
The one-way direction of time from past to future, often linked to entropy increase.The arrow of time is the concept that time has a one-way direction, flowing from the past to the future, unlike the symmetric laws of physics that work equally well forward and backward. The most widely accepted explanation ties this asymmetry to the second law of thermodynamics: entropy, a measure of disorder, tends to increase in an isolated system. This statistical tendency creates a clear distinction between past (lower entropy) and future (higher entropy), giving time its perceived direction.
The concept was popularized by Arthur Eddington in 1927, who coined the term 'arrow of time' and linked it to entropy increase, building on Ludwig Boltzmann's earlier statistical mechanics work.
At the microscopic level, the fundamental equations of physics are time-reversible — a movie of two particles colliding looks equally plausible played backward. However, when you have a system with many particles, the number of possible disordered states vastly outnumbers ordered ones. So, if you start in a low-entropy state, the natural evolution is toward higher entropy, simply because that's overwhelmingly more probable. This probabilistic drift from order to disorder is what we perceive as the forward flow of time.
The arrow of time is fundamental in cosmology and astrophysics, helping explain the evolution of the universe from the Big Bang to its eventual heat death, and it underpins our understanding of irreversible processes in thermodynamics.
The arrow of time is not absolute: in certain quantum experiments, researchers have observed temporary reversals of entropy increase on very small scales, though the overall macroscopic direction remains forward.
| x_t | variable | value of the process at time t — time series value |
| x_{t-1} | variable | value of the process at previous time — time series value |
| \phi | variable | autoregressive coefficient — autoregressive coefficient |
| \epsilon_t | variable | white noise innovation at time t — noise term |
| \sigma^2 | variable | variance of the noise — variance |












