{
  "id": "hubble-tension",
  "title": "Hubble Tension",
  "lang": "en",
  "source_lang": "en",
  "summary": "The Hubble tension is the discrepancy between certain measurements of the present cosmic expansion rate and values inferred from early-universe observations.",
  "category": "science",
  "subcategory": "astronomy",
  "type": "concept",
  "aliases": [
    "Hubble constant tension",
    "Hubble constant discrepancy"
  ],
  "generated_at": "2026-11-04",
  "model": "pro/gpt-6.1-sol",
  "url": "https://aiwiki.page/en/hubble-tension/",
  "translations": {
    "zh-Hans": {
      "url": "https://aiwiki.page/zh/hubble-tension/",
      "json": "https://aiwiki.page/zh/hubble-tension.json"
    }
  },
  "sections": [
    {
      "heading": "",
      "text": "The Hubble tension is a discrepancy between major methods of determining the present expansion rate of the universe, expressed by the Hubble constant, \\(H_0\\). Measurements based on distances to relatively nearby astronomical objects, especially Cepheid-calibrated Type Ia supernovae, generally favor a higher expansion rate than the value inferred from the cosmic microwave background under the standard Lambda–cold dark matter model, or \\(\\Lambda\\)CDM. Representative determinations are approximately \\(73\\) and \\(67.4\\ \\mathrm{km\\,s^{-1}\\,Mpc^{-1}}\\), respectively. Their difference is small compared with the historical uncertainty in cosmic distances, but large relative to the precision claimed by these particular analyses.\n\nThe discrepancy is important because it connects stellar astrophysics, astronomical distance measurement, and the physical history of the universe. It could reflect unrecognized measurement errors, incomplete modeling of astronomical objects, or physics beyond the standard cosmological model. “Tension” describes statistical inconsistency under specified assumptions; it does not establish that new physics has been discovered. Different observational methods do not divide cleanly into two unanimous groups, and their uncertainties and shared calibrations are central to interpreting the evidence."
    },
    {
      "heading": "The expansion rate and what is being compared",
      "text": "In an approximately homogeneous and isotropic universe, cosmic expansion is described by a scale factor \\(a(t)\\). The Hubble parameter is\n\n\\[H(t)=\\frac{\\dot a(t)}{a(t)},\\]\n\nand its value at the present epoch is \\(H_0\\). Although conventionally called a constant, \\(H_0\\) is the current value of a quantity that changes over cosmic history. Its usual unit, kilometers per second per megaparsec, expresses how recession velocity increases with distance. A megaparsec is approximately 3.26 million light-years. A value of \\(70\\ \\mathrm{km\\,s^{-1}\\,Mpc^{-1}}\\) corresponds to an inverse expansion timescale of about 14 billion years, although the actual age of the universe also depends on its expansion history.\n\nAt sufficiently low redshift, the leading approximation to the distance–redshift relation is\n\n\\[cz \\simeq H_0 d,\\]\n\nwhere \\(c\\) is the speed of light, \\(z\\) is cosmological redshift, and \\(d\\) is distance. This relation is not an exact formula for all redshifts. At greater distances, luminosity distance depends on the integrated expansion history, and observations require relativistic distance relations. Even nearby, galaxies have peculiar velocities caused by local gravitational fields in addition to the recession associated with expansion. These velocities can be hundreds of kilometers per second and must be modeled or reduced by selecting sufficiently distant objects.\n\nWithin spatially flat \\(\\Lambda\\)CDM, an illustrative expansion equation is\n\n\\[H^2(z)=H_0^2\\left[ \\Omega_r(1+z)^4+ \\Omega_m(1+z)^3+ \\Omega_\\Lambda \\right].\\]\n\nHere the \\(\\Omega\\) parameters denote present-day fractional contributions from radiation, matter, and a cosmological constant. Precision calculations also account for details such as the changing behavior of massive neutrinos. The equation follows from general relativity together with assumptions about the universe’s contents and large-scale geometry. Inferring \\(H_0\\) from early observations therefore means fitting a physical model, rather than directly observing the present expansion.\n\nThe central comparison is consequently between different measurement chains. A local distance ladder establishes absolute distances and relates them to redshifts, with relatively limited dependence on the universe’s early history. A microwave-background analysis measures patterns left by the primordial plasma and uses a cosmological model to infer present parameters. Other methods constrain different combinations of distance, expansion, and physical scales. Calling a measurement “early-universe” or “late-universe” identifies its main source of information, not the date on which the observation was made.\n\nThe tension is also distinct from the ordinary evolution of \\(H(t)\\). Standard cosmology already predicts that the expansion rate was much greater in the early universe. The dispute concerns values assigned to the same present-day parameter after each method’s data have been interpreted."
    },
    {
      "heading": "From historical disagreement to precision cosmology",
      "text": "The measurement problem predates the modern tension. Henrietta Swan Leavitt’s 1912 work established the relation between the pulsation periods and apparent brightnesses of Cepheid variables in the Small Magellanic Cloud. Because these stars were approximately at a common distance, the relation revealed a way to infer their relative intrinsic luminosities. Absolute calibration subsequently made Cepheids essential tools for measuring distances beyond the Milky Way.\n\nGeorges Lemaître connected an expanding relativistic universe with the observed velocity–distance relation in 1927. Edwin Hubble’s 1929 study provided influential observational evidence for that relation, drawing partly on galaxy velocities measured by Vesto Slipher. Early estimates of the expansion rate were around \\(500\\ \\mathrm{km\\,s^{-1}\\,Mpc^{-1}}\\), far above modern values. Misidentified stellar populations, confusion between bright stars and unresolved stellar associations, and an inaccurate distance scale contributed to the discrepancy. Walter Baade’s recognition of distinct stellar populations led to a major revision of the distance scale in the 1950s.\n\nFor much of the late twentieth century, competing distance scales produced estimates near \\(50\\) and \\(100\\ \\mathrm{km\\,s^{-1}\\,Mpc^{-1}}\\). These differences were much larger than today’s tension but arose when astronomical distance measurements had substantially greater uncertainties. The Hubble Space Telescope Key Project, led by Wendy Freedman and collaborators, reported \\(H_0=72\\pm8\\ \\mathrm{km\\,s^{-1}\\,Mpc^{-1}}\\) in 2001. That result combined Cepheid calibrations with several secondary distance indicators and represented a major reduction in uncertainty, rather than evidence for a conflict with early-universe cosmology.\n\nThe modern issue emerged as both measurement chains became more precise. The Wilkinson Microwave Anisotropy Probe, launched in 2001, and the Planck satellite, launched in 2009, mapped microwave-background anisotropies with increasing accuracy. Meanwhile, the SH0ES collaboration—Supernovae, \\(H_0\\), for the Equation of State of Dark Energy—developed a distance ladder designed to minimize differences between geometric calibrators, Cepheids, and supernova hosts. Planck’s initial cosmological results in 2013 made the contrast with the local ladder especially prominent.\n\nA frequently used benchmark is the Planck 2018 baseline result, published in its final cosmological-parameter analysis: \\(H_0=67.4\\pm0.5\\ \\mathrm{km\\,s^{-1}\\,Mpc^{-1}}\\) for flat, six-parameter \\(\\Lambda\\)CDM using temperature, polarization, and lensing information. In 2022, a SH0ES analysis reported \\(73.04\\pm1.04\\ \\mathrm{km\\,s^{-1}\\,Mpc^{-1}}\\), describing the difference as approximately five standard deviations. These are dated results with specific assumptions, not two immutable measurements. Their importance lies in showing how improvements in precision converted a modest discrepancy into a stringent consistency test."
    },
    {
      "heading": "The Cepheid–supernova distance ladder",
      "text": "The cosmic distance ladder transfers an absolute distance calibration across overlapping classes of astronomical objects. Its first rung uses geometric measurements. Important anchors include stellar parallaxes measured by the Gaia mission, detached eclipsing binary stars in the Large Magellanic Cloud, and the water-maser disk in NGC 4258. These anchors have different physical foundations and different systematic uncertainties. Their agreement tests whether an error in any one geometric technique could set the entire ladder’s distance scale incorrectly.\n\nThe next rung consists of Cepheid variables, pulsating stars whose periods correlate with luminosity. A practical calibration includes dependence on period, chemical composition, and extinction by dust. In simplified form,\n\n\\[M=a\\log_{10}P+b+\\gamma Z,\\]\n\nwhere \\(M\\) is absolute magnitude in a specified photometric construction, \\(P\\) is pulsation period, \\(Z\\) represents a metallicity indicator, and the coefficients are fitted. Near-infrared observations reduce sensitivity to dust and some stellar-population effects. Combinations of magnitudes known as Wesenheit magnitudes are designed to reduce extinction sensitivity, although they still depend on assumptions about the extinction law.\n\nCepheids can be observed in galaxies that have hosted a Type Ia supernova. These events are not perfectly identical standard candles. Their peak luminosities are standardized using light-curve shape, color, and sometimes host-galaxy information. Cepheids establish the distances to nearby supernova hosts, thereby calibrating the absolute magnitude of standardized supernovae. A much larger sample of supernovae farther into the smooth expansion flow then determines the relation between brightness and redshift.\n\nThe connection is expressed through the distance modulus,\n\n\\[\\mu=m-M=5\\log_{10}\\left(\\frac{d_L}{10\\ \\mathrm{pc}}\\right),\\]\n\nwhere \\(m\\) is apparent magnitude, \\(M\\) is absolute magnitude, and \\(d_L\\) is luminosity distance. For a fixed observed supernova relation, changing the calibrated \\(M\\) changes the inferred distance scale and hence \\(H_0\\). Moving from approximately \\(73\\) to \\(67.4\\ \\mathrm{km\\,s^{-1}\\,Mpc^{-1}}\\) requires increasing the corresponding distances by roughly 8 percent, equivalent to a distance-modulus shift of about 0.17 magnitude. This gives a useful scale for evaluating proposed calibration errors.\n\nThe ladder’s precision comes partly from observing different rungs with similar instruments, filters, and analysis procedures. Such consistency reduces errors associated with transferring photometric zero points between systems. It does not eliminate all astrophysical effects: Cepheids in distant hosts are more crowded than nearby calibrators, and the populations of supernovae used for calibration may differ from those in the expansion-flow sample. The determination therefore depends on a joint model of many stars and supernovae, their selection functions, and their correlated uncertainties, rather than on one simple average of independently measured distances."
    },
    {
      "heading": "Early-universe inference and the sound horizon",
      "text": "The microwave background preserves information from when the universe became transparent, approximately 380,000 years after the Big Bang. Before that epoch, photons and ordinary matter formed a tightly coupled plasma. Gravity compressed overdense regions while radiation pressure resisted compression, producing acoustic oscillations. Their imprint appears as a sequence of peaks in the angular power spectra of microwave-background temperature and polarization.\n\nA central physical scale is the sound horizon, the comoving distance an acoustic disturbance could travel before a specified epoch:\n\n\\[r_s(z_*)=\\int_{z_*}^{\\infty}\\frac{c_s(z)}{H(z)}\\,dz,\\]\n\nwhere \\(c_s(z)\\) is the sound speed and \\(z_*\\) denotes approximately the photon-decoupling epoch. The angular pattern strongly constrains\n\n\\[\\theta_*=\\frac{r_s(z_*)}{D_M(z_*)},\\]\n\nwith \\(D_M\\) the transverse comoving distance to that epoch. The microwave background measures this ratio exceptionally well. Converting it into \\(H_0\\), however, requires a model for both the physical sound horizon and the intervening expansion.\n\nAdditional properties of the peak pattern constrain the physical densities of baryons and dark matter, the primordial fluctuation spectrum, and other parameters. Within six-parameter \\(\\Lambda\\)CDM, those constraints are sufficiently tight to predict a narrow range for \\(H_0\\). Allowing extra parameters can broaden or move that range. Thus Planck’s small uncertainty is conditional on the model, not a completely theory-independent measurement error. Nevertheless, successful alternatives must reproduce much more than the angular location of one acoustic peak: they must also match relative peak heights, damping at small angular scales, polarization, and gravitational lensing.\n\nA related acoustic scale appears in the spatial clustering of galaxies as baryon acoustic oscillations, or BAO. The relevant ruler is usually the sound horizon at the baryon drag epoch, \\(r_d\\), when baryons ceased to be efficiently dragged by photons. It is close to, but distinct from, the sound horizon at photon decoupling. BAO measurements constrain combinations such as \\(D_M(z)/r_d\\) and \\(H(z)r_d\\); they do not determine an absolute Hubble constant without additional information about the ruler or the distance scale.\n\nCombining BAO with supernova relative distances and a physical calibration of \\(r_d\\) creates an inverse distance ladder. Calibration can use microwave-background data or constraints from Big Bang nucleosynthesis together with assumptions about pre-recombination physics. Such combinations generally favor a lower \\(H_0\\) under standard early-universe assumptions. Their relevance is that the lower value is not solely a consequence of one Planck data-processing pipeline. Independent microwave-background experiments also provide checks: an Atacama Cosmology Telescope Data Release 4 analysis combined with WMAP information found a \\(\\Lambda\\)CDM value close to \\(68\\ \\mathrm{km\\,s^{-1}\\,Mpc^{-1}}\\)."
    },
    {
      "heading": "Independent distance indicators and expansion probes",
      "text": "The tip of the red giant branch, or TRGB, provides a stellar distance indicator independent of the Cepheid period–luminosity relation. Low-mass red giants reach a characteristic luminosity before helium ignition in their cores. In suitable photometric bands, the sharp termination of their luminosity distribution serves as a standard candle, with corrections for population properties. Observations often target galaxy halos, where dust and crowding can be lower than in the star-forming regions containing Cepheids. A 2021 calibration by Freedman reported \\(H_0=69.8\\ \\mathrm{km\\,s^{-1}\\,Mpc^{-1}}\\), with quoted statistical and systematic uncertainties of \\(0.6\\) and \\(1.6\\), respectively. Different TRGB calibrations and analyses do not yield identical results, illustrating the importance of edge detection, field selection, and the absolute zero point.\n\nAstrophysical masers provide geometric distances through a different mechanism. Water molecules in nearly edge-on disks around massive black holes produce narrow microwave emission lines. Very-long-baseline interferometry resolves their angular distribution, while line-of-sight velocities and accelerations constrain the disk’s physical dimensions and dynamics. Comparing physical and angular scales yields distance. A 2020 Megamaser Cosmology Project analysis reported \\(H_0=73.9\\pm3.0\\ \\mathrm{km\\,s^{-1}\\,Mpc^{-1}}\\). Its precision was lower than that of the principal Cepheid ladder, but the method offered an important check based on different astrophysics.\n\nStrong gravitational lensing supplies another route. If a variable quasar is multiply imaged by a foreground galaxy, fluctuations reach the observer at different times because the light paths differ in length and gravitational potential. Time delays constrain a distance combination approximately proportional to \\(H_0^{-1}\\). The H0LiCOW collaboration’s 2019 analysis of six lenses obtained \\(73.3^{+1.7}_{-1.8}\\ \\mathrm{km\\,s^{-1}\\,Mpc^{-1}}\\) under its modeling assumptions. However, the lens mass distribution and matter along the line of sight are essential. Analyses allowing greater freedom in lens structure can yield substantially larger uncertainties, so a time-delay result cannot be interpreted separately from its mass-model assumptions.\n\nGravitational waves from compact-binary mergers can act as standard sirens. Their waveforms encode a luminosity distance without requiring stellar-luminosity calibration. A redshift must still be obtained from an identified host galaxy or a statistical association with possible hosts. The neutron-star merger GW170817 in 2017 supplied the first standard-siren determination of \\(H_0\\), with a central value near \\(70\\ \\mathrm{km\\,s^{-1}\\,Mpc^{-1}}\\) but broad uncertainty. Important limitations include the degeneracy between distance and binary inclination, detector calibration, and host peculiar velocity.\n\nOther approaches include surface-brightness fluctuations, Mira variables, and differential-age measurements of galaxies known as cosmic chronometers. These methods are valuable because their dominant errors differ from those of Cepheids or the microwave background. Their independence is nevertheless a matter of degree: two methods may share geometric anchors, a supernova sample, photometric calibration, or assumptions about stellar evolution. A collection of estimates is therefore not equivalent to the same number of independent tests."
    },
    {
      "heading": "Statistical significance and dependence on assumptions",
      "text": "The familiar significance of the tension is often calculated by approximating two parameter estimates as independent Gaussian distributions. If they have means \\(H_{0,A}\\), \\(H_{0,B}\\) and standard deviations \\(\\sigma_A\\), \\(\\sigma_B\\), a simple discrepancy statistic is\n\n\\[T=\\frac{|H_{0,A}-H_{0,B}|} {\\sqrt{\\sigma_A^2+\\sigma_B^2}}.\\]\n\nUsing \\(73.04\\pm1.04\\) and \\(67.4\\pm0.5\\ \\mathrm{km\\,s^{-1}\\,Mpc^{-1}}\\) gives \\(T\\simeq4.9\\). This explains the widely used description “five-sigma tension.” It applies to that comparison under its assumptions; it is not the significance shared by every local measurement or every cosmological model.\n\nIf the estimates are correlated, the denominator instead involves\n\n\\[\\sigma_\\Delta^2= \\sigma_A^2+\\sigma_B^2 -2\\,\\operatorname{Cov}(H_{0,A},H_{0,B}).\\]\n\nCorrelations are particularly important when combining several late-universe measurements. Two ladders may use the same geometric anchor or overlapping supernova samples. Several microwave-background analyses may share foreground assumptions or external information. Neglecting shared errors can make an apparent consensus look more precise than it is. Conversely, genuine agreement between methods with different dominant uncertainties provides stronger evidence than repeated analysis of the same observations.\n\nA quoted uncertainty is itself a product of modeling. Measurement uncertainty includes random variation and estimated systematic contributions, but an unrecognized bias does not automatically appear in the published error bar. In cosmological fits, nuisance parameters describe instrumental calibration, foreground emission, dust, or astrophysical populations. Their allowed ranges and correlations influence the inferred probability distribution of \\(H_0\\). Non-Gaussian distributions and extended parameter degeneracies may make a one-number “sigma” comparison inadequate.\n\nModel comparison introduces an additional distinction. Adding parameters often permits a higher \\(H_0\\) or broadens its allowed range, but compatibility achieved mainly by increasing uncertainty is not the same as a model predicting both datasets well. Tests can compare the change in best-fit likelihood, penalize extra freedom, or evaluate Bayesian evidence. These procedures answer related but different questions and may be sensitive to parameter priors. A proposed theory should be assessed against the full set of observations, not just whether its \\(H_0\\) error bar overlaps a local result.\n\nThere is also a useful distinction between disagreement in \\(H_0\\) and disagreement in the underlying calibrated quantities. The supernova ladder directly establishes a standardized absolute magnitude, while BAO analyses constrain distances relative to \\(r_d\\). Comparing these quantities can reveal whether a proposed solution truly changes the calibration or merely changes the mapping from observations to \\(H_0\\). The tension therefore cannot be resolved by averaging incompatible values or by treating all published estimates as interchangeable measurements of a model-free number."
    },
    {
      "heading": "Astrophysical and instrumental explanations",
      "text": "Cepheid crowding is a central systematic concern. At the distances of supernova-host galaxies, stars that are separable nearby can overlap within one telescope resolution element. Unresolved neighboring light can make a Cepheid appear brighter, biasing its distance downward and the inferred Hubble constant upward. Analyses use artificial-star tests, local-background measurements, and high-resolution observations to estimate this effect. The James Webb Space Telescope’s sharper near-infrared imaging provides a check on Cepheid photometry in fields previously observed with the Hubble Space Telescope. Published comparisons have tested substantial crowding-induced shifts in overlapping samples; extending such checks across the ladder remains important.\n\nMetallicity and dust introduce different challenges. Cepheid luminosities and colors depend on chemical composition, and their host galaxies span different abundance distributions. Dust changes both brightness and color, but the wavelength dependence of extinction need not be identical along every line of sight. Near-infrared measurements and reddening-reduced magnitude combinations limit these effects without making them disappear. Reliable tests require examining whether residual distances vary with abundance, period, color, local surface brightness, and position within the host galaxy.\n\nThe geometric anchors also have specific vulnerabilities. Gaia parallaxes require corrections for small zero-point offsets that depend on source properties and observation conditions. Cepheids are bright, variable objects, and some have companions that complicate astrometric interpretation. Detached eclipsing binaries require accurate stellar surface-brightness relations or atmosphere modeling to connect radii and observed fluxes to distance. Maser distances depend on the geometry and dynamics of their disks. Multiple anchors help because these errors do not have identical physical origins, although their combination still requires a consistent statistical treatment.\n\nSupernova standardization is another possible source of mismatch. Type Ia supernovae occur in populations spanning different stellar ages, metallicities, and dust environments. Their standardized brightness correlates with host properties, commonly represented by empirical corrections such as a host-mass step. A residual difference between nearby calibrators and more distant samples could shift the ladder. Selection effects also matter: surveys preferentially detect brighter objects near their limits, and spectroscopic follow-up can impose further biases. Simulations and cross-survey photometric calibration are used to estimate these effects, but population evolution cannot be represented adequately by a single universal correction without testing.\n\nOn the early-universe side, concerns include microwave foregrounds, instrumental beams, detector calibration, and assumptions about recombination. Observations at several frequencies help distinguish the cosmological signal from Galactic dust and other emission, while independent experiments test instrumental effects. Local structure has also been proposed as an explanation for a high nearby expansion rate: an underdense region can expand faster than the cosmic mean. However, producing the full principal discrepancy generally requires a much larger or deeper coherent local fluctuation than supported by standard structure-formation expectations and observations. Ordinary peculiar velocities remain important for individual objects, especially in small nearby samples."
    },
    {
      "heading": "Proposed new physics and discriminating observations",
      "text": "Many proposed physical solutions target the sound horizon. If expansion before recombination were faster, acoustic waves would travel a shorter comoving distance. A smaller ruler can allow a higher inferred \\(H_0\\) while preserving the observed angular acoustic scale. Because BAO and supernovae also constrain the ruler–distance relationship, analyses seeking to reconcile them with a value near \\(73\\ \\mathrm{km\\,s^{-1}\\,Mpc^{-1}}\\) commonly require a reduction in \\(r_d\\) of several percent, often roughly 7–9 percent depending on the datasets and model. The difficulty is achieving this without spoiling the rest of the microwave-background spectrum.\n\nEarly dark energy is one widely studied mechanism. In representative models, an additional field contributes a modest fraction of the total energy density around the era preceding recombination and then dilutes rapidly. It temporarily increases \\(H(z)\\), reducing the sound horizon. Viable implementations must control the epoch of this contribution, its peak fraction, and its subsequent evolution. They also alter gravitational potentials, the growth of matter fluctuations, and the acoustic peak structure. Consequently, a fit that improves the local-\\(H_0\\) comparison may face restrictions from galaxy clustering, weak lensing, or high-resolution microwave-background observations.\n\nAnother possibility is extra relativistic energy density, often parameterized by the effective number of relativistic species, \\(N_{\\rm eff}\\). In standard calculations including non-instantaneous neutrino decoupling, its value is approximately 3.044. Increasing it raises the early expansion rate, but also changes nucleosynthesis, photon diffusion, and the phase of acoustic oscillations. Particle interactions involving neutrinos or a dark sector can modify some of these signatures, at the cost of additional assumptions. Changes to atomic physics or the recombination history have also been investigated because they can alter the relation between the acoustic and diffusion scales.\n\nLate-time proposals modify dark energy, spatial curvature, or the laws governing gravity. Their appeal is that they can change the distance to the microwave-background surface without directly changing every early-universe process. However, BAO and supernova measurements strongly constrain the shape of the recent expansion history. Simple extensions, such as replacing the cosmological constant with a constant equation-of-state parameter different from \\(-1\\), generally do not provide a satisfactory common fit to all the relevant data. Very recent transitions and interacting dark-sector models require similarly careful checks against distances and structure growth.\n\nDiscriminating observations must separate these mechanisms rather than merely refine a single number. Improved polarization and small-scale microwave-background measurements test the acoustic phase and damping signatures of early physics. Galaxy surveys measure BAO distances and the growth of structure across redshift. Additional geometric anchors, resolved stellar observations, and independently calibrated red-giant and variable-star distances test the ladder’s absolute scale. Standard sirens and well-constrained lens time delays probe distances with substantially different calibration requirements. Each tests a different part of the chain linking primordial physical scales, stellar luminosities, and the present expansion rate."
    }
  ],
  "links": [
    "hubble-constant",
    "cosmic-microwave-background",
    "lambda-cdm-model",
    "hubble-parameter",
    "redshift",
    "speed-of-light",
    "neutrino",
    "general-relativity",
    "milky-way",
    "cosmic-distance-ladder",
    "stellar-parallax",
    "type-ia-supernova",
    "big-bang",
    "photon",
    "gravity",
    "polarization",
    "dark-matter",
    "baryon-acoustic-oscillations",
    "black-hole",
    "interferometry",
    "gravitational-lensing",
    "gravitational-wave",
    "measurement-uncertainty",
    "probability-distribution",
    "energy-density",
    "dark-energy"
  ],
  "sources": [],
  "license": "https://creativecommons.org/licenses/by/4.0/"
}