---
title: Doubly Special Relativity (DSR)
url: https://www.emergentmind.com/topics/doubly-special-relativity-dsr
type: topic
---

# Doubly Special Relativity (DSR)

Doubly Special Relativity (DSR) denotes a class of relativistic frameworks in which, besides the observer-independent speed scale \(c\), one postulates an additional observer-independent ultraviolet scale, usually a short length \(\ell_p\) or an equivalent high-energy scale \(\kappa\) or \(E_P\), while preserving a relativity principle among inertial observers [1003.3942]. In the formulation summarized by Amelino-Camelia, DSR is characterized by the Relativity Principle (RP), an observer-independent infrared light-speed scale \(c\), and an observer-independent short-length or high-momentum scale \(\ell_p\), introduced through an agreed measurement procedure \(M_{\ell_p}\) [1003.3942]. Operationally, DSR is implemented through nonlinear deformations of Lorentz transformations on energy-momentum space, modified Casimir invariants, and, in many realizations, nontrivial momentum-composition laws and noncommutative spacetime structures. The framework has been developed both as a formal generalization of special relativity and as a candidate kinematics for quantum-gravity phenomenology, especially near the Planck regime [1003.3942; 2507.13192; 2605.04422].

## 1. Foundational definition and scope

In the formulation explicitly discussed in the literature surveyed here, DSR is not merely any theory with two invariant scales. It requires that the new scale be a relativistic short-distance or high-momentum scale that enters the transformation laws between inertial frames, rather than a trivial invariant or an infrared geometric scale [1003.3942]. This distinguishes DSR from several superficially similar constructions.

Amelino-Camelia’s summary isolates three postulates: the Relativity Principle, an observer-independent speed-of-light scale \(c\) measured in the infrared limit, and an observer-independent ultraviolet scale \(\ell_p\) defined by an agreed measurement procedure [1003.3942]. A common illustrative realization uses a modified photon dispersion relation,
\[
E^2-c^2p^2 + f(E,p;\ell_p)=0,
\qquad
f(E,p;\ell_p)\approx \ell_p\,c\,p^2E
\]
at leading order [1003.3942]. However, the same source stresses that DSR does not logically require a deformed one-particle dispersion relation in every realization: twisted-Hopf constructions on canonical noncommutative spacetime can preserve the ordinary Casimir \(C=m^2=E^2-p^2\) while still supporting an invariant short scale \(\theta^{1/2}\) [1003.3942].

Several misconceptions are explicitly excluded by this criterion. Fock’s relativity or de Sitter relativity introduces a large infrared radius and leaves the dispersion relation undeformed, so it does not realize a second relativistic short-distance scale. Snyder spacetime introduces noncommuting coordinates but keeps Lorentz transformations undeformed, rendering the length scale “relativistically trivial” in the sense described in the source [1003.3942]. Likewise, the existence of a Hopf algebra, including \(\kappa\)-Poincaré, is treated as a candidate formalism rather than a sufficient condition for DSR; one must still verify the Relativity Principle and the absence of a preferred frame [1003.3942].

This suggests a precise delimitation of the subject: DSR is best understood as a program for constructing relativistic kinematics with two observer-independent scales, not as a single model.

## 2. Algebraic realizations, Casimirs, and modified dispersion relations

A large part of the DSR literature develops explicit realizations in terms of deformed Lorentz or Poincaré structures. One canonical example is the \(\kappa\)-Poincaré or \(\kappa\)-Minkowski framework. In the overview of Amelino-Camelia, the spacetime noncommutativity is
\[
[x^j,x^0]=i\lambda x^j,\qquad [x^j,x^k]=0,
\]
and the deformed Casimir takes the form
\[
\sim \left(\frac{2}{\lambda}\sinh\frac{\lambda P^0}{2}\right)^2-e^{\lambda P^0}P^2
\]
[1003.3942]. Li and Zhu analyze a DSR1 realization built from the \(\kappa\)-Poincaré algebra, with deformed brackets
\[
[N_i,P_j]=\delta_{ij}\,\frac{1-e^{-2\ell P_0}}{2\ell}-\ell P_iP_j,
\qquad
[N_i,P_0]=P_i,
\]
and Casimir
\[
C=\left(\frac{2}{\ell}\sinh\frac{\ell P_0}{2}\right)^2-e^{\ell P_0}\vec P^{\,2},
\]
which in \(1+1\) dimensions gives
\[
\frac{4}{\ell^2}\sinh^2\!\left(\frac{\ell E}{2}\right)-e^{\ell E}p^2=m^2
\]
[2605.04422].

For a massless particle in this DSR1 model,
\[
p=\frac{1-e^{-\ell E}}{\ell},
\qquad
E(p)=-\frac1\ell\ln(1-\ell p),
\]
and the group velocity is
\[
v(E)=\frac{dE}{dp}=e^{\ell E}\approx 1+\ell E+\dots
\]
[2605.04422]. In the subluminal case \(\ell<0\), this becomes \(v(E)\approx 1-|\ell|E<1\) [2605.04422].

Other standard realizations emphasize different sectors of the deformed mass shell. The Magueijo-Smolin (MS) form,
\[
E^2-p^2=m^2(1-\lambda E)^2,
\]
appears in the classification of DSR scenarios [1003.3942] and is also used in studies of Unruh radiation and relativistic oscillators [2403.20146; 2603.03572]. The Amelino-Camelia (AC) realization modifies the momentum sector through an energy-dependent prefactor, for example through
\[
\frac{E^2-m^2}{(1+E/2k)^2}=\varepsilon_n
\]
in the generalized Dirac oscillator analysis [2603.03572]. A first-order generalized modified dispersion relation discussed in several oscillator papers is
\[
p_0^2-\mathbf p^{\,2}-2l_p\alpha_2p_0^3+2l_p(\alpha_3-\alpha_1)p_0\mathbf p^{\,2}=m^2
\]
[2603.15632; 2602.22444; 2406.00074].

The survey literature also emphasizes that nonlinear redefinitions of generators do not automatically trivialize DSR. In a \(1+1\)-dimensional toy model, a nonlinear map can make the boost commutators appear undeformed while leaving the massless-particle velocity
\[
v\approx 1+2\lambda E'
\]
unchanged [1003.3942]. The explicit point is that physical observables can remain invariant under generator redefinitions even when algebraic expressions change.

## 3. Deformed boosts, momentum composition, and relative locality

DSR modifies not only one-particle Casimirs but also the action of boosts and, in multi-particle contexts, the composition of momenta. In the \(1+1\)-dimensional toy model summarized by Amelino-Camelia, the deformed boost generator
\[
N=x(P_1-\lambda P_1^2\Pi)+t(\Pi+\lambda P_1^2)
\]
implies
\[
[N,P_1]=\Pi+\lambda P_1^2,\qquad [N,\Pi]=P_1-2\lambda P_1\Pi
\]
[1003.3942]. In a leading-order DSR toy theory, an observer-independent decay law can use
\[
E_a=E_b+E_c+\lambda\,\vec p_b\!\cdot\!\vec p_c,
\]
\[
\vec p_a=\vec p_b+\vec p_c+\lambda(E_b\vec p_c+E_c\vec p_b),
\]
showing explicitly that conservation laws are deformed together with boosts and dispersion [1003.3942].

Carmona and collaborators give a systematic first-order construction in \(1+1\) dimensions using canonical variables \((x,t,\Pi,\Omega)\), deformed generators \(E,P,N\), and a Casimir
\[
C(\Pi,\Omega)=\Omega^2-\Pi^2+\frac{\alpha_1}{\Lambda}\Omega^3+\frac{\alpha_2}{\Lambda}\Omega\Pi^2
\]
[2207.03799]. They also discuss nonlinear composition laws
\[
(p\oplus q)_1=p_1+q_1+\frac{\gamma_1}{\Lambda}p_0q_1+\frac{\gamma_2}{\Lambda}p_1q_0,
\]
\[
(p\oplus q)_0=p_0+q_0+\frac{\beta_1}{\Lambda}p_0q_0+\frac{\beta_2}{\Lambda}p_1q_1
\]
[2207.03799].

Within this setting, relative locality is a central structural consequence. Interactions are implemented by boundary terms involving the deformed total momentum \(p^{(1)}\oplus p^{(2)}\oplus\cdots\), and an event that is local for one observer appears non-local for another translated observer [2207.03799]. The same theme appears in Smolin’s discussion of \(\kappa\)-Minkowski spacetime, where the boost transformation of coordinates depends on the particle momentum:
\[
\delta x^i=-\omega^it-\ell_p\epsilon^{ijk}\omega_jL_k+\Order(\ell_p^2),
\]
\[
\delta t=-\omega^ix_i+\ell_p\,t\,(\omega^ip_i)+\Order(\ell_p^2)
\]
[1004.0664]. Since the transformed shift depends on momentum, coincident worldlines in one frame can split in another, leading to the “classical locality paradox” [1004.0664].

In curved spacetime, relative locality has also been generalized beyond flat momentum-space constructions. In a \(2+1\)-dimensional de Sitter setting, a first-order transverse deformation with parameters \(a,b\), curvature scale \(H\), and Planck-length parameter \(\ell\) modifies the de Sitter algebra and yields a Casimir
\[
C
= E^2-(P_1^2+P_2^2)-2H(N_1P_1+N_2P_2)-H^2R^2
-2(a+b)\ell H\epsilon_{ij}EN_iP_j
\]
[2507.13192]. In this model, a hard photon acquires both a transverse spatial shift
\[
\Delta x_2^B=\frac{b\,\ell\,\Pi_{1,h}^B\,z\,(z+2)}{2H}
\]
and an angular deviation
\[
\Delta\theta\approx a\,\ell\,\Pi_{1,h}^B\ln(1+z)
\]
relative to a soft photon [2507.13192]. The authors interpret these as transverse relative-locality and dual-lensing effects in de Sitter spacetime [2507.13192].

## 4. Time delays, photon propagation, and phenomenology

Time-of-flight phenomenology has often been treated as a flagship DSR prediction, but the literature summarized here presents a more qualified picture. In Amelino-Camelia’s review, a simple modified velocity law,
\[
v(E)=\frac{dE}{dp}\approx 1-\lambda E,
\]
would give a gamma-ray-burst delay \(\Delta t\sim \lambda\,\Delta E\,T\) and motivate observational searches [1003.3942]. At the same time, the review stresses that DSR threshold anomalies are typically suppressed because deformations in the dispersion relation and conservation laws tend to cancel, and that DSR forbids photon decay thresholds that would be compatible with Lorentz-symmetry breaking [1003.3942].

Subsequent work sharpened the status of time delays. Carmona et al. show that in DSR the delay of massless particles is not determined by the modified dispersion relation alone, because deformed translations consistent with relative locality also contribute [2207.03799]. In their first-order formulation,
\[
\Delta t=L\,\Pi\,\frac{2}{3\Lambda}\Bigl(a_4+a_5-a_6-3(a_1+a_2-a_3)\Bigr)
\]
or equivalently
\[
\Delta t=L\,P\,\frac{2(w_1+w_2-w_3)}{3\Lambda}
\]
[2207.03799]. The leading delay vanishes if and only if
\[
a_4+a_5-a_6=3(a_1+a_2-a_3),
\qquad\text{equivalently}\qquad
w_1+w_2-w_3=0
\]
[2207.03799]. In particular, the classical basis of \(\kappa\)-Poincaré, the Magueijo-Smolin basis, and the DCL1 basis all give \(\Delta t=0\) for photons at leading order [2207.03799].

A related first-order classification introduces deformation parameters \(\alpha_1,\alpha_2,\alpha_3\) in the boost sector and obtains, for massless particles,
\[
v_{\rm gr}\simeq 1+(\alpha_1+\alpha_2-\alpha_3)\ell_PE
\]
and
\[
\Delta t=(\alpha_1+\alpha_2-\alpha_3)\frac{\ell_PE\,D}{c}
\]
[2406.00074]. The zero-delay condition is then
\[
\alpha_1+\alpha_2-\alpha_3=0,
\]
leaving a two-parameter family of nontrivial DSR models with \(v=c\) and \(\Delta t=0\) at \(\Order(\ell_p)\) [2406.00074]. This is reinforced by the broader astroparticle-physics summary, which notes that several DSR bases satisfy no-delay conditions while still supporting deformed kinematics and relative locality [2307.03462].

The same sources treat null time-delay searches as insufficient for excluding DSR. A plausible implication is that phenomenological constraints must combine time-of-flight, threshold, conservation-law, and relative-locality observables rather than relying on a single signal [2406.00074; 2207.03799].

Not all studies are conciliatory. Sasaki’s critique argues that variable-light-speed DSR, if it provides a family of energy-dependent signal speeds, would operationally single out an absolute rest frame by a Poincaré-style two-signal protocol, and therefore would be nonviable as a relativity theory [1009.3717]. Li and Zhu develop a related but more algebraically specific objection in the DSR1 model with observer-independent light-speed variation. For \(\ell<0\), they derive a critical rapidity \(\eta_c(E)\) such that a boosted box can overtake its own forward photon, while the physical rapidity window \(\eta_c<\eta<\eta_{\max}\) remains open [2605.04422]. They conclude that this produces tensions in asymptotic particle counting and inertial motion that do not appear to be removable by relative locality alone [2605.04422].

These disagreements document an active conceptual controversy within DSR research rather than a settled no-go theorem applying uniformly to all realizations.

## 5. Gravity, cosmology, and statistical mechanics

One important strand of the subject connects DSR to gravity and cosmology. In a “gravity’s rainbow” realization, the metric is taken as a one-parameter family
\[
ds^2=-f(\lambda E)^{-2}dt^2+a(t)^2g(\lambda E)^{-2}dx_idx^i,
\]
and in the constant-\(c\) choice one sets \(g(\lambda E)=f(\lambda E)\) [1606.00910]. The corresponding rainbow-cosmology Friedmann equation is
\[
H^2-2H\frac{\dot f}{f}+\left(\frac{\dot f}{f}\right)^2
=\frac{8\pi G_0}{3}f(\lambda E)^{-2}\rho
\]
[1606.00910]. By imposing the loop-quantum-cosmology effective Friedmann equation
\[
H^2=\frac{8\pi G_0}{3}\rho\left(1-\frac{\rho}{\rho_c}\right),
\]
using the radiation law \(\dot\rho+4H\rho=0\), and identifying \(\lambda E=\rho/\rho_c\), the rainbow function must satisfy
\[
4\rho\,\frac{df}{d\rho}+f=(1-\rho/\rho_c)^{-1/2}
\]
with solution
\[
f(\rho)=\frac14\left(\frac{\rho}{\rho_c}\right)^{-1/4}B_{\rho/\rho_c}\!\left(\frac14,\frac12\right)
\]
[1606.00910]. Rewriting in terms of \(\lambda E\), one obtains the unique rainbow function
\[
f(\lambda E)=g(\lambda E)
=\frac14(\lambda E)^{-1/4}B_{\lambda E}\!\left(\frac14,\frac12\right),
\]
which fixes the associated DSR model and the exact modified dispersion relation [1606.00910]. The same analysis yields a hard UV cutoff \(E\le E_P=\lambda^{-1}\), finite total microstates
\[
\Gamma=\frac{g_*V}{6\pi^2\lambda^3},
\]
maximum entropy density
\[
s_{\max}=\ln(V/\lambda^3)+\text{const.},
\]
and internal-energy bound
\[
\rho_{\max}=\frac{g_*}{8\pi^2\lambda^4}
\]
[1606.00910]. Matching \(\rho_{\max}\) to the loop-quantum-cosmology critical density fixes \(\lambda\) in terms of \( \gamma,\alpha_0,g_* \) and \(l_{\rm Pl}\) [1606.00910]. The paper interprets this as support for DSR as an appropriate flat limit of loop quantum gravity [1606.00910].

Thermodynamic consequences have also been developed directly from DSR-modified kinematics. In the DSR2 photon-gas analysis, the photon dispersion relation is
\[
p=\frac{E}{1+\lambda E},
\qquad
v_g(E)=\frac{dE}{dp}=(1+\lambda E)^2,
\]
with a deformed integration measure
\[
d^4p\rightarrow \frac{d^4p}{(1+\lambda E)^6}
\]
[1102.2613]. The grand partition function becomes
\[
\ln\Xi=\frac{\pi^2}{45}VT^3f(T),
\]
where \(f(T)\) is a dimensionless integral encoding the deformation [1102.2613]. The internal-energy density, pressure, entropy density, specific heat, and equation-of-state ratio are then deformed accordingly [1102.2613]. The numerical conclusions reported are sign-sensitive: for \(\lambda>0\), the grand partition function, energy density, specific heat, entropy, and pressure are smaller than in special relativity, while the photon velocity and \(P/u\) are larger; for \(\lambda<0\), the tendencies reverse [1102.2613]. The effects remain small until \(T\gtrsim 10^{-3}E_P\) [1102.2613].

For a Maxwell-Boltzmann ideal gas, a different DSR dispersion,
\[
\epsilon(p)^2-p^2=m^2\left(1-\frac{\epsilon(p)}{\kappa}\right)^2,
\]
together with the cutoff \(0\le p,\epsilon<\kappa\), leads to a one-particle partition function involving “Incomplete Modified Bessel functions” [1108.0896]. In that framework the massless and special-relativistic limits are non-perturbative, the internal energy saturates at high temperature because of the cutoff, the specific heat tends to zero, and the equation of state becomes stiffer [1108.0896].

The cosmological and thermodynamic literature therefore treats DSR not only as a deformation of inertial kinematics but as a source of modified statistical ensembles, bounded state counting, and altered early-universe equations of state.

## 6. Wave equations, electrodynamics, and exactly solvable relativistic systems

A second major line of work embeds DSR modifications into relativistic wave equations. Takka and Bouda derive first-order \(1/\kappa\) deformations of Maxwell’s equations and the Lorentz force from the \(\kappa\)-deformed phase space
\[
[x^\mu,x^\nu]=\frac{i\hbar}{\kappa}(x^\mu n^\nu-x^\nu n^\mu),
\qquad
[x^\mu,p^\nu]=i\hbar\left(-\eta^{\mu\nu}+\frac{n^\mu p^\nu}{\kappa}\right),
\qquad
[p^\mu,p^\nu]=0
\]
[2207.14531]. Their generalized field tensor depends on position and velocity, and the resulting Maxwell equations acquire explicit \(1/\kappa\) corrections depending on \(\dot x\) and derivatives of the zeroth-order field [2207.14531]. The equation of motion becomes
\[
m\ddot x^\mu
= qF^\mu{}_\nu\dot x^\nu+\Gamma^\mu{}_{\alpha\beta}\dot x^\alpha\dot x^\beta+G_{(1)}^\mu(x),
\]
with a velocity-squared “gravitational-type” Lorentz force proportional to \(qm/\kappa\) and tied to the electromagnetic field [2207.14531]. The authors explicitly contrast this with Fock’s nonlinear relativity, where analogous terms are independent of charge [2207.14531].

Bound-state and oscillator problems provide a complementary exactly solvable arena. In the \(2+1\)-dimensional two-body Dirac equation inspired by Amelino-Camelia DSR, the leading correction rescales spatial derivatives by
\[
\partial_x\to \lambda_k\,\partial_x,
\qquad
\lambda_k=1+\frac{E}{4E_p},
\]
and the radial equation becomes a deformed Bessel equation with solution
\[
\varphi_1(r)=A\,J_1\!\left(\frac{\lambda}{\lambda_k}r\right)
\]
[2511.05641]. For a Coulomb-type interaction, the effective fine-structure constant runs as
\[
\alpha_{\rm eff}(E)=\frac{\alpha}{1+E/(4E_p)},
\qquad
\frac{\alpha_{\rm eff}}{\alpha}\approx 1-\frac{E}{4E_p}
\]
[2511.05641]. Applied to positronium-like systems, the corresponding binding-energy shifts are exceedingly small, of order \(10^{-22}\,\mathrm{eV}\) in the ground state [2511.05641].

Several recent studies analyze generalized Dirac or Klein-Gordon oscillators in standard AC and MS DSR realizations and in generalized first-order Planck-length expansions. In the one-dimensional generalized Dirac oscillator, the undeformed spatial problem yields a real set \(\{\varepsilon_n\}\) through supersymmetric decoupling, while DSR modifies the algebraic reconstruction map to physical energies [2603.03572]. In the MS prescription,
\[
E^2-m^2\left(1-\frac{E}{k}\right)^2=\varepsilon_n,
\]
whereas in the AC prescription
\[
\frac{E^2-m^2}{(1+E/2k)^2}=\varepsilon_n
\]
with admissibility condition
\[
\varepsilon_n<4k^2
\]
[2603.03572]. For a pseudo-Hermitian complexified Morse interaction, the intrinsic finiteness of Morse bound states can be further reduced by the AC truncation [2603.03572].

In the three-dimensional Dirac oscillator, the undeformed invariant
\[
\Lambda_{Nj}^{(-)}=m\hbar\omega[2(N-j)+1],\qquad
\Lambda_{Nj}^{(+)}=m\hbar\omega[2(N+j)+3]
\]
is preserved as the kinematic label, while the AC, MS, or generalized DSR prescription deforms the algebraic relation between \(\Lambda_{Nj}^{(\pm)}\) and the relativistic energy [2603.15632]. The same pattern occurs for the three-dimensional Klein-Gordon oscillator, whose separability in spherical coordinates and Laguerre-spherical-harmonic structure are unchanged, while DSR modifies only the quantization condition \(N\leftrightarrow E\) and produces branch-dependent, excitation-enhanced shifts [2602.22444].

A linear-fractional or projective formulation of DSR in \((1+1)\) dimensions provides another exactly solvable oscillator model. Starting from
\[
\pi^\mu=\frac{p^\mu}{1+(a\!\cdot\!p)/E_p},
\qquad
C(p)=\frac{\eta_{mn}p^mp^n}{[1+(a\!\cdot\!p)/E_p]^2}=m^2,
\]
one obtains three inequivalent deformation geometries: time-like, space-like, and light-like [2602.09861]. For the time-like case, the deformed Dirac oscillator obeys
\[
E^2-m^2(1-E/E_p)^2=2m\omega n
\]
with exact branches
\[
E_n^{(\pm)}
=\frac{-m^2/E_p\pm\sqrt{m^2+2m\omega n(1-m^2/E_p^2)}}{1-m^2/E_p^2}
\]
[2602.09861]. The nonrelativistic expansions show that the deformation geometry controls whether the leading effect is a rest-energy shift, a renormalization of the oscillator spacing, or both [2602.09861].

These constructions show that DSR can be realized as a deformation of spectral reconstruction rather than a deformation of the spatial eigenfunctions themselves. This suggests a useful division between kinematic and spectral effects in solvable relativistic models.

## 7. Conceptual tensions and open problems

The literature represented here portrays DSR as technically productive but conceptually unsettled. Several open issues were already identified in the 2010 survey: the proper formalism for multi-particle states, the “soccer-ball problem,” the role of interactions, the extension to curved backgrounds, and the possibility that DSR symmetries are only approximate and cease to apply above the Planck scale [1003.3942].

Locality is a recurring tension. Smolin’s \(\kappa\)-Minkowski analysis argues that the classical splitting of events under momentum-dependent boosts should not be interpreted in a purely classical limit with \(\ell_p\neq0\), because finite \(\ell_p\) requires \(\hbar\neq0\) and therefore an intrinsically quantum spacetime description [1004.0664]. In that treatment, wavepacket spreading induced by the noncommutative structure can dominate the classical splitting and thereby “smear away” the paradox [1004.0664]. By contrast, the Li-Zhu thought experiment claims that overtaking and reabsorption of photons in covariant light-speed-variation DSR generates an asymptotic counting problem that relative locality does not resolve [2605.04422].

Another persistent issue concerns phenomenological interpretation. The existence of nontrivial DSR models with zero leading photon delay means that a null time-of-flight result does not by itself exclude deformations of Lorentz symmetry [2406.00074; 2207.03799]. Conversely, critiques of variable-light-speed DSR argue that a continuous family of fixed signal speeds would operationally reveal an ether-like rest frame [1009.3717]. The phenomenological situation is therefore model-dependent: some DSR realizations predict energy-dependent propagation effects, some do not at leading order, and some appear to face structural consistency problems.

At the same time, the subject has expanded into curved-spacetime relative locality, deformed electrodynamics, gravity’s rainbow, thermodynamics, Unruh radiation, and exactly solvable relativistic bound states [2507.13192; 2207.14531; 1606.00910; 1102.2613; 2403.20146; 2603.15632]. This breadth indicates that DSR functions less as a single universally accepted theory than as a family of deformation schemes for relativistic kinematics, with shared invariant-scale principles but non-equivalent dynamical and observational consequences.

Source: https://www.emergentmind.com/topics/doubly-special-relativity-dsr