An energy reservoir lets light emerge later than the returning mass that supplies it. The simplest useful calculation keeps the orbit model fixed, prescribes a heating rate, and lets stored energy escape over a chosen time. It makes the delay understandable without treating one telescope band as a direct mass-flow meter.
Separate mass, power and an observed flux
Prerequisites are the fallback calculation from the previous chapter and energy conservation. A mass rate M˙ has units of kilograms per second. A heating power H and an escaping luminosity L have units of watts. Stored energy E has units of joules. For comparison with astronomical tables, one watt is 107ergs−1.
The Lab assigns a deliberately restricted heating law. First convert a fraction η of the returning rest-mass energy into a candidate power, then cap the power admitted to the reservoir:
The second relation is the usual ionized-hydrogen electron-scattering scale. In a spherical static argument, outward radiation force on electrons balances inward gravity on the electron–proton pair. Real moving gas can differ substantially. Here the scale defines an imposed cap on heating; it is not a universal law that all tidal-disruption luminosities must obey.
The candidate power above the cap never enters this reservoir. Its physical destination lies outside the model. The efficiency is prescribed; calculating it would require a description of the shock or accretion flow that supplies the heat.
Solve the reservoir before drawing its light
Take an escape time τ>0. If a fraction dt/τ of stored energy escapes during a short interval dt, then L=E/τ. Conservation inside this defined reservoir gives
dtdE=H−L,L=τE.
Start with zero stored energy before heating begins. Multiplying the differential equation by the integrating factor et/τ and integrating gives
L(t)=∫0tτH(t′)e−(t−t′)/τdt′.
Every contribution comes from an earlier time. Heat deposited long ago has mostly escaped; recent heat is still partly stored. The exponential kernel integrates to one over all later times, so an isolated pulse ultimately releases all the energy admitted to the reservoir.
When heating is constant within a numerical bin of width Δt, the exact update between its endpoints is
Lend=H+(Lstart−H)e−Δt/τ.
A plotted value inside the bin must use the elapsed time from the bin’s start in that same update. For the geometric centre tc=tlefttright used here, that elapsed time is tc−tleft. It equals half the width only for an arithmetic centre. Assigning an endpoint value to an interior time would create a timing error. The limit τ→0 gives instantaneous escape, L=H, and is handled separately.
A pulse gives an independent check
Worked example. Feed the reservoir a constant H0=1044ergs−1 for twenty days, then switch it off. Use τ=10 days and a million-solar-mass black hole, whose imposed cap lies above this input. During heating, L/H0=1−e−t/τ. At day twenty the luminosity is 0.864665H0.
After switch-off, the stored energy decays exponentially. Twenty more days later, the luminosity is 0.864665e−2H0=0.117020H0. Integrating the full luminosity curve, including its tail, yields H0 times twenty days. A test that checks both these values and the total energy probes the implementation more strongly than copying its numerical update.
An analytic twenty-day heating pulse drives two reservoirs. A long escape time makes the rise slower and leaves a longer tail after heating stops. The input and output are powers; this is a teaching reservoir, not observed photometry or a radiation-transfer simulation.
Increasing the escape time while keeping the fallback record fixed changes the escaping light but leaves the debris orbits and mass-return curve unchanged. A change in the light curve can originate after orbital return. Identifying the actual storage process requires additional information about the gas.
A detector sees only part of the escaping spectrum
Bolometric luminosity means power added over all wavelengths. A telescope’s band flux is power per collecting area within its selected energy or wavelength range, after the relevant spectral modelling and absorption treatment. Even if total power followed fallback exactly, the fraction in one band could evolve.
The simple limit F=L/(4πD2) assumes isotropic emission, a suitable luminosity distance D and a bolometric flux. It does not convert an X-ray table into a bolometric luminosity without a spectral correction. Lodato and Rossi’s multiband calculation, Sections 3–4, illustrates how different bands can acquire different decline shapes from the same underlying accretion model.
The companion tutorial preserves ASASSN-14li’s published X-ray fluxes and errors. It normalizes the observed and model curves separately to ask a narrower question: how similar are their declining shapes under an explicit choice of time origin? This loses the absolute amplitude deliberately. A good shape match would still leave distance, spectrum, absorption and the mass-to-light process to be determined.
Exercises in causality and units
Heating switches off. Can the model luminosity remain positive? Where does the energy come from?
At fixed fallback, increase the efficiency until heating is capped throughout the peak. Does a further increase necessarily raise the peak light?
A programme feeds uncapped heating into a reservoir, then caps only the escaping light. Is it the same model as the one above?
An X-ray flux falls twice as fast as a normalized fallback curve. Give two physical quantities that could vary without changing the debris return times.
Answers with the reasoningShow the answer
Yes. Stored energy continues to escape. The tail must be included when checking the total admitted and emitted energy.
No. While the cap is active, the admitted heating stays fixed. The reservoir responds to that admitted power, not to the rejected candidate power.
No. It stores the excess energy and changes the later output. Its energy equation and treatment of capped emission would have to be defined separately.
The spectrum’s temperature can change the fraction inside the X-ray band; the intervening absorbing column can also change. Spectral observations can test these possibilities in an individual event.