The most tightly bound debris returns first because its ellipse has the shortest period. A known mass distribution in energy can therefore be converted into mass per unit return time. This conversion explains the familiar late-time decline and also shows why a real curve need not begin with that decline.
A bound fragment brings its own clock
Prerequisites are the energy assignment from the previous chapter, Kepler’s third law and the meaning of an integral as a sum of small contributions. Keep the black hole fixed at the origin. Each fragment has position r, velocity v and a saved mass weight. Specific orbital energy ε is measured in joules per kilogram; time is measured from the release instant unless another origin is stated.
For a bound Kepler ellipse of semimajor axis a, energy and period satisfy
ε=−2aGM∙,P=2πGM∙a3.
Eliminate a. The return-period scale is
P(ε)=(2∣ε∣)3/22πGM∙.
As the energy approaches zero from below, the period becomes arbitrarily long. The sign of energy decides whether a return exists; its magnitude sets the period. Angular momentum sets the shape and pericentre of the ellipse, but at fixed energy it does not change the Kepler period.
The period starts and ends at the same orbital phase. A fragment released on the incoming branch first makes a short trip to pericentre, then completes an orbit before its next pericentre passage. A saved-state replay can add that initial travel time explicitly. The following analytic derivation approximates disruption as occurring near pericentre and uses t≃P for the long return time. An exact saved-state calculation retains the release phase.
Change energy intervals into time intervals
Let tmin=P(−Δε) be the period of the most bound edge in the linear model. Write s=t/tmin, so that bound debris returning at time t has energy magnitude ∣ε∣=Δεs−2/3. The returning mass in a small time interval is the mass in the corresponding small energy interval. Taking the absolute value of the conversion gives
M˙(t)=dεdMε(t)dtdε.
The derivative contributes a factor proportional to t−5/3. If the energy distribution is flat across the full interval, its value is M∗/(2Δε). Therefore
M˙flat(t)=3tminM∗s−5/3,s≥1.
The same asymptotic slope follows for a smooth distribution with a finite, nonzero value at zero energy. It is the combination of a Kepler period and a locally regular energy distribution, not an extra force that acts on the gas. A surviving core or a different distribution can change the result.
The beginning remembers the star
Insert the uniform-sphere energy distribution derived in the previous chapter. The fallback curve becomes
M˙uniform(t)=2tminM∗(s−5/3−s−3),s≥1.
At the first return the infinitesimal stellar edge has zero mass per unit energy, so this curve starts from zero, rises and then declines. Setting its derivative to zero gives a peak at s=(9/5)3/4≃1.554. At late times, the second term fades faster and the usual power law remains.
Two analytic references share the same energy half-width and total stellar mass. A flat energy distribution begins with a finite return rate; a uniform-density sphere rises to a peak. Both approach the same late-time slope, with different normalizations. The logarithmic vertical axis omits the zero at the first return. These are analytic references, separate from observed light and hydrodynamic simulations.
Worked example. For the million-solar-mass black hole and solar star from the previous chapter, tmin=40.954 days. The uniform-sphere curve peaks at about 63.642 days, at 0.951 solar masses per year. The separate flat-energy reference begins at 2.97 solar masses per year. A different early curve can therefore share the same shortest-period scale and the same late-time exponent.
These numbers come from the analytic continuum, using the published constants stated in the first chapter. Finite three-dimensional particle states use the exact potential and a finite set of masses. The sampled edge position and the chosen bin widths affect their first return and binned peak.
Integrating the curve catches missing debris
Integrating a mass rate over time must give a mass. For the uniform-sphere model, the mass returned by a time T with S=T/tmin≥1 is
M∗Mret(T)=21−43S−2/3+41S−2.
It is zero at the first return and approaches one half at infinite time. At eight minimum periods, only 0.31640625 of the stellar mass has returned. The rest of the bound mass is still on long orbits. Stopping a plot at that time and declaring that the star lost all the unplotted mass would confuse a finite observing window with escape.
For a saved record, sum the bound fragment masses first. Then sum the mass assigned to each return-time bin and the bound mass beyond the final edge. Those two sums should recover the bound mass. Unbound fragments belong in a separate total. This is also a practical way to find dropped rows or an incorrect time-unit conversion.
Practice with a return curve
One fragment has four times the binding-energy magnitude of another. Compare their orbital periods.
Explain why a uniform-sphere curve begins at zero while the flat-energy reference begins at a finite value.
In the uniform-sphere idealization, calculate the bound mass still away at eight minimum periods for a solar-mass star.
A saved record has fewer than half its total mass in the plotted return bins. What two checks distinguish ordinary late returns from a numerical loss?
Answers with the reasoningShow the answer
The period scales as ∣ε∣−3/2. Multiplying the energy magnitude by four divides the period by eight.
The edge of a sphere contains vanishing mass per unit slice thickness; the most bound energy edge therefore has zero weight. A flat energy distribution assigns a finite weight all the way to its edge.
The bound total is half a solar mass. Subtracting 0.31640625 gives 0.18359375 solar masses still on long orbits.
Recompute each fragment’s energy and return time from its saved position and velocity. Then add the plotted returned mass to the bound mass beyond the last bin. A missing late tail is physical within the model; a failure to recover the saved total signals a record or binning error.
Sources
Stone, Sari and Loeb, arXiv:1210.3374v1, Section 2, for the energy-to-return-time framework. The flat and uniform-sphere formulas above are derived explicitly from their assumptions.
Law-Smith and colleagues, arXiv:2007.10996v3, Equations 3–4 and Section 3, for mass scaling and the stellar-structure dependence of hydrodynamic fallback. Their saved simulation curves are a distinct comparison in the method tutorial.