architecture

Mathematical Architecture

The formal structure of the P(t) master formula: integrand, integration domain, dimensional consistency, and the factorisation into physical stages.

Last updated: 2026-07-20

This chapter sets out the formal architecture of the Schubart Master Formula: what is being integrated, over what domain, and why the resulting expression is dimensionally consistent. Having motivated the framework physically in the earlier chapters, the aim here is to make its mathematical skeleton fully explicit, so that any subsequent numerical or analytic work rests on an unambiguous definition.

The expanded physical representation

The canonical Schubart Master Formula is the compact volume form $P(t) = \eta \cdot \int_V \Phi_{\text{eff}}(\mathbf{r},t) \cdot \sigma_{\text{eff}}(E)\,dV$ . To make its mathematical architecture fully explicit, this chapter works with its expanded physical representation,

P(t) = \iint \Phi(E,\theta,t)\,\sigma(E)\,\Delta p\,v_{\text{ph}}\,\eta_{\text{conv}}\;dE\,d\theta .

Here $P(t)$ is the instantaneous usable electrical power [W]; $\Phi(E,\theta,t)$ is the radiation flux density; $\sigma(E)$ is the scattering cross-section; $\Delta p$ is the directed momentum transfer; $v_{\text{ph}}$ is the phonon velocity; and $\eta_{\text{conv}}$ is the conversion efficiency. The double integral runs over the energy variable $E$ and the angular variable $\theta$ . This is the expanded physical representation; the same expression, together with the standard symbol glossary, is reproduced at the equation and the notation reference.

Reading the integrand as a product of stages

The integrand factorises into a chain of physical stages, and this factorisation is the organising idea of the whole framework. Reading left to right: $\Phi$ supplies the incident resource; $\sigma$ sets how strongly that resource couples to the material; $\Delta p$ converts coupling into directed momentum; $v_{\text{ph}}$ transports that momentum; and $\eta_{\text{conv}}$ rectifies it into usable power. Because these appear as a product, the whole integrand vanishes if any factor vanishes — the mathematical expression of the “chain” argument developed in non-equilibrium energy conversion. The product form also means the factors are, to first order, independently tunable: improving transport does not change coupling, and vice versa, which is what makes the framework useful for isolating bottlenecks.

The factorisation can be made precise. Only two factors carry the integration variables — the flux $\Phi(E,\theta,t)$ depends on $E$ and $\theta$ , and the cross-section $\sigma(E)$ depends on $E$ — while $\Delta p$ , $v_{\text{ph}}$ , and $\eta_{\text{conv}}$ are, in the basic form, treated as effective constants representative of the converter. Under that assumption the constant factors pull out of the integral,

P(t) = \Delta p\, v_{\text{ph}}\, \eta_{\text{conv}} \iint \Phi(E,\theta,t)\,\sigma(E)\;dE\,d\theta ,

so the entire spectral and angular content is concentrated in the overlap integral $\iint \Phi\,\sigma\,dE\,d\theta$ . This separation is what licenses the “isolate the bottleneck” methodology: one can study the environment-and-material overlap independently of the transport-and-rectification prefactor. When any of the three prefactors is in fact energy- or angle-dependent — for instance $\Delta p(E,\theta)$ from the microscopic average — it must remain inside the integral, and the separation above holds only approximately. The framework states the general form with all factors inside the integral precisely so that this generalisation requires no change of structure.

Differentiating the factorised form gives the sensitivity of the output to each stage. Because the dependence is multiplicative, the relative sensitivities simply add:

\frac{dP}{P} = \frac{d(\Delta p)}{\Delta p} + \frac{dv_{\text{ph}}}{v_{\text{ph}}} + \frac{d\eta_{\text{conv}}}{\eta_{\text{conv}}} + \frac{d\!\left(\iint \Phi\sigma\right)}{\iint \Phi\sigma} .

A ten-percent improvement in any single factor yields a ten-percent improvement in the output, and the worst factor dominates the achievable gain. This logarithmic-derivative structure is the quantitative backbone of the bottleneck argument.

The integration domain

The domain of integration deserves care. The energy integral runs over the range of energies present in the ambient spectrum and to which the converter can couple; formally one may take it from zero to infinity, with the understanding that the integrand is negligible outside the support of $\Phi(E,\theta,t)\,\sigma(E)$ . The angular integral runs over the solid angle subtended by the incident flux at the converter surface. In many treatments $\theta$ is a polar angle and a full solid-angle integration would introduce the usual $\sin\theta$ measure; the framework keeps $\theta$ explicit precisely so that whatever angular measure is appropriate to a given geometry can be inserted without changing the overall structure. The important formal point is that the limits are set by the physical support of the integrand, not chosen arbitrarily.

Time dependence

The output carries an explicit time argument because the flux does. Time enters through $\Phi(E,\theta,t)$ ; the coupling and rectification factors are treated as quasi-static, meaning they respond effectively instantaneously to changes in flux on the timescales of interest. Consequently $P(t)$ inherits its time dependence from the environment, and the framework naturally describes a converter whose output tracks a fluctuating ambient field. If any of the material parameters were themselves time-dependent — for example if $\eta_{\text{conv}}$ drifted with temperature — they too could be given a time argument without altering the form of the integral.

The quasi-static assumption can be stated as a separation of timescales. Let $\tau_{\text{env}}$ be the characteristic time over which the ambient flux varies and $\tau_{\text{resp}}$ the internal response time of the converter (set by scattering and collection). The instantaneous form of the master formula is valid when

\tau_{\text{resp}} \ll \tau_{\text{env}} ,

so that at every instant the converter has effectively reached its steady response to the current flux. When this holds, one may average the output over an observation window $T$ to obtain a mean power,

\langle P \rangle = \frac{1}{T}\int_0^T P(t)\;dt ,

which is the quantity most experiments actually report. Should the separation fail — a flux varying faster than the converter can respond — the framework would require a convolution with a response kernel in place of the instantaneous product, an extension that leaves the spectral–angular structure untouched but replaces the quasi-static factors with memory functions.

Order of integration and separability

Because the integrand is a product of a spectral–angular part and a set of quasi-constant factors, Fubini’s theorem guarantees that the double integral may be evaluated in either order, $\int dE \int d\theta$ or $\int d\theta \int dE$ , whenever the integrand is absolutely integrable — which it is, since $\Phi$ and $\sigma$ are non-negative and the integrand has finite support. This freedom is practically useful: one may integrate over angle first to obtain an effective spectral flux $\Phi_{\text{eff}}(E)=\int \Phi(E,\theta,t)\,d\theta$ and then perform a single spectral overlap integral,

P(t) = \Delta p\, v_{\text{ph}}\, \eta_{\text{conv}} \int_0^\infty \Phi_{\text{eff}}(E,t)\,\sigma(E)\;dE ,

or, conversely, integrate over energy first to obtain an angular power pattern. The two routes must agree, and checking that they do is a simple but effective self-consistency test for any numerical implementation. When a factor genuinely couples the two variables — for instance an angle-dependent cross-section $\sigma(E,\theta)$ — separability is lost and the full double integral must be retained, but the interchange of integration order remains valid. This is another reason the framework is written with both integrals explicit rather than pre-collapsed into an effective one-dimensional form.

Dimensional consistency

A framework of this kind is only credible if it is dimensionally sound, so it is worth checking. The flux density $\Phi$ has units of energy per area, per time, per solid angle, per energy — that is, $\mathrm{W\,m^{-2}\,sr^{-1}\,J^{-1}}$ when the energy density is expressed per joule. The cross-section $\sigma$ contributes $\mathrm{m^{2}}$ , the momentum transfer $\Delta p$ contributes $\mathrm{kg\,m\,s^{-1}}$ , the phonon velocity $v_{\text{ph}}$ contributes $\mathrm{m\,s^{-1}}$ , and $\eta_{\text{conv}}$ is dimensionless. The integration measure $dE\,d\theta$ contributes energy times solid angle, $\mathrm{J\,sr}$ . In any concrete realisation of the framework the product of these units, together with whatever proportionality constant relates the geometric factors to power, must reduce to watts. The framework requires that this reduction hold; a specific application must supply the constants that make the bookkeeping close exactly. This discipline — insisting that the units balance before any number is trusted — is what keeps the formalism from drifting into unfalsifiable territory.

Positivity and boundedness

Two structural properties reassure us that the formula is well-behaved. First, positivity: the flux $\Phi$ is non-negative by definition, the cross-section $\sigma$ is non-negative, $v_{\text{ph}}$ is positive, and $\eta_{\text{conv}}$ lies in $[0,1]$ . The only factor that can carry a sign is the directed momentum transfer $\Delta p$ , which is positive when the net push is toward the collection axis and would be negative for a push against it. In a properly oriented converter $\Delta p \ge 0$ , so the integrand is non-negative and $P(t) \ge 0$ : the framework cannot report negative delivered power, as physical sense demands. Second, boundedness: because $\eta_{\text{conv}} \le 1$ , the delivered power can never exceed the total rectifiable power crossing the boundary,

P(t) \;\le\; \iint \Phi(E,\theta,t)\,\sigma(E)\,\Delta p\,v_{\text{ph}}\;dE\,d\theta ,

which is itself bounded by the incident flux. There is no configuration of the parameters that lets the output exceed the resource supplied by the environment. Positivity and boundedness together are the mathematical guarantees that correspond, respectively, to “no spontaneous reverse current” and “no output beyond the available resource” — the two things a credible energy-conversion model must never violate.

From structure to evaluation

With the architecture fixed, the remaining work is interpretation and evaluation. The meaning and units of each factor are catalogued in variable definitions; the control surface across which the balance is drawn is specified in system boundaries; the full step-by-step construction is given in full derivation; and the governing relations that follow from the structure are collected in governing relations.