Thermodynamic Completeness Test

• Thermodynamic completeness test is a framework to determine if a chosen set of thermodynamic variables or records uniquely and sufficiently describes equilibrium and response functions.
• It integrates criteria like uniqueness, integrability, and detailed balance across diverse settings including solids, quantum systems, and Markovian thermal processes.
• The test distinguishes complete descriptions from those missing hidden records, ensuring that observable dynamics accurately reflect the full thermodynamic behavior.

Searching arXiv for recent and foundational uses of “thermodynamic completeness” and closely related formulations. The thermodynamic completeness test denotes a family of criteria for deciding whether a chosen thermodynamic description is sufficient to determine the relevant equilibrium properties, response functions, reachable states, or transport statistics of a system. In the equilibrium thermodynamics of solids, completeness is formulated in terms of thermodynamic coordinates created by constraints and validated by the fundamental relation of equilibrium (Shirai, 2018). In quantum statistical mechanics, completeness is the uniqueness of the entropy-maximizing ergodic state for each macroscopic variable set (Sewell, 2018). In open-system, Markovian, and non-Markovian settings, the same expression is used for tests of whether a reduced generator, state trajectory, or memory kernel determines thermodynamic observables, or whether additional current or measurement records are indispensable (Dann et al., 2020, Tian, 4 May 2026, Tian, 7 May 2026). Collectively, these formulations suggest that thermodynamic completeness is a sufficiency-and-uniqueness requirement imposed on the variables or records retained by a theory.

1. Recurring structure of the completeness problem

Across the literature, thermodynamic completeness is posed as a question about whether a reduced description loses information that is thermodynamically operative. In one class of formulations, the retained variables must determine a unique equilibrium state and all equilibrium response functions. In another, the retained dynamical object must determine all observables of interest, rather than only the reduced state evolution. In a third, completeness is tested through integrability, Jacobian conditions, or compatibility with the Gibbsian or GKLS thermodynamic structure [(Cooper et al., 2011); (Dann et al., 2020)].

A concise comparison is useful.

Setting	Retained object	Completeness condition
Solids	$\{X_j\}$ and $U$	“Uniqueness” and “Sufficiency”
Quantum statistical mechanics	macroscopic observables $\{q_j\}$	$\forall q\in Q,\ \exists !\,p_q\in Ex$
Markovian thermal processes	path of populations $p(t)$	$p(0)\to_{MTP}p(t_f)\iff p(0)\gg_\gamma p(t_f)$
Open quantum dynamics	GKLS generator	commutation, detailed balance, thermal fixed point
Majorana or Markovian transport	island kernel or state trajectory	observable must be invariant on fibers of the projection

The central misconception addressed by these works is that a formally exact reduced state equation is automatically thermodynamically complete. Several papers show that this is false: identical state dynamics may coexist with different thermodynamic records, different lead-resolved noise, or different current statistics (Tian, 7 May 2026, Tian, 4 May 2026). Conversely, in equilibrium settings the works emphasize that completeness is not merely a matter of minimizing the number of variables; it is a matter of retaining all variables needed for uniqueness, integrability, and the recovery of thermodynamic derivatives [(Shirai, 2018); (Cooper et al., 2011)].

2. Thermodynamic coordinates and completeness in solids

In Shirai’s formulation, equilibrium is defined for an isolated system by the condition that no work can be extracted from it by any cyclic process whose only external effect is lifting a weight; equivalently, under its given constraints and fixed total energy $U$, there is one and only one state of maximum entropy. A thermodynamic coordinate (TC) $X_k$ is the time-average of a microscopic property $x_k(t)$ made well defined by a constraint $\xi_k$, formally

$U$0

subject to $U$1 or $U$2 (Shirai, 2018).

The construction procedure begins by enumerating constraints $U$3, such as container walls, dislocation pinning, or energy barriers trapping each atom near its local minimum. One then identifies the corresponding observable $U$4, defines $U$5, and removes dependent variables until an independent set $U$6 remains. The resulting fundamental relation of equilibrium is single-valued in either entropy or energy representation,

$U$7

with differential form

$U$8

The completeness test has two parts. The first is uniqueness: there is exactly one equilibrium state for each choice of $U$9, $\{q_j\}$0, and $\{q_j\}$1, equivalently the map from constraints to $\{q_j\}$2 is one-to-one. The second is sufficiency: all measurable response functions, including specific heats, elastic constants, and susceptibilities, can be obtained by appropriate second derivatives of the fundamental relation. In particular,

$\{q_j\}$3

If some observed temperature dependence cannot be reproduced unless an additional coordinate is added, the set is incomplete. The Hessian conditions supply the local stability criterion: the matrix of second derivatives of $\{q_j\}$4 must be negative definite, or equivalently that of $\{q_j\}$5 positive definite.

The paper’s most distinctive conclusion is that the mean values of all the atom positions of a given solid together with the internal energy constitute a commensurate set of state variables. Contrary to the conventional view, an infinite number of the atom positions and their microscopic characters do not conflict with the principles of thermodynamics. The decisive requirement is not finiteness but uniqueness of their equilibrium values against random motions. Jaynes’s max-entropy construction is used to support this point: the number of meaningful constraints equals the number of thermodynamic coordinates, and the expectation values of statistical variables are the information needed to determine the probability distribution of states.

Two examples make the criterion concrete. For a harmonic solid, a crystal of $\{q_j\}$6 atoms with equilibrium positions $\{q_j\}$7 has normal coordinates $\{q_j\}$8 with frequencies $\{q_j\}$9, free energy

$\forall q\in Q,\ \exists !\,p_q\in Ex$0

phonon entropy

$\forall q\in Q,\ \exists !\,p_q\in Ex$1

and heat capacity at fixed $\forall q\in Q,\ \exists !\,p_q\in Ex$2 given by $\forall q\in Q,\ \exists !\,p_q\in Ex$3. If defect arrangements or strains shift $\forall q\in Q,\ \exists !\,p_q\in Ex$4, new $\forall q\in Q,\ \exists !\,p_q\in Ex$5 must be treated as additional TCs. For point defects, the total interstitial number $\forall q\in Q,\ \exists !\,p_q\in Ex$6 can itself become a real TC through the configurational entropy

$\forall q\in Q,\ \exists !\,p_q\in Ex$7

with conjugate

$\forall q\in Q,\ \exists !\,p_q\in Ex$8

If $\forall q\in Q,\ \exists !\,p_q\in Ex$9 is omitted, the predicted $p(t)$0 misses the defect-formation contribution.

3. Equilibrium state-space, entropy differentiability, and integrability

A distinct equilibrium formulation appears in Sewell’s quantum statistical treatment of macroscopic observables. An infinite macroscopic system is represented by a $p(t)$1-algebra of quasi-local observables together with commuting extensive conserved quantities $p(t)$2, where $p(t)$3. For translationally invariant ergodic states, the intensive observables are the densities

$p(t)$4

and the thermodynamic entropy function is

$p(t)$5

A chosen set of macroscopic observables is thermodynamically complete iff for each $p(t)$6 there exists a unique ergodic state $p(t)$7 with $p(t)$8 and $p(t)$9 (Sewell, 2018).

This definition makes uniqueness, rather than mere parameter counting, the decisive criterion. Under the algebraic assumptions summarized in the paper, thermodynamic completeness implies differentiability of $p(0)\to_{MTP}p(t_f)\iff p(0)\gg_\gamma p(t_f)$0 on $p(0)\to_{MTP}p(t_f)\iff p(0)\gg_\gamma p(t_f)$1. The argument uses Legendre duality, the reduced pressure

$p(0)\to_{MTP}p(t_f)\iff p(0)\gg_\gamma p(t_f)$2

global thermodynamic stability, and the uniqueness of the modular automorphism group. If $p(0)\to_{MTP}p(t_f)\iff p(0)\gg_\gamma p(t_f)$3 were non-differentiable at $p(0)\to_{MTP}p(t_f)\iff p(0)\gg_\gamma p(t_f)$4, the subgradient $p(0)\to_{MTP}p(t_f)\iff p(0)\gg_\gamma p(t_f)$5 would contain at least two distinct vectors; the same equilibrium state would then satisfy two different KMS conditions, which contradicts the uniqueness of modular dynamics. The Ising ferromagnet below $p(0)\to_{MTP}p(t_f)\iff p(0)\gg_\gamma p(t_f)$6 supplies the standard illustration: $p(0)\to_{MTP}p(t_f)\iff p(0)\gg_\gamma p(t_f)$7 alone fails completeness because two ergodic pure phases have the same energy density but opposite magnetization, whereas $p(0)\to_{MTP}p(t_f)\iff p(0)\gg_\gamma p(t_f)$8 with $p(0)\to_{MTP}p(t_f)\iff p(0)\gg_\gamma p(t_f)$9 restores uniqueness.

A more geometric and axiomatic version is given by Cooper and Russell. There the state space $U$0 is a measure space equipped with two preorderings, “warmer-than” and “adiabatically accessible-from,” induced by empirical temperature $U$1 and empirical entropy $U$2. After canonical recalibration one obtains absolute temperature $U$3 and absolute entropy $U$4 so that the map $U$5 pushes the measure forward to Lebesgue measure. In local coordinates this is the Jacobian condition

$U$6

From this follow the Maxwell relations and the existence of the thermodynamic potentials $U$7, $U$8, $U$9, and $X_k$0 (Cooper et al., 2011).

The completeness-test procedure in that framework is explicit. One identifies candidate functions $X_k$1 and $X_k$2, computes

$X_k$3

recalibrates if $X_k$4, verifies the Maxwell relations, checks the integrability of the fundamental $X_k$5-form, and constructs the potentials by line integration. If $X_k$6 on the domain, the description is incomplete and no exact potentials exist. The ideal gas and van der Waals gas satisfy the test in the examples given.

Taken together, these equilibrium formulations separate two issues that are often conflated. A set of variables may parametrize states, yet still fail completeness if it does not single out a unique entropy maximizer or if it does not support a globally integrable thermodynamic structure.

4. Continuous thermomajorization and complete laws for Markovian thermal processes

Lostaglio and Korzekwa formulate completeness for Markovian thermal processes as a reachability problem on energy populations. For a non-degenerate Hamiltonian with Gibbs distribution $X_k$7, a state is identified with its energy-population vector $X_k$8. Thermomajorization is defined by the Lorenz curve $X_k$9 built from the $x_k(t)$0-ordering of the ratios $x_k(t)$1, and continuous thermomajorization,

$x_k(t)$2

means that there exists a continuous path $x_k(t)$3 from $x_k(t)$4 to $x_k(t)$5 such that for all $x_k(t)$6, one has $x_k(t)$7 (Lostaglio et al., 2021).

The central theorem is

$x_k(t)$8

This gives necessary and sufficient conditions for the existence of a Markovian thermal process between the given initial and final energy distributions. The framework also yields an infinite family of generalized entropy-production inequalities. For any convex $x_k(t)$9, the $\xi_k$0-divergence

$\xi_k$1

is non-decreasing along any Markovian thermal-process trajectory. A distinguished exhaustive family is obtained from

$\xi_k$2

Although these conditions are stated as an infinite family, the paper shows that a finite check is sufficient. The argument is based on canonical sequences of adjacent swaps between $\xi_k$3-orderings and on the fact that a change of ordering occurs only when adjacent slopes become equal. The resulting finite second-laws conditions reduce the problem to finitely many thermomajorization checks after $\xi_k$4 full two-level thermalizations along a canonical path. The constructive side of the theory is equally important: the allowed transformations can be realized by protocols built from elementary thermalizations, and an explicit algorithm maintains a frontier of undominated states under full-swap steps. The authors report that, despite super-factorial worst-case growth in principle, pruning keeps the frontier small enough in practice to handle $\xi_k$5 in minutes–hours on a laptop.

This is a different notion of completeness from equilibrium-state uniqueness, but it addresses the same structural question: whether the chosen laws exhaust the admissible thermodynamic transformations. A plausible implication is that, in dynamical resource-theoretic settings, completeness means not merely monotonicity of one entropy production functional, but a criterion that is both necessary and sufficient for reachability.

5. Open-system compatibility, complete positivity, and thermodynamic admissibility

For open quantum systems, thermodynamic completeness appears as a test of whether an approximate master equation is compatible with thermodynamics. Dann and Kosloff derive the compatibility conditions from strict energy conservation between system and bath,

$\xi_k$6

which implies that the dissipative map $\xi_k$7 commutes with the free unitary propagator,

$\xi_k$8

or infinitesimally $\xi_k$9 with $U$00 (Dann et al., 2020).

From spectral analysis one obtains the general thermodynamically compatible GKLS form. The jump operators must be energy-eigenoperators $U$01 or combinations inside degenerate Bohr sets, and the rates must satisfy Kubo–Martin–Schwinger detailed balance,

$U$02

Residual dephasing must lie in the unitary-invariant subspace and can be written as a double commutator with Hermitian operators that are analytic functions of $U$03. The paper organizes this as a six-step checklist: semigroup/CPTP property, unique thermal fixed point, commutation with free unitary, energy-eigenoperator jump structure, detailed balance, and admissible dephasing. The two-coupled-qubit example distinguishes a local ansatz, which fails because $U$04 once the inter-qubit coupling is nonzero, from the global energy-eigenbasis master equation, which passes the test. The same formal structure is used to argue that the nonequilibrium steady state in heat transport is diagonal in the global energy basis and that thermodynamically compatible dynamics has no exceptional points.

Argentieri, Benatti, Floreanini, and Pezzutto connect a related completeness question to complete positivity and the second law in a driven two-level micro-circuit. They compare a non-completely-positive Redfield generator with a completely positive Davies/GKSL generator and define the internal entropy production rate by

$U$05

Whenever the generator is CPTP, one also has

$U$06

The non-positive Kossakowski matrix of the Redfield generator permits states with $U$07 and repeated second-law violations, whereas the GKSL generator preserves $U$08 (Argentieri et al., 2015).

Their proposed thermodynamic completeness test is operational. The physical current operator is proportional to $U$09, the asymptotic current is directly measurable, and the protocol compares the experimentally observed long-time current with the distinct predictions of the Redfield and GKSL dynamics. In that usage, thermodynamic completeness functions as a reliability test for the reduced dynamics: a current observable is used to discriminate a completely positive, thermodynamically admissible master equation from one that violates the second law.

6. Reduced dynamics, hidden records, and thermodynamic incompleteness

Recent work sharpens the distinction between state evolution and thermodynamic record. In non-Markovian Majorana transport, the island density matrix obeys a memory-kernel equation

$U$10

with $U$11 and

$U$12

The crucial observation is that the island equation depends only on the sums over reservoir channels $U$13 and has no knowledge of how that sum is distributed among channel pairs. By contrast, charge-current averages, zero-frequency noise, and mixed charge-energy cumulants require the detailed channel labels and therefore the tilted kernels with counting fields introduced before the sum (Tian, 7 May 2026).

The thermodynamic completeness criterion in that setting is: a transport observable $U$14 can be read off from the island memory-kernel equation iff $U$15 is invariant under any redistribution of the underlying channel-pair kernels that leaves their sum unchanged. Equivalently, if $U$16 is the linear projection sending the detailed reservoir kernel to the island kernel, then completeness for $U$17 requires

$U$18

If two microscopic assignments have the same island memory kernel but different cumulant-generating functions, the island equation is thermodynamically incomplete for that observable. The two energy-filtered devices in the paper exhibit exactly this phenomenon: they have identical island tomography and relaxation, yet different charge noise measured separately in the leads, heat noise, and mixed charge-energy correlations. The geometry of the projection from reservoir records to island kernels and the topology of the network of tunnel contacts identify which transport information is absent from island-state dynamics.

An analogous but more general result is developed for quantum and classical Markovian dynamics through a path-space action formalism in which the state trajectory and the thermodynamic record are distinct components of a Markovian path. The state $U$19 and current record $U$20 satisfy

$U$21

and the action is

$U$22

At a fixed state $U$23, with admissible current space $U$24 and linear map $U$25, a probe $U$26 is thermodynamically complete iff the following equivalent conditions hold:

$U$27

$U$28

$U$29

Thus incomplete probes satisfy

$U$30

This formulation shows that an unconditioned density-matrix generator does not determine heat-current, particle-transfer, photon-counting, spin-transfer, or continuous-measurement statistics; the quantum instrument and the thermodynamic increment assigned to each record outcome must also be specified. In the classical representation, the same test identifies hidden exchange, reaction, transport, and kinetic current records eliminated by projection to the state trajectory. The geometric and topological language is explicit: the current-space Hessian projects to a quotient state geometry, while graph cycles, divergence-free currents, and harmonic currents span directions invisible to state observations (Tian, 4 May 2026).

These two papers recast thermodynamic completeness as a projection problem. A reduced equation may be exact for the state, but incomplete for observables that vary along the fibers of the projection from records to states.

7. Diagnostics and specialized extensions

A diagnostic variant appears in complex Langevin simulations. There the proposed criterion is configurational temperature, derived from the gradient and Hessian of the complex action. In a configurational setting with $U$31 one has the Rugh–Butler formula

$U$32

where $U$33 and $U$34. In the Euclidean field-theory analogy, if the algorithm samples $U$35 instead of $U$36, the measured inverse temperature is $U$37 rather than unity. The estimator used in the paper is

$U$38

The authors argue that this probes thermodynamic consistency directly, rather than merely the drift behavior. In one-dimensional PT-symmetric models it reproduces the input temperature with high precision and sensitively detects algorithmic errors, step-size artifacts, and incomplete thermalization (Joseph et al., 28 Jan 2026).

A specialized integrable-systems usage is provided by the open XXX spin-$U$39 chain with arbitrary boundary fields. There the “thermodynamic completeness test” refers to verifying that the off-diagonal Bethe ansatz equations are complete at finite $U$40 and reduce correctly in the thermodynamic limit. Numerical diagonalizations for $U$41 with $U$42, $U$43, and $U$44 show that the Bethe equations with the minimal choice of $U$45 produce exactly $U$46 distinct solutions whose energies coincide with all $U$47 levels from exact diagonalization. As the relative angle between the boundary fields tends to zero, the extended $U$48–$U$49 relations and Bethe equations reduce to the conventional parallel-boundary form, establishing a one-to-one correspondence between roots in the two parameterizations. In the thermodynamic limit, the boundary-angle contribution to the energy is of order $U$50 (Jiang et al., 2013).

These diagnostic and specialized uses broaden the term without altering its structural role. The common issue remains whether the retained data are sufficient to reconstruct the thermodynamic content relevant to the problem at hand. In some settings that content is equilibrium entropy and response; in others it is reachability under thermal processes, admissibility of a master equation, lead-resolved transport statistics, or the reliability of a numerical sampling scheme.