Statistical Mechanics

Status: Stable core; foundations (irreversibility, quantum thermalization) actively researched. Last updated: 2026-06-08

Statistical mechanics is the bridge layer between microscopic, deterministic, time-reversible dynamics (Hamiltonian classical-mechanics flow or quantum-mechanics unitary evolution) and the phenomenological, irreversible laws of thermodynamics. Its central achievement is to derive the bulk behavior of matter built from $\sim 10^{23}$ constituents not by integrating their trajectories — which is intractable and, for chaotic systems, useless beyond microscopic times — but by replacing them with probability distributions over microstates and extracting the overwhelmingly typical macroscopic outcome.

Scope

The domain partitions into five strata, all of which this page treats:

1. Equilibrium ensemble theory — the microcanonical, canonical, and grand-canonical ensembles; partition functions; free energies; and the systematic derivation of thermodynamic relations as derivatives of $\ln Z$.
2. Fluctuations and linear response — the fluctuation–dissipation theorem (FDT), Kubo/Green–Kubo formulas, and their modern far-from-equilibrium extensions (Jarzynski, Crooks).
3. Collective phenomena — phase transitions, critical exponents, scaling, universality, and Wilson's renormalization group (RG).
4. Non-equilibrium foundations — the Boltzmann equation, the $H$-theorem, the origin of macroscopic irreversibility from reversible microdynamics, ergodicity, mixing, and typicality.
5. Quantum statistical mechanics — Fermi–Dirac and Bose–Einstein statistics, and the eigenstate thermalization hypothesis (ETH) as the mechanism by which isolated quantum systems thermalize.

The recurring conceptual question that organizes all five is: why does an ensemble average predict the behavior of a single individual macroscopic system? OPEN — see Open problems.

Core formalism

Phase space, microstates, macrostates

A classical system of $N$ particles in $d$ dimensions has phase space $\Gamma = \mathbb{R}^{2dN}$ with point $x=(q_1,\dots,q_{dN},p_1,\dots,p_{dN})$. The microstate is $x$; dynamics is the Hamiltonian flow $\Phi_t$ generated by $H(x)$ through $\dot q_i = \partial H/\partial p_i$, $\dot p_i = -\partial H/\partial q_i$. Liouville's theorem ESTABLISHED states the flow preserves the symplectic volume $d\Gamma = \prod_i dq_i\,dp_i$, so the phase-space density $\rho(x,t)$ obeys the Liouville equation

$\partial_t \rho = \{H,\rho\} = -\sum_i\left(\frac{\partial H}{\partial p_i}\frac{\partial\rho}{\partial q_i}-\frac{\partial H}{\partial q_i}\frac{\partial\rho}{\partial p_i}\right),\qquad \frac{d\rho}{dt}=0 \ \text{along trajectories.}$

A macrostate $M$ is an equivalence class of microstates indistinguishable at the level of a few coarse observables (energy, volume, particle number, magnetization). The combinatorial heart of the subject is that the overwhelming majority of phase-space volume — quantum-mechanically, of Hilbert-space dimension — belongs to a single macrostate, the equilibrium one ESTABLISHED.

Quantum-mechanically the microstate is a ray $|\psi\rangle \in \mathcal{H}$; the statistical object is the density matrix $\hat\rho$, evolving by the von Neumann equation $i\hbar\,\partial_t\hat\rho = [\hat H,\hat\rho]$, with $S_{\mathrm{vN}}=-k_B\,\mathrm{Tr}(\hat\rho\ln\hat\rho)$.

The three equilibrium ensembles

Microcanonical (NVE). Isolated system, energy fixed in a shell $[E,E+\Delta]$. The postulate of equal a priori probabilities gives

$\rho_{mc}(x)=\frac{1}{\Omega(E)}\,\delta\!\big(H(x)-E\big),\qquad \Omega(E)=\int_\Gamma d\Gamma\,\delta(H(x)-E),$

with the Boltzmann entropy $S_B(E)=k_B\ln[\Omega(E)\,\Delta]$ (equivalently $k_B\ln\Sigma(E)$ with $\Sigma$ the phase volume below $E$; the two agree to $O(\ln N)$). Temperature, pressure, and chemical potential follow from $dS=\tfrac{1}{T}dE+\tfrac{P}{T}dV-\tfrac{\mu}{T}dN$ ESTABLISHED.

Canonical (NVT). System in weak thermal contact with a reservoir at temperature $T$, $\beta=1/k_BT$. Maximizing Gibbs–Shannon entropy at fixed $\langle H\rangle$ — equivalently, tracing out a large short-range-coupled reservoir — yields the Boltzmann–Gibbs distribution

$\rho_c(x)=\frac{1}{Z}e^{-\beta H(x)},\qquad Z(\beta,V,N)=\frac{1}{N!\,h^{dN}}\int_\Gamma d\Gamma\, e^{-\beta H(x)}\quad(\text{quantum: }Z=\mathrm{Tr}\,e^{-\beta\hat H}).$

The Helmholtz free energy is $F=-k_BT\ln Z$, with $dF=-S\,dT-P\,dV+\mu\,dN$. All equilibrium thermodynamics follows from derivatives of $\ln Z$ ESTABLISHED:

$\langle E\rangle=-\partial_\beta\ln Z,\qquad C_V=\partial_T\langle E\rangle=k_B\beta^2\,\langle(\Delta E)^2\rangle,\qquad S=-\partial_T F.$

The relation $C_V=k_B\beta^2\langle(\Delta E)^2\rangle$ is the first instance of a fluctuation–response identity: a static susceptibility ($C_V$) equals an equilibrium variance.

Grand canonical ($\mu VT$). Energy and particle exchange with a reservoir, fugacity $z=e^{\beta\mu}$:

$\rho_{gc}=\frac{1}{\Xi}e^{-\beta(H-\mu N)},\qquad \Xi(\mu,V,T)=\sum_{N=0}^\infty z^N Z(N,V,T)=\mathrm{Tr}\,e^{-\beta(\hat H-\mu\hat N)},$

grand potential $\Phi_G=-k_BT\ln\Xi=-PV$. For ideal quantum gases, $\ln\Xi=\mp\sum_k\ln(1\mp z\,e^{-\beta\epsilon_k})$, giving the Bose–Einstein ($-$) and Fermi–Dirac ($+$) occupations

$\langle n_k\rangle=\frac{1}{e^{\beta(\epsilon_k-\mu)}\mp 1}.$

These furnish blackbody radiation, Bose–Einstein condensation, degeneracy pressure, and the electronic theory of metals ESTABLISHED.

Ensemble equivalence. In the thermodynamic limit ($N,V\to\infty$ at fixed $N/V$) the three ensembles yield identical intensive predictions, because energy/number fluctuations are $O(1/\sqrt N)$ relative to the mean; large-deviation theory makes this a statement about Legendre duality of rate functions ESTABLISHED. Equivalence fails at first-order transitions, for long-range interactions, and in finite systems ESTABLISHED — see Where it breaks down.

Entropy: Boltzmann versus Gibbs/Shannon

• Boltzmann entropy $S_B(M)=k_B\ln|\Gamma_M|$ is a function of the macrostate — the log phase-space volume of microstates consistent with $M$. It is defined for an individual system and can increase along a single trajectory as the system migrates to larger macro-cells. This is the entropy of the second law INFERENCE — within the typicality program.
• Gibbs/Shannon entropy $S_G=-k_B\int\rho\ln\rho\,d\Gamma$ is a functional of the distribution. Under Liouville evolution $S_G$ is exactly constant, because $\rho$ is merely transported ESTABLISHED. It equals the thermodynamic entropy at equilibrium but cannot, by itself, describe approach to equilibrium without coarse-graining.

The reconciliation: the fine-grained Gibbs entropy is conserved, while a coarse-grained $\bar\rho$ has non-decreasing entropy, matching $S_B$. Jaynes recast equilibrium distributions as maximum-entropy (MaxEnt) distributions consistent with known macroscopic constraints, making statistical mechanics a special case of Bayesian inference CONTESTED — whether MaxEnt is physics or epistemology. See information-theory.

The Ising model and exact paradigms

The Ising Hamiltonian $H=-J\sum_{\langle ij\rangle}s_is_j-h\sum_i s_i$, $s_i=\pm1$, is the canonical interacting model. In $d=1$ the transfer matrix

$T=\begin{pmatrix}e^{\beta(J+h)}&e^{-\beta J}\\ e^{-\beta J}&e^{\beta(J-h)}\end{pmatrix}, \qquad Z=\mathrm{Tr}\,T^N,$

has a unique largest eigenvalue for all $T>0$, so there is no finite-temperature transition ESTABLISHED. Onsager's 1944 exact solution of the $d=2$ zero-field model gives a continuous transition at $\sinh(2\beta_c J)=1$; Yang's spontaneous magnetization $m=[1-\sinh^{-4}(2\beta J)]^{1/8}$ yields the exponent $\beta=1/8$, the first rigorous proof of non-mean-field critical behavior ESTABLISHED.

Phase transitions, scaling, universality, RG

Near a continuous transition the correlation length $\xi\sim|t|^{-\nu}$ diverges, $t=(T-T_c)/T_c$. Critical exponents $(\alpha,\beta,\gamma,\delta,\nu,\eta)$ satisfy scaling relations — Rushbrooke $\alpha+2\beta+\gamma=2$, Fisher $\gamma=\nu(2-\eta)$, and hyperscaling $2-\alpha=d\nu$ ESTABLISHED. The singular free-energy density takes the homogeneous form

$f_s(t,h)=b^{-d}\,f_s(b^{y_t}t,\,b^{y_h}h),$

so exponents are fixed by the two RG eigenvalues $y_t,y_h$. Universality ESTABLISHED: exponents depend only on spatial dimensionality, order-parameter symmetry, and interaction range — not microscopic detail. Wilson's renormalization group explains why ESTABLISHED: iterated coarse-graining plus rescaling defines a flow in coupling space; critical points are unstable fixed points, and exponents are eigenvalues of the linearized flow. The $\epsilon=4-d$ expansion and momentum-shell RG (1971–72) put this on a calculational footing, with upper critical dimension $d_c=4$ above which mean-field exponents hold. The RG is the deepest formal bridge to quantum-field-theory.

Fluctuation–dissipation and linear response

Equilibrium fluctuations encode the response to small perturbations. The classical FDT relates the response function $\chi(t)$ to the equilibrium autocorrelation $C(t)=\langle A(t)A(0)\rangle$ via $\chi(t)=-\beta\,\theta(t)\,\dot C(t)$. In frequency space, the quantum (Kubo) form reads

$\chi''(\omega)=\frac{1}{2\hbar}\big(1-e^{-\beta\hbar\omega}\big)\,\tilde S(\omega),$

with $\tilde S(\omega)$ the power spectrum of fluctuations ESTABLISHED. Special cases: the Einstein relation $D=\mu k_BT$, Nyquist noise $S_V=4k_BTR$, and Green–Kubo formulas for transport coefficients. Modern far-from-equilibrium extensions — the Jarzynski equality $\langle e^{-\beta W}\rangle=e^{-\beta\Delta F}$ and the Crooks fluctuation theorem — are exact for arbitrarily fast driving given a well-defined initial equilibrium state ESTABLISHED; by Jensen's inequality they recover $\langle W\rangle\ge\Delta F$, the second law as a corollary.

Boltzmann equation, $H$-theorem, irreversibility

For a dilute gas the one-particle distribution $f(\mathbf r,\mathbf v,t)$ obeys

$\partial_t f+\mathbf v\!\cdot\!\nabla_{\mathbf r}f+\tfrac{\mathbf F}{m}\!\cdot\!\nabla_{\mathbf v}f=\int d\mathbf v_1\,d\Omega\,|\mathbf v-\mathbf v_1|\,\sigma\,(f'f_1'-ff_1).$

With $H(t)=\int f\ln f\,d\mathbf v\,d\mathbf r$, Boltzmann's $H$-theorem gives $dH/dt\le0$ (so $S=-k_BH$ increases), with equality only at the Maxwell–Boltzmann distribution. The decisive non-mechanical input is the Stosszahlansatz (molecular chaos): pre-collision velocities are uncorrelated, $f_2(\mathbf v,\mathbf v_1)=f(\mathbf v)f(\mathbf v_1)$. This breaks time-reversal symmetry by hand ESTABLISHED that this is the locus of asymmetry. Loschmidt's reversibility paradox (momentum reversal yields $dH/dt>0$) and Zermelo's recurrence paradox (Poincaré recurrence for bounded systems) prove the theorem cannot follow from mechanics alone. The resolution — Boltzmann's, sharpened by Lanford's 1975 rigorous short-time derivation — is that the $H$-theorem is statistical/typical: the overwhelming majority of microstates compatible with a low-entropy macrostate evolve toward higher entropy; recurrence times scale as $\sim e^{N}$ (astronomically long); and the arrow of time is inherited from a low-entropy initial condition (the "Past Hypothesis"), not from the dynamics INFERENCE, broad consensus on the structure; the cosmological source is OPEN.

Quantum thermalization (ETH)

For a closed quantum system, unitary evolution preserves $S_{\mathrm{vN}}$, yet local observables relax to thermal values. The eigenstate thermalization hypothesis (Deutsch 1991, Srednicki 1994) posits that matrix elements of few-body operators in energy eigenstates take the form

$\langle E_n|\hat O|E_m\rangle = O(\bar E)\,\delta_{nm} + e^{-S(\bar E)/2}\,f_O(\bar E,\omega)\,R_{nm},$

with $\bar E=(E_n+E_m)/2$, $\omega=E_n-E_m$, smooth $O,f_O$, and $R_{nm}$ effectively random of unit variance. Each individual eigenstate then already encodes the microcanonical prediction, so the long-time average of a local observable is thermal [INFERENCE — strong numerical and random-matrix support, unproven for any realistic interacting Hamiltonian]. ETH provably fails for integrable systems (which relax to a Generalized Gibbs Ensemble) and for many-body-localized (MBL) systems. See quantum-mechanics.

Foundational assumptions

See ASSUMPTIONS_LEDGER for the cross-domain register and EPISTEMICS for marker definitions.

Domain of validity

Equilibrium ensemble theory is rigorously controlled and quantitatively predictive for systems with many degrees of freedom ($N\gg1$, ideally $N\to\infty$ for sharp statements); short-range, additive interactions (so energy and entropy are extensive and ensembles equivalent); systems that have equilibrated (times long compared to microscopic relaxation yet still subject to the relevant constraints); and weak reservoir coupling for the canonical/grand-canonical forms ESTABLISHED. The RG treatment of criticality is valid where $\xi$ greatly exceeds the lattice spacing. Linear-response/FDT holds for small perturbations about equilibrium; Jarzynski/Crooks extend exact statements arbitrarily far from equilibrium but still require a well-defined initial equilibrium state and Hamiltonian (or detailed-balance Markov) dynamics. The Boltzmann equation is valid in the Boltzmann–Grad dilute limit (density $\to0$, $N\to\infty$ at fixed $N\sigma^2$) for times short compared to recurrence. Quantum statistics (Fermi–Dirac, Bose–Einstein, the partition-function trace) apply wherever the microdescription is a many-body quantum Hamiltonian with a well-defined trace — essentially all condensed matter; ETH governs thermalization for generic non-integrable, non-localized closed systems. See CONSTANTS_AND_SCALES for $k_B$, $h$, and the relevant scale separations.

Where it breaks down

• Long-range / non-additive interactions (self-gravitating clusters and galaxies, unscreened Coulomb systems, dipolar systems). Energy ceases to be extensive; the thermodynamic limit is ill-defined absent special (Kac) scaling; microcanonical and canonical ensembles become inequivalent. Self-gravitating systems exhibit negative specific heat (which the canonical ensemble cannot represent), the gravothermal catastrophe, and — for the idealized point-mass model — no global maximum-entropy equilibrium, the entropy being unbounded above until a short-distance cutoff is imposed (a property of the unregularized UV, distinct from the long-range non-additivity; cf. GC-17) ESTABLISHED. A genuine domain-of-validity breakdown, not an inconsistency — see cosmology and general-relativity.
• Small / nanoscale / single-molecule systems. $O(1/\sqrt N)$ fluctuations are no longer negligible; ensemble equivalence fails; the Boltzmann-vs-Gibbs (surface-vs-volume) entropy distinction and the work/heat split become operationally important. Quantities are non-self-averaging and must be replaced by full distributions (stochastic thermodynamics). Phase transitions are smooth crossovers ESTABLISHED.
• The arrow of time. Not a logical inconsistency but the deepest conceptual incompleteness: the $H$-theorem's monotonic decrease cannot follow from time-symmetric mechanics (Loschmidt) and conflicts with Poincaré recurrence (Zermelo). The framework delivers irreversibility only after a time-asymmetric input (molecular chaos / a low-entropy past), pushing the explanation onto an unexplained boundary condition it cannot itself justify OPEN. See GAPS_AND_CONTRADICTIONS.
• Integrable and many-body-localized quantum systems. ETH fails: extensively many (quasi-)conserved quantities prevent relaxation to the standard Gibbs ensemble. Integrable systems relax to a Generalized Gibbs Ensemble with one Lagrange multiplier per conserved charge; MBL systems retain initial-condition memory indefinitely. Whether MBL is a true stable phase in $d\ge2$ in the thermodynamic limit is CONTESTED.
• Glasses, granular media, active matter. Ergodicity is broken or the system is driven; phase space is not explored; configurational entropy, aging, and history-dependence appear. Equilibrium ensembles do not apply; replica/cavity methods, effective temperatures, or fully non-equilibrium descriptions are required. No complete first-principles statistical mechanics of the glass transition exists; whether it is a true thermodynamic transition is CONTESTED.
• First-order transitions and metastability. Ensemble equivalence breaks at coexistence; the canonical free energy is the convex hull (Maxwell construction), losing the metastable branches, surface tension, and nucleation kinetics the system physically exhibits ESTABLISHED.
• Negative absolute temperature / bounded spectra. Population inversion in upper-bounded-spectrum systems (nuclear spins, certain optical lattices) gives formally negative $T$, exposing the Boltzmann-vs-Gibbs entropy dispute over whether negative temperatures and Carnot efficiencies $>1$ are physical CONTESTED, unsettled post-2013.

Open problems (internal)

• Why ensemble averages predict an individual system. OPEN/CONTESTED Competing programs — ergodic theory, typicality/measure-concentration (Boltzmann–Lebowitz–Goldstein), Jaynesian MaxEnt, and quantum canonical typicality / ETH — each work in some regime; none is a universally accepted single derivation. The pragmatic answer (it works; fluctuations are tiny) is undisputed; the principled one is unresolved.
• Origin of the thermodynamic arrow and the Past Hypothesis. OPEN That an asymmetric boundary condition is needed is INFERENCE bordering on ESTABLISHED; why the early universe had extraordinarily low (especially gravitational) entropy is open and bleeds into cosmology. Penrose's Weyl-curvature hypothesis is one SPECULATIVE proposal.
• First-principles theory of the glass transition. OPEN/CONTESTED Whether an underlying ideal thermodynamic transition exists (Kauzmann temperature, random first-order transition theory from the replica/cavity solution of mean-field spin glasses) or the transition is purely kinetic is disputed. Mean-field/infinite-dimensional results are rigorous; their relevance to $d=3$ is contested.
• Rigorous ETH for a realistic Hamiltonian. OPEN Strongly supported numerically and by random-matrix arguments but unproven for any realistic interacting model. The boundaries among thermalizing, integrable (GGE), prethermal, quantum-scarred, and MBL behavior are under active study and partly CONTESTED.
• A predictive variational principle for non-equilibrium steady states. OPEN Large-deviation theory, the Gallavotti–Cohen fluctuation theorem, and macroscopic fluctuation theory give powerful exact results, and stochastic thermodynamics is rigorous for Markov dynamics, but no universal organizing principle exists (min/max entropy-production proposals are of limited or CONTESTED validity).
• Rigorous hydrodynamic limit. OPEN Lanford derives the Boltzmann equation only for short times in the dilute limit; a full rigorous derivation of Navier–Stokes and finite transport coefficients from interacting Hamiltonian microdynamics remains a major problem (related to Hilbert's sixth problem).
• Boltzmann vs Gibbs entropy. CONTESTED A live post-2013 debate over which entropy obeys the second law and equipartition for small or bounded-spectrum systems; consensus is absent and it matters operationally for nanoscale thermodynamics.

See OPEN_PROBLEMS and HYPOTHESES.

Connections to other frameworks

• thermodynamics — Statistical mechanics is the microscopic foundation that derives the phenomenological laws: $F=-k_BT\ln Z$ reproduces the thermodynamic potentials, $dS=\delta Q/T$ emerges from coarse-grained $S_B$, and the third law connects to ground-state degeneracy. The reduction is subtle: the second law's strict irreversibility is statistical/typical, not absolute, and requires the extra Past-Hypothesis input thermodynamics simply posits.
• classical-mechanics and quantum-mechanics — Supply the reversible, entropy-conserving microdynamics (Liouville/unitary). The central tension is that these conserve fine-grained entropy while statistical mechanics produces irreversible relaxation — resolved only via coarse-graining plus initial conditions.
• quantum-field-theory / particle-physics — The Euclidean path integral at finite temperature is a partition function with imaginary-time period $\beta\hbar$ (Matsubara/thermal field theory). Wilsonian RG, born in critical phenomena, became the organizing principle of QFT; critical exponents are operator anomalous dimensions, and the $\epsilon$-expansion is shared technology.
• information-theory — Shannon entropy is formally identical to Gibbs entropy; Jaynes recast the subject as MaxEnt inference. Landauer's principle ($k_BT\ln 2$ per bit erased) and Bennett's resolution of Maxwell's demon tie thermodynamic cost to computation. Whether entropy is fundamentally physical or epistemic is CONTESTED.
• cosmology / general-relativity — Gravity's non-additivity breaks standard ensemble theory; black-hole thermodynamics ($S=A/4$ in Planck units) imports statistical-mechanical concepts into gravity and raises the still-open microstate-counting question (string theory, holography). The arrow of time connects to the cosmological initial condition.
• mathematics — Ergodic theory, mixing, and Lyapunov/chaos theory underpin the ensemble postulates; large-deviation theory provides the rigorous language for fluctuations, free energies as rate functions, and ensemble equivalence. KAM's demonstration of generic non-ergodicity forced the foundational shift to typicality.

See THEORY_MAP, UNIFICATION_LANDSCAPE, and UNIFYING_PRINCIPLES.

Key references

Full entries in BIBLIOGRAPHY.

• L. D. Landau & E. M. Lifshitz, Statistical Physics, Part 1 (Course of Theoretical Physics, Vol. 5), 3rd ed. — canonical treatment of ensembles, fluctuations, and phase transitions.
• K. Huang, Statistical Mechanics, 2nd ed. (Wiley, 1987) — Boltzmann equation, $H$-theorem, ensembles, quantum gases, Ising/transfer matrix.
• R. K. Pathria & P. D. Beale, Statistical Mechanics, 3rd ed. — quantum statistics, BEC, FDT, introduction to criticality and RG.
• L. E. Reichl, A Modern Course in Statistical Physics — kinetic theory, linear response, FDT, non-equilibrium.
• N. Goldenfeld, Lectures on Phase Transitions and the Renormalization Group (Addison-Wesley, 1992) — scaling, universality, Wilsonian RG, $\epsilon$-expansion.
• L. Onsager, "Crystal Statistics. I.", Phys. Rev. 65, 117 (1944) — exact 2D Ising solution.
• K. G. Wilson & J. Kogut, "The renormalization group and the $\epsilon$ expansion", Phys. Rep. 12, 75 (1974).
• J. L. Lebowitz, "Boltzmann's Entropy and Time's Arrow", Physics Today 46(9), 32 (1993) — typicality resolution of the irreversibility paradoxes.
• J. M. Deutsch, Phys. Rev. A 43, 2046 (1991); M. Srednicki, Phys. Rev. E 50, 888 (1994) — ETH.
• C. Jarzynski, Phys. Rev. Lett. 78, 2690 (1997); G. E. Crooks, Phys. Rev. E 60, 2721 (1999) — non-equilibrium work/fluctuation relations.
• R. Kubo, J. Phys. Soc. Jpn. 12, 570 (1957) — linear response and the FDT.
• E. T. Jaynes, "Information Theory and Statistical Mechanics", Phys. Rev. 106, 620 (1957).
• O. E. Lanford III, "Time evolution of large classical systems", Lecture Notes in Physics 38 (Springer, 1975).
• A. I. Khinchin, Mathematical Foundations of Statistical Mechanics (Dover, 1949).
• L. D'Alessio, Y. Kafri, A. Polkovnikov & M. Rigol, "From quantum chaos and eigenstate thermalization to statistical mechanics and thermodynamics", Adv. Phys. 65, 239 (2016).

See also

• domains/thermodynamics.md
• domains/classical-mechanics.md
• domains/quantum-mechanics.md
• domains/quantum-field-theory.md
• domains/information-theory.md
• domains/cosmology.md
• domains/general-relativity.md
• domains/mathematics.md
• THEORY_MAP.md · OPEN_PROBLEMS.md · GAPS_AND_CONTRADICTIONS.md · ASSUMPTIONS_LEDGER.md · CONSTANTS_AND_SCALES.md · EPISTEMICS.md

References

See BIBLIOGRAPHY.md for full citation details.