atomic meter states encode through relative quantum phase and can become entangled with the field. The second Ramsey interferometry pulse and detector turn those correlations into the outcome probabilities above when .
Tracing out the atom multiplies a field density matrix entry by . Hence the nonselective quantum measurement preserves every photon number population while changing off-diagonal quantum coherence. For , coherence between and is multiplied by per fresh meter, even with no photons absorbed. Field quantum phase information can therefore be disturbed by an energy-preserving probe. No universal self-adjoint phase operator or naive canonical number–phase commutator is needed for this conclusion.
One possible finite-contrast quantum instrument makes the distinction sharper. Define diagonal Kraus operators and . After their ideal readout, flip the reported label with probability . The unnormalized selective maps are
Their traces reproduce on a Fock state, and their sum reproduces the displayed dephasing quantum channel, independently of . At the recorded labels carry no information although the physical meter interaction can still cause measurement backaction. This is an illustrative instrument, not a reconstruction of all laboratory imperfections; equal likelihoods alone do not determine the disturbance.
Photon-recoil lag correlation Created 2026-10-18 Updated 2026-10-20
A photon-recoil lag correlation asks how often a photon record occurs at a specified time displacement from an electron record. Let be observed separate-channel rates, the rate of detected genuine pairs before a timing cut, and their normalized timing probability density function. For an original dilute, stationary model with observation time , neglecting end effects and detector dead time, the expected pair density is
This counts all pairs of records, rather than at most one detector trigger per electron. It uses independent unrelated detections for the flat term and an added same-event contribution. A normal distribution for timing spread is a chosen approximation, not a claim about the historical instrument. Intrinsic emission delay, counter response and readout uncertainty can all affect the observed probability density function.
Define the accepted lag interval by : is its full width. Integrating the flat baseline gives the existing accidental coincidence expectation . The additional area is . Moving a window of the same full width away from the peak tests the baseline under the model's stationarity assumption. It does not remove correlated instrumental backgrounds or changing rates automatically.
Figure 1.
Lag-dependent pairs and a joint probability distribution carry information absent from singles totals
. Original chosen models, not historical counts or fitted data. Left: a normalized normal distribution of genuine-pair lags with standard deviation is added to an independent-pair baseline. Shaded intervals have full width , centered at zero and . Right: a separate exact ideal-bin model ignores timing spread; the two populations have identical singles expectations but different joint expectations. Details.
For the lag model, choose , , , and timing standard deviation . A full width gives expected accidental coincidences and accepted genuine pairs, using the error function. The total is approximately , whereas the displaced interval from to has approximately expected pairs. A finite sample fluctuates around these expectations; these values are not measured counts or a significance calculation.
The right panel isolates the information lost by retaining singles totals alone. In a separate idealized model, stipulate independent bins, with Bernoulli distribution indicators for electron and photon detection in a bin. Set and . Independence predicts , hence one expected double detection. Alternatively choose
These nonnegative probabilities sum to one and preserve both marginal probabilities, yet predict double detections in expectation. Their covariance is . This is an exact joint probability distribution illustrating pairing; it does not model the timing spread in the left panel or reconstruct either historical hypothesis quantitatively. Its lesson is why two singles totals cannot decide whether the observed events share a cause.
For specified exposure periods with constant death hazard functions , the integrated hazard through the endpoint is . The displayed event probability follows from the survivor function identity. For different latent exposure schedules mixed with probabilities , average the schedule-specific probabilities, , rather than exponentiating the average integrated hazard. When the hazards are small, gives a person-time approximation. A comparison must specify the entire custody or treatment schedule; later changes to exposure cannot be inferred from an endpoint rate alone.
Pierre de Fermat Created 2026-10-13 Updated 2026-11-05
Pierre de Fermat (1601–1665) was a lawyer and government official whose mathematical work helped shape number theory, early calculus and probability. His 1654 correspondence with Blaise Pascal on games of chance helped turn questions about fair division of stakes into systematic probability calculations.[ref]
Fermat's little theorem states that for a prime number and an integer ; it gives a reusable arithmetic relation, not a guarantee that every number satisfying one modular test is prime. Fermat's last theorem is the separate assertion that has no positive-integer solution for . The two statements illustrate the range of his arithmetic questions and must not be conflated with one another or with a proof of the latter assertion by Fermat himself.
Figure 1.
Portrait of Pierre de Fermat
. Rolland Lefebvre (1608–1677); portrait in Musée des Beaux-Arts de Narbonne; reproduction via source listed on Commons. Public domain (life-plus-100, pre-1931 U.S. publication) and faithful reproduction / PD-Art. Source and usage information on the linked Wikimedia file page. Source.
Pinsker's inequality Created 2026-09-24 Updated 2026-11-05
With Total variation distance defined by and relative entropy computed using natural logarithms, Pinsker's inequality states
If is not absolutely continuous with respect to , the relative entropy is infinite and the bound is vacuous. A small relative entropy therefore forces small discrepancies in probabilities of all events. For example, and have Total variation distance and relative entropy , giving an upper bound about .
A short proof reduces to a binary event. Grouping the probability mass inside and outside , the log-sum inequality gives , with and . For , the binary relative entropy as a function of has minimum zero at and second derivative . Integrating this derivative bound twice gives . Endpoint cases follow by limits. Taking the supremum over proves the inequality. Using base-two logarithms changes the constant by a factor of .
Plotting position Created 2026-10-14 Updated 2026-11-05
A plotting position assigns a probability level to an order statistic for a Q-Q plot. The displayed formula is one convention, not a universal definition. R's ppoints uses , with by default for , which gives our choice at ; for its default is . NIST's normal probability plot instead describes an approximation to uniform distribution order statistic medians. State the convention when comparing plots, especially their extreme quantiles. [ref][ref]
Poisson arrivals see time averages Created 2026-09-24 Updated 2026-10-14
Poisson arrivals see time averages says that an exogenous, homogeneous Poisson process samples prearrival states with the same long-run frequencies as observation over time, provided the observed system cannot anticipate future arrivals and the relevant time averages exist. The system may depend on all past arrivals: an ordinary queue does. The required independence is between future increments of the Poisson process and the history available to the system, including its service information. Assume measurable state observations with ordinary piecewise continuous functions as sample paths; write for the state immediately before time .
For a state set , let count offered arrivals by time and let be their times. With a positive finite constant arrival rate, the PASTA equality is
almost surely when the time limit exists. The indicator functions count visits on the left and time spent on the right. For a stationary process with ergodicity, the right side is the stationary probability of . It matters that the arriving customer has not yet changed the observed state.
In the M/M/1 utilization and delay example, server utilization is . Thus PASTA says that 80 percent of offered arrivals find the server busy and 20 percent find the whole queue empty. If an offered arrival seeing population is admitted with conditional probability , using fresh randomness independent of that state, the accepted fraction is , where is the stationary distribution. Multiplying this fraction by the offered rate gives the admitted queue throughput when every admitted customer eventually completes service. Accepted arrivals are selected by state and need not themselves form a Poisson process or see the unconditioned time frequencies.
Arbitrary arrivals need not see time averages. Suppose a customer arrives every two minutes and its service time is exactly one minute, with one server and no other customers. The server is busy half the time, but every arrival finds it idle. A uniformly random initial phase can even make the occupancy a stationary process, without repairing the disagreement. Nevertheless Little's law still gives , with rate measured in customers per minute. This separates Little's law from PASTA.
Ronald W. Wolff's 1982 paper gave a general PASTA theorem under nonanticipation, rather than requiring the observed system to be independent of its driving Poisson process. Its Theorem 1 compares the two long-run fractions; stationarity is a way to establish convergence, not the central sampling condition. [ref]. [ref].
The sum of independent Bernoulli random variables with possibly unequal success probabilities has a Poisson binomial distribution. Its probability generating function is : expand the product to read off the probability mass function. Linearity of expectation and additivity of variance for independent random variables give and . Equal recover the binomial distribution.
For three trials with success probabilities , expanding gives masses at . Their expected value is and variance is . Replacing all three success probabilities by their average preserves the expected value but gives binomial distribution variance , so it does not preserve the law.
Poisson-binomial random variable Created 2026-10-20 Updated 2026-11-05
A sum of independent Bernoulli random variables whose success probabilities may differ. Its distribution is the Poisson binomial distribution.
Poisson count-arrival joint central limit theorem Created 2026-10-25 Updated 2026-10-31
For a rate-one Poisson process with Poisson arrival times , suppose and . Then
where and denotes the bivariate normal distribution with standard normal marginal distributions, including the singular cases. In particular, equal scales give and separated scales give .
To prove this, fix real and put for large . The event is exactly . Counts at the two deterministic times are sums of a shared count on and an independent count on the remaining interval. The central limit theorem for a centered Poisson random variable of diverging parameter follows by writing a count at a real time as its count at , a sum of independent unit increments, plus an independent remainder with expected value at most one. Markov's inequality makes that remainder negligible after division by .
Apply this central limit theorem to the two independent count blocks whenever their lengths diverge. A block with bounded length, or length negligible compared with the scale of its coordinate, contributes a term tending to zero in probability. The standardized pair of counts therefore has a joint normal limit whose correlation coefficient is
This includes : the difference block then has negligible variance after normalization. In the arrival event the standardized count at must be at least . Reversing this inequality changes the sign of the second limiting normal random variable, giving the asserted negative correlation coefficient. Continuity of the marginal distributions permits the moving thresholds and the weak inequalities, including at the singular limits.
Every is realized by with , and zero by . Conversely, even when has no limit, it has a subsequence converging in the extended interval . If the original pair has a joint bivariate normal distribution limit, that subsequence has the same limit, forcing its correlation coefficient into .
The normalized coherent state has number amplitude , hence the displayed Poisson law with mean and variance . A product of coherent modes has factorized number probabilities.
Poisson distribution Created 2026-09-24 Updated 2026-10-14
For , a Poisson random variable has probability mass function
The exponential function series verifies normalization. At , define with certainty. The equality of mean and variance is a distinguishing model property, though it does not by itself characterize the law.
When counting events in a homogeneous Poisson process, distinguish its rate from the distribution parameter: an interval of duration has . A rate measured per minute times a duration measured in minutes gives a dimensionless expected count. This process assumes independent increments, a constant rate, and a negligible chance of multiple events relative to interval length in a sufficiently short interval. A rate alone does not establish a Poisson distribution. See Bertsekas and Tsitsiklis, chapter 6 summary, pages 46-47.
For a constructed arrival model with events per minute over minutes, . The mean and variance are both three, and
The probability of at least one event is . These results describe the specified process, not every arrival mechanism with the same average rate.
The Poisson limit theorem, often called the law of rare events, connects this law to the binomial distribution. Keep fixed and let for integers . For each fixed nonnegative integer and sufficiently large ,
There are more opportunities, each less likely to succeed, while the mean stays and the variance approaches . For , the probability of three successes is at , at , and at , approaching the Poisson distribution value .
Figure 1.
Binomial counts approach Poisson counts while their mean stays fixed at three
. Details.
The panels show the same count scale and the same Poisson distribution reference. At , is not especially small and the difference is visible. No universal trial-count cutoff guarantees an adequate approximation for every event.
The independence assumption matters even when every success is rare. For , if all indicator random variables equal one shared Bernoulli random variable of parameter , their sum is zero with probability and with probability . Its mean is three, yet its zero-count probability tends to one, not . Some dependent counts still admit approximation through the Stein-Chen method, but this requires additional control of dependence.
For historical orientation, the BnF records Poisson's original probability treatise, Recherches sur la probabilité des jugements, as published in Paris in 1837. This bibliographic fact alone does not establish when the modern distribution terminology arose. See BnF record of Poisson's 1837 treatise.
Poisson germ-grain model 2026-10-14
A Poisson germ-grain model assigns a measurable region to each point of a Poisson point process and takes the union of those regions. A location remains uncovered exactly when no Poisson point has a region containing it, so its coverage probability is determined by the intensity of the corresponding set of germs and marks.
Poisson interarrival times Created 2026-10-14 Updated 2026-10-25
Let be the th arrival time, and set . In a Homogeneous Poisson process of rate , the gaps are independent and identically distributed random variables with an exponential distribution:
Here uses the rate convention. Under the scale convention , the same probability density function is ; a simulation function accepting a scale must receive . Conversely, summing independent rate- exponential random variables to form and counting those sums constructs a Homogeneous Poisson process.[ref]
The first-gap formula follows immediately from and the Poisson distribution. Independence of all successive gaps is the additional interval-theorem assertion; deterministic-time independent increments alone should not simply be substituted at random arrival times without justification. The memorylessness of the exponential distribution gives
At the toy rate of two per minute, the mean gap is half a minute, the median is minutes, and the probability of waiting longer than half a minute is . Here is the natural logarithm. Already having waited a minute without an arrival leaves this same conditional probability of waiting another half minute. The exponential distribution describes a model with no aging of the remaining wait; a scheduled arrival with a fixed gap has no such property.
Poissonization Created 2026-10-20 Updated 2026-11-05
Poissonization replaces a fixed number of independent observations by an independent Poisson count. It often turns dependent category counts into independent ones. Let labels have probabilities and draw independent of those labels. The category counts then satisfy
Indeed their joint Probability generating function is . This proves both the marginal laws and independence. Conditioning their sum back to recovers the Multinomial distribution for observations.
A rate- Poisson process gives a dynamic version. Independently marking its arrivals by labels splits it into independent Poisson processes of rates . The first arrival of each label is an independent exponential random variable, which simplifies coupon-collection calculations. Returning to a fixed-sample problem requires conditioning or a separate comparison argument; merely changing a clock does not preserve every probability. For an integrable stopping count determined by the labels and independent clock intervals, Wald's equation gives expected elapsed time equal to the expected count divided by .
Poisson limit theorem Created 2026-09-24 Updated 2026-11-05
Let have the binomial distribution with trial count and success Probability . If , then
This is the Poisson limit theorem: many independent rare trials retain a finite total expected count. For a proof using characteristic functions,
The logarithmic expansion error vanishes because . The limit is the Poisson distribution characteristic function, so the Lévy continuity theorem proves convergence. The case is the degenerate count zero; it also follows directly from Markov's inequality.[ref]
For example, is approximated by , and its zero-count Probability is about , close to . The limiting mean is the limit of the product, not either individual parameter. independent unequal-probability Bernoulli trials have a related Poisson approximation when their largest success Probability vanishes and their total mean converges; dependence requires additional control, as in the Stein-Chen method.
Poisson Metropolis chain 2026-10-25
A Poisson Metropolis chain samples a Poisson distribution by proposing an upward or downward unit step with equal probability and reflecting a downward proposal at zero into a self-proposal. Its upward and downward transition probabilities are the displayed , with , and the diagonal entry is . For , adjacent stationary fluxes equal . Hence detailed balance holds. Positive adjacent probabilities imply an irreducible Markov chain; implies aperiodicity. The normalized invariant Poisson distribution implies positive recurrence, so the countable-state Markov chain convergence theorem applies. The boundary proposal probability is essential: forcing every proposal at zero to go upward would require a different acceptance ratio.
Poisson-multinomial conditioning Created 2026-09-24 Updated 2026-11-05
Let be independent Poisson random variables of means , with for finitely many categories. Conditional on , the count vector has the multinomial distribution with probabilities :
Divide the independent Poisson joint mass by the Poisson distribution mass of . The exponential factors cancel, leaving the formula. Before conditioning the counts are independent; after a positive fixed total, distinct positive-probability categories have covariance .
For means , conditioning on total gives the first count a binomial distribution with parameters . A zero-mean category is always zero. If , every count is zero almost surely; conditioning on total zero is degenerate, while conditioning on a positive total has zero probability and does not define the displayed law.