The Inflaton and Slow-Roll Dynamics

Standard hot Big Bang cosmology, built on the FLRW metric and the Friedmann equations, is a triumph — yet it hands us its most striking features as unexplained initial conditions. Why is the cosmic microwave background the same temperature in patches never in causal contact (the horizon problem)? Why is space so nearly flat that the early density had to match the critical value to one part in $10^{60}$ (the flatness problem)? Where are the magnetic monopoles grand unified theories insist were made in profusion? Inflation answers all three at once with a single stroke: an early epoch of accelerated expansion, $\ddot a>0$ . By the acceleration equation this demands a source with $\rho+3p<0$ , a pressure more negative than $-\rho/3$ — nothing ordinary will do. The simplest engine that delivers it is a scalar field, the inflaton $\phi$ , minimally coupled to gravity through a potential $V(\phi)$ . Displace the field high onto a flat stretch and let it roll: if the potential is gentle enough, cosmic expansion drags on the field like thick friction, kinetic energy stays tiny, and the potential energy sits almost frozen — a temporary cosmological constant that inflates space quasi-exponentially. The whole story is governed by two dimensionless slow-roll parameters, $\epsilon$ and $\eta$ , built from the slope and curvature of $V$ . While both stay far below one, inflation runs; when the field reaches the steep part and $\epsilon$ climbs to unity, slow-roll fails, the field oscillates, and its energy reheats the universe into the hot Big Bang. The same $\epsilon$ and $\eta$ fix the tilt of the primordial spectrum, turning the shape of $V(\phi)$ into numbers a satellite can measure.

Key takeaways

The inflaton is a scalar field $\phi$ whose potential energy $V(\phi)$ dominates the cosmic budget, acting like a slowly draining cosmological constant that drives quasi-exponential expansion $a\propto e^{Ht}$ .
Its motion obeys a Klein–Gordon equation with Hubble friction, $\ddot\phi+3H\dot\phi+V'(\phi)=0$ , coupled to the Friedmann equations for $H$ .
Slow-roll holds when kinetic energy is negligible beside $V$ , quantified by small slow-roll parameters $\epsilon=\tfrac12 M_{\rm Pl}^2(V'/V)^2$ and $\eta=M_{\rm Pl}^2 V''/V$ ; then $w\simeq-1$ and the e-fold count $N=\int H\,\mathrm dt$ must reach $50$ – $60$ to solve the horizon and flatness problems.
The potential's shape is observable: $n_s-1=2\eta-6\epsilon$ and $r=16\epsilon$ . Planck finds $n_s\approx0.965$ , $r<0.06$ , ruling out $V=\tfrac12 m^2\phi^2$ ( $r\approx0.13$ ) and favouring flat plateaus.
Inflation ends when $\epsilon\to1$ ; the field oscillates and reheats the cosmos, seeding the radiation era, while its quantum fluctuations become the CMB anisotropies.
Key idea: a scalar field stuck high on a flat potential behaves like a slowly draining cosmological constant — its potential energy dominates, its pressure is nearly $-\rho$ , and spacetime inflates quasi-exponentially until the field rolls off the flat region and slow-roll breaks down.

The core insight

A scalar field stuck high on a flat potential behaves like a slowly draining cosmological constant: its potential energy dominates, the pressure is nearly $-\rho$ , and spacetime inflates quasi-exponentially until the field rolls off the flat region and slow-roll breaks down.

Where this lives today

Inflation is now the working framework of observational cosmology. Planck measured the scalar spectral index to $n_s=0.9649\pm0.0042$ — a significant tilt away from perfect scale-invariance, exactly the small departure slow-roll predicts. The tightest bound on primordial gravitational waves, from the BICEP/Keck experiments combined with Planck, is $r<0.036$ (the round $r<0.06$ is still often quoted), already excluding power-law potentials like $m^2\phi^2$ and $\lambda\phi^4$ and steering theory toward flat plateaus such as the Starobinsky and Higgs-inflation models. The great open target is that tensor signal: a detection of primordial B-mode polarization in the CMB would fix the energy scale through $V^{1/4}\approx1.9\times10^{16}\,(r/0.06)^{1/4}$ GeV and probe physics near the GUT scale. Next-generation experiments — the Simons Observatory, CMB-S4, and LiteBIRD — aim for $r\sim10^{-3}$ . The same slow-roll mathematics is recycled to model dark energy as a slowly rolling "quintessence" field, and reheating remains an active frontier connecting inflation to the origin of matter.

The physical picture

Picture a ball rolling on a very gently sloped hill — but immersed in thick honey. The honey is Hubble friction, the term $3H\dot\phi$ , so viscous that the ball reaches terminal velocity almost at once and only creeps downhill. While it creeps, its height barely changes, so $V(\phi)$ acts as a nearly fixed reservoir of vacuum energy — and that reservoir inflates space. Because $V$ is almost constant, the pressure $p=\tfrac12\dot\phi^2-V$ sits close to $-V$ while the density $\rho=\tfrac12\dot\phi^2+V$ sits close to $+V$ , giving $w\approx-1$ . Pressure gravitates in GR, and sufficiently negative pressure makes gravity repulsive on cosmological scales — the engine of acceleration. The first figure is the landscape itself: the plateau where the field slow-rolls, and the minimum where inflation dies.

Figure

The inflaton rolls slowly down a nearly flat plateau where ε = (M_Pl²/2)(V′/V)² ≪ 1, so H² ≈ V/3M_Pl² stays almost constant and w ≈ −1; when the field reaches the steep wall ε grows to 1, inflation ends, and φ oscillates about the minimum to reheat the universe

Hubble friction is itself proportional to the expansion rate, set by $V$ , so a higher, flatter potential both stores more energy and drags harder on the field — a self-reinforcing arrangement that makes slow-roll an attractor: a wide spread of initial velocities $\dot\phi$ relaxes onto the same trajectory within a fraction of an e-fold, which is why we can predict anything without knowing the field's starting speed. The number of e-folds $N=\ln(a_{\rm end}/a)$ measures how much space stretched, and the shape of $V$ fixes it — and fixes what we see: quantum jitters in $\phi$ , stretched beyond the horizon, freeze into the density ripples imprinted on the CMB, whose tilt away from scale-invariance is set by the slope and curvature of $V$ . The second figure is the observational verdict — where different potentials land in the plane of the two headline numbers.

Figure

Slow-roll predictions in the n_s–r plane against Planck 2018: the plateau model R² lands at n_s ≈ 0.965, r ≈ 0.004 inside the allowed region, while chaotic V = ½m²φ² predicts r ≈ 0.13 for N between 50 and 60 and is excluded by the bound r < 0.06

The core equations

Everything begins with an action: the inflaton is a canonical scalar minimally coupled to Einstein gravity, and both its equation of motion and stress-energy fall out of varying this one functional. The reduced Planck mass $M_{\rm Pl}=(8\pi G)^{-1/2}$ is the natural unit, signature $(-,+,+,+)$ .

S=\int\mathrm d^4x\,\sqrt{-g}\left[\frac{M_{\rm Pl}^2}{2}R-\frac12 g^{\mu\nu}\partial_\mu\phi\,\partial_\nu\phi-V(\phi)\right] \qquad (1)

where: $M_{\rm Pl}$ — reduced Planck mass, $M_{\rm Pl}=(8\pi G)^{-1/2}\approx2.4\times10^{18}\,\mathrm{GeV}$ ; $R$ — Ricci scalar; $\phi$ — the inflaton scalar field; $V(\phi)$ — its potential energy density; $g$ — determinant of $g_{\mu\nu}$ . The first term is Einstein–Hilbert gravity, the last two the field's kinetic and potential energy. Varying $S$ with respect to $\phi$ on a homogeneous FLRW background gives a damped oscillator equation. The damping is cosmological expansion itself — the friction coefficient $3H$ is not a material viscosity.

\ddot\phi+3H\dot\phi+V'(\phi)=0 \qquad (2)

where: $\dot\phi$ — time derivative $\mathrm d\phi/\mathrm dt$ ; $H=\dot a/a$ — Hubble parameter; $3H\dot\phi$ — Hubble friction; $V'(\phi)=\mathrm dV/\mathrm d\phi$ — the driving slope. Varying $S$ with respect to the metric gives the field's stress-energy — for a homogeneous $\phi$ , a perfect fluid whose density and pressure split by the sign of $V$ .

\rho_\phi=\tfrac12\dot\phi^2+V(\phi),\qquad p_\phi=\tfrac12\dot\phi^2-V(\phi) \qquad (3)

where: $\rho_\phi$ — energy density of the inflaton fluid; $p_\phi$ — its pressure; $\tfrac12\dot\phi^2$ — kinetic energy density. The potential adds to $\rho$ but subtracts from $p$ — the origin of negative pressure. Feeding this fluid into the Friedmann equations closes the system. The second relation carries a message: for a real scalar, $\dot H\le0$ always.

H^2=\frac{1}{3M_{\rm Pl}^2}\!\left(\tfrac12\dot\phi^2+V\right),\qquad \dot H=-\frac{\dot\phi^2}{2M_{\rm Pl}^2} \qquad (4)

where: $H^2$ — squared expansion rate, set by the total density; $\dot H$ — its rate of change, non-positive since $\dot\phi^2\ge0$ (the null energy condition holds), so inflation is at best quasi-de Sitter: $H$ only falls. The equation of state follows from (3), interpolating between potential domination ( $w\to-1$ ) and kinetic domination ( $w\to+1$ ).

w_\phi=\frac{p_\phi}{\rho_\phi}=\frac{\tfrac12\dot\phi^2-V}{\tfrac12\dot\phi^2+V} \qquad (5)

where: $w_\phi$ — equation-of-state parameter. Acceleration requires $w_\phi<-\tfrac13$ ( $\dot\phi^2<V$ ); the deep inflationary limit $\dot\phi^2\ll V$ gives $w_\phi\to-1$ . The two slow-roll parameters distil the whole approximation into the shape of the potential alone — pure numbers built from the first and second derivatives of $V$ .

\epsilon\equiv\frac{M_{\rm Pl}^2}{2}\!\left(\frac{V'}{V}\right)^2,\qquad \eta\equiv M_{\rm Pl}^2\,\frac{V''}{V} \qquad (6)

where: $\epsilon$ — steepness parameter; inflation requires $\epsilon<1$ , and $\epsilon=1$ marks its end. $\eta$ — curvature parameter; $|\eta|<1$ is needed for the slow-roll equation of motion. Both are small for many e-folds. Integrating the slow-roll motion converts the expansion into a geometric integral over the field range — no time variable survives, only $V$ 's profile.

N\equiv\ln\frac{a_{\rm end}}{a}=\int_t^{t_{\rm end}}\!H\,\mathrm dt\simeq\frac{1}{M_{\rm Pl}^2}\int_{\phi_{\rm end}}^{\phi}\frac{V}{V'}\,\mathrm d\phi \qquad (7)

where: $N$ — number of e-folds between field value $\phi$ and the end of inflation; $\phi_{\rm end}$ — where $\epsilon=1$ . The horizon and flatness problems need $N\gtrsim50$ – $60$ for observable scales. Finally, the same parameters predict the two headline observables — the spectrum's tilt and its gravitational-wave content — the bridge from a microscopic potential to a satellite measurement.

n_s-1=2\eta-6\epsilon,\qquad r=16\epsilon \qquad (8)

where: $n_s$ — scalar spectral index (perfect scale-invariance is $n_s=1$ ); $r$ — tensor-to-scalar ratio, the amplitude of primordial gravitational waves relative to density perturbations. Both are evaluated at horizon exit of the CMB scales.

Derivation AThe slow-roll equations from the inflaton stress-energy

The goal here is to show, starting only from the acceleration equation and the scalar stress-energy of equation (3), exactly when a rolling field inflates space and what its equation of motion collapses to in that regime. We reach two working results — the acceleration condition $\dot\phi^2<V$ and the overdamped slow-roll pair $3H\dot\phi\simeq-V'$ , $H^2\simeq V/3M_{\rm Pl}^2$ — and see which terms are dropped and why small $\epsilon$ and $\eta$ license dropping them.

Step 1. Write the second Friedmann (acceleration) equation, $\ddot a/a=-\tfrac{1}{6M_{\rm Pl}^2}(\rho+3p)$ . Acceleration $\ddot a>0$ demands $\rho+3p<0$ . Why: gravity in GR responds to $\rho+3p$ , not $\rho$ alone, so pressure can flip the sign of the cosmic force — the whole reason a scalar can push space apart.

Step 2. Insert the inflaton fluid from (3): $\rho_\phi+3p_\phi=2\dot\phi^2-2V$ , so the condition $\rho_\phi+3p_\phi<0$ becomes simply $\dot\phi^2<V$ . Why: it reduces the abstract energy condition to a transparent competition — kinetic versus potential energy — showing inflation is nothing but "potential wins".

Step 3. Equivalently $w_\phi<-\tfrac13$ . In strong slow-roll we impose $\dot\phi^2\ll V$ , forcing $w_\phi\to-1$ and, through (4), $\dot H\to0$ , i.e. nearly constant $H$ . Why: only when the kinetic term is far smaller than $V$ does the expansion become genuinely quasi-exponential rather than barely accelerating.

Step 4. Drop the acceleration term in (2): assuming $|\ddot\phi|\ll|3H\dot\phi|$ turns the second-order equation into a first-order balance, $3H\dot\phi\simeq-V'(\phi)$ . Why: with strong friction the field rides at terminal velocity, slope balancing drag — the honey-on-a-hill limit made quantitative.

Step 5. Drop the kinetic term in (4), leaving $H^2\simeq V(\phi)/3M_{\rm Pl}^2$ . Why: the potential now single-handedly sets the expansion rate, so a flatter, higher plateau means a larger, steadier $H$ .

Step 6. Check self-consistency: the two approximations are controlled by $\epsilon\ll1$ (justifying $\dot\phi^2\ll V$ , since $\dot\phi^2/V=\tfrac23\epsilon$ ) and $|\eta|\ll1$ (justifying $|\ddot\phi|\ll|3H\dot\phi|$ ). Why: slow-roll is a regime with an explicit small parameter, so we always know when it is trustworthy and when it is about to break. *As a consistency check, substituting the reduced slow-roll equations into the kinematic parameter $\epsilon_H\equiv-\dot H/H^2$ reproduces $\epsilon_H\simeq\tfrac12 M_{\rm Pl}^2(V'/V)^2=\epsilon$ , confirming that the potential and Hubble slow-roll parameters agree at leading order.*

Go deeper · Derivation B · 1 minCounting e-folds from the shape of the potential

Derivation BCounting e-folds from the shape of the potential

Here the aim is to turn the definition of e-folds, an integral of $H$ over time, into an integral over the field value alone, so the total expansion becomes a property of the potential's profile and nothing else. The payoff is equation (7) and its rewriting in terms of $\epsilon$ , which together let us read off how far and how flatly a field must roll for the required $50$ – $60$ e-folds, used directly in the worked examples.

Step 1. Start from $N=\int_t^{t_{\rm end}}H\,\mathrm dt$ , the logarithm of how much $a$ grows to the end of inflation. Why: $N=\ln(a_{\rm end}/a)$ is the natural clock — physical scales grow by $e^N$ , so $N$ counts directly against the horizon and flatness requirements.

Step 2. Change variable from $t$ to $\phi$ using $\mathrm dt=\mathrm d\phi/\dot\phi$ , giving $N=\int_\phi^{\phi_{\rm end}}(H/\dot\phi)\,\mathrm d\phi$ . Why: the field rolls monotonically in slow-roll, so $\phi$ is a good clock and the swap avoids solving for $a(t)$ .

Step 3. Substitute $\dot\phi\simeq-V'/3H$ and $3H^2\simeq V/M_{\rm Pl}^2$ : then $H/\dot\phi\simeq-3H^2/V'=-V/(M_{\rm Pl}^2 V')$ . Why: this injects the potential's shape — the ratio $V/V'$ measures how gently $V$ slopes, and gentle slopes buy many e-folds.

Step 4. Integrate, ordering the limits so $N>0$ for a field rolling toward $\phi_{\rm end}$ : $N\simeq\dfrac{1}{M_{\rm Pl}^2}\displaystyle\int_{\phi_{\rm end}}^{\phi}\dfrac{V}{V'}\,\mathrm d\phi$ . Why: this is equation (7) — the total expansion is a definite integral over the field range, computable the instant you name a potential.

Step 5. Rewrite using $\sqrt{2\epsilon}=M_{\rm Pl}|V'/V|$ , giving $N=\displaystyle\int\dfrac{\mathrm d\phi}{M_{\rm Pl}\sqrt{2\epsilon}}$ . Why: the smaller $\epsilon$ is, the more e-folds accumulate per unit field displacement, so flat potentials are e-fold-efficient. *As a check, applying this formula to $V=\tfrac12 m^2\phi^2$ gives $N=(\phi^2-\phi_{\rm end}^2)/4M_{\rm Pl}^2$ , the standard chaotic-inflation result, with $\phi_{\rm end}=\sqrt2\,M_{\rm Pl}$ from $\epsilon=1$ .*

Go deeper · Derivation C · 1 minWhy the equation of state sits just above minus one

Derivation CWhy the equation of state sits just above minus one

The purpose of this short derivation is to make the near-vacuum behaviour of the inflaton quantitative: to prove that the equation of state during slow-roll is not merely "close to $-1$ " but precisely $w=-1+\tfrac23\epsilon_H$ , so a single parameter $\epsilon$ controls both the acceleration and the departure from a true cosmological constant. It ties equation (5), the Friedmann relations (4), and the parameter (6) into one line.

Step 1. From (3), $\rho_\phi+p_\phi=\dot\phi^2$ exactly — all the $V$ terms cancel. Why: for a scalar, $\rho+p$ is pure kinetic energy, the cleanest diagnostic of how far the field is from frozen.

Step 2. Divide by $\rho_\phi$ : $1+w_\phi=(\rho_\phi+p_\phi)/\rho_\phi=\dot\phi^2/\rho_\phi$ . Why: this isolates the deviation of $w$ from $-1$ as the ratio of kinetic to total energy, which slow-roll keeps small.

Step 3. Use $\rho_\phi=3M_{\rm Pl}^2 H^2$ and $\epsilon_H\equiv-\dot H/H^2=\dot\phi^2/(2M_{\rm Pl}^2H^2)$ . Then $\dot\phi^2/\rho_\phi=\dot\phi^2/(3M_{\rm Pl}^2H^2)=\tfrac23\epsilon_H$ . Why: it converts a ratio of energies into the slow-roll parameter, closing the loop between the fluid and geometric pictures.

Step 4. Therefore $w_\phi=-1+\tfrac23\epsilon_H\simeq-1+\tfrac23\epsilon$ . Why: the master relation — as $\epsilon\to0$ the inflaton becomes an exact cosmological constant, and every departure from de Sitter is first-order in $\epsilon$ . *Two limiting cases confirm the result: at $\epsilon=1$ (end of inflation) this gives $w=-1/3$ , exactly the borderline $\ddot a=0$ ; and $\epsilon\to0$ gives $w\to-1$ , the de Sitter limit — both consistent with equation (5) evaluated directly.*

Limiting cases

The inflaton's behaviour is bracketed by four regimes, from frozen field to fast roll:

Case	Result	Meaning
Exact de Sitter ( $\dot\phi\to0$ )	$\rho=-p=V_0,\ H=\mathrm{const},\ a\propto e^{Ht}$	Frozen field: a perfect cosmological constant, $w=-1$ , pure exponential expansion. Inflation never ends in this idealized limit.
Slow-roll ( $\epsilon,\|\eta\|\ll1$ )	$3H\dot\phi\simeq-V',\ H^2\simeq V/3M_{\rm Pl}^2$	Quasi-de Sitter: $H$ drifts slowly, $w\approx-1+\tfrac23\epsilon$ , e-folds accumulate. The physically relevant inflationary regime.
Kinetic domination ( $\dot\phi^2\gg V$ )	$w\to+1,\ \rho\propto a^{-6},\ a\propto t^{1/3}$	Stiff fluid: kinetic energy dilutes faster than anything else. No acceleration; slow-roll badly violated.
End of inflation ( $\epsilon\to1$ )	$\ddot a=0,\ \tfrac12\dot\phi^2\sim V$	Acceleration ceases; the field rolls fast, oscillates about the minimum, and reheating begins.

Common trapThe slow-roll parameters describe the potential, not a single instant of motion

A common confusion is to treat $\epsilon$ and $\eta$ as snapshots of the field's velocity and acceleration. They are not: the potential slow-roll parameters in equation (6) are built purely from the shape of $V(\phi)$ — its slope $V'/V$ and curvature $V''/V$ — and are defined even for a field momentarily at rest. A second, genuinely kinematic family, the Hubble parameters $\epsilon_H=-\dot H/H^2$ and $\eta_H$ , is defined from the expansion history; the two coincide only to leading order, and quoting one where the other is meant is a real error in precision work. A related trap: slow-roll is an approximation, obtained by dropping $\ddot\phi$ from equation (2), justified only while $|\eta|\ll1$ . Near $\epsilon=1$ it fails and you must integrate the full second-order equation — the formulae $n_s-1=2\eta-6\epsilon$ and $r=16\epsilon$ are leading-order results at horizon exit, not exact identities.

Worked examples

1 · Chaotic inflation with V = ½m²φ² (fully worked)

The simplest large-field model takes a free massive scalar, $V=\tfrac12 m^2\phi^2$ . Find the field value at which inflation ends, the field value $60$ e-folds earlier, the slow-roll parameters and observables there, and the inflaton mass fixed by the measured perturbation amplitude. Then say why the model is ruled out.

Step 1. Slow-roll parameters. With $V'=m^2\phi$ and $V''=m^2$ , equation (6) gives $\epsilon=\tfrac{M_{\rm Pl}^2}{2}(V'/V)^2=\tfrac{M_{\rm Pl}^2}{2}(2/\phi)^2=2M_{\rm Pl}^2/\phi^2$ and $\eta=M_{\rm Pl}^2 V''/V=2M_{\rm Pl}^2/\phi^2$ . For this potential $\epsilon=\eta=2M_{\rm Pl}^2/\phi^2$ — a helpful coincidence.

Step 2. End of inflation. Setting $\epsilon=1$ gives $\phi_{\rm end}^2=2M_{\rm Pl}^2$ , so $\phi_{\rm end}=\sqrt2\,M_{\rm Pl}\approx1.41\,M_{\rm Pl}$ . The field must sit far above this while inflating.

Step 3. Field value at $N=60$ . The e-fold integral (derivation 2) gives $N=(\phi^2-\phi_{\rm end}^2)/4M_{\rm Pl}^2$ , so $\phi^2=(4N+2)M_{\rm Pl}^2=242\,M_{\rm Pl}^2$ and $\phi\approx15.6\,M_{\rm Pl}$ — a super-Planckian field range, the hallmark of large-field models.

Step 4. Observables. There $\epsilon=\eta=2/242=8.26\times10^{-3}$ . Equation (8) gives $n_s=1-4\epsilon=\mathbf{0.967}$ and $r=16\epsilon=\mathbf{0.132}$ . The tilt is fine, but the tensor ratio is huge.

Step 5. Fix the mass. The measured amplitude $A_s=V/(24\pi^2 M_{\rm Pl}^4\,\epsilon)\approx2.1\times10^{-9}$ pins the one free parameter: $m\approx5.8\times10^{-6}\,M_{\rm Pl}\approx\mathbf{1.4\times10^{13}\ GeV}$ , with energy scale $V^{1/4}\approx\mathbf{1.9\times10^{16}\ GeV}$ and Hubble rate $H\approx9\times10^{13}$ GeV — all near the grand-unification scale.

Step 6. The verdict. Planck combined with BICEP/Keck requires $r<0.06$ , but this model predicts $r\approx0.13$ across $N=50$ – $60$ . $V=\tfrac12 m^2\phi^2$ is excluded — a genuinely falsified inflationary model, and a triumph of the slow-roll formalism turning a shape into a killable number.

2 · A plateau model that survives (Starobinsky R²)

Starobinsky's $R^2$ gravity is equivalent to a scalar on a flat plateau, $V=V_0(1-e^{-\sqrt{2/3}\,\phi/M_{\rm Pl}})^2$ , with large- $N$ predictions $n_s\simeq1-2/N$ and $r\simeq12/N^2$ . Evaluate at $N=55$ and compare with data.

Step 1. Spectral index: $n_s=1-2/55=\mathbf{0.964}$ , squarely inside the Planck band $n_s=0.965\pm0.004$ .

Step 2. Tensor ratio: $r=12/55^2=\mathbf{3.97\times10^{-3}}$ , more than an order of magnitude below the $r<0.06$ bound. This model is favoured — the flat plateau ( $\epsilon\sim1/N^2$ , far below the $1/N$ of $m^2\phi^2$ ) suppresses gravitational waves.

Step 3. Energy scale. Since $V^{1/4}=(\tfrac32\pi^2 A_s r)^{1/4}M_{\rm Pl}$ , the tiny $r\approx0.004$ gives $V^{1/4}\approx8\times10^{15}$ GeV — the same neighbourhood as example 1 but with a hundredfold weaker gravitational-wave signal. The lesson: $n_s$ and $r$ together discriminate potential shapes, and data point to plateaus, not power laws.

Physics of the portrait

Inflation is one of the rare cosmological ideas with a nearly datable birth — it grew from an attempt to solve a particle-physics embarrassment, the overproduction of magnetic monopoles, and became the framework explaining why the universe looks as it does on the largest scales.

Origins · 1 minAlan Guth and how this was discovered

Alan Guth

born 1947 · the inflationary universe and slow-roll cosmology

Alan Guth was a young particle theorist, not a cosmologist, when in December 1979 he had what his notebook calls a "spectacular realization": a brief period of exponential expansion driven by a supercooled false vacuum would flatten space, dilute monopoles to invisibility, and stretch a tiny causal patch across the whole sky. His 1981 paper The Inflationary Universe named the idea and framed the three problems it solved — horizon, flatness, and monopoles. Guth's original "old inflation," though, was stuck: it trapped the field in a false vacuum that escaped only by bubble nucleation, and the bubbles never merged into a smooth universe (the graceful-exit problem). Within a year Andrei Linde, and independently Andreas Albrecht with Paul Steinhardt, rescued it by letting the field roll slowly down a gentle potential rather than tunnel — the slow-roll paradigm this article develops, a scalar creeping across a flat plateau as a decaying cosmological constant. Linde pushed it further with chaotic inflation in 1983, showing even the humble $V=\tfrac12 m^2\phi^2$ inflates if the field starts high enough. The deepest payoff arrived almost as a bonus: the inflaton's quantum fluctuations, worked out at the 1982 Nuffield workshop, freeze into a nearly scale-invariant spectrum — the seeds of every galaxy, later measured by COBE and Planck. Guth, an American physicist at MIT, thus turned a monopole puzzle into a predictive theory of cosmic structure.

History

1980

— Starobinsky proposes an R² gravity model with an early de Sitter phase

1981

— Guth introduces inflation to solve the horizon, flatness, and monopole problems

1982

— Linde, and Albrecht with Steinhardt, replace it with slow-roll new inflation

1983

— Linde proposes chaotic inflation driven by a simple V = ½m²φ²

1992

— COBE detects the CMB anisotropies that inflation had predicted

Going further

Texts. Misner, Thorne & Wheeler, Gravitation (1973) · Wald, General Relativity (1984), Ch. 5 · Carroll, Spacetime and Geometry (2004), Ch. 8 · Hartle, Gravity (2003), Ch. 18–19 · Schutz, A First Course in General Relativity, 2nd ed. (2009) · Mukhanov, Physical Foundations of Cosmology (2005) · Baumann, TASI Lectures on Inflation (arXiv:0907.5424) · Liddle & Lyth, Cosmological Inflation (2000). Key relations used throughout: $\rho_\phi+3p_\phi=2(\dot\phi^2-V)$ ; under slow-roll $\epsilon_H=-\dot H/H^2=\tfrac12 M_{\rm Pl}^2(V'/V)^2$ ; the e-fold integral for $V=\tfrac12 m^2\phi^2$ gives $N=(\phi^2-\phi_{\rm end}^2)/4M_{\rm Pl}^2$ ; the full $m^2\phi^2$ chain at $N=60$ yields $\phi\approx15.6\,M_{\rm Pl}$ , $\epsilon=8.26\times10^{-3}$ , $n_s=0.967$ , $r=0.132$ , $m\approx1.4\times10^{13}$ GeV, $V^{1/4}\approx1.9\times10^{16}$ GeV; and the Starobinsky predictions are $n_s=1-2/N$ , $r=12/N^2$ . Convention throughout: $M_{\rm Pl}^2=(8\pi G)^{-1}$ , signature $(-,+,+,+)$ ; observational inputs are Planck 2018 ( $n_s=0.9649\pm0.0042$ , $A_s\approx2.1\times10^{-9}$ ) with $r<0.06$ from Planck+BICEP/Keck.

Prerequisites. The FLRW Metric · The Friedmann Equations · The Scale Factor and Hubble Parameter · The Cosmological Fluid and Equation of State · The Cosmological Constant and Dark Energy. Leads to. The Radiation-Dominated Era · The Dark-Energy-Dominated Era · Critical Density and the Density Parameters.

Ask the tutor: Give me a potential $V(\phi)$ and I will walk you from its slope and curvature all the way to the $n_s$ and $r$ a CMB satellite would measure — and tell you whether the data have already ruled it out.

Why does negative pressure cause the universe to accelerate rather than collapse?

Because in general relativity it is the combination $\rho+3p$ , not the density alone, that sources cosmic deceleration. The acceleration equation reads $\ddot a/a=-\tfrac{1}{6M_{\rm Pl}^2}(\rho+3p)$ , so if $p<-\rho/3$ the right-hand side flips sign and $\ddot a>0$ . An inflaton with $\dot\phi^2\ll V$ has $p\approx-V\approx-\rho$ , comfortably satisfying this. Pressure gravitates in GR, so a large negative pressure carries a large repulsive gravitational effect — not a mechanical push on a boundary, but a change in how spacetime curves. Ordinary matter and radiation ( $p\ge0$ ) always decelerate the expansion; only sufficiently negative pressure can accelerate it.

What exactly is an e-fold and why do we need about sixty of them?

An e-fold is one factor of $e$ of growth in the scale factor: $N=\ln(a_{\rm end}/a)$ , so $60$ e-folds means space stretched by $e^{60}\approx10^{26}$ . The number is set by the horizon and flatness problems. To explain why widely separated CMB patches share a temperature, the whole observable universe must have fit inside a single causal region before inflation, and undoing the subsequent expansion takes roughly $50$ – $60$ e-folds. Flatness needs a similar amount: each e-fold drives the curvature term $k/a^2$ down relative to the inflaton energy, pushing $\Omega\to1$ . The precise figure depends on the reheating temperature, hence " $50$ to $60$ ."

How can quantum fluctuations of a single field seed galaxies?

During inflation the field $\phi$ carries quantum zero-point fluctuations on every scale. As space expands, a given wavelength is stretched until it exceeds the Hubble radius, at which point the fluctuation can no longer evolve and "freezes," leaving a small, nearly constant curvature perturbation. Because $H$ is almost the same when successive scales exit the horizon, the spectrum is almost scale-invariant, with a slight tilt set by how $H$ and $\epsilon$ drift — precisely the $n_s\approx0.965$ Planck measures. After inflation these frozen perturbations re-enter the horizon and seed the matter clumping that becomes the cosmic web. The same mechanism produces tensor modes whose amplitude relative to the scalars is $r=16\epsilon$ .

What is the difference between the potential and Hubble slow-roll parameters?

They answer the same question — "how close to de Sitter are we?" — from different data. The potential parameters $\epsilon=\tfrac12 M_{\rm Pl}^2(V'/V)^2$ and $\eta=M_{\rm Pl}^2 V''/V$ come from the shape of $V(\phi)$ , convenient because most models are specified by a potential. The Hubble parameters $\epsilon_H=-\dot H/H^2$ and $\eta_H=-\ddot\phi/(H\dot\phi)$ come from the actual expansion history and are the "true" measures of departure from exponential. In slow-roll they agree at leading order, $\epsilon_H\simeq\epsilon$ , so the distinction is often glossed over — but they differ at next order, and the exact acceleration condition is $\epsilon_H<1$ , not $\epsilon<1$ . For percent-level CMB predictions you track which is which.

Does inflation happen at a definite energy, and can we measure it?

Yes in principle, through the tensor-to-scalar ratio. The amplitude of primordial gravitational waves is fixed by the Hubble rate during inflation, hence by the energy density: $V^{1/4}\approx1.9\times10^{16}\,(r/0.06)^{1/4}$ GeV, so measuring $r$ measures the energy scale directly. For $r$ near the current bound this is about $10^{16}$ GeV — the grand-unification scale, far beyond any accelerator, making inflation a rare window onto ultra-high-energy physics. If $r$ is much smaller (many plateau models predict $r\sim10^{-3}$ or below), inflation happened at a lower scale. A confirmed detection of primordial B-mode polarization would be the measurement; none exists yet, only upper limits.

Why does inflation end, and what happens next?

Inflation ends because no realistic potential is perfectly flat forever. As the field rolls, it reaches a region where $V$ steepens, $\epsilon$ grows, and when $\epsilon=1$ acceleration stops ( $\ddot a=0$ ). By then the kinetic energy has caught up with the potential, slow-roll is over, and the field falls toward the minimum, overshooting and oscillating about it. Those oscillations are the handoff: coupled to other fields, the oscillating inflaton decays, transferring its energy into a hot bath of particles. This is reheating, and it sets the initial temperature of the radiation era — the "hot" in hot Big Bang. Everything afterward (nucleosynthesis, recombination, structure formation) proceeds from the state reheating leaves behind, which is why inflation provides the initial conditions for the conventional universe.