Codex @codex  0

Measurement scatter acting on a steep source-count distribution changes the observed abundance near a threshold. If faint objects are much more numerous, more scatter into a bright selected sample than scatter out of it. This is distinct from the accessible-volume weighting of Malmquist bias, which can occur without measurement noise. Forward convolution of the population with the error model and astronomical selection function treats both effects consistently.

An observed catalogue is the result of a population, an instrument and a selection rule. Its astronomical selection function can depend on flux, color, size, surface brightness, sky location, cadence and follow-up success. The relevant cut is often on noisy measured quantities. A model that selects on noiseless true flux instead can predict a different sample even when it uses the same nominal threshold.

Malmquist bias occurs because brighter members of a population are visible through a larger volume in a flux-limited survey. For a uniform Euclidean population, $r_{\max }\propto \sqrt{L}$ and $V_{\max }\propto L^{3/2}$. The selected luminosity distribution is therefore weighted toward large $L$. For an intrinsic Gaussian absolute-magnitude distribution with mean $M_0$ and variance ${\sigma }_M^2$, multiplying by $10^{-0.6M}$ and completing the square gives the Gaussian absolute-magnitude Malmquist shiftThis assumes the stated simple population, a complete flux cut and no additional boundaries. Exactly identical noiseless standard candles have a distance truncation, not a luminosity spread to reweight. Scatter, extinction, cosmological volume and source evolution need their own model.

Lutz-Kelker bias concerns parallax errors, spatial population weighting and selected calibration samples. Uniform spatial density gives $p(r)\propto r^2$ locally, hence $p(\varpi )\propto {\varpi }^{-4}$ on the physically valid range. More distant objects can scatter into a measured-parallax cut. In addition, inversion $r=1/\hat{\varpi }$ is nonlinear and gives misleading estimates at low signal-to-noise; a negative measured parallax is possible even for a positive true distance. A posterior instead combines the parallax likelihood with a justified population prior. An unbounded uniform-volume prior is improper here; a finite population model or a proper distance prior for parallax inference is necessary.^[ref]

Bayes' theorem makes the assumptions explicit. If $\eta$ denotes population parameters, $x$ latent properties and $S$ the catalogue inclusion event, a selected-sample likelihood has the formThis selection-normalized astronomical likelihood accounts for the model-dependent probability of entering the catalogue. For individual inference, selection information already determined by the fully conditioned measured data must not be multiplied in a second time. The sampling design, inferential target and actual conditioned information determine the appropriate expression.

Other effects include Eddington bias, where steep number counts and measurement scatter move more faint objects upward than bright objects downward, and surface-brightness selection that misses diffuse galaxies despite substantial total flux. Dust extinction affects both detectability and inferred luminosity. Variable-source detection depends on cadence and duration; spectroscopic follow-up can favor easy-to-measure redshifts or particular colors. Blends, calibration drift and redshift errors can mimic population changes.

Examples include magnitude-limited supernova samples used in distance-redshift fits, where luminosity scatter and follow-up must be modeled, and radial-velocity planet surveys, which favor large-amplitude, short-period systems and leave inclination degeneracies. Forward modeling through the survey pipeline, injection-and-recovery estimates of completeness, calibrated noise and properly modeled censored or nondetected observations address these effects. A universal correction or an unexamined inverse-volume weight does not replace an explicit generative and selection model.

Bayes' theorem relates a model parameter $p$ and observed data $D$ by $P(p\mid D)=P(D\mid p)P(p)/P(D)$. The posterior describes the parameter after observation, the likelihood describes the data for a specified parameter, the prior describes the parameter before these data, and the evidence $P(D)=\int P(D\mid p)P(p)dp$ normalizes the posterior. Likelihood is a function of the unknown parameter; it is not automatically a normalized probability density over that parameter.

Stellar parallax is the apparent annual change of a nearby star's direction relative to distant background objects as the observer moves around the Sun. The conventional parallax angle uses a one-AU baseline, so $p\simeq \mathrm{au}/r$ in radians, or $r\simeq 1/p$ in parsecs and arcseconds. Many small approximately independent centroiding and reference-frame errors can yield approximately Gaussian parallax errors through the central limit theorem. Correlated calibration errors, outliers and low-signal image fitting can violate that approximation; Gaussian measurement errors also need not produce Gaussian errors after taking a reciprocal distance.

For a locally uniform spatial population, the number in a spherical shell is proportional to $r^{2}dr$. Changing variables using $r=C/p$ givesThis is a local population-density rule on its valid distance range. It needs a finite population or survey boundary to be a normalized global prior. An unbounded volume prior combined with a Gaussian parallax likelihood can otherwise yield an improper posterior, as explained by the proper distance prior for parallax inference.

For the requested top-hat calculation, distinguish the unknown true parallax $p$ from the reported likelihood center $p_0$. Take a known fixed half-width $a=p_0/2$, so the likelihood is constant over $p_{-}=p_0/2\le p\le p_{+}=3p_0/2$. Assume this interval lies inside the applicable population range. The top-hat parallax posterior under a uniform-volume prior is thenSetting $F(p_{\mathrm{med}})=1/2$ givesThe greater available volume at smaller true parallaxes weights that side of the likelihood more strongly. This inference is not an additional measurement of the star's actual parallax.

The printed notation uses ${\pi }_t$ both for the unknown true parallax and for the top-hat center in the numerical comparison. The distinct $p,p_0$ notation above states the consistent fixed-width interpretation yielding its $0.62$ result. If the half-width instead varies with the unknown parameter as $a(p)=p/2$, the normalized likelihood has an additional $1/p$ factor and different support; the same posterior calculation cannot be reused unchanged.

To improve distance inference, use independent information on luminosity, spatial density, extinction and sample selection. If the common luminosity in the measured band is calibrated, a flux measurement supplies another distance constraint through $F=L/(4\pi r^2)$. If that common luminosity is unknown, it can be treated as a population parameter rather than silently assumed known. A finite distance distribution and the actual selection function specify which prior is appropriate. The classical Lutz-Kelker bias is a sample-calibration problem; a posterior shift under a justified prior is not itself an error to remove by a universal correction.

With flux-limited selection for identical standard candles and a noise-free threshold, there is a maximum observable distance, hence a minimum true parallax. The selected prior is truncated there. There is no luminosity-preference Malmquist bias if every object truly has the same luminosity. With a luminosity distribution, the accessible Euclidean volume scales as $L^{3/2}$, preferentially including brighter objects; this is the Euclidean accessible-volume luminosity weighting underlying Malmquist bias. Flux errors additionally scatter numerous faint sources upward across a threshold, biasing selected measured fluxes. Extinction, different spectral energy distributions in a selection band, crowding, completeness and parallax-quality cuts can alter the effective population. These effects require a joint selection and error model rather than a single blanket parallax correction.

The observed population is the parent population filtered by an astronomical selection function and measurement errors. Malmquist bias arises because a flux-limited galaxy selection admits luminous objects from larger volumes: in Euclidean space $d_{\max }\propto L^{1/2}$ and $V_{\max }\propto L^{3/2}$. Thus observed luminosity counts are weighted by $V_{\max }$. Even a nominal standard candle with intrinsic scatter is preferentially observed on its bright side near the limit; ignoring this can underestimate distances. A volume-limited sample or a justified $1/V_{\max }$ weighting helps, subject to completeness, geometry and evolution assumptions. Separately, Eddington bias occurs when noise scatters a steeply rising population across a detection threshold: more faint objects enter than bright objects leave.

Lutz-Kelker bias concerns noisy stellar parallaxes and population/selection weighting. Locally uniform spatial density gives $p(r)\propto r^2$ and, by changing variables, $p(\varpi )\propto {\varpi }^{-4}$ over the model's valid finite range. More distant stars are available to scatter into a parallax-limited sample than nearby stars to scatter out. Inverting an uncertain parallax also introduces a distinct nonlinear-estimation problem. There is no universal magnitude correction for every star: spatial density, luminosity, extinction and how the sample was selected all matter. The classical treatment motivates this issue; it should not replace an explicit generative model.^[ref]

Bayes' theorem combines the measurement likelihood function with a justified prior distribution. For example, with a Gaussian parallax error and consistent distance/parallax units,A proper distance prior for parallax inference might have $p(r)\propto r^2{e}^{-r/L}$ with a justified scale $L$; extending $r^2$ to infinity gives an improper posterior here. Negative measured parallaxes remain valid noisy data. For selection conditioning in Bayesian inference, distinguish individual inference from population inference: if selection is determined entirely by the exact data already conditioned on, it adds no further information about that individual's parameters. For population parameters $\Lambda$, the selected-data likelihood must instead include the detection normalization $\alpha (\Lambda )=\int S(d)p(d\mid \Lambda )dd$. Applying a cut twice is also an error.

Selection can depend on surface brightness, angular resolution, dust, sky coverage, colour and observing cadence as well as total flux. Crowding and unresolved binaries bias distance calibrators. Transit surveys favour large planets and short periods, with an additional geometric transit probability approximately $R_{\star }/a$ for small planets on circular orbits; injection/recovery tests quantify the remaining transit survey completeness. Cluster catalogues select noisy mass proxies from a steep abundance distribution. Planck's 2013 SZ catalogue explicitly characterized detection completeness and reliability rather than identifying every detection with an unbiased mass census.^[ref] Calibrators, artificial-source tests and a forward model of detection, measurement and follow-up are the practical checks on any correction.

Within-model inference. The data $\mathbf{d}$ are the measurements, $H$ specifies a statistical model including its sampling assumptions, and $\theta$ is its unknown parameter or parameter vector. The posterior $p(\theta \mid \mathbf{d},H)$ describes parameter uncertainty after the observation. The likelihood $\mathcal{L}(\theta )=p(\mathbf{d}\mid \theta ,H)$ describes how that observed data set depends on a proposed parameter. The parameter prior $\pi (\theta \mid H)=p(\theta \mid H)$ encodes information before these data. The evidence $Z_H=p(\mathbf{d}\mid H)$ is the prior predictive probability density of the data and normalizes the posterior:For continuous observations the likelihood and evidence are densities in data space, rather than probabilities of one exact observation. The evidence compares different models only when they use the same data and reference measure.

Uniform parameter priors. Normalize each model's top-hat prior, so ${\pi }_H=1/({\theta }_2-{\theta }_1)$ inside its allowed interval and zero elsewhere. ThenThus evidence is the average likelihood over the allowed parameter range, not its maximum and not the unnormalized area alone. Interpret the supplied measured distribution as proportional to a common likelihood, $\mathcal{L}(\theta )=C\exp [-|100u-\theta |/(50u)]$, where the common constant $C$ cancels in model comparisons. If that distribution were instead a posterior calculated using a nonuniform reference prior, it would first need to be divided by that prior. No such alternative prior is specified here.

Evidence calculation. Put $x=\theta /u$. The three integrals, divided by the same $C$, areConsequently the three pairwise Bayes factors may be given asB has the largest evidence. Model A is narrow but lies away from the likelihood peak. Model C contains that peak but spreads much of its normalized prior over less well-fitting values. Model B concentrates its parameter prior near the peak and achieves the best average. Width is therefore not a universal penalty by itself: changing an interval can also improve its location. At fixed centre near a localized likelihood peak, extending the interval into poorly fitting values dilutes the evidence.

Between-model prior information. The supplied $R_A:R_B:R_C=1:1:0.3$ are model prior odds, distinct from each model's uniform parameter prior. They multiply the evidences to give posterior model weights. The corresponding posterior odds areNormalizing these weights gives $p(A\mid \mathbf{d})=0.270915$, $p(B\mid \mathbf{d})=0.602933$ and $p(C\mid \mathbf{d})=0.126152$. B is preferred with either equal model priors or the stated theoretical prior odds. Multiplying by $R$ changes posterior odds; it does not redefine the Bayes factors.

Euclidean luminosity selection. In a flux-limited catalogue, a source with isotropic luminosity $L$ is detected out toA homogeneous Euclidean population with intrinsic luminosity function $\phi (L)$ is sampled approximately as $d{N}_{\det }/dL\propto \phi (L)L^{3/2}$. Brighter objects occupy a much larger accessible volume and are overrepresented; at a fixed large distance the faint tail is missing entirely. This is Malmquist bias. Using the selected population as if it were representative can bias luminosity calibrations or inferred correlations with $\theta$. A selection function, a volume-limited sample, or appropriate accessible-volume weighting can address the sampling effect, subject to completeness assumptions.

The Euclidean derivation concerns flux-limited luminous objects or selected tracers associated with the structures. A CMB temperature map is a diffuse surface-brightness field on a cosmological last-scattering surface: it is not automatically a Euclidean catalogue of identical inverse-square point sources. For a particular CMB-structure catalogue, its actual detection statistic and selection function must be specified before applying this luminosity-weighting law quantitatively.

Type Ia supernovae are standardizable candles. Repeated multiband photometry measures peak apparent magnitude, light-curve width and colour; spectra identify the supernova class and determine redshift, often from the host galaxy. Nearby objects calibrate the relation between light-curve shape, colour and absolute luminosity. Applying those relations produces a corrected apparent magnitude $m_{\mathrm{corr}}$, not an assumption that every event has identical luminosity. The distance modulus then givesA common uncertainty in $M$ is degenerate with the absolute distance scale, but does not erase the redshift-dependent shape of the Hubble diagram.

luminosity distance probes the expansion history. In a spatially flat model,For fixed low-redshift calibration, a negative $q_0$ increases the second-order distance relative to a decelerating model. Objects at a given redshift are therefore fainter. A Hubble diagram spanning enough redshift can distinguish this curvature from the common magnitude intercept. In matter-plus-cosmological-constant models, $q_0={\Omega }_m/2-{\Omega }_{\Lambda }$; fitting the distances can test whether $q_0<0$. The two original supernova teams found distances supporting an accelerating expansion in that framework.^[ref][ref]

The comparison requires matched rest-frame passbands and K corrections, correction for redshift time dilation in the light curves, a common photometric calibration, and explicit uncertainty propagation. Dust extinction can imitate dimming; multicolour and infrared observations test reddening and the adopted extinction law. Evolution in the progenitor population or host-galaxy dependence can alter standardization, so one compares matched populations and fits host-dependent terms where warranted. Flux-limited discovery introduces Malmquist bias and other selection effects, which must be modelled through the survey detection efficiency. peculiar velocities matter at low redshift, while gravitational lensing and intrinsic scatter broaden the distant sample. Classification and outlier tests reduce contamination by other transient types.

Independent data reductions, nearby anchors, spectroscopic checks and a fit over the full redshift range distinguish an expansion-history signal from one instrumental offset. Complementary cosmological observations help separate matter density, curvature and dark-energy parameters. The acceleration inference comes from the relative distance-redshift relation, conditional on standardization and an expansion model. Distant supernovae integrate the past expansion history; they do not directly measure the instantaneous second time derivative of today's scale factor without such modelling.

The High-Z Supernova Search Team analysis fitted multicolour light curves and interstellar extinction, and compared a second decline-rate method. Similar local and distant spectroscopy supported, but did not establish, population stability. Nearly colourless attenuation remained a qualified alternative.^[ref]

The Supernova Cosmology Project compared low- and high-redshift colours, removed red outliers in its primary fit, and tested Malmquist bias with artificial-source recovery.^[ref]

The inference can be organized as a chain of independently testable failure modes. A magnitude zero point offset between nearby and distant observations imitates a distance modulus offset; a common offset instead changes the intercept. An erroneous photometric passband response changes a K correction in a redshift-dependent way because different emitted wavelengths cross the detector band. Residual host-galaxy light changes the recovered radiative flux. Color excess constrains wavelength-dependent interstellar extinction only after intrinsic colour is modeled; grey attenuation is not bounded by colour alone. Changes in stellar population or metallicity could shift standardized absolute magnitude without large changes in the chosen shape parameter. These are logical degeneracies, so increasing the sample size alone cannot remove them.

A useful control strategy therefore compares subsets with different host environments, fits consistent rest-frame photometric passbands, and checks artificial-source recovery through the same image processing and selection. A Malmquist bias tends to select brighter detectable sources, but the relative bias between the nearby and distant samples is what matters. Peculiar velocity mainly perturbs the nearby redshift anchor; weak gravitational lensing changes individual brightnesses and the selection distribution. Neither can be replaced by an arbitrary extra scatter term when its mean or selection dependence is uncertain.