Boundary Conditions for Stellar Models

The four equations of stellar structure — mass continuity, hydrostatic equilibrium, energy generation, and energy transport — are first-order ordinary differential equations, and by themselves they are useless. They describe how the pressure, radius, luminosity, and temperature of a star change as you move outward through its mass, but a differential equation with no boundary data admits infinitely many solutions, the way "the slope is 2" describes a whole family of parallel lines. What turns this open-ended system into a single physical star is the imposition of boundary conditions, and their placement is peculiar: two conditions belong at the center and two at the surface. This two-point boundary value problem — that split — is the source of everything hard and interesting about computing stellar models. At the center the conditions are pure regularity: a mathematical point encloses no volume and no energy source, so the radius and interior luminosity must both vanish, $r=0$ and $L=0$ . Crucially, the central pressure and temperature are not handed to us — they are eigenvalue-like numbers the global solution must discover. At the surface the naive move is to set pressure, density, and temperature to zero at the stellar radius, but a real star has no sharp edge; it fades into a photosphere at optical depth $\tau\approx 2/3$ where photons finally escape carrying the imprint of the effective temperature. Because the constraints live at opposite ends, no single integration can thread them — the whole apparatus of shooting and relaxation methods exists to solve exactly this. This article closes the stellar-structure system that its prerequisites opened.

Key takeaways

The four stellar-structure ODEs form a two-point boundary value problem: two conditions at the center, two at the surface — you cannot integrate them in one pass from either end.
The central conditions are regularity statements: as $m\to 0$ , $r=0$ and $L=0$ , because a point holds no volume and no enclosed source. Central $P_c$ and $T_c$ are unknowns fixed by the global fit.
The crude surface conditions $P=T=\rho=0$ at $r=R$ are analytically handy (polytropes) but unphysical for the outer layers.
Realistic surface conditions live at the photosphere, $\tau=2/3$ , where the Eddington gray law gives $T=T_{\rm eff}$ and hydrostatic balance gives $P_{\rm phot}\approx \tfrac{2}{3}\,g/\kappa$ .
Because data are split between two points, the system is solved by shooting (integrate from both ends, match at a fitting point) or by Henyey relaxation (linearize the whole grid and iterate).
Together with the equation of state, opacity, and energy generation, the boundary conditions select a unique model for a given mass and composition — the content of the Vogt–Russell theorem.
Key idea: two regularity conditions at the center ( $r=0$ , $L=0$ ) and two physical conditions at the photosphere ( $T=T_{\rm eff}$ , $P\approx\tfrac{2}{3}g/\kappa$ ) close the four stellar-structure ODEs as a two-point boundary value problem with a unique solution.

The core insight

Two regularity conditions at the center ( $r=0$ , $L=0$ ) and two physical conditions at the surface — a photospheric $T=T_{\rm eff}$ and $P\approx\tfrac{2}{3}g/\kappa$ , or the crude $T\to 0$ , $P\to 0$ — close the four stellar-structure ODEs as a two-point boundary value problem.

Where this lives today

Every modern stellar-evolution code — MESA, the Geneva and Padova grids, GARSTEC — is at bottom a solver for this two-point boundary value problem, re-run as the composition slowly changes over a star's life. The workhorse is Henyey relaxation: the four ODEs are discretized on a mass grid of hundreds to thousands of shells, the boundary conditions are appended as extra equations, and the whole block-tridiagonal system is Newton-iterated to convergence at each timestep. The surface is where the physics is still refined: instead of the one-line Eddington relation, current codes attach precomputed non-gray model atmospheres (PHOENIX, MARCS, ATLAS grids in $T_{\rm eff}$ and $\log g$ ) at a fitting depth $\tau\sim 10$ – $100$ , capturing line blanketing and near-surface convection the gray law omits. This matters observationally: it sets the predicted radii and colors that asteroseismology (Kepler, TESS) and interferometry (CHARA, VLTI) now measure to better than a percent. The same problem, ported to the outer layers of neutron stars and the irradiated atmospheres of hot Jupiters, remains an active frontier.

The physical picture

Think of the star as a rope stretched between two anchors. The four differential equations describe the shape the rope takes under tension, but a shape is only pinned down once you nail both ends. Half the nails are at the center — pure regularity, the demand that nothing blow up at the origin — and half at the surface, where the star meets vacuum and photons leak out. The center speaks geometry (a point holds zero mass and zero light); the surface speaks physics (photons escape where the gas turns transparent). The first figure lays out this two-anchor structure and shows why you must integrate from both ends toward a meeting point in the middle.

Figure

Two-point boundary value problem for a star: two regularity conditions anchor the center (r = 0, L = 0) and two physical conditions anchor the photosphere at optical depth τ = 2/3 (T = T_eff = 5772 K, P = (2/3) g/κ, with g = 274 m/s² for the Sun); shooting integrates from both ends toward a fitting shell where r, P, L, T must all match

The center is not where the drama is; the surface is. Deep inside, the pressure is so enormous — of order $10^{16}\ \mathrm{Pa}$ in the Sun — that whether you set the surface pressure to zero or to its true photospheric value of $\sim 10^4\ \mathrm{Pa}$ makes a difference of one part in $10^{12}$ for the interior. That is why the crude "zero" surface condition is fine for the deep structure and for polytropes, yet fatal for the radius and near-surface temperature: the outermost layers, where a photon finally random-walks to freedom, are exactly the ones it mishandles. Photons escape not from a mathematical edge but from the photosphere, where the optical depth reaches order unity — precisely $\tau=2/3$ in the gray approximation, and there the local temperature equals the effective temperature. The second figure shows the temperature law that pins this down.

Figure

The Eddington gray-atmosphere law T⁴ = (3/4) T_eff⁴ (τ + 2/3): the local temperature equals the effective temperature exactly at τ = 2/3, marked at T = 5772 K for the Sun, while the true surface τ = 0 is cooler, T = (1/2)^(1/4) T_eff = 0.841 T_eff = 4854 K

The core equations

The starting point is the closed set of four structure equations in Lagrangian form, with the mass $m$ interior to radius $r$ as the independent variable. The mechanical pair balances gravity against pressure and relates radius to enclosed mass; write them together.

\frac{\mathrm{d}r}{\mathrm{d}m}=\frac{1}{4\pi r^2\rho},\qquad \frac{\mathrm{d}P}{\mathrm{d}m}=-\frac{Gm}{4\pi r^4} \qquad (1)

where: $r$ — radius enclosing mass $m$ ; $m$ — interior mass, the independent variable; $\rho$ — local density; $P$ — total pressure (gas plus radiation); $G$ — gravitational constant. Reading: the first is geometry (radius grows as you add shells); the second is hydrostatic equilibrium (each shell's weight held up by the pressure gradient). The thermal pair carries the energy budget: luminosity is generated by the enclosed sources, and temperature falls outward at a rate set by whichever transport mechanism — radiation or convection — actually carries the flux.

\frac{\mathrm{d}L}{\mathrm{d}m}=\varepsilon,\qquad \frac{\mathrm{d}T}{\mathrm{d}m}=-\frac{Gm\,T}{4\pi r^4 P}\,\nabla \qquad (2)

where: $L$ — luminosity crossing the shell at $m$ ; $\varepsilon$ — specific energy generation rate (nuclear minus neutrino, plus gravothermal terms); $T$ — temperature; $\nabla\equiv\mathrm{d}\ln T/\mathrm{d}\ln P$ , equal to $\nabla_{\rm rad}$ if radiative, $\approx\nabla_{\rm ad}$ if convective. Reading: four coupled first-order ODEs need four boundary conditions. At the center, regularity fixes two of them. A mathematical point encloses zero volume and zero source, so both the radius and the interior luminosity must vanish as the enclosed mass goes to zero.

\text{as } m\to 0:\qquad r\to 0,\qquad L\to 0 \qquad (3)

where: $r\to 0$ — a point encloses no volume; $L\to 0$ — a point encloses no source. Reading: notice what is not fixed — the central $P_c$ and $T_c$ are unknowns, eigenvalue-like numbers the global solution must select so the surface conditions come out right. The surface reference is the optical depth, counted inward from space, which measures how many photon mean free paths of material lie above a given layer.

\tau(r)=\int_r^{\infty}\kappa\,\rho\,\mathrm{d}r' \qquad (4)

where: $\tau$ — optical depth measured inward from outside; $\kappa$ — Rosseland mean opacity per unit mass. Reading: $\tau=0$ is deep space, $\tau$ grows inward, and $\tau\approx 2/3$ is the photosphere from which the emergent light effectively originates. The two physical surface conditions then live at the photosphere. The temperature there equals the effective temperature, and the luminosity is fixed by the blackbody area law.

\tau_{\rm phot}=\frac{2}{3}\ \Rightarrow\ T=T_{\rm eff},\qquad L=4\pi R^2\sigma T_{\rm eff}^4 \qquad (5)

where: $T_{\rm eff}$ — effective temperature; $R$ — photospheric radius; $\sigma$ — Stefan–Boltzmann constant. Reading: this replaces the crude $T(R)=0$ with a finite, observable temperature at finite radius. The fourth condition is the photospheric pressure, obtained by integrating hydrostatic balance across the thin surface layers down to $\tau=2/3$ ; Derivation 3 supplies it.

P_{\rm phot}\simeq \frac{2}{3}\,\frac{g}{\kappa},\qquad g=\frac{GM}{R^2} \qquad (6)

where: $g$ — surface gravity; $M$ — total stellar mass; $\kappa$ — surface-layer opacity. Reading: the surface pressure is how much material one mean free path of opacity holds up against gravity — it ties the outer boundary to $M$ , $R$ , and $\kappa$ .

Derivation AWhy the center forces r = 0 and L = 0, and the near-center series

The goal is to show that the two central boundary conditions are not choices but forced consequences of regularity, and then to extract the leading behavior near the origin — the small-mass expansions that supply the starting values for an outward integration. We assume only that the central density, pressure, temperature, and energy-generation rate are finite, and follow the mass-continuity and energy equations into the limit $m\to 0$ , watching what stays finite and what would have to diverge.

Step 1. Near the center the density is nearly uniform at its central value $\rho_c$ , so the enclosed mass of a small sphere is $m\approx\tfrac{4}{3}\pi r^3\rho_c$ . As $r\to 0$ this drives $m\to 0$ too, and inverting gives $r\approx(3m/4\pi\rho_c)^{1/3}$ . Why: with finite $\rho_c$ the radius and the enclosed mass must vanish together — a point holds no volume, so $r=0$ at $m=0$ is forced, not assumed.

Step 2. Integrate the energy equation with $\varepsilon\approx\varepsilon_c$ constant over the innermost shells: $L(m)=\int_0^m\varepsilon\,\mathrm{d}m'\approx\varepsilon_c\,m$ , so $L\to 0$ as $m\to 0$ . Why: a mathematical point encloses no source; a nonzero central luminosity would demand an infinite energy-generation rate, so $L=0$ is likewise forced.

Step 3. Feed the radius expansion into hydrostatic balance: $\mathrm{d}P/\mathrm{d}m=-Gm/4\pi r^4=-\tfrac{G}{4\pi}\left(4\pi\rho_c/3\right)^{4/3}m^{-1/3}$ . Integrating from the center gives $P(m)\approx P_c-\tfrac{3G}{8\pi}\left(4\pi\rho_c/3\right)^{4/3}m^{2/3}$ . Why: the pressure approaches its central value smoothly as $m^{2/3}$ , with zero slope at the origin — the regular behavior a physical core must have.

Step 4. Combine the radiative gradient with $L\propto m$ to get the temperature series, $T^4(m)\approx T_c^4-\tfrac{1}{2ac}\tfrac{\kappa_c\varepsilon_c}{(4\pi)^2}\left(4\pi\rho_c/3\right)^{4/3}m^{2/3}$ , where $a$ is the radiation constant and $c$ the speed of light. Why: this confirms $T\to T_c$ smoothly and, with $P(m)$ , gives the four starting values a shooting code needs to lift off from the singular point $m=0$ . The result reproduces the standard central expansions (Kippenhahn, Weigert & Weiss; Hansen, Kawaler & Trimble): $L\propto m$ and $P_c-P\propto m^{2/3}$ are the leading regular terms, and any other choice at $m=0$ demands a divergent density or source.

Go deeper · Derivation B · 1 minThe Eddington gray-atmosphere temperature law

Derivation BThe Eddington gray-atmosphere temperature law

The goal is to derive the surface temperature condition from first principles: to show that a plane-parallel, gray, radiative-equilibrium atmosphere obeys $T^4(\tau)=\tfrac{3}{4}T_{\rm eff}^4(\tau+2/3)$ , so the local temperature equals the effective temperature at exactly $\tau=2/3$ . This is what justifies placing the outer boundary at the photosphere rather than at a fictitious zero-temperature edge. We work with the first two angular moments of the radiative transfer equation and close them with the Eddington approximation.

Step 1. In radiative equilibrium the total flux carried by radiation is constant with depth and defines the effective temperature, $F=\sigma T_{\rm eff}^4=\text{const}$ . Why: no energy is created or destroyed in the atmosphere, so whatever flux leaves the top must pass through every layer — this constant flux is the anchor of the whole derivation.

Step 2. The first moment of the transfer equation gives $\mathrm{d}P_{\rm rad}/\mathrm{d}\tau=F/c$ , and the Eddington closure relates radiation pressure to temperature, $P_{\rm rad}=\tfrac{1}{3}\tfrac{4\sigma}{c}T^4=\tfrac{a}{3}T^4$ . Why: the moment hierarchy of transfer never closes on its own; the Eddington approximation (isotropic-enough radiation, so $P_{\rm rad}=\tfrac{1}{3}u$ ) supplies the missing relation and makes the system solvable.

Step 3. Substitute and integrate in $\tau$ : $\tfrac{4\sigma}{3c}T^4=\tfrac{F}{c}\tau+C$ . The integration constant is set by the surface condition on the emergent radiation, which in the Eddington limit gives $C=\tfrac{2}{3}F/c$ (the Hopf constant $2/3$ ). Why: the surface is where half the sky is dark space, and enforcing that half-open geometry is exactly what pins the constant — the boundary condition on the atmosphere feeds the boundary condition on the star.

Step 4. Collect terms: $T^4(\tau)=\tfrac{3}{4}T_{\rm eff}^4\left(\tau+\tfrac{2}{3}\right)$ . Setting $T=T_{\rm eff}$ gives $1=\tfrac{3}{4}(\tau+2/3)$ , i.e. $\tau=2/3$ ; setting $\tau=0$ gives the true surface temperature $T(0)=(1/2)^{1/4}T_{\rm eff}\approx 0.841\,T_{\rm eff}$ . Why: this is the payoff — the photosphere is not put in by hand at $\tau=2/3$ , it emerges as the depth where the gray atmosphere reaches the effective temperature. This is the standard gray-atmosphere result (Carroll & Ostlie Ch. 9). The Hopf constant $2/3$ is exact only in the Eddington approximation; the true value from the Hopf function is $0.7104$ , a few-percent correction detailed atmospheres carry.

Go deeper · Derivation C · 1 minPhotospheric pressure from hydrostatic balance in optical depth

Derivation CPhotospheric pressure from hydrostatic balance in optical depth

The goal is to derive the fourth boundary condition, the photospheric pressure $P_{\rm phot}\approx\tfrac{2}{3}g/\kappa$ , and show why it is the natural partner of the temperature condition. The trick is to change the integration variable in the hydrostatic equation from radius to optical depth, whereupon the density cancels and the equation integrates almost trivially across the thin, high-gravity surface layers, tying the outer pressure directly to surface gravity and atmospheric opacity.

Step 1. Write hydrostatic balance in the plane-parallel surface layers as $\mathrm{d}P/\mathrm{d}r=-\rho g$ , with $g=GM/R^2$ effectively constant because the atmosphere is a razor-thin skin compared with the stellar radius. Why: near the surface the enclosed mass is essentially the whole mass $M$ and the radius is essentially $R$ , so gravity stops varying and the geometry flattens.

Step 2. Change variable using the optical-depth definition $\mathrm{d}\tau=-\kappa\rho\,\mathrm{d}r$ . Dividing the two relations, the density cancels: $\dfrac{\mathrm{d}P}{\mathrm{d}\tau}=\dfrac{g}{\kappa}$ . Why: this is the key simplification — the unknown run of density with depth drops out entirely, and pressure becomes a simple function of optical depth controlled only by gravity and opacity.

Step 3. Integrate from the top ( $\tau=0$ , $P\approx 0$ ) to $\tau=2/3$ , treating $g$ and $\kappa$ as roughly constant across the thin layer: $P_{\rm phot}=\displaystyle\int_0^{2/3}\frac{g}{\kappa}\,\mathrm{d}\tau\approx\frac{2}{3}\frac{g}{\kappa}$ . Why: this delivers a finite, physically motivated surface pressure that replaces $P(R)=0$ and, paired with $T=T_{\rm eff}$ , completes the outer boundary conditions. The result is standard (Kippenhahn, Weigert & Weiss Ch. 11). For the Sun, $g=274\ \mathrm{m/s^2}$ and a photospheric $\kappa\approx 0.02\ \mathrm{m^2/kg}$ give $P_{\rm phot}\approx 9\times 10^3\ \mathrm{Pa}$ , within a factor of the observed $\sim 10^4\ \mathrm{Pa}$ — the discrepancy is the crudeness hiding in a single constant $\kappa$ .

Limiting cases

The surface treatment can be posed at increasing levels of fidelity; the center barely notices the difference, but the radius and colors depend on it critically.

Case	Result	Meaning
Zero (simple) surface conditions	$P(R)=0,\ T(R)=0,\ \rho(R)=0$	Analytically convenient (polytropes, Lane–Emden); fine for the deep interior, but misplaces the radius and the near-surface temperature
Eddington gray photosphere	$T^4(\tau)=\tfrac{3}{4}T_{\rm eff}^4\!\left(\tau+\tfrac{2}{3}\right)$	At $\tau=2/3$ gives $T=T_{\rm eff}$ ; a clean analytic surface temperature condition
Photospheric pressure condition	$P_{\rm phot}=\tfrac{2}{3}\,g/\kappa$	Hydrostatic balance integrated to $\tau=2/3$ ; ties surface $P$ to gravity and opacity
Full model atmosphere	$\{T,P\}\big\vert_{\rm fit}$ from a non-gray integration	Non-gray, line-blanketed, partly convective atmosphere matched at $\tau_{\rm fit}\sim 10$ – $100$ ; most accurate
Center (always)	$r\to 0,\ L\to 0$	Regularity; $P_c,\ T_c$ remain free eigenvalues fixed by the global fit

Common trapThe surface is a hard edge where everything drops to zero

The naive conditions $T=P=\rho=0$ at $r=R$ are seductive because they are simple and make polytropes solvable — but they describe a star with a sharp vacuum boundary, which no star has. A real star fades into a photosphere at $\tau\approx 2/3$ where the temperature is a finite $T_{\rm eff}$ and the pressure a finite $\tfrac{2}{3}g/\kappa$ ; setting them to zero corresponds to $\tau=0$ , an unphysical transparent edge above the layers that actually radiate. The damage is specific: the zero condition misplaces the stellar radius and distorts the temperature stratification of the outer envelope — precisely where convection sets in and the emergent spectrum forms. For the deep interior the error is negligible (surface pressure is $\sim 10^{-12}$ of central), so the sin is venial for the core and mortal for the atmosphere. Fit at $\tau\approx 2/3$ , not $\tau=0$ .

Worked examples

1 · The effective temperature of the Sun from the surface flux condition (fully worked)

The Sun radiates $L_\odot=3.828\times 10^{26}\ \mathrm{W}$ from a photospheric radius $R_\odot=6.957\times 10^8\ \mathrm{m}$ . Use the photospheric boundary condition $L=4\pi R^2\sigma T_{\rm eff}^4$ to find $T_{\rm eff}$ , and then locate the photosphere and the true surface temperature in the Eddington law.

Step 1. Solve the surface luminosity condition for the effective temperature: $T_{\rm eff}=\left(L/4\pi R^2\sigma\right)^{1/4}$ . The emitting area is $4\pi R_\odot^2=6.08\times10^{18}\ \mathrm{m^2}$ , and with $\sigma=5.670\times10^{-8}\ \mathrm{W\,m^{-2}K^{-4}}$ the denominator is $4\pi R^2\sigma=3.45\times10^{11}\ \mathrm{W\,K^{-4}}$ .

Step 2. Divide and take the fourth root: $T_{\rm eff}^4=3.828\times10^{26}/3.45\times10^{11}=1.11\times10^{15}\ \mathrm{K^4}$ , so $T_{\rm eff}=(1.11\times10^{15})^{1/4}=\mathbf{5772\ K}$ . This is the number the boundary condition sets at the photosphere — not a free parameter but the temperature the surface flux demands.

Step 3. Place the photosphere. In the Eddington law $T^4=\tfrac{3}{4}T_{\rm eff}^4(\tau+2/3)$ , setting $T=T_{\rm eff}$ gives $\tau=2/3$ exactly. The true top of the atmosphere ( $\tau=0$ ) is cooler: $T(0)=(1/2)^{1/4}T_{\rm eff}=0.841\times 5772=\mathbf{4854\ K}$ — so the gas the crude condition would set to $T=0$ is in fact at nearly 4900 K. T_eff = 5772 K at τ = 2/3, with the τ = 0 surface at 4854 K — never zero.

2 · The photospheric pressure of the Sun (fully worked)

Estimate the pressure at the base of the solar photosphere using $P_{\rm phot}\approx\tfrac{2}{3}g/\kappa$ , taking a Rosseland mean opacity of $\kappa\approx 0.02\ \mathrm{m^2/kg}$ for the H⁻-dominated surface layers.

Step 1. Surface gravity: $g=GM_\odot/R_\odot^2=(6.674\times10^{-11})(1.989\times10^{30})/(6.957\times10^8)^2$ . The numerator is $1.327\times10^{20}$ and the denominator is $4.84\times10^{17}$ , so $g=\mathbf{274\ m/s^2}$ — about 28 times Earth's gravity.

Step 2. Photospheric pressure: $P_{\rm phot}=\tfrac{2}{3}\,g/\kappa=\tfrac{2}{3}\times 274/0.02=\mathbf{9.1\times10^{3}\ Pa}$ , roughly one-tenth of an Earth atmosphere. The observed value is $\sim 10^4\ \mathrm{Pa}$ , so the estimate lands within tens of percent.

Step 3. Read the sensitivity: every uncertainty rides in the single opacity $\kappa$ , so halving it doubles $P_{\rm phot}$ — the whole reason detailed models replace this one-line condition with a tabulated non-gray atmosphere. Note too how far this is from the center: with $P_c\sim 2\times10^{16}\ \mathrm{Pa}$ , the ratio $P_{\rm phot}/P_c\sim 5\times10^{-13}$ . P_phot ≈ 9×10³ Pa, about 10⁻¹² of the central pressure — negligible for the core, decisive for the radius.

Physics of the portrait

The boundary conditions that close a stellar model are, more than anything, the legacy of one person's decision to treat a star's interior as a problem in radiation and its surface as a problem in atmospheres — and to insist the two match.

Origins · 1 minArthur Eddington and how this was discovered

Arthur Eddington

1882–1944 · the gray atmosphere and the internal constitution of the stars

Arthur Stanley Eddington turned stellar structure from speculation into physics. His 1926 masterwork The Internal Constitution of the Stars established that a star is a self-regulating ball of gas in which radiation pressure and radiative transport — not just gas pressure — govern the balance, and that the run of temperature is set by how photons diffuse outward against opacity. Two of his ideas sit at the heart of this article. First, the Eddington approximation for a gray atmosphere, which closes the transfer moment equations and yields $T^4=\tfrac{3}{4}T_{\rm eff}^4(\tau+2/3)$ — the relation that lets us place the outer boundary at the photosphere $\tau=2/3$ where $T=T_{\rm eff}$ , rather than at an unphysical zero-temperature edge. Second, his mass–luminosity relation showed that once the interior and surface physics are fixed, a star's luminosity follows from its mass — an early glimpse of the uniqueness later formalized as the Vogt–Russell theorem. Eddington computed by hand what would later take supercomputers, and he did it by taking the boundary conditions seriously: regularity at the center, a real radiating atmosphere at the top. He was also the observer who, in 1919, tested general relativity at a solar eclipse.

History

1870

— Lane treats the Sun as a gaseous sphere in hydrostatic equilibrium, opening the theory of stellar interiors

1906

— Schwarzschild formulates radiative equilibrium in a stellar atmosphere and the criterion for convection

1926

— Eddington publishes The Internal Constitution of the Stars with the gray-atmosphere approximation

1926

— Vogt and Russell argue that mass and composition fix a unique equilibrium structure

1959

— Henyey and collaborators introduce the relaxation scheme for the stellar-structure boundary value problem

Going further

Texts. Carroll & Ostlie, An Introduction to Modern Astrophysics, 2nd ed., Ch. 9–11 (atmospheres, structure equations, boundary conditions) · Kippenhahn, Weigert & Weiss, Stellar Structure and Evolution, 2nd ed., Ch. 10–11 · Hansen, Kawaler & Trimble, Stellar Interiors, 2nd ed. · Prialnik, An Introduction to the Theory of Stellar Structure and Evolution, 2nd ed. · Mihalas, Stellar Atmospheres, for the non-gray surface treatment.

Provenance. The structure equations and central/surface conditions follow Carroll & Ostlie Ch. 10–11 and Kippenhahn, Weigert & Weiss Ch. 10–11. A verifier should confirm: (1) the central regularity conditions $r=L=0$ and the leading expansions $L\propto m$ , $P_c-P\propto m^{2/3}$ ; (2) the Eddington law $T^4=\tfrac{3}{4}T_{\rm eff}^4(\tau+2/3)$ giving $T=T_{\rm eff}$ at $\tau=2/3$ and $T(0)=0.841\,T_{\rm eff}$ ; (3) $P_{\rm phot}\approx\tfrac{2}{3}g/\kappa$ from integrating hydrostatic balance in $\tau$ ; (4) the solar checks $g_\odot=274\ \mathrm{m/s^2}$ , $T_{\rm eff,\odot}=5772\ \mathrm{K}$ from $L=4\pi R^2\sigma T_{\rm eff}^4$ , and $P_{\rm phot}\sim 10^4\ \mathrm{Pa}$ . The Hopf constant $2/3$ is exact only in the Eddington approximation (true value $0.7104$ ). No fabricated numerical values are used.

Prerequisites. Equations of Stellar Structure · Hydrostatic Equilibrium in Stars · Mass Continuity Equation · Opacity and Radiative Transfer in Stars · Equation of State for Stellar Matter. Leads to. Radiative Temperature Gradient · Convective Energy Transport · Mixing-Length Theory.

Ask the tutor: Ask me to walk through a shooting-method iteration by hand, or to show exactly how the Eddington 2/3 turns into a photospheric temperature for a star of your choice.

Why exactly two conditions at the center and two at the surface?

Because the system is four first-order ODEs, and a well-posed problem needs exactly four boundary conditions — no more, no fewer. The physics decides where they go. The center offers two: regularity forbids a nonzero radius or interior luminosity at a point, so $r=0$ and $L=0$ . The surface offers the other two: the star must radiate a definite flux ( $T=T_{\rm eff}$ , i.e. $L=4\pi R^2\sigma T_{\rm eff}^4$ ) and support a definite atmospheric pressure ( $P\approx\tfrac{2}{3}g/\kappa$ ). The central pressure and temperature are left unspecified — they are the unknowns the global solution solves for, which is why the problem is genuinely two-sided.

What actually goes wrong if I just set P and T to zero at the surface?

For the deep interior, essentially nothing: the true surface pressure is about $10^{-12}$ of the central pressure, so zeroing it barely perturbs the core. The trouble is in the outermost layers. The zero condition corresponds to $\tau=0$ , the top of the atmosphere, whereas the light you see comes from $\tau\approx 2/3$ , where the gas is at $T_{\rm eff}$ and a finite pressure. Anchoring at $\tau=0$ therefore misplaces the radius and distorts the temperature run of the envelope — exactly the region that governs convection and the emergent spectrum. So the zero condition is fine for a polytrope or a first estimate, and wrong for predicting a star's size or color.

Why is the photosphere at τ = 2/3 rather than τ = 1?

The factor $2/3$ falls out of the gray atmosphere, not out of a hand-waving "one mean free path." Solving the transfer moments with the Eddington closure gives $T^4=\tfrac{3}{4}T_{\rm eff}^4(\tau+2/3)$ , and asking where the local temperature equals $T_{\rm eff}$ returns $\tau=2/3$ precisely. Physically, the emergent flux is a weighted average of the source function over a range of depths, corresponding to looking down to about $\tau=2/3$ . A more exact transfer calculation (the Hopf function) shifts the constant to $0.7104$ , a few-percent correction. So $2/3$ is the clean analytic answer; $\tau=1$ is only a rough mnemonic for "where the star becomes opaque."

What is the difference between the shooting method and Henyey relaxation?

Both solve the same two-point BVP from opposite philosophies. Shooting treats it as a root-find over the unknown boundary data: guess the missing values (central $P_c,T_c$ ; surface $R,L$ ), integrate from both ends toward a fitting mass, and adjust by Newton–Raphson until the two integrations agree. It is simple but can be numerically stiff. Henyey relaxation instead discretizes all four equations on the entire mass grid at once, appends the boundary conditions as extra rows, linearizes about a trial model, and solves the block-tridiagonal system globally, iterating to convergence. Relaxation is more robust and is the standard for evolution, because it handles the whole star at once and marches stably over billions of years of timesteps.

How does the Vogt–Russell theorem connect to boundary conditions?

The Vogt–Russell theorem states that, given the constitutive physics (equation of state, opacity, energy generation as functions of state), the total mass and the chemical composition profile determine a unique equilibrium structure. The boundary conditions are the mechanism by which that uniqueness is enforced: they are what remove the residual freedom in the ODE solutions. Two central and two surface conditions pin down the otherwise free constants, so for a given $M$ and composition there is one self-consistent set of $\{P_c,T_c,R,L\}$ that satisfies them all. The theorem is not a rigorous mathematical proof — counterexamples exist for certain evolved configurations — but as a working principle it is why a single model emerges from the calculation rather than a family.

Does the central pressure really come out of nowhere, or is it constrained?

It is fully constrained, just not locally at the center. Unlike $r$ and $L$ , which regularity fixes at $m=0$ , the central pressure and temperature are global quantities: their correct values are the ones that make the outward integration satisfy the surface conditions, so you can think of them as eigenvalues. Guess them wrong and the model misses the photosphere; the iteration adjusts them until it lands. Order-of-magnitude they follow the virial-like scaling $P_c\sim GM^2/R^4$ (about $2\times10^{16}\ \mathrm{Pa}$ for the Sun), but the precise value comes only from solving the full BVP — the deep reason the problem is two-point and not an initial-value march.