18.1 Can Deterministic Systems Be Unpredictable?
18.1 Can Deterministic Systems Be Unpredictable?
Section titled “18.1 Can Deterministic Systems Be Unpredictable?”Throughout this book, we have developed mathematical models to understand and predict the behavior of biological systems. Once a model has been constructed, the underlying equations uniquely determine how the system evolves over time. Given the initial state of the system and the values of all model parameters, there should be only one possible future.
This property is known as determinism.
At first glance, determinism appears to imply predictability. If the governing equations are known exactly and the current state of the system is measured accurately, it seems reasonable to expect that the future can also be predicted accurately.
For many systems, this intuition is correct.
A pendulum swings predictably, radioactive decay follows well-defined statistical laws, and the trajectories of planets can be calculated decades or even centuries into the future with extraordinary precision. Similarly, many biological models introduced in the previous chapters converge toward stable equilibria or stable oscillations that can be predicted reliably.
It is therefore tempting to conclude that deterministic models are always predictable.
Surprisingly, this conclusion is wrong.
A simple thought experiment
Section titled “A simple thought experiment”Imagine performing the following numerical experiment.
You simulate a biological model twice.
Both simulations use
- exactly the same mathematical equations,
- exactly the same parameter values,
- and exactly the same numerical algorithm.
The only difference is the initial condition.
Suppose one simulation starts with an initial protein concentration of
while the second simulation starts with
The difference between the two initial conditions is almost immeasurably small.
Since both simulations begin from nearly identical states, one would naturally expect their trajectories to remain nearly identical throughout the simulation.
After all, if two systems start almost the same, why should they end up behaving completely differently?
The surprising observation
Section titled “The surprising observation”Now let both simulations evolve.
Initially, the trajectories are almost indistinguishable.
Their differences are so small that they may not even be visible in a graph.
As time passes, however, the trajectories begin to separate.
The tiny initial difference becomes larger.
After a while, the simulations no longer resemble one another.
Eventually, the two trajectories appear completely unrelated, despite being generated by exactly the same deterministic equations.
Nothing random has been added to the model.
No parameters have changed.
No external disturbances have been introduced.
The only difference was a minute change in the initial condition.
A paradox
Section titled “A paradox”This observation creates an apparent contradiction.
If the equations are completely deterministic, why does such a tiny difference produce completely different outcomes?
Does this mean that the system is actually random?
Or does it mean that our intuition about determinism is incomplete?
These questions puzzled mathematicians and physicists for decades.
The answer ultimately led to one of the most important discoveries of modern nonlinear dynamics:
Determinism does not necessarily imply predictability.
The limits of prediction
Section titled “The limits of prediction”In practice, every measurement contains some uncertainty.
No experimental instrument can determine the exact state of a biological system with infinite precision.
If a model amplifies these tiny uncertainties over time, long-term prediction eventually becomes impossible.
Importantly, this loss of predictability is not caused by imperfections in the measuring instruments.
It is an intrinsic property of the dynamical system itself.
Even with perfect mathematical equations, prediction may be fundamentally limited.
This realization completely changed our understanding of complex systems.
Rather than asking whether a system is deterministic or random, we must instead ask a different question:
How does the system respond to small perturbations?
The answer to this question determines whether long-term prediction is possible.
As we will see throughout this chapter, some nonlinear systems amplify tiny differences so dramatically that accurate long-term prediction becomes fundamentally impossible, even though the underlying dynamics remain perfectly deterministic.
This phenomenon is known as deterministic chaos.
Key concepts
Section titled “Key concepts”- Deterministic systems evolve according to fixed mathematical rules.
- Determinism does not automatically imply predictability.
- Tiny differences in initial conditions may grow dramatically over time.
- Prediction is fundamentally limited by the system dynamics rather than by measurement errors.
- Deterministic chaos emerges when nonlinear dynamics amplify small perturbations.
Summary
Section titled “Summary”Deterministic models are often expected to produce predictable behavior. However, some nonlinear systems behave very differently. Two simulations that differ only by an infinitesimal change in their initial conditions may eventually evolve into completely different trajectories. This surprising phenomenon demonstrates that deterministic equations can generate behavior that is practically unpredictable, introducing the central idea of deterministic chaos.
Self-check questions
Section titled “Self-check questions”- What is meant by a deterministic dynamical system?
- Why is determinism often associated with predictability?
- In the thought experiment, what is the only difference between the two simulations?
- Why is the divergence of the two trajectories surprising?
- Why is long-term unpredictability not necessarily caused by measurement errors?
- What central question motivates the study of deterministic chaos?