Equipartition theorem¶
In classical thermal equilibrium, each eligible independent quadratic energy term has mean energy k_B T/2.
Core Idea¶
The equipartition theorem says that an eligible independent quadratic energy term in a classical thermal equilibrium ensemble contributes an average energy of \(k_B T/2\), where \(T\) is absolute temperature and \(k_B\) is Boltzmann's constant. An ideal-gas momentum term \(p_x^2/(2m)\) and an equilibrated capacitor's stored-energy term \(CV^2/2\) have different physical carriers but the same positive quadratic form. When their continuous equilibrium distributions are applicable, their coefficients affect the spread of the variable, not the mean energy assigned to that one term.[1][2]
The word average is load-bearing. An individual gas molecule need not have equal kinetic energy in its three directions at a given instant; a capacitor's instantaneous voltage need not equal its root-mean-square fluctuation. Equipartition is a statement about an ensemble or long-run equilibrium average, not an equal allocation within every microstate. Classical thermal equilibrium and a genuinely available quadratic mode are preconditions. A quantum mode whose energy levels are too widely separated relative to \(k_B T\) does not receive an unconditional classical share.[1][3]
Structural Signature¶
Sig role-phrases: equilibrium temperature → eligible quadratic mode → Boltzmann-weighted averaging → \(k_B T/2\) term mean → conditional observable inference.
- Equilibrium temperature. A thermal equilibrium setting fixes \(T\) and supplies the statistical weights needed to average across accessible microstates. The theorem cannot be used to assign an instantaneous value, or transferred unchanged to a driven system that lacks the relevant equilibrium distribution.[1]
- Eligible quadratic mode. One independent squared coordinate or momentum appears with a positive coefficient in the energy, such as \(p_x^2/(2m)\), \(kx^2/2\) or \(CV^2/2\). Boundary restrictions, coupled variables and quantum discreteness must be handled rather than counted blindly as extra terms. A harmonic oscillator contributes two quadratic terms—kinetic and potential—when both are classically active.[1][3]
- Boltzmann-weighted average. The equilibrium probability falls with energy, but the variable can fluctuate. Integrating its quadratic contribution over the allowed classical continuum makes its mean \(k_B T/2\); the physical coefficient shifts the typical amplitude instead of changing this mean. Berkeley explicitly recovers the capacitor result from the Gaussian Boltzmann integral.[1][2]
- Equal modal mean. The result attaches to each qualifying term. Three independent translational momentum components sum to \(3k_B T/2\) per ideal monatomic particle; one capacitor-voltage term yields \(\langle CV^2/2\rangle=k_B T/2\). These are comparable energy averages, not identical measured variables.[1][2]
- Conditional observable inference. If the modeled terms exhaust the relevant temperature-dependent energy, one may sum them to estimate an idealized internal energy and differentiate for a heat capacity. For a single quadratic displacement or voltage, one can instead infer fluctuation variance. These inferences require their own model assumptions and must not be used as a general spectral-noise or real-material law.[3][2]
What It Is Not¶
It is not the claim that every degree of freedom under every description carries \(k_B T/2\). What counts is a classically accessible, independently handled quadratic term in the energy; one vibrational mode can have both kinetic and potential terms, while a quantum-frozen mode contributes less than a naive classical count. Nor does an arbitrary nonquadratic term meet this version of the theorem simply because it is a named coordinate.[1][3]
It is not instantaneous equal sharing, a guarantee that collisions have already equilibrated a system, or a universal heat-capacity formula. The textbook monatomic-gas and idealized-solid counts need their specified constant-volume and harmonic models. In actual diatomic gases, rotational and vibrational contributions change as temperature crosses their relevant quantum energy scales. The classical rule remains meaningful as a limit and diagnostic of where those models cease to work.[3]
It is not the information-theoretic asymptotic equipartition property, which concerns typical sequence probabilities, nor the live Partition Function itself. A partition function encodes equilibrium statistical weights; this theorem extracts a special quadratic-mode expectation from such weights. The similar word “equipartition” is not sufficient to merge the identities.
Scope of Application¶
In classical ideal-gas translation, the three momentum components enter as \(p_x^2/(2m)+p_y^2/(2m)+p_z^2/(2m)\). MIT's lecture derives \(k_B T/2\) in each direction and \(3k_B T/2\) total per particle. For a monatomic ideal gas at constant volume, with no additional temperature-dependent internal modes in the model, that supports a molar heat capacity \(3R/2\). It does not imply that all gases at all temperatures have the same heat capacity.[1][3]
For a classically active harmonic displacement \(x\) with potential \(kx^2/2\), the same rule implies \(\langle x^2\rangle=k_B T/k\) if the equilibrium coordinate is centered and the quadratic model is valid. Adding its quadratic kinetic counterpart gives \(k_B T\) mean energy per full harmonic oscillator. The equation is a variance statement about a specified coordinate, not a promise that all physical springs remain harmonic at any amplitude or temperature.[1]
In electrical engineering, a capacitor thermally coupled to a resistor stores \(CV^2/2\). Berkeley's instructor notes apply equipartition and independently evaluate the Boltzmann integral to obtain \(\langle V^2\rangle=k_B T/C\). This is a total equilibrium voltage variance under the circuit model. A frequency-dependent noise spectrum requires additional circuit and fluctuation analysis; equipartition alone does not give it.[2]
Clarity¶
The theorem replaces an ambiguous intuition—“heat spreads equally”—with a typed test: which term is quadratic, what ensemble averages it, and what does the physical variable mean? Equal mean energy per term does not mean equal mean speeds, equal voltage and velocity fluctuations, or equal energy at each moment. In a gas, mass influences \(\langle v_x^2\rangle\) even though \(\langle mv_x^2/2\rangle\) remains \(k_B T/2\). In a capacitor, capacitance changes \(\langle V^2\rangle\) while \(\langle CV^2/2\rangle\) remains the same classical mean.[1][2]
It also separates a theorem from its model-limited consequences. The heat capacity of a diatomic gas cannot be predicted by blindly counting all nominal rotational and vibrational coordinates as fully active at low temperature. OpenStax's hydrogen example shows distinct plateaus as quantum-level accessibility changes. A mismatch can expose a bad mode count or a regime where classical assumptions fail, rather than disproving the conditional theorem.[3]
Manages Complexity¶
Instead of solving every collision or microscopic trajectory, the analyst can identify eligible quadratic terms and sum their equilibrium averages. That compresses a distribution over many microstates into a modal energy prediction. The same calculation can then be reinterpreted as a gas's translational energy, a harmonic coordinate's thermal spread or a capacitor's noise variance, with the coefficient restored when converting energy back to the observed variable.[1][2]
The compression loses detail: equipartition does not specify fluctuation time correlations, frequency spectra, nonquadratic response, the approach to equilibrium or which quantum levels are populated. It helps locate the right average but does not replace the statistical distribution or dynamics needed for those other questions. The source textbooks' quantum heat-capacity failures and Berkeley's separate RC noise-spectrum derivation make this limit concrete.[3][2]
Abstract Reasoning¶
First identify the equilibrium ensemble and temperature. Write the relevant energy terms explicitly, diagonalizing coupled positive quadratic forms or stating why a coordinate is independently countable. Check that the mode is physically accessible and well described by a continuous classical integral at the temperature in question. Assign \(k_B T/2\) only to qualifying terms. For a claimed observable, map the energy expectation back through its coefficient and state any further assumptions: ideal-gas constant volume for heat capacity, quadratic stiffness for displacement variance, or an equilibrated RC model for voltage variance.[1][3][2]
If the predicted value disagrees with evidence, ask which link failed. Was the system out of equilibrium, the term nonquadratic, a mode coupled or constrained, or its spectrum quantum-discrete on the thermal scale? Did the model omit interactions or treat a finite-bandwidth measurement as a total variance? Each question targets a different part of the reasoning; simply adding or subtracting a degree of freedom is not a general repair.
Knowledge Transfer¶
Transfer from gas translation to capacitor fluctuations is literal at the statistical-mechanical level. The former integrates a Gaussian momentum component; the latter integrates a Gaussian voltage coordinate under stored-energy \(CV^2/2\). Both preserve equilibrium temperature, one accessible quadratic term and an ensemble mean of \(k_B T/2\). Their outputs differ: momentum variance converts to molecular speed and gas energy, while voltage variance converts to an electrical noise amplitude. A gas heat-capacity claim does not transfer to an RC spectrum, and the circuit's capacitance dependence does not change the equal-energy theorem.[1][2]
The theorem remains domain-specific even across these fields: both settings are physical thermal systems with Boltzmann statistics. Treating organizational resources or information bits as if they each must receive \(k_B T/2\) would be metaphor without a temperature, Hamiltonian and equilibrium measure.
Examples¶
Translational energy of an ideal monatomic gas¶
MIT's classical description writes a particle's translational energy as three separable squared momentum components. At thermal equilibrium, each has mean \(k_B T/2\), so the particle's mean translational energy is \(3k_B T/2\). In the ideal monatomic constant-volume model, this yields a molar heat-capacity contribution \(3R/2\). The result does not assert that any one molecule's three components are equal in a snapshot.[1][3]
Mapped back: Equilibrium temperature = gas thermal state at \(T\); eligible quadratic modes = \(p_x^2/(2m)\), \(p_y^2/(2m)\), \(p_z^2/(2m)\); Boltzmann-weighted averaging = classical momentum distribution; equal modal mean = \(k_B T/2\) for each component; conditional observable inference = \(3k_B T/2\) per particle and, under ideal monatomic constant-volume assumptions, \(3R/2\) molar heat capacity.
Equilibrium voltage fluctuations of an RC circuit¶
In Berkeley's RC example, a resistor at temperature \(T\) thermally agitates the charge on a capacitor. The capacitor energy is \(CV^2/2\), so its one quadratic voltage coordinate has mean energy \(k_B T/2\). Consequently \(\langle V^2\rangle=k_B T/C\). Berkeley also derives this result from the Boltzmann distribution. The example illustrates a variance, not an instantaneous voltage or a complete frequency spectrum.[2]
Mapped back: Equilibrium temperature = resistor–capacitor system at \(T\); eligible quadratic mode = \(CV^2/2\); Boltzmann-weighted averaging = Gaussian equilibrium distribution over \(V\); equal modal mean = \(\langle CV^2/2\rangle=k_B T/2\); conditional observable inference = total mean-square capacitor voltage \(k_B T/C\) under the ideal circuit model.
Structural Tensions¶
- T1: Short mode count versus physical-model fidelity. Counting quadratic terms efficiently predicts thermal means, but a heat-capacity or variance claim can fail if relevant interactions, constraints or additional energy terms were omitted. Diagnostic: Which temperature-dependent terms does the proposed count actually cover, and what further assumptions connect their sum to the observable?[1][3]
- T2: Classical continuum versus quantum-accessible spectrum. Classical Gaussian integration yields the equal share, while levels spaced too widely relative to \(k_B T\) are not populated as a continuum; the diatomic heat-capacity plateaus expose this regime shift. Diagnostic: Is the selected mode's relevant energy gap small on the thermal scale?[3]
- T3: Common energy mean versus different readouts. Gas momentum and capacitor voltage share \(k_B T/2\) as an energy average, but their variable variances depend on mass and capacitance, and their time-dependent measurements need additional analysis. Diagnostic: What coefficient and observation rule convert modal energy to the reported speed, voltage, heat capacity or spectrum?[1][2]
Structural–Framed Character¶
The theorem is structurally mathematical within physics. Evaluative weight is absent from the claim: a quadratic term either has the conditional ensemble mean or it does not, independent of whether the heat or noise is desirable. Human-practice dependence enters through how a model selects coordinates and validates equilibrium, not through a social convention determining \(k_B T/2\). Institutional origin in thermodynamics and statistical mechanics explains the language; it is not an extra physical premise.[1]
Vocabulary travel from molecular motion to electrical noise is legitimate because temperature, energy and Boltzmann averaging occupy the same roles in both. It does not authorize the information-theoretic use of “equipartition” as the same theorem. Import versus recognition means one can recognize equipartition in a newly modeled RC circuit by verifying its quadratic energy and thermal equilibrium; one cannot import the formula into a nonthermal system by analogy alone. Its character: a conditional, structurally precise thermal-average theorem with robust cross-setting use inside classical statistical physics and explicit quantum/model boundaries.
Structural Core vs. Domain Accent¶
Portable skeleton. The shared structure is equilibrium measure + independent eligible quadratic term → \(k_B T/2\) average energy, followed by a setting-specific conversion to an observable. It is not merely “equal shares” in an untyped resource allocation.[1][2]
Domain-bound mechanism. Physical temperature, Boltzmann weights, Hamiltonian energy and classical accessible modes make the theorem work. Gas translation uses mass-weighted momenta; RC noise uses capacitance-weighted voltage. Gas heat capacity and circuit voltage variance are accents with extra modeling assumptions; neither is the theorem's entire identity. Quantum discreteness limits both whenever the continuous approximation fails.[1][3]
Prime boundary. The named theorem does not travel outside thermal statistical mechanics without replacing its central \(k_B T\) and energy-measure machinery. Degrees of Freedom covers independent parameters more broadly, but does not distribute thermal energy among them. The independent prime-level pattern is weaker than, and cannot substitute for, this exact domain-specific theorem.
Instantiates / Related Primes¶
This entry presupposes Thermodynamic Equilibrium. Classical equipartition requires an equilibrium thermal state at temperature T, then adds quadratic-mode and ensemble-average conditions.
Relationships to Other Abstractions¶
Current abstraction Equipartition theorem Domain-specific
Parents (1) — more general patterns this builds on
-
Equipartition theorem presupposes Thermodynamic Equilibrium Prime
Classical equipartition requires an equilibrium thermal state at temperature T, then adds quadratic-mode and ensemble-average conditions.The live Thermodynamic Equilibrium prime supplies the macroscopic stationary/no-net-flow condition required for the canonical thermal ensemble of this theorem. The theorem additionally requires eligible quadratic terms and suitable classical statistical weights to derive a mean k_B T/2 per term.
Hierarchy paths (3) — routes to 3 parentless roots
- Equipartition theorem → Thermodynamic Equilibrium → Equilibrium → Fixed Point
- Equipartition theorem → Thermodynamic Equilibrium → Entropy (Thermodynamic Sense)
- Equipartition theorem → Thermodynamic Equilibrium → Second Law of Thermodynamics
Neighborhood in Abstraction Space¶
Equipartition theorem sits in a moderately populated region (58th percentile for distinctiveness): it has near-neighbors but no dense thicket of look-alikes.
Family — Statistical Mechanics & Particle Phenomena (15 abstractions)
Nearest neighbors
- Exponential Stability — 0.87
- Partition Function — 0.87
- Langevin Dynamics — 0.86
- Canonical Ensemble — 0.85
- Carnot's theorem (thermodynamics) — 0.85
Computed from structural-signature embeddings · 2026-10-08
Not to Be Confused With¶
- Equal instantaneous energy: equipartition is an expectation across accessible equilibrium states, not a snapshot identity.[1]
- All coordinates or all named degrees of freedom: only eligible classical quadratic terms receive this simple assignment; a quantum-frozen or hard-constrained coordinate cannot be counted automatically.[3]
- Universal heat capacity: a derived estimate under a specific idealized mode inventory; quantum activation, interactions and additional modes change actual values.[3]
- Full Johnson–Nyquist noise spectrum: Berkeley obtains \(k_B T/C\) variance by equipartition, then needs circuit response and bandwidth analysis for spectral predictions.[2]
- Asymptotic equipartition property: typical sequence probabilities in information theory, not thermal modal energy.
References¶
[1] MIT OpenCourseWare 5.62, Lecture 10: “Quantum vs. Classical. Equipartition. Internal Degrees of Freedom”, Spring 2008, original instructor notes, pp.1–4. registry ↩a ↩b ↩c ↩d ↩e ↩f ↩g ↩h ↩i ↩j ↩k ↩l ↩m ↩n ↩o ↩p ↩q ↩r ↩s ↩t
[2] UC Berkeley EE142, “Physical Origin of Electrical Noise”, original instructor slides, pp.10, 19–24, RC noise variance and Boltzmann integral. registry ↩a ↩b ↩c ↩d ↩e ↩f ↩g ↩h ↩i ↩j ↩k ↩l ↩m ↩n
[3] OpenStax, University Physics Volume 2, §2.3, “Heat Capacity and Equipartition of Energy”, original publisher textbook, hydrogen and idealized-solid discussions. registry ↩a ↩b ↩c ↩d ↩e ↩f ↩g ↩h ↩i ↩j ↩k ↩l ↩m ↩n ↩o