A closed timelike curve is a path in spacetime that returns to the same event while pointing toward the local future at every point. Because its tangent is everywhere timelike, it represents the kind of trajectory that a massive observer could locally follow without exceeding the speed of light. Its closure nevertheless permits an observer to encounter events in the observer’s own past. The concept is therefore more precise than the general idea of backward time travel: it concerns the causal geometry of an entire spacetime, not merely a clock running backward or an unusual choice of coordinates.
Closed timelike curves occur in some exact solutions of general relativity, including rotating cosmological models and certain idealized extensions of rotating black-hole geometries. Their mathematical existence does not establish that they occur in nature or can be constructed. Their importance lies in exposing a distinction between the local laws of relativity, which constrain motion within light cones, and the global organization of those cones, which need not automatically exclude causal loops.
Research on these curves connects differential geometry, gravitational dynamics, the initial-value problem, and quantum field theory. Central questions include whether initially causal conditions can evolve into a region containing causal loops, whether quantum effects prevent their formation, and whether physical laws can consistently describe interactions in their presence.
Geometric definition and physical meaning
In a time-oriented Lorentzian manifold, a closed timelike curve can be represented by a smooth map whose tangent is everywhere timelike and consistently future-directed. With metric signature , the timelike requirement is
Equivalently, a parametrized curve on an interval returns to its starting event, , with compatible tangent directions at the join. Discussions also permit piecewise-smooth curves with physically interpretable changes of velocity. The important feature is not the particular parametrization but the existence of a continuous, locally future-directed timelike route around the loop.
A timelike traveler measures positive elapsed proper time along every segment. If a closed curve is traversed once, the accumulated proper time is
There is no requirement that the traveler’s clock reverse. The conflict is between positive local aging and return to the same spacetime event. Nor does an arbitrary traveler have to repeat an identical physical state after one circuit: closure of a geometric trajectory and consistency of the matter carried along it are separate questions. A complete physical model must make those conditions compatible.
The word “event” is essential. Returning to the same spatial location on a later date is an ordinary occurrence, not a closed timelike curve. Likewise, a periodic orbit around a planet is spatially closed but remains open in spacetime because its successive revolutions occur at different times. A coordinate time that decreases along part of a trajectory is also insufficient evidence. Coordinates may behave poorly, whereas the timelike character and closure of a curve are coordinate-independent properties.
A closed timelike curve need not be a geodesic. A geodesic describes free fall, while a general timelike curve may require acceleration supplied by engines, supports, or other forces. Establishing that a spacetime contains closed timelike curves therefore does not establish that an unpowered observer could follow one. Conversely, the absence of closed timelike geodesics does not guarantee the absence of accelerated closed timelike trajectories.
Closed null curves are related but distinct. Their tangent is everywhere lightlike, so they describe possible light propagation rather than massive observers. The distinction matters at boundaries where causal behavior first degenerates: closed or repeatedly returning null trajectories may arise even where no closed timelike curve passes through the boundary itself.
Causal order and the hierarchy of spacetime conditions
The geometric language of causal futures makes chronology violation precise. The chronological future of an event consists of all events reachable from along future-directed timelike curves. A closed timelike curve exists through exactly when
A spacetime satisfies the chronology condition if this never happens. This condition expresses a basic form of causal order: no event lies strictly to its own timelike future. The set of points through which closed timelike curves pass is called the chronology-violating set. It is open, reflecting the fact that timelike relationships persist under sufficiently small perturbations.
Chronology is only one level in a hierarchy of causal conditions. The causality condition also excludes closed null curves. Strong causality rules out causal paths that leave and return to arbitrarily small neighborhoods in a way that defeats a well-behaved local causal ordering. Stable causality requires that causal loops remain absent under a suitable slight widening of the light cones. This is stronger than merely finding no loop in one exactly specified geometry.
A particularly useful characterization of stable causality is the existence of a globally defined time function that increases along every future-directed causal curve. Under the standard smooth spacetime assumptions, a smooth temporal function with timelike gradient can be used. If a function increases strictly along a future-directed timelike path, that path cannot return to its starting point: closure would require the final and initial values of to agree. Thus a global temporal ordering supplies a direct obstruction to closed timelike curves.
The stronger condition of global hyperbolicity is central to predictive physics. One standard characterization combines causality with compactness of causal diamonds , where include lightlike as well as timelike connections. Another uses the existence of a Cauchy surface, a spacelike hypersurface intersected exactly once by every inextendible causal curve. Such a surface allows the entire spacetime history, within the relevant domain, to be organized around initial data.
These distinctions explain why “no time travel” and “a well-posed global evolution” are not interchangeable claims. A spacetime may have no closed timelike curves yet still lack a Cauchy surface because of a missing boundary, singular behavior, or causal influences entering from elsewhere. Conversely, any spacetime containing a closed timelike curve fails to be globally hyperbolic: a loop is incompatible with a hypersurface that every inextendible causal history crosses exactly once.
Rotating universes, cylinders, and cosmic strings
Kurt Gödel’s 1949 rotating cosmological solution provided a decisive example of chronology violation within general relativity. It solves the Einstein field equations with pressureless matter and a negative cosmological constant. The matter congruence has nonzero rotation, and the geometry is homogeneous. Although each observer has an ordinary local light cone, the arrangement of those cones over sufficiently large distances permits closed timelike paths. Gödel’s universe contains such curves through every event, so its chronology violation is not confined to a distant exceptional region.
The solution matters because its causal loops do not require a traveler to cross a local light-speed barrier. A timelike path can remain within the allowed future cone at each point while the global geometry brings it back to its origin. Descriptions in which light cones “tip over” can convey this behavior in a chosen coordinate system, but the invariant statement is the existence of closed future-directed timelike curves. Gödel’s model is not an observational description of the expanding universe: it is a nonexpanding rotating cosmology, unlike the standard large-scale cosmological models used to interpret observations.
Rotating cylindrical systems supply another family of examples. Willem Jacob van Stockum studied an exact solution for rotating dust in 1937. Analyses of rotating cylinders, including work by Frank Tipler in 1974, showed how sufficiently extreme rotational geometries can admit closed timelike curves. In an axisymmetric metric with periodic angular coordinate , an orbit at fixed values of the other coordinates becomes timelike wherever
Because returns to the same value after , such an orbit is then a closed timelike curve. This test is useful in several rotating geometries, although it identifies only one possible family of curves.
The familiar rotating-cylinder constructions rely heavily on idealization, particularly infinite length or nonordinary global boundary conditions. They do not establish that a finite cylinder made from physically achievable material can function as a time machine. The distinction between an exact solution and a realizable object is especially important here: removing the infinite extent changes the gravitational problem rather than merely making the apparatus smaller.
In 1991, J. Richard Gott proposed a configuration involving two rapidly moving cosmic strings. An ideal straight string produces a conical geometry, and the combined spacetime identifications associated with moving strings can allow causal loops under appropriate conditions. Subsequent work showed why constructing such a configuration from ordinary causal initial conditions is a separate and restrictive problem. Cosmic strings themselves remain hypothetical, and infinitely extended ideal strings do not provide a demonstrated engineering route to chronology violation.
Rotating black holes and analytic extensions
The Kerr metric, discovered by Roy Kerr in 1963, describes an uncharged rotating black hole in vacuum. In geometrized units, its parameters are the mass and the specific angular momentum . For , the ideal stationary solution has outer and inner horizon radii
The outer horizon is the familiar event horizon. The inner horizon is a Cauchy horizon in the maximally extended exact solution, beyond which data from an ordinary exterior initial region no longer uniquely determine the continuation in the same way.
The maximal analytic extension of Kerr spacetime contains regions with closed timelike curves. In Boyer–Lindquist coordinates, the angular metric coefficient is
In parts of the negative- extension near the ring singularity, this coefficient becomes negative. The periodic angular direction is then timelike, producing closed timelike circles. This is a statement about the global extension of an exact mathematical solution, not about the ordinary exterior environment of an astrophysical black hole.
Several distinctions prevent this result from being interpreted as an available route into the past. The event horizon is not itself a chronology horizon, and crossing it does not by itself place a traveler on a causal loop. The ergoregion outside the outer horizon is also not a time-travel region: within it, rotation prevents observers from remaining static relative to infinity, but future-directed timelike motion need not close. Chronology violation in the analytic extension occurs much deeper in the idealized structure.
The principal physical difficulty is the behavior of the inner horizon under perturbations. Radiation moving through the interior can become enormously blueshifted near the Cauchy horizon. Work by Eric Poisson and Werner Israel in 1990 developed the mass-inflation picture, in which interacting streams produce rapid growth of an internal mass parameter and curvature. Later analyses show a variety of singular behaviors whose details depend on the system and assumptions.
Consequently, the maximally extended stationary Kerr solution should not be treated as a literal map of an interior produced by stellar collapse. The physical status of its deeper regions is tied to questions of stability, singularities, and strong cosmic censorship. Observational evidence for rotating black holes supports the relevance of Kerr-like exterior geometries; it does not establish the existence of the chronology-violating portions of their ideal mathematical extensions.
Traversable wormholes and temporal offsets
A wormhole is a spacetime connection between regions that would otherwise be separated by a longer route through the surrounding geometry. Not every wormhole is traversable. The bridge in the maximally extended Schwarzschild solution, for example, does not allow a traveler to pass from one exterior region to the other. Discussions of wormhole time machines instead concern hypothetical traversable throats that remain open long enough for matter and signals to pass through them.
Michael Morris and Kip Thorne analyzed traversable wormhole geometries in 1988, and work that year by Morris, Thorne, and Ulvi Yurtsever connected them to chronology violation. A commonly studied static, spherically symmetric line element is
where is a redshift function and is a shape function. A throat occurs at a minimum areal radius satisfying . Avoiding a horizon and obtaining the required outward flaring place significant restrictions on the geometry and the matter supporting it.
A time-machine construction adds a temporal offset between two mouths. If one mouth undergoes a relativistic journey while the other remains relatively stationary, time dilation can produce unequal elapsed times along their external histories. A difference in gravitational potential can also produce differential aging. In an idealized model, the internal wormhole connection relates mouth events in such a way that this external aging difference becomes a shortcut not only through space but through external time.
Suppose the mouth separation in an approximately stationary exterior is , the temporal offset is , and traversal time through the throat is negligible for the estimate. A causal loop can become possible when
A traveler can then combine passage through the wormhole with a sufficiently rapid, but subluminal, return through ordinary space. Finite throat traversal time and nontrivial exterior geometry modify this simple criterion. The invariant requirement remains that the combined route be future-directed and timelike everywhere while returning to its initial event.
The supporting matter is a major obstacle. In classical general relativity, the standard Morris–Thorne throat requires violation of the null energy condition near the throat, meaning that for some null vector . Quantum fields can exhibit negative renormalized energy densities in restricted settings, but that fact does not establish the existence of a stable macroscopic traversable wormhole. Nor does a temporal offset normally grant access to arbitrarily remote history: in these constructions, the accessible past is limited by the history of the mouths and the establishment of their time shift.
Formation from initial data and chronology protection
The existence of a spacetime containing closed timelike curves is different from the formation of such curves from initially causal conditions. Gödel’s universe, for example, already contains chronology violation throughout its history. It does not describe an apparatus built in an ordinary region that later changes the causal structure. The latter question requires a dynamical account of matter, geometry, and the boundary separating predictable evolution from any chronology-violating extension.
The initial-value formulation of general relativity begins with geometric and matter data on a spacelike hypersurface, subject to constraint equations. Standard results associated with Yvonne Choquet-Bruhat and Robert Geroch establish the existence and uniqueness, up to the appropriate equivalence, of a maximal globally hyperbolic development under suitable assumptions. That development cannot contain closed timelike curves. A proposed time machine must therefore involve a continuation beyond the globally hyperbolic region, a failure of an assumption, or initial conditions that were not globally causal in the required sense.
A Cauchy horizon marks a boundary of the domain determined by initial data. A chronology horizon refers to a boundary associated with the onset of chronology violation. These notions often meet in time-machine models, but they are not synonymous. A Cauchy horizon can arise without closed timelike curves, and the boundary of a chronology-violating region need not have all the properties assumed in a particular theorem. Care is needed when transferring conclusions between different constructions.
One important class consists of compactly generated horizons. Roughly, their null generators, followed into the past, eventually remain within a compact region. This captures aspects of a localized time machine, rather than a spacetime whose unusual causal behavior depends on infinite structures or remote boundary conditions. Theorems involving such horizons relate their possible formation to focusing of light rays, energy conditions, completeness, and genericity assumptions. Their force depends on those hypotheses; no single classical theorem excludes every spacetime with causal loops.
Stephen Hawking proposed the chronology protection conjecture in 1992: physical laws might prevent the appearance of closed timelike curves. One motivation was that radiation circulating through a nearly closed null route could return repeatedly, become blueshifted, and generate a large stress-energy contribution that alters the geometry before a time machine forms. This is a proposed physical obstruction, not a general proof. The conjecture remains distinct both from the established mathematical existence of chronology-violating solutions and from the stronger claim that every possible quantum theory of gravity must forbid them.
Self-consistency, paradoxes, and classical dynamics
The best-known objection to travel into the past is the grandfather paradox: an action prevents a condition necessary for that same action to occur. In geometric terms, however, general relativity supplies a spacetime and local equations of motion, not a rule allowing a history to be established and then overwritten. If an interaction lies on a closed causal structure, every occurrence of the relevant event must belong to one mutually compatible solution. An inconsistent proposed sequence is not automatically a solution of the field and matter equations.
The Novikov self-consistency principle, associated with Igor Novikov’s work from the 1980s onward, makes this requirement explicit: only globally self-consistent histories are physically admissible in a spacetime containing causal loops. A traveler’s intervention in the past would already be part of that past. The principle does not prove that closed timelike curves exist or that consistent solutions always exist. It specifies the kind of consistency a history would need if the underlying spacetime permitted them.
Classical collision models illustrate the issue without relying on human intentions. Joseph Polchinski suggested a billiard-ball scenario in which a ball passes through a wormhole, emerges at an earlier exterior time, and collides with its younger self so as to prevent the original entry. Analyses by Fernando Echeverria, Gunnar Klinkhammer, and Kip Thorne in 1991 found self-consistent trajectories in model settings, including glancing collisions that cause the younger ball to enter the wormhole along the trajectory required for the older ball to emerge.
Such examples replace an apparent contradiction with a boundary-value problem, but they do not restore ordinary determinism. Data specified outside the chronology-violating region can admit more than one globally consistent continuation. In other systems or with other boundary requirements, existence and uniqueness require separate analysis. Local equations may remain unchanged while their global solution space becomes radically different. A consistency condition is therefore not equivalent to an ordinary initial-value law with a unique prediction.
Causal loops can also produce bootstrap situations, such as information transmitted to the past and later retransmitted without an identifiable first creation event. A closed chain of explanations is unusual, but it is not itself a violation of a local conservation equation. Whether a particular loop is physically possible depends on the dynamics of its carrier, degradation, interactions, and boundary conditions. The second law of thermodynamics does not provide a simple stand-alone prohibition: thermodynamic irreversibility involves statistical assumptions and boundary conditions that must themselves be examined in a causally nonstandard spacetime.
Quantum fields, fixed points, and proposed quantum models
Quantum field theory is most straightforward on globally hyperbolic backgrounds, where initial data and commutation relations support a controlled evolution. Closed timelike curves obstruct that standard framework. Fields can encounter multiple causal routes between events, and specifying their state near a chronology horizon may fail to produce the short-distance structure needed to define observables. The problem is not merely that quantum particles behave unpredictably; it concerns whether the mathematical quantities used in semiclassical physics remain well defined.
In semiclassical gravity, the metric responds to an expectation value of quantum stress-energy:
Some calculations find that vacuum polarization or circulating radiation becomes large near a chronology horizon, supporting a backreaction mechanism for chronology protection. A divergent result on a prescribed background, however, does not by itself determine the final geometry. Once backreaction becomes important, the original background may no longer be a valid approximation.
A major result by Bernard Kay, Marek Radzikowski, and Robert Wald in 1997 concerns quantum fields at compactly generated Cauchy horizons. Under the theorem’s assumptions, an initially suitable state cannot maintain the required Hadamard short-distance behavior at certain horizon points. Standard point-splitting methods then fail to define the renormalized stress-energy tensor there. This is more precise than the claim that a computable stress tensor simply becomes infinite everywhere. It shows a breakdown of the usual semiclassical description in an important class of candidate time-machine settings.
A separate line of research treats a closed timelike curve as an abstract information-processing resource. David Deutsch proposed a quantum consistency model in 1991. If an ordinary system with density operator interacts unitarily with a system on a loop, the loop state is required to satisfy
The equation imposes a fixed point: after the interaction, the loop system’s reduced state must match the state supplied to that interaction. In finite dimensions, fixed-point existence follows from properties of continuous maps on the compact convex set of density operators.
Deutsch’s model and alternatives based on postselected quantum processes are not interchangeable theories. They can lead to different predictions, and their effective behavior need not preserve the linear input-output structure of ordinary isolated quantum mechanics. Laboratory demonstrations can emulate selected consistency equations using conventional quantum devices, entanglement, and postselection. Such demonstrations test information-processing models; they do not produce a spacetime loop or show a signal arriving in an actual earlier event.
Observational status and tests of physical realizability
No observation has established the existence of a closed timelike curve. Tests of general relativity measure phenomena such as gravitational time dilation, orbital dynamics, gravitational lensing, and gravitational waves. They constrain particular geometries and dynamical processes, usually in regions with ordinary causal structure. Agreement with general relativity in those regimes does not verify every global extension admitted by its equations.
The same limitation applies to black-hole observations. Measurements consistent with rotating black holes concern the exterior geometry, radiation from surrounding matter, or the dynamics of mergers. They do not directly probe a negative- analytic extension or a chronology-violating region beyond an unstable inner horizon. Inferring such regions from exterior agreement would require additional assumptions about the entire interior solution and its stability.
Cosmological evidence also distinguishes an exact mathematical counterexample from a viable model of the universe. Gödel’s homogeneous rotating solution demonstrates that Einstein’s equations alone do not guarantee chronology, but its nonexpanding geometry is not the standard observational cosmology. Searches for cosmic strings constrain hypothetical defects; they have not established the rapidly moving ideal configurations required by the Gott construction. Traversable wormholes likewise remain hypothetical rather than observed channels of communication.
Apparent backward propagation in experiments requires a different analysis. Anomalous pulse velocities, delayed-choice experiments, quantum correlations, and postselected simulations do not by themselves demonstrate a controllable message reaching its sender’s past. A claim of a closed timelike curve would need a physically meaningful account of the relevant events and their causal connections, not simply an unusual ordering of coordinate labels or a reconstructed signal peak.
For theoretical proposals, physical realizability is assessed through several independent questions: whether the stress-energy is physically allowed, whether the configuration can arise from admissible initial data, whether it survives perturbations, whether quantum observables remain definable, and whether a complete causal route can actually be traversed. A solution can satisfy the local gravitational equations while failing one or more of these tests. Research in quantum gravity addresses regimes where the classical or semiclassical description breaks down, including the high curvature and unstable horizons encountered in many chronology-violating constructions; a universal, experimentally established rule governing those regimes is not yet available.