Why is the energy of empty space so small, and what would it mean if there were no cosmological constant at all?

There is a peculiar mismatch at the heart of modern physics. General relativity allows empty space to gravitate. Quantum field theory tells us that empty space is not really empty: fields have vacuum states, fluctuations, and phase-dependent contributions to their energy. Yet the gravitational effect associated with the vacuum is extraordinarily small.

The number inferred from cosmology is not merely smaller than the energy scales of particle physics. It is smaller by roughly 120 orders of magnitude when compared with the fourth power of the reduced Planck scale. And it is not exactly zero. The observed value is small enough for galaxies to form, but large enough to dominate the expansion of the Universe today.

This is the cosmological constant problem. It is often described as the worst fine-tuning problem in theoretical physics, but that phrase can hide several distinct questions. Why is the vacuum energy so small? Why is it not exactly zero? Why does it have the value it does? Why is it becoming cosmologically important only around the present epoch? And is the thing we call dark energy really a cosmological constant?

These questions are connected, but they are not identical. A proposal can address one while leaving the others untouched. The goal here is to separate them, derive the anthropic argument in a useful form, and examine the main families of proposed resolutions.

1. What is the cosmological constant?

Einstein’s field equations can contain a term proportional to the metric:

$$ G_{\mu\nu} + \Lambda g_{\mu\nu} = 8\pi G T_{\mu\nu}. $$

Here, \$ G_{\mu\nu} \$ describes spacetime curvature, \$T_{\mu\nu}\$ describes matter and radiation, and \$\Lambda\$ is the cosmological constant. It is not an extra force acting on objects. It is a term in the gravitational field equations that changes the relation between geometry and matter.

In a homogeneous and isotropic universe, the acceleration equation is

$$ \frac{\ddot a}{a} = -\frac{4\pi G}{3}(\rho + 3p), $$

where \$a(t)\$ is the scale factor, \$\rho\$ is the total energy density, and \$p\$ is the total pressure. A cosmological constant is equivalent to a fluid with

$$ \rho_\Lambda = \frac{\Lambda}{8\pi G}, \qquad p_\Lambda = -\rho_\Lambda. $$

Its pressure is negative enough that \$\rho_\Lambda + 3p_\Lambda < 0\$. When it dominates the cosmic energy budget, it therefore causes accelerated expansion. Unlike matter and radiation, its density does not dilute as the Universe expands.

This is why the cosmological constant is associated with dark energy. But the two expressions should not be treated as synonyms. A cosmological constant is one particular model of dark energy: it has exactly constant energy density and an equation-of-state parameter

$$ w \equiv \frac{p_{\rm DE}}{\rho_{\rm DE}} = -1. $$

Dark energy is the broader name for whatever is responsible for the observed late-time acceleration, assuming general relativity is the correct theory of gravity on cosmological scales. A dynamical field or a modification of gravity could produce similar expansion history without a fundamental \$\Lambda\$ term. 1

2. The observational fact: the vacuum scale is tiny

The late-time Universe is well described, to a good approximation, by the \$\Lambda\$CDM model: cold dark matter, ordinary matter, radiation, and a cosmological constant. Type Ia supernovae, the cosmic microwave background, baryon acoustic oscillations, and large-scale structure all contribute to the evidence for this picture. In the standard framework, roughly 70% of the present cosmic energy density is attributed to dark energy. 2

The measured dark-energy density corresponds to an energy scale of about

$$ \rho_{\Lambda,\mathrm{obs}}^{1/4} \simeq 2.26 \times 10^{-3}\ {\rm eV}. $$

To compare it with particle-physics scales, use natural units, \$\hbar = c = 1\$, and express everything in powers of GeV. Since \$1\ {\rm eV} = 10^{-9}\ {\rm GeV}\$,

$$ \begin{aligned} \rho_{\Lambda,\mathrm{obs}} &= (2.26 \times 10^{-3}\ {\rm eV})^4 \\ &= (2.26 \times 10^{-12}\ {\rm GeV})^4 \\ &\simeq 2.6 \times 10^{-47}\ {\rm GeV}^4. \end{aligned} $$

Now compare this with the reduced Planck mass,

$$ M_{\rm Pl} = \frac{1}{\sqrt{8\pi G}} \simeq 2.435 \times 10^{18}\ {\rm GeV}. $$

A Planck-scale estimate of vacuum energy is of order \$M_{\rm Pl}^4\$:

$$ M_{\rm Pl}^4 = (2.435 \times 10^{18}\ {\rm GeV})^4 \simeq 3.5 \times 10^{73}\ {\rm GeV}^4. $$

The ratio is therefore

$$ \begin{aligned} \frac{M_{\rm Pl}^4}{\rho_{\Lambda,\mathrm{obs}}} &\simeq \frac{3.5 \times 10^{73}} {2.6 \times 10^{-47}} \\ &\simeq 1.3 \times 10^{120}. \end{aligned} $$

That is the origin of the famous 120-order-of-magnitude discrepancy. It is a comparison of energy densities. Taking the fourth root gives

$$ \frac{M_{\rm Pl}}{\rho_{\Lambda,\mathrm{obs}}^{1/4}} \simeq 1.1 \times 10^{30}, $$

so the corresponding energy-scale hierarchy is about 30 orders of magnitude, not 120. Both statements describe the same hierarchy, but refer to different quantities. 3 4

A small number is not automatically a problem. Physics contains many small dimensionless parameters. The problem is that quantum field theory seems to generate contributions to vacuum energy that are vastly larger, and there is no known symmetry in the observed low-energy theory that explains why their sum should remain so close to zero.

3. Why quantum field theory makes this difficult

In quantum field theory, a field can be viewed, roughly, as an infinite collection of harmonic oscillators, one for each momentum mode. Even in its lowest-energy state, each oscillator has zero-point energy. For a simple free field, the formal vacuum-energy density contains an integral of the form

$$ \rho_{\rm vac} = \frac{1}{2} \int^{k_{\rm max}} \frac{d^3 k}{(2\pi)^3} \sqrt{k^2 + m^2}. $$

At large momentum, the integral scales as the fourth power of the cutoff. For a massless field and a sharp three-momentum cutoff \$K\$, the integral can be evaluated explicitly:

$$ \begin{aligned} \rho_{\rm vac} &= \frac{1}{2}\int_0^K \frac{4\pi k^2\,dk}{(2\pi)^3}\,k \\ &= \frac{1}{4\pi^2}\int_0^K k^3\,dk \\ &= \frac{K^4}{16\pi^2}. \end{aligned} $$

The numerical coefficient depends on the field content and regulator, but the characteristic scaling is

$$ \rho_{\rm vac} \sim K^4. $$

For \$K = M_{\rm Pl}\$, the dimensional estimate without loop factors is \$M_{\rm Pl}^4\$, giving the ratio of about \$1.3 \times 10^{120}\$ calculated above. Keeping the \$1/(16\pi^2)\$ from this particular sharp-cutoff integral instead gives

$$ \frac{M_{\rm Pl}^4}{16\pi^2\rho_{\Lambda,\mathrm{obs}}} \simeq 8.5 \times 10^{117}. $$

That is about 118 orders of magnitude. It is still conventionally summarized as a roughly 120-order problem. The cutoff estimate is not a regulator-independent prediction of quantum field theory; its purpose is to show the scale of the naturalness problem. 3 4

There are also lower-energy contributions, which are important because they do not rely on extrapolating a field theory all the way to the Planck scale.

Electroweak scale. The electroweak symmetry-breaking scale is of order \$M_{\rm EW} \sim 200\ {\rm GeV}\$. Its fourth power is

$$ (200\ {\rm GeV})^4 = 1.6 \times 10^9\ {\rm GeV}^4. $$

Compared with the observed density,

$$ \frac{(200\ {\rm GeV})^4}{\rho_{\Lambda,\mathrm{obs}}} \simeq \frac{1.6 \times 10^9}{2.6 \times 10^{-47}} \simeq 6.1 \times 10^{55}. $$

Thus the electroweak contribution is roughly 55–56 orders of magnitude larger, before any cancellation with the bare cosmological constant. The precise coefficient depends on the Higgs potential and the convention used for the electroweak scale. 3 4

QCD scale. The QCD vacuum changes across the chiral-symmetry-breaking transition. A characteristic strong-interaction scale is \$M_{\rm QCD} \sim 0.3\ {\rm GeV}\$. The corresponding dimensional estimate is

$$ \rho_{\rm QCD} \sim (0.3\ {\rm GeV})^4 = 8.1 \times 10^{-3}\ {\rm GeV}^4. $$

The ratio is

$$ \frac{\rho_{\rm QCD}}{\rho_{\Lambda,\mathrm{obs}}} \sim \frac{8.1 \times 10^{-3}}{2.6 \times 10^{-47}} \simeq 3.1 \times 10^{44}. $$

So the QCD scale alone gives a mismatch of about 44 orders of magnitude. Using \$0.2\ {\rm GeV}\$ instead gives about \$6 \times 10^{43}\$, or roughly 44 orders. This is the commonly quoted “40- to 45-order” problem: it comes from a known low-energy scale, not from assuming Planck-scale physics. 3 4

Strictly speaking, this estimates the change in vacuum energy associated with the QCD transition, not an independently measurable absolute vacuum energy. A counterterm can shift the overall zero of energy. But once gravity is included, the total renormalized vacuum energy must still be small, so this known phase-transition contribution has to be included in the required cancellation. 3 4

There is an important qualification here. The quartically divergent cutoff integral is not, by itself, a direct observable prediction of quantum field theory. Its value depends on the regularization and renormalization prescription. One can absorb cutoff-dependent terms into a bare cosmological constant. It would therefore be misleading to say that quantum field theory has made a clean prediction of a huge vacuum energy and experiment has simply contradicted it.

The deeper problem survives renormalization. In a gravitational effective field theory, the cosmological constant is an allowed parameter. The renormalized value is a sum of a bare term and contributions from quantum fields, symmetry breaking, and phase transitions. Schematically,

$$ \rho_{\Lambda,\mathrm{eff}} = \rho_{\Lambda,\mathrm{bare}} + \rho_{\mathrm{loops}} + \rho_{\mathrm{phase\ transitions}} + \cdots . $$

The observed value requires this sum to be tiny. But known physics gives large contributions to the individual terms. Even finite contributions from massive particles, not just a formal Planck-scale cutoff, are enormous compared with the measured result. The top-quark contribution alone illustrates the severity of the radiative-stability problem. 4

The issue is not that a counterterm cannot be chosen to cancel the contributions. It can. The issue is that the cancellation must be extraordinarily precise, and it must keep working whenever quantum corrections are added or the vacuum changes phase.

This is the naturalness version of the cosmological constant problem: why is the small observed value stable under the corrections that the theory itself seems to require?

3.1 Why phase transitions matter

The problem is not confined to virtual particles at extremely high energy. The vacuum state of a field can change when the Universe passes through a phase transition. The Higgs field, for example, sits in a nonzero vacuum configuration today. Quantum chromodynamics also has a nontrivial vacuum structure.

A change in the vacuum state changes the energy attributed to that state. If gravity responds to vacuum energy in the ordinary way, these changes should affect the cosmological constant. Yet the late-time value remains tiny. A successful explanation must account not only for an initial cancellation, but also for why later changes in the vacuum do not leave behind a much larger effective cosmological constant. 4

3.2 The old problem and the coincidence problem

It is useful to distinguish two problems.

The old cosmological constant problem is the smallness problem: why is the effective vacuum energy so close to zero compared with particle-physics scales?

The coincidence problem asks why the vacuum-energy density is becoming comparable to the matter density at roughly the present epoch. Matter dilutes as

$$ \rho_m(a) = \rho_{m,0} a^{-3}, $$

whereas a true cosmological constant remains constant:

$$ \rho_\Lambda(a) = \rho_{\Lambda,0}. $$

There is no obvious reason why these very different components should have comparable densities now. The first problem concerns an enormous hierarchy. The second concerns a special-looking cosmic epoch. A theory that solves one does not automatically solve the other. 5

4. The anthropic argument

The anthropic approach is often caricatured as saying that the cosmological constant has its value because otherwise we would not be here. That statement is too vague to be useful. The serious version is a selection argument: if different regions or universes realize different values of the vacuum energy, observers can only measure values compatible with the formation of structures that eventually produce observers.

The argument does not require the Universe to be designed for life. It requires an ensemble of physically realized possibilities, a distribution of values across that ensemble, and a selection effect that changes what observers are likely to measure. Weinberg showed that galaxy formation places a quantitative upper bound on a positive cosmological constant. 6

4.1 Why a large positive value prevents galaxies

In the early Universe, matter dominates over vacuum energy. Small primordial density fluctuations can then grow under gravity. In the matter-dominated era, the linear growth of a density perturbation is approximately proportional to the scale factor:

$$ \delta(a) \propto a. $$

Eventually, a sufficiently overdense region becomes nonlinear, collapses, and forms a gravitationally bound object. Galaxies and their stars require this process.

A positive cosmological constant changes the story. Since matter density falls as \$a^{-3}\$ while vacuum density remains constant, the vacuum eventually dominates. Once this happens, the expansion accelerates and the growth of density perturbations becomes strongly suppressed. Existing bound objects survive, but new large-scale structures become much harder to form.

Define \$a_\Lambda\$ as the scale factor at which matter and vacuum densities are equal. Then

$$ \rho_{m,0} a_\Lambda^{-3} = \rho_\Lambda, \qquad a_\Lambda = \left(\frac{\rho_{m,0}}{\rho_\Lambda}\right)^{1/3}. $$

Increasing \$\rho_\Lambda\$ makes \$a_\Lambda\$ smaller. Vacuum domination begins earlier, leaving less time for primordial fluctuations to grow. For a region to collapse before growth is suppressed, its linearly extrapolated density contrast must reach the collapse threshold,

$$ \delta_{\rm lin}(t_\Lambda) \gtrsim \delta_c, \qquad \delta_c \simeq 1.686. $$

This is the basic physical mechanism behind the anthropic upper bound. A sufficiently large positive vacuum energy prevents ordinary density fluctuations from producing enough gravitationally bound structures. 6

The bound can be made more quantitative. If galaxy formation has already begun by redshift \$z_c\$, the matter density at that time is

$$ \rho_m(z_c) = \rho_{m,0}(1+z_c)^3. $$

A vacuum density much larger than the matter density at that epoch would have dominated too early and strongly inhibited collapse. Weinberg’s classic estimate gives, for a spatially flat model,

$$ \rho_\Lambda \lesssim \frac{\pi^2}{3}(1+z_c)^3 \rho_{m,0}. $$

Taking \$z_c = 4\$ gives an upper bound of order \$4 \times 10^2 \rho_{m,0}\$. This is an order-of-magnitude argument, not a precision prediction of the observed value. It says that a cosmological constant vastly larger than the present matter density would make galaxy formation difficult. 6

A large negative cosmological constant creates a different problem. It eventually causes the Universe to recollapse. If the recollapse happens before galaxies, stars, heavy elements, and observers can develop, that value is also excluded by the selection condition. The positive and negative sides of the allowed interval therefore have different physical bounds.

4.2 From a bound to a probability distribution

An upper bound alone does not tell us what value an observer should expect. For that, we need a probability distribution.

Suppose a theory admits many vacua with different values of \$\rho_\Lambda\$. Let

  • \$p_{\rm prior}(\rho_\Lambda)\$ be the probability density before applying observer selection;
  • \$n_{\rm obs}(\rho_\Lambda)\$ be the expected number of observers, or a proxy such as the amount of matter ending up in suitable galaxies.

The probability density for an observer to measure a particular value is then

$$ P_{\rm obs}(\rho_\Lambda) = \frac{ p_{\rm prior}(\rho_\Lambda)n_{\rm obs}(\rho_\Lambda) }{ \displaystyle \int d\rho'_\Lambda\, p_{\rm prior}(\rho'_\Lambda)n_{\rm obs}(\rho'_\Lambda) }. $$

This is the central equation of the anthropic argument. It is a conditional probability: the prior distribution is reweighted by how many observers can arise in each kind of universe.

Now make two assumptions.

First, suppose the underlying theory produces a dense set of possible vacuum energies near zero, and the prior is nearly constant across the very narrow range compatible with galaxy formation:

$$ p_{\rm prior}(\rho_\Lambda) \simeq \mathrm{constant}. $$

Second, suppose the number of observers is roughly proportional to the amount of matter that ends up in galaxies capable of forming stars and planets. Then \$n_{\rm obs}\$ falls rapidly when the vacuum energy is so large that structure formation is suppressed.

Under these assumptions, the observer-weighted distribution is largest at values small enough to allow structure formation, but it need not be peaked at exactly zero. A typical observer can find a positive value that is not far below the upper anthropic range. This is the essence of Weinberg’s prediction. 7

The probability of observing a value no larger than ours is

$$ P(\rho_\Lambda \leq \rho_{\Lambda,\rm obs}) = \int_{-\infty}^{\rho_{\Lambda,\rm obs}} P_{\rm obs}(\rho_\Lambda)\,d\rho_\Lambda. $$

The result depends on the prior, the model of galaxy formation, and the choice of observer proxy. In one semi-analytic galaxy-formation study, assuming a flat prior and weighting observers by stellar mass, the observed value lay in the lower 6.7% of the predicted distribution. That is not a parameter-free prediction, but it illustrates that the selection argument can be made quantitative. 8

4.3 What the anthropic argument does and does not explain

The argument is impressive because it turns a vague statement about observers into a calculation involving structure formation. It also explains why a positive vacuum energy should not be arbitrarily large.

But it has demanding prerequisites.

There must be an ensemble of vacua or regions in which the vacuum energy actually varies. The values must scan densely enough that at least some lie within the narrow structure-forming range. The prior distribution must be understood, or at least be smooth over that range. Finally, we must have a defensible way to count observers or to use galaxies as their proxy.

String theory’s landscape is sometimes invoked as a possible source of such an ensemble. Different compactifications and flux choices may produce many metastable vacua with different low-energy parameters. If the resulting values of vacuum energy densely sample the region near zero, anthropic selection could explain why observers see a small value. But the existence and cosmological realization of the required ensemble, as well as the measure used to assign probabilities, remain unsettled. 9

There is also a conceptual limitation. The anthropic argument does not show why the vacuum energy of an individual vacuum is small. It says that observers are selected from a larger set of possibilities. It explains an observation only if the underlying ensemble and probability measure are part of a credible physical theory.

In that sense, anthropic reasoning is not a dynamical cancellation mechanism. It is an explanation based on selection.

5. Candidate resolutions

There is no accepted solution to the cosmological constant problem. Different proposals attack different parts of it. Some try to cancel vacuum energy, some alter its gravitational effect, some make it dynamical, and others explain why observers should measure a small value.

5.1 Supersymmetry and other symmetries

Supersymmetry relates bosons and fermions. In an exactly supersymmetric vacuum, their zero-point contributions cancel, and the vacuum energy can vanish. This is attractive because a symmetry can protect a small quantity against quantum corrections.

The difficulty is that supersymmetry is not observed as an exact symmetry of the low-energy world. If it is broken, the cancellation is incomplete. A breaking scale at or above the TeV scale typically gives a vacuum-energy contribution enormously larger than the observed value. Supersymmetry therefore illustrates the kind of protection one would like, but known four-dimensional versions do not solve the problem simply by being present. 4

Other proposed symmetries include scale symmetries and more exotic relations between positive- and negative-energy sectors. These ideas face their own problems: the symmetry must survive quantum corrections, be compatible with gravity, and avoid instabilities or unwanted new degrees of freedom.

5.2 Dynamical relaxation and adjustment mechanisms

Another idea is that the vacuum energy is not fixed once and for all. A new field could evolve in response to vacuum energy and drive the effective cosmological constant toward a small value.

The simplest versions introduce a scalar field whose potential energy changes over cosmic time. In principle, the field can roll toward a region of small effective vacuum energy. This is the intuition behind relaxation and adjustment mechanisms.

The obstacle is that a field which cancels an arbitrary constant must continue doing so after quantum corrections and phase transitions. Many simple local adjustment mechanisms run into no-go results or produce unacceptable cosmological behavior. More sophisticated constructions use extra dimensions, nonlocal ingredients, special couplings, or time-dependent solutions to evade the assumptions behind those results. 1 4

5.3 Quintessence: perhaps dark energy is not constant

Quintessence replaces the cosmological constant as the source of acceleration with a slowly rolling scalar field \$\phi\$. Its energy density and pressure are

$$ \rho_\phi = \frac{1}{2}\dot\phi^2 + V(\phi), \qquad p_\phi = \frac{1}{2}\dot\phi^2 - V(\phi). $$

If the potential energy dominates over the kinetic energy, then \$p_\phi \simeq -\rho_\phi\$, and the field behaves much like a cosmological constant. But unlike a true \$\Lambda\$, its density can change with time.

Quintessence can address the question of what drives the observed acceleration if future data show that \$w\$ differs from \$-1\$ or changes with time. It does not, by itself, solve the old cosmological constant problem. A scalar field can roll on top of a large vacuum energy, and the theory must still explain why that baseline contribution is so small. 10

This is a recurring distinction: explaining the dynamics of dark energy is not the same as explaining the smallness of vacuum energy.

5.4 Modified gravity and self-tuning

Perhaps the problem is not the vacuum energy, but our assumption about how gravity responds to it. General relativity may be an excellent approximation on Solar-System and astrophysical scales while requiring modification on cosmological scales.

Scalar-tensor theories, braneworld models, massive gravity, and more general modifications of the gravitational action have all been studied. Some attempt to make the gravitational response insensitive to a constant vacuum-energy contribution, while still allowing ordinary matter to gravitate normally. This is often called self-tuning.

The difficulty is severe. A viable theory must reproduce the observed expansion history, the growth of cosmic structure, gravitational lensing, gravitational-wave propagation, and the many successful local tests of general relativity. Some self-tuning constructions also encounter instabilities, strong coupling, or hidden fine-tuning. Modified gravity is a broad research program, not a single established solution. 11

5.5 Unimodular gravity: Lambda as an integration constant

In unimodular gravity, the gravitational field equations are written in a trace-free form. Taking their divergence and using energy-momentum conservation recovers Einstein’s equations with a cosmological constant that appears as an integration constant.

This changes the status of \$\Lambda\$. Instead of being a parameter inserted directly into the action, it arises from the integration of the field equations. That is conceptually interesting: the value of \$\Lambda\$ can be viewed as part of the initial or global data of the theory.

But changing the place where a constant enters the equations is not automatically the same as explaining why its value is small. Whether unimodular gravity addresses the radiative problem depends on the formulation, the global constraints, and how quantum corrections are treated. It is therefore a candidate framework, not a settled solution. 12

5.6 Vacuum-energy sequestering

Sequestering proposals attempt to make gravity insensitive to vacuum energy generated by matter loops. They introduce global variables or constraints into the action. These constraints can cancel contributions that would otherwise appear as a large cosmological constant in the gravitational field equations.

The attraction is that the cancellation is designed to survive quantum corrections from a protected matter sector, rather than requiring a separate fine-tuning after every correction. The price is structural: the theory uses global or nonlocal ingredients, and one must understand their consistency, cosmological implications, and interpretation. Sequestering is one of the more direct attempts to address the old problem itself. 13

5.7 Causal sets and a fluctuating Lambda

Causal set theory proposes that spacetime has a fundamentally discrete structure, with the number of elementary spacetime events related to spacetime volume. In some versions, the cosmological constant is not perfectly fixed but fluctuates.

The characteristic estimate is

$$ \Delta\Lambda \sim \frac{1}{\sqrt{V}}, $$

where \$V\$ is a spacetime four-volume. For a cosmological region, \$V\$ is roughly of order \$H^{-4}\$, so

$$ \Delta\Lambda \sim H^2. $$

This is intriguing because the observed cosmological constant is of the same order as the present Hubble scale squared. It offers a possible connection between the smallness of \$\Lambda\$ and the enormous spacetime volume of the observable Universe.

The idea remains speculative. Specific ever-present-\$\Lambda\$ models face observational constraints, including constraints from the isotropy of the cosmic microwave background. The scaling is suggestive, but it is not yet a demonstrated derivation of the observed value. 14

5.8 Quantum cosmology and vacuum selection

There are also proposals in which the value of the cosmological constant is selected by a quantum-cosmological probability distribution. Older ideas involving Euclidean path integrals and wormholes, for example, suggested an enhancement of the probability near zero vacuum energy.

These proposals are conceptually ambitious, but their conclusions depend on difficult questions: how to define the gravitational path integral, which boundary conditions to use, how to interpret the resulting probabilities, and whether the calculation describes the real Universe. They are interesting possibilities, not established predictions. 1

6. Does the cosmological constant actually exist?

The question is ambiguous, because it can mean several different things.

First, perhaps there is no fundamental cosmological-constant term in the gravitational action. That is logically possible. A theory such as unimodular gravity can place the effective constant in an integration constant rather than a coupling. But the observable gravitational equations can still contain an effective term with exactly the behavior of a cosmological constant.

Second, perhaps the observed acceleration is real, but is not caused by a constant vacuum energy. A slowly evolving scalar field, an interacting dark sector, or modified gravity could produce a similar expansion history. In that case, \$\Lambda\$ might be zero while the effective dark-energy component is not.

Third, perhaps there is no cosmic acceleration at all, and the apparent acceleration is an artifact of treating the Universe as homogeneous when it is not. Inhomogeneous cosmologies and backreaction proposals explore versions of this possibility. The challenge is to reproduce the full set of observations, not just the supernova distance-redshift relation. Within the standard homogeneous cosmology and general relativity, the evidence for late-time acceleration is strong. 3

The current observational situation is worth stating carefully. As of November 2026, DESI baryon-acoustic-oscillation measurements by themselves remain broadly consistent with flat \$\Lambda\$CDM. When combined with cosmic microwave background and supernova data, some analyses prefer a time-dependent dark-energy equation of state. A November 2026 review describes this as a moderate preference, not a definitive detection of dynamical dark energy. 2

The details depend on the data combination and the supernova calibration. For example, a 2026 reanalysis of the Dark Energy Survey supernova sample reduced one previously reported preference for evolving dark energy from \$4.2\sigma\$ to \$3.2\sigma\$, corresponding to Bayesian odds of only about 5:1 in favor of the more flexible model. That is interesting evidence, but it is not a discovery. 15

Even if future observations establish that dark energy evolves, the old cosmological constant problem will remain. A dynamical field does not explain why the vacuum contribution from quantum fields is so small. It changes the description of cosmic acceleration; it does not automatically solve the vacuum-energy hierarchy.

So the most defensible statement is not that the cosmological constant has been proved to exist, nor that it has been disproved. The data strongly support late-time acceleration within general relativity. Whether the source is a true constant, a dynamical component, or a modification of gravity is still an empirical question.

7. What would count as a solution?

A convincing solution should do more than reproduce the observed number by choosing a counterterm. It should explain why the small value is stable under quantum corrections and vacuum phase transitions. It should be compatible with the Standard Model and general relativity where those theories have been tested. It should also give a coherent account of the observed cosmic acceleration and, ideally, make predictions that distinguish it from competing models.

This gives us a way to evaluate proposals.

A symmetry-based solution should identify the symmetry and show how it protects the vacuum energy after the symmetry is broken. A dynamical solution should explain why the field evolves to the right value and why it does not spoil the early Universe. A modified-gravity solution must pass local and cosmological tests. An anthropic explanation must provide a physical ensemble, a sufficiently dense scan of vacuum energies, and a defensible probability measure. A sequestering or unimodular construction must show precisely how quantum corrections enter its global constraints.

No proposal has yet met all of these requirements in a way accepted by the community.

8. The deeper lesson

The cosmological constant problem sits at an unusual intersection. General relativity tells us that spacetime responds to energy and pressure. Quantum field theory tells us that vacuum states carry energy and that their contributions change when the microscopic theory changes. Cosmology tells us that the effective vacuum energy is tiny, positive, and dynamically important today.

The tension is not simply that a number is small. It is that the quantity appears to be sensitive to every energy scale in the theory, while observation asks for an almost perfect cancellation that remains stable across the history of the Universe.

Anthropic reasoning offers a striking possibility: perhaps the observed value is not typical of all vacua, but typical among the rare vacua where galaxies and observers can exist. Dynamical fields, modified gravity, unimodular theories, sequestering, and causal-set ideas offer different ways to change the underlying physics. None has yet become the accepted answer.

There is also a possibility that the question itself is pointing in the wrong direction. Perhaps vacuum energy is not what gravitates in the naive way suggested by combining quantum field theory with general relativity. Perhaps the cosmological constant is emergent, or the observed acceleration is not a cosmological constant at all.

That is why the problem remains so important. It may be telling us that one of our most successful theories is incomplete, not because it fails where we have tested it, but because we do not yet understand how its most basic ingredients fit together.

References


  1. S. Weinberg, “The Cosmological Constant Problem,” Reviews of Modern Physics 61, 1–23 (1989). https://doi.org/10.1103/RevModPhys.61.1 ↩︎ ↩︎ ↩︎

  2. X. Zhang, “Diagnosing Dark Energy in the Post-DESI Era: From Effective Dynamics to Multi-Probe Consistency Tests,” Science China Physics, Mechanics & Astronomy 69, 120401 (2026). https://doi.org/10.1007/s11433-026-3116-4 ↩︎ ↩︎

  3. S. M. Carroll, “The Cosmological Constant,” Living Reviews in Relativity 4, 1 (2001). https://doi.org/10.12942/lrr-2001-1 ↩︎ ↩︎ ↩︎ ↩︎ ↩︎ ↩︎

  4. N. Craig, “Naturalness: Past, Present, and Future,” European Physical Journal C 83, 825 (2023). https://doi.org/10.1140/epjc/s10052-023-11928-7 ↩︎ ↩︎ ↩︎ ↩︎ ↩︎ ↩︎ ↩︎ ↩︎ ↩︎

  5. P. J. E. Peebles and B. Ratra, “The Cosmological Constant and Dark Energy,” Reviews of Modern Physics 75, 559–606 (2003). https://doi.org/10.1103/RevModPhys.75.559 ↩︎

  6. S. Weinberg, “Anthropic Bound on the Cosmological Constant,” Physical Review Letters 59, 2607–2610 (1987). https://doi.org/10.1103/PhysRevLett.59.2607 ↩︎ ↩︎ ↩︎

  7. S. Weinberg, “A Priori Probability Distribution of the Cosmological Constant,” Physical Review D 61, 103505 (2000). https://doi.org/10.1103/PhysRevD.61.103505 ↩︎

  8. T. Sudoh, T. Totani, R. Makiya, and M. Nagashima, “Testing Anthropic Reasoning for the Cosmological Constant with a Realistic Galaxy Formation Model,” Monthly Notices of the Royal Astronomical Society 464, 1563–1568 (2017). https://doi.org/10.1093/mnras/stw2401 ↩︎

  9. R. Bousso and J. Polchinski, “Quantization of Four-form Fluxes and Dynamical Neutralization of the Cosmological Constant,” Journal of High Energy Physics 2000 (06), 006 (2000). https://doi.org/10.1088/1126-6708/2000/06/006 ↩︎

  10. E. J. Copeland, M. Sami, and S. Tsujikawa, “Dynamics of Dark Energy,” International Journal of Modern Physics D 15, 1753–1936 (2006). https://doi.org/10.1142/S021827180600942X ↩︎

  11. “Modified Gravity Approaches to the Cosmological Constant Problem,” Universe 9, 63 (2023). https://doi.org/10.3390/universe9020063 ↩︎

  12. P. Jiroušek, “Unimodular Approaches to the Cosmological Constant Problem,” Universe 9, 131 (2023). https://doi.org/10.3390/universe9030131 ↩︎

  13. N. Kaloper and A. Padilla, “Vacuum Energy Sequestering: The Framework and Its Cosmological Consequences,” Physical Review D 90, 084023 (2014). https://doi.org/10.1103/PhysRevD.90.084023 ↩︎

  14. S. Das, A. Nasiri, and Y. K. Yazdi, “Aspects of Everpresent Lambda. Part I: A Fluctuating Cosmological Constant from Spacetime Discreteness,” Journal of Cosmology and Astroparticle Physics 2023 (10), 047 (2023). https://doi.org/10.1088/1475-7516/2023/10/047 See also J. D. Barrow, “Strong Constraint on Ever-Present Lambda,” Physical Review D 75, 067301 (2007). https://doi.org/10.1103/PhysRevD.75.067301 ↩︎

  15. DES Collaboration, “The Dark Energy Survey Supernova Program: A Reanalysis of Cosmology Results and Evidence for Evolving Dark Energy with an Updated Type Ia Supernova Calibration,” Monthly Notices of the Royal Astronomical Society 548, stag632 (2026). https://doi.org/10.1093/mnras/stag632 ↩︎