Comparing evolutionary models with populations

Tracks and isochrones select different things

An evolutionary track follows a model star through time for selected initial conditions and physical prescriptions. Its changing location in luminosity and effective temperature describes that model’s sequence. Other stars at different locations are not automatically the same object seen at successive ages.

An isochrone is a common-age model locus assembled from stars of different initial masses under specified composition and other assumptions. The word means “equal time”. It selects the state of each applicable model at that age rather than tracing one star’s movement along the entire line.

The construction requires models whose phases are matched consistently across an initial-mass grid. Interpolation and the treatment of short or rapidly changing phases can affect the resulting locus. A smooth drawn curve is not itself evidence that every part was directly observed or that every physical input is known.

Observed colour and magnitude also differ from intrinsic luminosity and effective temperature. Distance, extinction, atmosphere models, instrumental bands and unresolved companions enter the comparison. A match in a diagram depends on how the model was transformed into the observed quantities.

A locus and a count are different predictions

A model locus specifies where stars with the supplied age and inputs could lie. Predicting how many stars occupy each part additionally requires an initial distribution of masses and a formation history. A stellar initial mass function describes a number distribution under an explicitly chosen mass measure.

Counts at a later time depend on evolutionary duration as well as birth counts. A short-lived stage can be sparsely populated even when many progenitors form. The number per interval of mass, the number per interval of logarithmic mass and the mass contained in an interval are distinct quantities.

An observing process then recovers only some objects. Brightness limits, crowding and measurement quality can make recovery probability depend on the type of star. A missing detection does not by itself show that the star has disappeared or that an evolutionary prediction failed.

The following original calculation fixes birth counts and gives an independent recovery probability for each eligible star in a mass category. It deliberately omits contamination, unresolved binaries, mass exchange and an earlier formation phase. It is an arithmetic teaching population, not an observed cluster or a fitted initial mass function.

Specify the clock and the selection before counting

Let m = M/M₀ be a positive dimensionless initial mass, with arbitrary positive mass unit M₀. Stipulate accessible energy E_av = 100mE₀ and constant expenditure P = m³E₀/T₀. Here E₀ and T₀ are arbitrary positive units, and the simplified energy budget neglects other reservoirs. The main-sequence duration is τ_MS = 100m⁻²T₀.

For this calculation only, t = 0 is a common idealized main-sequence starting time and an object is eligible for the main-sequence count when t < τ_MS. There is no pre-main-sequence clock in this model. Leaving that count means leaving the stipulated stage, not ceasing to exist.

Choose dimensionless masses m = 1, 2 and 4 with fixed birth counts 40, 20 and 10. Their supplied main-sequence durations are 100T₀, 25T₀ and 6.25T₀. The respective per-object recovery probabilities are 0.5, 0.8 and 0.9 for an eligible star.

These mass and lifetime relations have been stipulated for the example. They do not provide one universal stellar mass–luminosity law or convert the arbitrary time unit to years. The table and conclusions retain those restrictions.

Account for eligible categories and incomplete recovery

At t = 10T₀, masses m = 1 and 2 remain eligible, while m = 4 has passed its stipulated main-sequence duration. The expected recovered count is 40 × 0.5 + 20 × 0.8 = 36. It is an expectation over repeated independent observing trials of this fixed population, not a required integer outcome of each trial.

For N fixed independent Bernoulli recoveries with a known common probability p in a category, the count variance is Np(1 − p). Adding independent eligible categories gives variance 40 × 0.5 × 0.5 + 20 × 0.8 × 0.2 = 13.2 at t = 10T₀.

At t = 30T₀, only m = 1 remains eligible. The expected recovered count is 20 and its variance is 10. These variances describe the supplied recovery trials; they do not include uncertain birth counts, uncertain lifetimes, contamination or correlations.

A lower observed count could result from an older population, a different birth distribution or poorer recovery, among other possibilities. The example separates those inputs so their effects can be calculated. An actual age inference must constrain or marginalize the relevant unknowns rather than attributing every missing bright star to age.

Keep calibration, selection and validation distinct

In an actual population diagram, the main-sequence turnoff is the region where stars of the inferred common age are leaving that stage. It can constrain age in conjunction with composition, distance, extinction and the adopted models. It is not a universal brightness-to-age conversion.

A cluster or association must also be identified through evidence of membership and formation history. A chance foreground or background object can occupy a misleading location. The next chapter examines companions, shared origins, membership and population-age inference more directly.

Some model inputs are calibrated with selected observations. Testing the same measurements assesses the calibration fit; additional independent measurements test other predictions. Changing the input physics and comparing the resulting predictions can reveal model sensitivity, while numerical convergence checks whether a chosen computation is stable.

The useful conclusion is conditional and testable: given a population history, physical model and observing process, which distributions should be recovered? A credible comparison preserves those conditions, reports tensions and limitations, and combines evidence beyond a visually attractive curve.

Worked examples

Try the reasoning

1. Does one isochrone trace the time history of a single star?

   Show the answer and reasoning

   No.

   It selects model stars of different initial masses at one common age under stated inputs.

2. Does removal from the example’s main-sequence count mean that a star ceases to exist?

   Show the answer and reasoning

   No.

   The eligibility rule concerns one evolutionary stage; later states are outside that count.

3. Does the variance 13.2 include uncertainty in the supplied birth counts and lifetimes?

   Show the answer and reasoning

   No.

   Those quantities are fixed in the example; only the stipulated independent recovery trials contribute.

4. Does an isochrone alone specify how many stars an observer will detect along it?

   Show the answer and reasoning

   No.

   Initial numbers, evolutionary durations, formation history and the selection process are also required.

What to carry forward

• Tracks follow model objects through time; isochrones compare different initial masses at a common age.
• A population’s recovered counts require both a birth model and an observing model.
• Agreement with a curve must be assessed alongside calibration, selection, uncertainty and independent evidence.