The numbers of arrangements in the simulation above were counted by enumerating every combination of oscillator quantum numbers and comparing the results with the combinatorial formula. The checks used total energies of 1, 6, and 12ε, B containing 1, 3, or 6 oscillators, and A having either the same frequency or 2 times the frequency. The displayed statistics sample 8400 equally probable arrangements; a Python check used the same sample size. Steps 3 and 5 calculated the logarithms of the combinatorial counts exactly. In step 7, 2000 oscillators started with an average of 2 quanta each, followed by a total of 40 × ten thousand random transfers of 1 quantum at a time. The resulting quantum-number distribution was compared with the Boltzmann distribution, which here is a geometric distribution.
| The simulation above (equal frequencies) | For A's quantum numbers 0–6, the numbers of arrangements are 28, 21, 15, 10, 6, 3, 1 (84 in total). P(0) = 33.33%; 8400 samples gave 33.58%. |
| The simulation above (A has twice the frequency) | For A's quantum numbers 0–3, the numbers of arrangements are 28, 15, 6, 1 (50 in total). P(0) = 56.00%. |
| The energy division with the most arrangements | A has 60 quanta and B has 40, with oscillator counts in the ratio 300: 200. |
| The slopes at this energy division | The slope of ln WA is 1.7821; the slope of ln WB is 1.7773, each differentiated with respect to its own quantum number. |
| The fraction of oscillators with 0 quanta | Numerical experiment: 0.3375; Boltzmann distribution: 0.3333 |
| The fraction of oscillators with 1 quantum | Numerical experiment: 0.2235; Boltzmann distribution: 0.2222 |
| The fraction of oscillators with 2 quanta | Numerical experiment: 0.1440; Boltzmann distribution: 0.1481 |
| The fraction of oscillators with 3 quanta | Numerical experiment: 0.0975; Boltzmann distribution: 0.0988 |
| The fraction of oscillators with 4 quanta | Numerical experiment: 0.0650; Boltzmann distribution: 0.0658 |