Even more

 The Broader Kester Family: From One Operator to an Entire Theory of Self-Referential Function Dynamics

The original Kester function was defined by a remarkably simple rule:

Kf(x)=
\begin{cases}
2f(x),&f\text{ is continuous at }x,\\
f(x),&f\text{ is discontinuous at }x.
\end{cases}

At first sight, this looks almost too elementary to generate substantial mathematics.

But the important feature is not the number 2.

The important feature is that the function determines the set on which the function is modified.

The operator first asks:

Is f continuous at x?

and then uses the answer to decide what value f should have at x.

That self-reference is what produces the dynamics.

Once this structure is isolated, the number 2 can be replaced, generalized, complexified, made spatially dependent, or even replaced by an entire family of transformations.

This suggests that the Kester function should perhaps be viewed not as a single isolated construction, but as the first member of a much broader class of continuity-conditioned operators.

1. The General Kester Operator

Let X be a topological space and let

f:X\to V,

where V is a normed vector space, such as

\mathbb R,\qquad \mathbb C,\qquad \mathbb R^n,

or a more general topological vector space.

Write

C_f=\{x\in X:f\text{ is continuous at }x\}

for the continuity set of f.

The original Kester operator can then be written

K_2f(x)
=
2^{\mathbf 1_{C_f}(x)}f(x).

Equivalently,

K_2f(x)=
\begin{cases}
2f(x),&x\in C_f,\\
f(x),&x\notin C_f.
\end{cases}

The subscript 2 emphasizes something important:

the factor 2 is only one possible choice.

We can replace it with an arbitrary scalar \lambda.

2. The \lambda-Kester Family

Define

\boxed{
K_\lambda f(x)=
\begin{cases}
\lambda f(x),&x\in C_f,\\
f(x),&x\notin C_f.
\end{cases}}

or, more compactly,

\boxed{
K_\lambda f
=
\lambda^{\mathbf 1_{C_f}}f.
}

The original Kester operator is simply

K=K_2.

This immediately creates several qualitatively different dynamical regimes.

Real expanding regime

\lambda>1.

Continuous points are amplified.

Real contracting regime

0<\lambda<1.

Continuous points are contracted.

Sign-reversing regime

\lambda<0.

Continuity points are multiplied by a negative number, potentially producing alternating behavior.

Identity regime

\lambda=1.

Nothing happens:

K_1f=f.

Zeroing regime

\lambda=0.

Every continuity point is sent to zero:

K_0f(x)=
\begin{cases}
0,&x\in C_f,\\
f(x),&x\notin C_f.
\end{cases}

Complex rotational regime

\lambda\in\mathbb C,\qquad |\lambda|=1.

The magnitude is preserved while the phase changes.

This last case opens the door to a completely different kind of Kester dynamics.

3. The Universal Itinerary Formula

The fundamental bookkeeping identity survives unchanged.

Let

f_0=f,

and recursively define

f_{n+1}=K_\lambda f_n.

Let

C_n=C_{f_n}.

Define the continuity itinerary of x by

\omega_n(x)
=
\left(
\mathbf 1_{C_0}(x),
\mathbf 1_{C_1}(x),
\dots,
\mathbf 1_{C_{n-1}}(x)
\right).

Its number of continuity events is

N_n(x)
=
\sum_{j=0}^{n-1}\mathbf 1_{C_j}(x).

Then

\boxed{
f_n(x)=\lambda^{N_n(x)}f(x).
}

This is the central algebraic law of the generalized theory.

The function itself supplies the original value f(x), while the continuity history supplies the exponent.

Thus the evolution can be separated into two pieces:

\boxed{
\text{value at time }n
=
\text{initial value}
\times
\text{history-dependent multiplier}.
}

This means that the Kester system possesses an unusual form of memory.

The current value alone is not enough to reconstruct its history.

One must know how many times the point has previously been continuous.

4. Kester Dynamics as a Memory System

For the original operator,

f_n(x)=2^{N_n(x)}f(x).

For the generalized operator,

f_n(x)=\lambda^{N_n(x)}f(x).

Consequently, two points x and y can have identical current continuity status but completely different values because their historical itineraries differ.

For example,

\omega_5(x)=(1,1,0,1,0)

has

N_5(x)=3.

Another point might have

\omega_5(y)=(1,0,1,0,1),

which also gives

N_5(y)=3.

These two points have different histories but the same multiplier:

\lambda^3.

Meanwhile,

\omega_5(z)=(1,1,1,1,1)

produces

\lambda^5.

Thus the natural state variable is not merely

x\in C_n?

but rather

(x,\omega_n(x)).

This is why the theory begins to resemble symbolic dynamics.

5. Symbolic Dynamics Hidden Inside the Kester Operator

Every point generates a binary sequence

\omega(x)
=
(\omega_0(x),\omega_1(x),\omega_2(x),\ldots),

where

\omega_n(x)=
\begin{cases}
1,&x\in C_n,\\
0,&x\notin C_n.
\end{cases}

Therefore every point has a symbolic itinerary

\omega(x)\in\{0,1\}^{\mathbb N}.

The Kester system therefore creates a map

x\longmapsto \omega(x).

This map records the entire continuity history of the point.

The exponent is simply

N_n(x)=
\omega_0(x)+\cdots+\omega_{n-1}(x).

Hence

f_n(x)=
\lambda^{\omega_0(x)+\cdots+\omega_{n-1}(x)}f(x).

The original real-valued function has therefore generated a symbolic dynamical system in the background.

6. The Growth Rate of a Point

Suppose

\lambda\neq0.

Taking absolute values gives

|f_n(x)|
=
|\lambda|^{N_n(x)}|f(x)|.

If f(x)\neq0, we can examine the logarithmic growth rate:

\frac1n\log\frac{|f_n(x)|}{|f(x)|}
=
\frac{N_n(x)}n\log|\lambda|.

If the limiting frequency of continuity exists,

\rho(x)
=
\lim_{n\to\infty}\frac{N_n(x)}n,

then

\boxed{
\lim_{n\to\infty}
\frac1n\log\frac{|f_n(x)|}{|f(x)|}
=
\rho(x)\log|\lambda|.
}

This provides a natural analogue of a pointwise Lyapunov exponent.

We could call it the Kester continuity exponent:

\boxed{
\chi_f(x)
=
\rho(x)\log|\lambda|.
}

It measures the asymptotic growth or decay caused by the continuity history.

7. Three Fundamental Dynamical Regimes

The magnitude of \lambda now becomes fundamental.

Regime I: |\lambda|>1

If

\rho(x)>0

and

f(x)\neq0,

then

|f_n(x)|\to\infty.

The more frequently x becomes continuous, the faster its value grows.

In this regime, continuity acts as an amplification event.

Regime II: |\lambda|<1

Now every continuity event reduces the magnitude.

If

\rho(x)>0,

then

|f_n(x)|\to0.

Continuity becomes a damping mechanism.

This is mathematically fascinating because the same topological condition—continuity—has completely opposite dynamical consequences depending on |\lambda|.

Regime III: |\lambda|=1

Now

|f_n(x)|=|f(x)|.

There is no radial growth at all.

The dynamics move entirely into the phase.

This produces the complex Kester system.

8. The Complex Kester Operator

Let

\lambda=e^{2\pi i\alpha}.

Then

|\lambda|=1.

The generalized operator becomes

K_\lambda f(x)
=
\begin{cases}
e^{2\pi i\alpha}f(x),&x\in C_f,\\
f(x),&x\notin C_f.
\end{cases}

After n iterations,

\boxed{
f_n(x)
=
e^{2\pi i\alpha N_n(x)}f(x).
}

Therefore

|f_n(x)|=|f(x)|.

The Kester dynamics no longer expand or contract the function.

Instead, continuity events rotate it around a circle in the complex plane.

9. Rational Phase Rotation

Suppose

\alpha=\frac pq

with p,q relatively prime.

Then

\lambda^q
=
e^{2\pi ip}
=
1.

Therefore the possible phases belong to the finite set

\left\{
1,\lambda,\lambda^2,\ldots,\lambda^{q-1}
\right\}.

For a point whose continuity count increases indefinitely, its value moves around a finite collection of points on the circle

|z|=|f(x)|.

For example, with

\lambda=i,

we have

1,\quad i,\quad -1,\quad -i.

The point therefore follows the cycle

f(x)
\longrightarrow
if(x)
\longrightarrow
-f(x)
\longrightarrow
-if(x)
\longrightarrow
f(x).

Thus the Kester operator can generate genuine periodic phase dynamics.

10. Irrational Phase Rotation

Now suppose

\alpha\notin\mathbb Q.

Then

e^{2\pi i\alpha n}

does not repeat periodically.

If the continuity count visits all sufficiently large integers, then the sequence

e^{2\pi i\alpha N_n(x)}

has dense behavior on the unit circle.

Consequently,

\{f_n(x):n\ge0\}

can have orbit closure

\boxed{
\overline{\{f_n(x):n\ge0\}}
=
\{z\in\mathbb C:|z|=|f(x)|\}.
}

Thus a purely topological question—

How often is x continuous?

—can determine whether its complex value follows a periodic orbit or a dense orbit.

That is a striking bridge between topology and dynamical systems.

11. The Zero Set Becomes Dynamically Special

There is another universal feature.

If

f(x)=0,

then for every \lambda,

f_n(x)=\lambda^{N_n(x)}0=0.

Therefore

\boxed{
f(x)=0\implies f_n(x)=0\quad\forall n.
}

The point’s value is completely pinned.

Its continuity behavior, however, need not be pinned.

This distinction is extremely important.

The value dynamics may be trivial while the topological dynamics remain complicated.

A zero can therefore act as an anchor around which surrounding values evolve.

12. A Broader Two-Factor Kester Family

The next generalization is even more natural.

Instead of multiplying continuous points by \lambda and leaving discontinuous points unchanged, introduce two factors:

\boxed{
K_{\alpha,\beta}f(x)
=
\begin{cases}
\alpha f(x),&x\in C_f,\\
\beta f(x),&x\notin C_f.
\end{cases}}

Equivalently,

K_{\alpha,\beta}f(x)
=
\alpha^{\mathbf1_{C_f}(x)}
\beta^{1-\mathbf1_{C_f}(x)}
f(x).

The original Kester operator is recovered with

\alpha=2,\qquad \beta=1.

This two-parameter family dramatically expands the possible dynamics.

After n steps,

\boxed{
f_n(x)
=
\alpha^{N_n(x)}
\beta^{\,n-N_n(x)}
f(x).
}

Thus both continuity and discontinuity events now carry dynamical weights.

13. The Ratio Between Continuity and Discontinuity Dynamics

For the two-factor operator,

f_n(x)
=
\alpha^{N_n(x)}
\beta^{n-N_n(x)}
f(x).

Rewrite this as

f_n(x)
=
\beta^n
\left(\frac{\alpha}{\beta}\right)^{N_n(x)}
f(x),

assuming \beta\neq0.

This separates the dynamics into:

1. a global factor \beta^n, and
2. a continuity-dependent factor

\left(\frac{\alpha}{\beta}\right)^{N_n(x)}.

This reveals that what really matters for the topology-sensitive part of the dynamics is the ratio

\boxed{\frac{\alpha}{\beta}}.

The original Kester operator has

\frac{\alpha}{\beta}=2.

14. A Continuous Weight Function

We can go even further.

Why should the multiplier be constant across the entire domain?

Let

a:X\to\mathbb C

be a function.

Define

\boxed{
K_a f(x)=
\begin{cases}
a(x)f(x),&x\in C_f,\\
f(x),&x\notin C_f.
\end{cases}}

Then

K_af(x)
=
a(x)^{\mathbf1_{C_f}(x)}f(x).

This is a spatially weighted Kester operator.

Now two continuity points can evolve differently.

One point might be multiplied by

2,

another by

3,

while a third might undergo a complex rotation such as

e^{i\pi/3}.

Continuity still controls the switch, but the response to continuity depends on location.

15. Matrix-Valued Kester Dynamics

The same idea works for vector-valued functions.

Let

f:X\to\mathbb R^d

and let

A

be a d\times d matrix.

Define

K_Af(x)=
\begin{cases}
Af(x),&x\in C_f,\\
f(x),&x\notin C_f.
\end{cases}

Then the iteration becomes

f_n(x)=A^{N_n(x)}f(x).

This is remarkable because the scalar exponent N_n(x) now controls matrix dynamics.

If A has eigenvalues larger than 1, certain directions expand.

If its eigenvalues have modulus less than 1, those directions contract.

If its eigenvalues lie on the unit circle, rotational or quasi-periodic behavior can occur.

Thus the Kester idea can naturally connect to linear dynamical systems.

16. Operator-Valued Kester Dynamics

There is no fundamental reason to stop with matrices.

Suppose

T:V\to V

is a linear operator.

Define

K_Tf(x)=
\begin{cases}
T(f(x)),&x\in C_f,\\
f(x),&x\notin C_f.
\end{cases}

If T is applied repeatedly whenever the point is continuous, then formally

\boxed{
f_n(x)=T^{N_n(x)}f(x).
}

This transforms the Kester construction from scalar dynamics into an operator-dynamical framework.

The continuity history becomes a clock telling the system how many times T should be applied.

17. Nonlinear Kester Operators

We can even abandon linearity.

Let

\Phi:V\to V

be any transformation.

Define

K_\Phi f(x)=
\begin{cases}
\Phi(f(x)),&x\in C_f,\\
f(x),&x\notin C_f.
\end{cases}

Now continuity determines whether the nonlinear dynamical system \Phi advances.

The Kester mechanism has therefore become:

\boxed{
\text{topological state}
\longrightarrow
\text{dynamical update}.
}

For example,

\Phi(z)=z^2

would produce

f_{n+1}(x)=
\begin{cases}
f_n(x)^2,&x\in C_n,\\
f_n(x),&x\notin C_n.
\end{cases}

This is no longer simply a scaling system.

It becomes a continuity-gated nonlinear dynamical system.

18. Kester Dynamics and the Logistic Map

Take

\Phi(z)=rz(1-z).

Then

K_\Phi f(x)=
\begin{cases}
r f(x)(1-f(x)),&x\in C_f,\\
f(x),&x\notin C_f.
\end{cases}

At continuous points, the value advances one step through the logistic map.

At discontinuous points, it waits.

The resulting system therefore combines:

• topology,
• continuity,
• nonlinear dynamics,
• memory,
• and local iteration.

The continuity structure acts as a gate controlling when the nonlinear dynamics are allowed to evolve.

19. A General “Topological Clock”

This suggests an especially useful conceptual interpretation.

Define

\tau_n(x)=N_n(x).

Then \tau_n(x) counts the number of times the point has been granted permission to evolve.

For a linear Kester system,

f_n(x)=T^{\tau_n(x)}f(x).

The ordinary iteration count is n.

The Kester iteration count is instead

\tau_n(x).

Thus different points experience different amounts of dynamical time even though the entire function is being iterated simultaneously.

This gives us a useful phrase:

The Kester operator creates a topology-dependent internal clock.

Two points can experience ten global iterations while one has experienced ten dynamical updates and another has experienced only three.

20. Local Time Versus Global Time

This distinction becomes particularly important for the staircase constructions.

Suppose

x\in C_0,

but

x\notin C_1.

Then during the first two global iterations, the point has experienced only one effective multiplication.

Its internal clock is

N_2(x)=1.

Another point may satisfy

x\in C_0,\qquad x\in C_1.

For that point,

N_2(x)=2.

So after the same two global iterations,

f_2(x)=\lambda f(x)

for the first point, while

f_2(y)=\lambda^2f(y)

for the second.

The Kester operator therefore creates asynchronous dynamics from a synchronous iteration process.

21. A Continuity-Driven Dynamical Spectrum

We can now classify generalized Kester systems according to the multiplier.

Multiplier	Dominant behavior
\lambda=0	Continuous points collapse to zero
0<|\lambda|<1	Continuity produces decay
|\lambda|=1	Magnitude preserved; phase dynamics dominate
|\lambda|>1	Continuity produces amplification
\lambda<0	Sign/phase alternation
\lambda\in\mathbb C	Expansion/contraction + rotation

The original K_2 system occupies only one location in this much larger parameter space.

22. Fixed Points of the Generalized Operator

A function f is a fixed point when

K_\lambda f=f.

At a continuity point,

\lambda f(x)=f(x).

Hence

(\lambda-1)f(x)=0.

Therefore, if

\lambda\neq1,

every continuity point of a fixed function must satisfy

\boxed{f(x)=0.}

At discontinuity points, the operator leaves the value unchanged automatically.

Thus a generalized fixed point must satisfy the structural condition

\boxed{
f(x)=0
\quad\text{for every }x\in C_f
}

whenever

\lambda\neq1.

This gives a very broad fixed-point principle.

The Thomae function is an especially elegant example because it vanishes precisely on its dense continuity set of irrationals.

23. A Whole Family of Fixed Points

This observation suggests that Thomae’s function is not necessarily an isolated curiosity.

Suppose f satisfies

f|_{C_f}=0.

Then

K_\lambda f=f

for every nonzero \lambda, provided \lambda\neq1 is handled in the obvious way.

Indeed:

• at continuity points, f(x)=0, so multiplication changes nothing;
• at discontinuity points, the Kester rule leaves the value unchanged.

Thus the fixed-point problem can be reframed as:

Find functions whose entire continuity set lies inside their zero set.

This is a potentially large class.

24. The Zero-Continuity Principle

For \lambda\neq1, fixed points satisfy

C_f\subseteq Z_f,

where

Z_f=\{x:f(x)=0\}.

So the geometry of the zero set becomes central to the fixed-point theory.

This creates a beautiful triangle:

\boxed{
\text{continuity set}
\quad\leftrightarrow\quad
\text{zero set}
\quad\leftrightarrow\quad
\text{fixed-point structure}.
}

The original Kester function therefore naturally leads into questions about the topology of zero sets.

25. Periodic Points

Fixed points are only the first level.

Suppose

K_\lambda^m f=f.

At a point whose continuity itinerary contains N_m(x) continuity events,

f_m(x)
=
\lambda^{N_m(x)}f(x).

Therefore, whenever

f(x)\neq0,

a necessary pointwise condition is

\boxed{
\lambda^{N_m(x)}=1.
}

For

\lambda=e^{2\pi i\alpha},

this becomes

\alpha N_m(x)\in\mathbb Z.

Thus the possible periodic behavior is controlled simultaneously by:

1. the arithmetic of \lambda,
2. the continuity itinerary,
3. the zero set of f.

This is where the topology and number theory of the Kester system begin to interact.

26. The Kester Family as a Research Program

The broader theory can now be organized into several major directions.

Direction I — Topological dynamics

Study the evolution

C_0,C_1,C_2,\ldots

and determine which sequences of continuity sets are possible.

Direction II — Symbolic dynamics

Study the admissible itineraries

\omega(x)\in\{0,1\}^{\mathbb N}.

Direction III — Asymptotic dynamics

Study

\lim_{n\to\infty}\frac{N_n(x)}n.

Direction IV — Complex dynamics

Study

\lambda^{N_n(x)}

when

\lambda\in\mathbb C.

Direction V — Fixed points

Classify functions satisfying

K_\lambda f=f.

Direction VI — Periodic points

Classify solutions of

K_\lambda^m f=f.

Direction VII — Generalized operators

Replace multiplication by an arbitrary transformation

\Phi.

Direction VIII — Spatially varying dynamics

Replace the constant multiplier with

\lambda(x).

27. The Most General Continuity-Gated Operator

All of these examples suggest an abstract definition.

Let X be a topological space, V a state space, and

\Phi,\Psi:V\to V

two transformations.

Define

\boxed{
\mathcal K_{\Phi,\Psi}(f)(x)
=
\begin{cases}
\Phi(f(x)),&x\in C_f,\\
\Psi(f(x)),&x\notin C_f.
\end{cases}}

The ordinary Kester operator corresponds to

\Phi(z)=2z,
\qquad
\Psi(z)=z.

The complex Kester operator corresponds to

\Phi(z)=\lambda z,
\qquad
\Psi(z)=z.

The two-factor version corresponds to

\Phi(z)=\alpha z,
\qquad
\Psi(z)=\beta z.

The nonlinear version corresponds to an arbitrary pair of maps.

This gives us the broadest formulation so far:

A Kester-type operator is a function transformation whose local update rule is selected by whether the input function is continuous at that point.

28. The Central Idea

The striking thing about this framework is that the operator does not merely transform values.

It transforms values according to topology.

Ordinary iteration asks:

f\mapsto \Phi(f).

Kester iteration asks:

f
\mapsto
\text{inspect continuity of }f
\mapsto
\text{choose a local evolution rule}
\mapsto
\text{produce a new function}.

The topology of the function therefore becomes part of the dynamical state.

That is the central mathematical idea behind the entire Kester family.

29. Toward a General Kester Dynamical System

We can summarize the emerging theory schematically:

\boxed{
f
\longrightarrow
C_f
\longrightarrow
\omega(x)
\longrightarrow
N_n(x)
\longrightarrow
f_n(x)
\longrightarrow
C_{f_n}.
}

The process then feeds back into itself:

C_{f_n}
\longrightarrow
C_{f_{n+1}}
\longrightarrow\cdots

This creates a closed feedback loop.

The value determines continuity.

Continuity determines the next value.

The new value determines the next continuity set.

And so on.

So the Kester operator is not merely an unusual function transformation.

It is a feedback dynamical system on functions.

30. A New Conjectural Landscape

The generalized framework suggests several questions that are much broader than the original K_2 problem.

Conjecture/Question 1: Itinerary Realisability

Which binary sequences

\omega\in\{0,1\}^{\mathbb N}

can actually occur as continuity itineraries of a point?

Not every abstract binary sequence is necessarily topologically realizable.

Question 2: Continuity-Set Dynamics

Given an arbitrary sequence of sets

C_0,C_1,C_2,\ldots,

when can it arise from Kester iteration?

This is a much deeper inverse problem.

Question 3: Entropy

If many different itineraries are possible, can one define a Kester topological entropy measuring the complexity of the continuity dynamics?

One possible starting point would be the growth of admissible finite itineraries:

\mathcal I_n
=
\{\omega_n(x):x\in X\}.

Then one might investigate

h_K
=
\limsup_{n\to\infty}
\frac1n\log |\mathcal I_n|.

Question 4: Ergodic Kester Dynamics

If a probability measure is placed on X, can we study the statistical frequency

\rho(x)
=
\lim_{n\to\infty}\frac{N_n(x)}n?

If this exists almost everywhere, the multiplier dynamics acquire a statistical description.

Question 5: Complex Kester Spectrum

For

\lambda=e^{2\pi i\alpha},

what phase orbit structures can be generated by admissible continuity itineraries?

Periodic?

Dense?

Finite unions of circles?

More complicated orbit closures?

31. The Bigger Picture

The original definition

Kf(x)=
\begin{cases}
2f(x),&f\text{ continuous at }x,\\
f(x),&f\text{ discontinuous at }x
\end{cases}

looks like a tiny modification of a familiar function.

But the deeper interpretation is much more interesting.

It introduces a feedback mechanism:

\boxed{
\text{topology controls dynamics}.
}

And after iteration:

\boxed{
\text{dynamics reshape topology}.
}

So the relationship runs both ways:

\text{Topology}
\longrightarrow
\text{Dynamics}
\longrightarrow
\text{Topology}
\longrightarrow
\text{Dynamics}
\longrightarrow\cdots

That feedback loop is the real object of study.

The number 2 is merely the simplest example of a much larger family.

The Kester function can therefore be viewed as the seed of a general theory of topology-gated dynamical operators—operators in which a local topological property of a function determines how that function evolves.

And once complex scalars, nonlinear transformations, matrices, variable weights, symbolic itineraries, asymptotic continuity frequencies, fixed points, and periodic orbits are admitted, the original question becomes only the beginning:

\boxed{
\text{What dynamical systems can be generated when continuity itself becomes the switch controlling evolution?}
}

That is the broader Kester problem.