Mimetic Dark Matter: Theory and Extensions

• Mimetic dark matter is a reformulation of general relativity that isolates the conformal mode, introducing a constrained scalar field whose dynamics mimic pressureless dust.
• The framework reproduces key cosmological observations by yielding cold dark matter behavior in FRW spacetimes and compatible linear scalar perturbations.
• Extensions of the original model include higher-derivative terms and vector couplings to address stability issues, leading to potential applications in inflation and bouncing cosmologies.

Mimetic dark matter is a reformulation of general relativity in which the conformal degree of freedom of the metric is isolated covariantly and promoted into a constrained scalar mode whose stress tensor is exactly of pressureless-dust type. In its original form, due to Chamseddine and Mukhanov, the construction does not add particle dark matter; instead, the extra gravitational degree of freedom mimics cold dark matter at the level of the Einstein equations (Chamseddine et al., 2013). In later literature, the term also covers a broader family of constrained scalar, vector, scalar–vector–tensor, and k-essence-like extensions that preserve the central mimetic mechanism while modifying stability, sound speed, or cosmological phenomenology (Chamseddine et al., 2014).

1. Foundational construction

The original construction introduces an auxiliary metric $\tilde g_{\mu\nu}$ and a scalar field $\phi$, and defines the physical metric by

$$g_{\mu\nu} \equiv P\,\tilde g_{\mu\nu}, \qquad P \equiv \tilde g^{\alpha\beta}\partial_\alpha\phi\,\partial_\beta\phi.$$

Because under $\tilde g_{\mu\nu}\to \Omega^2\tilde g_{\mu\nu}$ one has $P\to P$, the physical metric is invariant under Weyl rescalings of the auxiliary metric. The transformation is therefore singular: only the conformal class of $\tilde g_{\mu\nu}$ matters, and the conformal mode is carried by $\phi$ rather than by an independent component of $\tilde g_{\mu\nu}$ (Chamseddine et al., 2013).

A direct consequence is the normalization condition

$g^{\mu\nu}\partial_\mu\phi\,\partial_\nu\phi = 1.$

With signature $(+,-,-,-)$ this means that $\phi$0 is a unit timelike covector. In mostly-plus conventions the same condition is written with the opposite sign, $\phi$1; the difference is purely conventional (Chamseddine et al., 2013).

The conceptual point emphasized early in the literature is that the mimetic scalar is not introduced as ordinary matter. Rather, it is the conformal mode of gravity rendered dynamical by a non-invertible field redefinition. A common misconception is to interpret the model as a standard scalar–tensor theory. In the original proposal, the Einstein–Hilbert action is still built from the physical metric alone; the novelty is that the metric itself is parametrized in a singular, conformally invariant way (Chamseddine et al., 2013).

2. Field equations and effective dust

Varying the Einstein–Hilbert action written in terms of the physical metric gives the modified field equations

$\phi$2

together with the scalar equation

$\phi$3

Here $\phi$4 and $\phi$5. The right-hand side can be identified as an effective mimetic energy–momentum tensor,

$\phi$6

Defining $\phi$7, the constraint gives $\phi$8, so the tensor takes the pressureless-dust form

$\phi$9

The flow is geodesic because $$g_{\mu\nu} \equiv P\,\tilde g_{\mu\nu}, \qquad P \equiv \tilde g^{\alpha\beta}\partial_\alpha\phi\,\partial_\beta\phi.$$0 is a gradient with fixed norm, and the scalar equation becomes the dust continuity equation (Chamseddine et al., 2013).

An equivalent and widely used formulation introduces a Lagrange multiplier,

$$g_{\mu\nu} \equiv P\,\tilde g_{\mu\nu}, \qquad P \equiv \tilde g^{\alpha\beta}\partial_\alpha\phi\,\partial_\beta\phi.$$1

Variation with respect to $$g_{\mu\nu} \equiv P\,\tilde g_{\mu\nu}, \qquad P \equiv \tilde g^{\alpha\beta}\partial_\alpha\phi\,\partial_\beta\phi.$$2, $$g_{\mu\nu} \equiv P\,\tilde g_{\mu\nu}, \qquad P \equiv \tilde g^{\alpha\beta}\partial_\alpha\phi\,\partial_\beta\phi.$$3, and $$g_{\mu\nu} \equiv P\,\tilde g_{\mu\nu}, \qquad P \equiv \tilde g^{\alpha\beta}\partial_\alpha\phi\,\partial_\beta\phi.$$4 gives

$$g_{\mu\nu} \equiv P\,\tilde g_{\mu\nu}, \qquad P \equiv \tilde g^{\alpha\beta}\partial_\alpha\phi\,\partial_\beta\phi.$$5

Taking the trace identifies $$g_{\mu\nu} \equiv P\,\tilde g_{\mu\nu}, \qquad P \equiv \tilde g^{\alpha\beta}\partial_\alpha\phi\,\partial_\beta\phi.$$6. Golovnev showed that this Lagrange-multiplier action is dynamically equivalent to the original singular-metric formulation and makes the dust interpretation transparent (Golovnev, 2013).

Even in vacuum, the mimetic density need not vanish. For $$g_{\mu\nu} \equiv P\,\tilde g_{\mu\nu}, \qquad P \equiv \tilde g^{\alpha\beta}\partial_\alpha\phi\,\partial_\beta\phi.$$7 one has $$g_{\mu\nu} \equiv P\,\tilde g_{\mu\nu}, \qquad P \equiv \tilde g^{\alpha\beta}\partial_\alpha\phi\,\partial_\beta\phi.$$8, so the conformal mode becomes dynamically nontrivial and gravitates as an effective cold dark matter component (Chamseddine et al., 2013).

3. Cosmological dynamics

In a spatially flat FRW spacetime,

$$g_{\mu\nu} \equiv P\,\tilde g_{\mu\nu}, \qquad P \equiv \tilde g^{\alpha\beta}\partial_\alpha\phi\,\partial_\beta\phi.$$9

the constraint is solved by $\tilde g_{\mu\nu}\to \Omega^2\tilde g_{\mu\nu}$0. The conservation equation then gives

$\tilde g_{\mu\nu}\to \Omega^2\tilde g_{\mu\nu}$1

or, more generally, $\tilde g_{\mu\nu}\to \Omega^2\tilde g_{\mu\nu}$2. At the homogeneous level the mimetic component is therefore indistinguishable from cold pressureless matter, and when ordinary matter and a cosmological constant are included the background expansion reproduces the standard $\tilde g_{\mu\nu}\to \Omega^2\tilde g_{\mu\nu}$3CDM history (Chamseddine et al., 2013).

This dust interpretation extends to linear scalar perturbations in the minimal model. The effective four-velocity is geodesic, the sound speed vanishes, and the continuity equation reproduces the standard growth of a pressureless component. The same dust-like limit appears in later phenomenological implementations, for example in models where a Lagrange multiplier $\tilde g_{\mu\nu}\to \Omega^2\tilde g_{\mu\nu}$4 obeys $\tilde g_{\mu\nu}\to \Omega^2\tilde g_{\mu\nu}$5 whenever $\tilde g_{\mu\nu}\to \Omega^2\tilde g_{\mu\nu}$6 (Matsumoto, 2016).

Once a potential $\tilde g_{\mu\nu}\to \Omega^2\tilde g_{\mu\nu}$7 is added, the mimetic sector can mimic much more than cold dark matter. The background dynamics can be reduced to a linear equation for a power of the scale factor, and this was used to construct mimetic inflation, quintessence-like evolution, and nonsingular bouncing cosmologies (Chamseddine et al., 2014). A separate line of work used a specific potential,

$\tilde g_{\mu\nu}\to \Omega^2\tilde g_{\mu\nu}$8

to split the mimetic sector into a dust-like dark matter contribution $\tilde g_{\mu\nu}\to \Omega^2\tilde g_{\mu\nu}$9 and a dark-energy-like contribution $P\to P$0, while preserving standard growth in the regime $P\to P$1 (Matsumoto, 2016).

The minimal theory, however, has vanishing scalar sound speed. To obtain propagating scalar modes during inflation, higher-derivative terms were introduced. In one such extension the added term $P\to P$2 generates a sound-speed parameter

$P\to P$3

which allows quantum scalar perturbations and CMB-scale inflationary phenomenology (Saadi, 2014).

4. Variational structure and Hamiltonian interpretation

A central clarification due to Golovnev is that the mimetic substitution changes the class of admissible variations. After replacing the metric by a derivative-dependent expression in $P\to P$4, one is no longer varying over the full space of metrics. In that restricted variational problem, the trace of Einstein’s equations becomes an identity and the missing trace information reappears as the scalar continuity equation. This is why the theory is not dynamically equivalent to ordinary general relativity despite being built from the Einstein–Hilbert action (Golovnev, 2013).

The Hamiltonian analysis sharpened the distinction between the dust interpretation and genuine stability. For healthy mimetic scalar–tensor theories, the mimetic contribution to the Hamiltonian is bounded from below if and only if the mimetic energy density $P\to P$5 is positive. In the shift-symmetric mimetic dark matter model, the conserved current implies that $P\to P$6 is time-independent in unitary gauge, so positivity of $P\to P$7 is preserved by suitable initial and boundary conditions. In models without shift symmetry, positivity can be enforced by redefining $P\to P$8 (Ganz et al., 2018).

The same analysis also established that mimeticization does not cure an unhealthy seed theory. In particular, adding the mimetic constraint to a theory with a higher-derivative $P\to P$9 term does not remove the Ostrogradsky ghost, and mimetic $\tilde g_{\mu\nu}$0 gravity is healthy only if the usual $\tilde g_{\mu\nu}$1, $\tilde g_{\mu\nu}$2 conditions hold together with $\tilde g_{\mu\nu}$3 (Ganz et al., 2018).

A related Hamiltonian treatment of the original scalar model concluded that it describes regular pressureless dust only as long as the dust energy density remains positive under time evolution. For certain inhomogeneous configurations the density can evolve through zero, and because the scalar Hamiltonian is linear in the corresponding momentum, the theory can then become unstable (Chaichian et al., 2014).

5. Major extensions of the mimetic framework

A large fraction of the subsequent literature is devoted to giving the mimetic sector a controlled deviation from exact dust. In the effective-theory treatment of cosmological perturbations, the scalar quadratic action was written as

$\tilde g_{\mu\nu}$4

The minimal higher-derivative mimetic model has a gradient instability, but adding an intrinsic–extrinsic curvature mixing term $\tilde g_{\mu\nu}$5 and a tuned $\tilde g_{\mu\nu}$6 term can yield a stable “imperfect dark matter” phase with $\tilde g_{\mu\nu}$7 and $\tilde g_{\mu\nu}$8 (Hirano et al., 2017).

Another major branch replaces the scalar-only sector by vector or scalar–vector structures. In the tensor–vector and tensor–vector–scalar models, the physical metric is built from a vector norm, or from a vector norm modulated by a scalar-dependent factor $\tilde g_{\mu\nu}$9, rather than from the scalar gradient alone. The original scalar model was found to be conditionally stable, whereas both the vector and tensor–vector–scalar versions were argued to be free of ghost instabilities (Chaichian et al., 2014). In the later MiTeVeS construction, a constrained timelike vector coupled derivatively to a scalar generates a conserved Noether current; on homogeneous FRW backgrounds this produces a dark-matter component

$\phi$0

alongside a stiff component $\phi$1 and a cosmological-constant-like term (Benisty et al., 2021).

The mimetic idea was also generalized to a Weyl-invariant scalar–vector–tensor theory that is on-shell equivalent to general relativity plus k-essence. In that construction the physical metric is $\phi$2 with $\phi$3 of Weyl weight $\phi$4, and the field equations reduce to those of a k-essence Lagrangian $\phi$5, where the overall normalization $\phi$6 is a global integration constant. A distinctive result is that the reconstruction fails at $\phi$7: the mimetic fluid cannot classically cross an ultra-relativistic equation of state (Jiroušek et al., 2022).

More formal extensions include a supergravity embedding in which a constrained chiral multiplet coupled to $\phi$8 supergravity and a Lagrange multiplier multiplet spontaneously breaks supersymmetry, leaving a graviton, a massive gravitino, and two scalar fields representing mimetic dark matter (Chamseddine, 2021).

6. Criticisms, limitations, and recent directions

The minimal model behaves exactly like cold dust, and it inherits dust’s nonlinear limitations. In particular, caustics are expected in multistream regions, and the original brief note did not provide a regularization mechanism. Later higher-derivative cures can improve sound speed or gradient stability, but they also raise sign and consistency questions that must be checked case by case (Chamseddine et al., 2013).

A distinct criticism concerns cosmological initial conditions. A detailed study of the basic mimetic dark matter model argued that it has no natural mechanism to generate adiabatic initial conditions. In that formulation both the background and first-order perturbations contain arbitrary functions, and matching CMB and matter-power-spectrum data requires functional fine-tuning. Using a modified version of [CLASS](https://www.emergentmind.com/topics/colorado-learning-attitudes-about-science-survey-class), the analysis found that Planck data already disfavor $\phi$9 deviations in mimetic velocity initial conditions, while Euclid-like surveys would probe percent-level departures (Khalifeh et al., 2019).

Recent work has focused directly on these two issues: stability and production. One 2025 proposal introduced a second scalar $\tilde g_{\mu\nu}$0 and imposed $\tilde g_{\mu\nu}$1 with a positive function $\tilde g_{\mu\nu}$2, arguing that the model is ghost free when $\tilde g_{\mu\nu}$3 and caustic free because the effective dark matter acquires a small pressure. In that construction the effective dark-matter equation of state is

$\tilde g_{\mu\nu}$4

when $\tilde g_{\mu\nu}$5, so the model remains close to $\tilde g_{\mu\nu}$6CDM while deviating from exact dust (Kanambaye, 5 Jul 2025).

A separate 2026 direction couples the mimetic field to the Gauss–Bonnet invariant during inflation. For the choice $\tilde g_{\mu\nu}$7 and $\tilde g_{\mu\nu}$8, the resulting dust-like component at the end of inflation has

$\tilde g_{\mu\nu}$9

and matter–radiation equality occurs at

$g^{\mu\nu}\partial_\mu\phi\,\partial_\nu\phi = 1.$0

For $g^{\mu\nu}\partial_\mu\phi\,\partial_\nu\phi = 1.$1, $g^{\mu\nu}\partial_\mu\phi\,\partial_\nu\phi = 1.$2, and $g^{\mu\nu}\partial_\mu\phi\,\partial_\nu\phi = 1.$3, the corresponding $g^{\mu\nu}\partial_\mu\phi\,\partial_\nu\phi = 1.$4 is stated to be consistent with observations (Chamseddine et al., 9 Jan 2026).

This suggests that the contemporary status of mimetic dark matter is bifurcated. The original theory remains important as a clean demonstration that a constrained conformal mode can reproduce dust-like gravity without particle dark matter. At the same time, most current work is directed toward establishing whether a mimetic sector can be made simultaneously ghost free, caustic free, adiabatic, and observationally competitive once perturbations and nonlinear dynamics are treated in full detail.