Extended Kester function

Absolutely. Below is a blog-friendly, very long version that folds the earlier Kester-function material together with the general multiplier a, the complex-valued version, phase dynamics, fixed points, periodic orbits, itineraries, staircases, and the open oscillation problem.

The Kester Function: A Dynamical Theory of Continuity, Discontinuity, and Iteration

An extended exploration of a function operator defined by whether the original function is continuous at each point

Introduction

A surprisingly simple idea can lead to a surprisingly complicated dynamical system.

Start with an ordinary function

f:\mathbb R\to\mathbb R.

At every point x, ask a very basic question:

Is f continuous at x?

Now define a new function by treating continuous and discontinuous points differently:

Kf(x)=
\begin{cases}
2f(x),&f\text{ is continuous at }x,\\[4pt]
f(x),&f\text{ is discontinuous at }x.
\end{cases}

I call this the Kester function or, more precisely, the Kester operator.

The remarkable feature is that the definition of Kf depends on a property of f, namely its continuity at x, rather than simply on the numerical value f(x).

This means that applying the operator changes the function, which can change its continuity set, which changes the way the operator acts on the next iteration.

Consequently,

f,\quad Kf,\quad K^2f,\quad K^3f,\ldots

form a dynamical system.

The central question becomes:

What happens when continuity itself determines the evolution of a function?

This post develops that question from several directions.

1. The Basic Definition

Let

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

be the continuity set of f.

Then the Kester operator can be written compactly as

\boxed{
Kf=(1+\mathbf 1_{C_f})f
}

where \mathbf 1_{C_f} is the indicator function of the continuity set:

\mathbf 1_{C_f}(x)=
\begin{cases}
1,&x\in C_f,\\
0,&x\notin C_f.
\end{cases}

Therefore

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

The operator is therefore self-referential.

It asks about the continuity of f in order to decide how to modify f.

2. The First Important Observation

The Kester operator does not simply multiply the whole function by 2.

Instead, it divides the domain into two dynamically different regions:

C_f

and

D_f=\mathbb R\setminus C_f.

On C_f,

f\mapsto2f.

On D_f,

f\mapsto f.

Thus continuity produces growth, while discontinuity produces freezing.

This gives the operator an intuitive interpretation:

Continuity causes evolution; discontinuity causes inertia.

But this is only the beginning.

The crucial question is what happens to the continuity set after the transformation.

3. A First Surprising Example

Consider

f(x)=
\begin{cases}
2,&x\in B,\\
1,&x\notin B,
\end{cases}

where

B=\{0\}\cup\left\{\frac1n:n\in\mathbb N\right\}.

The set B is closed and has empty interior.

Every point outside B has a neighbourhood avoiding B, so f is locally constant there.

Thus

C_f=\mathbb R\setminus B.

Every point of B is discontinuous because its function value is 2, while nearby points outside B have value 1.

Now apply K.

For x\notin B,

Kf(x)=2(1)=2.

For x\in B,

Kf(x)=f(x)=2.

Therefore

\boxed{Kf(x)=2\quad\text{for every }x.}

So

\boxed{C_{Kf}=\mathbb R.}

The Kester operator has completely repaired every discontinuity.

This is one of the simplest demonstrations that the operator can transform a discontinuous function into a continuous one.

4. Continuity Can Be Created

This leads to a fundamental phenomenon.

Normally, when one modifies a discontinuous function, there is no reason to expect the modification to repair its discontinuities.

The Kester operator does something very particular.

At a discontinuity, it leaves the value unchanged.

At surrounding continuity points, it doubles the values.

If the surrounding values were exactly half the discontinuous value, then doubling the surrounding values makes them match.

That is precisely what happened above:

1\longrightarrow2

on the continuity region, while

2\longrightarrow2

on the discontinuity set.

The discontinuity disappears.

5. The Foundational Itinerary Formula

Now iterate the operator.

Let

f_0=f

and

f_{n+1}=Kf_n.

Let

C_n=C_{f_n}.

At every stage,

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

Define the continuity itinerary exponent

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

In words:

N_n(x) counts how many times x has been a continuity point during the first n stages.

Then we obtain the central algebraic identity

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

This follows directly by induction.

For n=1,

f_1(x)=2^{\mathbf1_{C_0}(x)}f(x).

Assuming

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

we have

f_{n+1}(x)
=
2^{\mathbf1_{C_n}(x)}f_n(x)

and therefore

f_{n+1}(x)
=
2^{\mathbf1_{C_n}(x)}
2^{N_n(x)}
f(x).

Hence

f_{n+1}(x)
=
2^{N_n(x)+\mathbf1_{C_n}(x)}f(x)

and therefore

\boxed{
f_{n+1}(x)=2^{N_{n+1}(x)}f(x).
}

This formula is arguably the central bookkeeping device for the entire theory.

6. The Continuity Itinerary

It is useful to record not merely how many times a point has been continuous, but its entire history.

Define

\omega_n(x)
=
\left(
\mathbf1_{C_0}(x),
\mathbf1_{C_1}(x),
\dots,
\mathbf1_{C_{n-1}}(x)
\right).

This is a binary string.

For example,

(1,1,1,0,0,1)

means that x was:

• continuous at stage 0,
• continuous at stage 1,
• continuous at stage 2,
• discontinuous at stage 3,
• discontinuous at stage 4,
• continuous again at stage 5.

The exponent N_n(x) is simply the number of 1’s in this string.

Thus

N_n(x)=\|\omega_n(x)\|_1.

The Kester dynamics can therefore be viewed as a system of evolving binary itineraries.

7. Finite Kester Staircases

One of the most interesting constructions is a function that heals one discontinuity after another.

Take distinct points

x_1,\dots,x_m.

Define

f(x)=
\begin{cases}
2^j,&x=x_j,\\
1,&x\notin\{x_1,\dots,x_m\}.
\end{cases}

Initially,

C_0
=
\mathbb R\setminus\{x_1,\dots,x_m\}.

The ordinary points double:

1\to2.

The exceptional points remain frozen:

2^j\to2^j.

Therefore x_1, whose value is 2, now matches the surrounding value.

At the next stage, the background doubles again:

2\to4.

Now x_2, whose value is 4, is repaired.

The process continues.

Thus one can obtain

C_0\subsetneq C_1\subsetneq C_2
\subsetneq\cdots\subsetneq C_m=\mathbb R.

So for every positive integer m, there are functions requiring exactly m Kester iterations to become globally continuous.

This gives the finite Kester staircase phenomenon.

8. Infinite Pointwise Healing

The staircase idea can be extended indefinitely.

Consider

f(n)=2^n

at positive integers and

f(x)=1

elsewhere.

Initially,

C_0=\mathbb R\setminus\mathbb N.

The background evolves:

1\to2\to4\to8\to16\to\cdots.

The point 1 becomes compatible with the background after one iteration.

The point 2 becomes compatible after two.

The point 3 after three.

And so on.

At every finite stage, however, infinitely many integer discontinuities remain.

Thus

C_n\neq\mathbb R

for every finite n.

Nevertheless,

\bigcup_{n=0}^{\infty}C_n=\mathbb R.

This distinguishes two very different notions:

Global eventual continuity

There exists some finite n such that

C_n=\mathbb R.

Pointwise eventual continuity

For every x, there exists some finite n=n(x) such that

x\in C_n.

The second can occur without the first.

That distinction is one of the most interesting features of Kester dynamics.

9. A Nontrivial Fixed Point

The operator also possesses nonconstant fixed points.

A particularly beautiful example is Thomae’s function, sometimes called the popcorn function:

T(x)=
\begin{cases}
1/q,&x=p/q\in\mathbb Q
\text{ in lowest terms},\\
0,&x\notin\mathbb Q.
\end{cases}

Its continuity set is

C_T=\mathbb R\setminus\mathbb Q.

At every irrational,

T(x)=0.

Since irrational points are continuity points,

KT(x)=2T(x)=0=T(x).

At rational discontinuity points,

KT(x)=T(x).

Hence

\boxed{KT=T.}

Therefore

\boxed{K^nT=T}

for every n\ge1.

So the Kester operator possesses nonconstant fixed points whose continuity and discontinuity structures remain permanently intertwined.

10. The Operator Is Not Simply a Smoothing Operator

It would be tempting to say:

“Kester dynamics gradually smooths functions.”

That is not correct.

The operator can:

• repair discontinuities,
• preserve discontinuities,
• create new discontinuities,
• produce finite healing staircases,
• produce pointwise but not globally finite healing,
• and possess nontrivial fixed points.

The geometry of the continuity set is therefore essential.

11. The Boundary Phenomenon

A particularly important local fact concerns points lying on the boundary of the continuity set.

Suppose

x\in C_f\cap\partial C_f.

Then x is continuous for f, but arbitrarily close to x there are discontinuity points.

At x,

Kf(x)=2f(x).

But nearby discontinuity points are not multiplied.

If f(x)\neq0, this creates a mismatch.

This gives a mechanism by which the Kester operator can actually create a new discontinuity at a point that was originally continuous.

That observation is extremely important for understanding the dynamics.

The operator therefore does not preserve the continuity set.

12. The Local Oscillation Problem

This raises a deeper question.

Can a point follow an itinerary such as

C_0,\quad D_1,\quad C_2,\quad D_3,\quad C_4,\ldots?

In other words:

Can a point repeatedly enter and leave the continuity set?

This is much harder than constructing a simple one-step repair.

The reason is that a point’s value evolves according to its history:

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

But nearby points can have different histories.

Therefore the local behaviour of f_n is controlled by an entire collection of itinerary classes.

For a two-step analysis, there are four possible histories:

(1,1),\quad
(1,0),\quad
(0,1),\quad
(0,0).

Their multipliers are respectively

4,\quad2,\quad2,\quad1.

So if a point has itinerary

(1,0),

then its value at stage 2 is

2f(x).

For that point to become continuous, nearby points must have appropriately chosen initial values and itineraries so that

2^{N_2(t)}f(t)\to2f(x).

This is a highly constrained local matching problem.

13. The Itinerary View Changes the Problem

The original question was:

Is K^n f continuous?

The more refined question is:

How are continuity histories distributed around each point?

This converts the problem from ordinary analysis into something resembling a symbolic dynamical system.

Each 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}

The value evolution is then controlled by the number of 1’s.

This suggests a possible future theory of Kester symbolic dynamics.

14. Generalising the Factor 2

There is nothing mathematically sacred about the number 2.

We can define a more general operator

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

For the original Kester operator,

a=2.

The iteration formula becomes

\boxed{
K_a^n f(x)=a^{N_n(x)}f(x).
}

This immediately produces a much larger family of dynamical systems.

For example:

a=3

gives

K_3f(x)=
\begin{cases}
3f(x),&f\text{ continuous at }x,\\
f(x),&f\text{ discontinuous at }x.
\end{cases}

The staircase construction then uses values

3,\;3^2,\;3^3,\ldots

instead of

2,\;2^2,\;2^3,\ldots.

More generally, for any a,

1,\;a,\;a^2,\;a^3,\ldots

becomes the natural scale of the dynamics.

15. The Complex Kester Operator

The really interesting generalisation comes when the functions are complex-valued.

Let

f:X\to\mathbb C.

Define

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

where

\lambda\in\mathbb C.

The iteration formula is

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

This transforms Kester dynamics from purely amplitude-based behaviour into potentially genuine phase dynamics.

16. Real Multipliers

Suppose

\lambda>1.

Then every continuity event increases the magnitude.

For example,

\lambda=3

gives

f_n(x)=3^{N_n(x)}f(x).

If a point is continuous at every stage, then

f_n(x)=3^nf(x).

The value grows exponentially.

If a point is never continuous, then

f_n(x)=f(x).

It remains frozen.

Thus the operator separates points into different growth histories.

17. Contracting Multipliers

Now consider

0<|\lambda|<1.

The dynamics reverse.

Every continuity event decreases the magnitude.

For example,

\lambda=\frac12.

Then

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

A point that is continuously updated infinitely often can be driven toward zero.

Thus the same topological mechanism can generate either:

• exponential growth when |\lambda|>1,
• exponential decay when |\lambda|<1,
• or magnitude preservation when |\lambda|=1.

18. The Unit Circle: Pure Phase Dynamics

The most interesting case is

|\lambda|=1.

Write

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

Then

K_\lambda^n f(x)
=
e^{2\pi i\alpha N_n(x)}f(x).

Since

\left|e^{2\pi i\alpha N_n(x)}\right|=1,

we obtain

\boxed{
|f_n(x)|=|f(x)|.
}

The magnitude no longer changes.

Only the phase changes.

This produces an entirely different dynamical picture.

19. Rational Phase Rotation

Suppose

\alpha=\frac pq

with p,q\in\mathbb Z.

Then

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

Because

\lambda^q=1,

the phase evolution is periodic.

If a point experiences infinitely many continuity events, its possible phase values belong to the finite set

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

Thus the orbit lies on a finite subset of the circle

|z|=|f(x)|.

For example, take

\lambda=i.

Then

i^4=1.

A continuously updated point can follow

f(x)
\to if(x)
\to -f(x)
\to -if(x)
\to f(x)
\to\cdots

This produces a four-cycle.

20. Irrational Phase Rotation

Now suppose

\alpha\notin\mathbb Q.

Then

e^{2\pi i\alpha}

is an irrational rotation of the unit circle.

If the continuity history supplies every integer exponent sufficiently often—in particular, if

N_n(x)

takes all sufficiently large integer values—then

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

is dense in the unit circle.

Thus the orbit of a nonzero value can become dense on

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

This gives a striking interpretation:

The Kester operator can turn continuity events into angular rotations.

The topology of the continuity set determines when rotations occur.

The arithmetic of \alpha determines the geometry of the resulting orbit.

21. Kester Dynamics as a Skew Product

This suggests an abstract viewpoint.

Each point x carries two pieces of information:

1. its original value f(x),
2. its continuity itinerary.

The itinerary determines the exponent

N_n(x).

The exponent determines the multiplier

\lambda^{N_n(x)}.

So the system can be viewed schematically as

x
\longrightarrow
\omega(x)
\longrightarrow
N_n(x)
\longrightarrow
\lambda^{N_n(x)}
\longrightarrow
f_n(x).

This separates the topological dynamics from the algebraic dynamics.

22. Fixed Points of the Complex Kester Operator

Consider

K_\lambda f=f.

At a discontinuity point,

K_\lambda f(x)=f(x)

automatically.

Therefore discontinuity points impose no algebraic restriction.

At a continuity point,

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

For this to equal f(x),

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

Hence

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

Therefore, if

\lambda\neq1,

then every continuity point of a fixed function must satisfy

\boxed{f(x)=0.}

This gives a useful fixed-point criterion.

23. Fixed Points for \lambda\neq1

For a nontrivial multiplier,

\lambda\neq1,

a fixed function must satisfy

f(x)=0

at every continuity point.

Its nonzero values can therefore survive only on its discontinuity set.

Thomae’s function provides a beautiful example because

T(x)=0

at every irrational continuity point.

Thus

K_\lambda T=T

for every complex \lambda.

This is stronger than the original factor-2 example.

24. Periodic Points

Fixed points are only period one.

We can ask for

K_\lambda^m f=f.

At a point that experiences N_m(x) continuity events during the cycle,

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

Therefore, for a nonzero value,

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

Thus roots of unity naturally generate periodic behaviour.

For example,

\lambda=i.

A point that is continuously updated four times obtains

i^4f(x)=f(x).

This creates a natural route to periodic Kester dynamics.

25. A Caution About Periodicity

There is an important distinction here.

The condition

\lambda^m=1

does not automatically guarantee that the entire function has period m.

The continuity sets themselves can change.

For a genuine periodic orbit of functions, we need

K_\lambda^m f=f

as functions, not merely pointwise phase rotation under a predetermined itinerary.

Therefore periodicity requires both:

• algebraic compatibility of the multiplier,
• and dynamical compatibility of the continuity sets.

This is another reason the problem is richer than ordinary multiplication.

26. The General a=3 Kester Function

Returning to your suggestion of replacing 2 by 3, define

K_3f(x)=
\begin{cases}
3f(x),&f\text{ is continuous at }x,\\
f(x),&f\text{ is discontinuous at }x.
\end{cases}

Then

K_3^n f(x)=3^{N_n(x)}f(x).

A staircase can now be built using

3,\;9,\;27,\;81,\ldots

rather than

2,\;4,\;8,\;16,\ldots.

For example, define isolated exceptional points x_1,x_2,x_3 with

f(x_1)=3,\qquad
f(x_2)=9,\qquad
f(x_3)=27,

and let the background equal 1.

Then the background evolves

1\to3\to9\to27.

The exceptional points can therefore be incorporated one layer at a time.

The same topological staircase mechanism survives.

27. Generalisation to Arbitrary Multipliers

The natural definition is therefore

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

This creates an entire family of Kester operators

\{K_\lambda:\lambda\in\mathbb C\}.

The original Kester operator is simply

K=K_2.

The associated iteration formula is

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

This may be a more natural mathematical object to study than a single operator.

28. Three Dynamical Regimes

The multiplier naturally divides the theory into three broad regimes.

Expanding regime

|\lambda|>1.

Continuity events increase magnitude.

Contracting regime

|\lambda|<1.

Continuity events decrease magnitude.

Neutral regime

|\lambda|=1.

Continuity events preserve magnitude and may rotate phase.

This creates a surprisingly familiar dynamical-systems structure.

29. The Zero Set

The set

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

is particularly important.

At a zero,

K_\lambda f(x)=0

regardless of whether x is continuous or discontinuous.

Thus the algebraic update cannot move the value away from zero.

This makes zero-valued points special candidates for complicated continuity behaviour.

The zero set therefore deserves to be treated as a dynamical object in its own right.

30. Pointwise Healing Time

For each point x, define its Kester healing time

\tau_f(x)
=
\inf\{n\ge0:x\in C_n\},

with

\tau_f(x)=\infty

if x never becomes continuous.

This provides a map

\tau_f:\mathbb R\to\mathbb N_0\cup\{\infty\}.

A function can therefore be viewed as carrying a “continuity clock” at every point.

For a finite staircase,

\tau_f(x_j)=j.

For the infinite integer example,

\tau_f(n)=n.

For Thomae’s function,

\tau_f(x)=0

for irrational x, while rational points have

\tau_f(x)=\infty.

This single function records an enormous amount of dynamical information.

31. Three Basic Classes

The theory naturally suggests three broad classes.

Class I — Already continuous

C_f=\mathbb R.

Then

K_\lambda f=\lambda f

and therefore

K_\lambda^n f=\lambda^n f.

Class II — Eventually globally continuous

There exists finite m such that

C_m=\mathbb R.

After that point,

K_\lambda^{m+r}f
=
\lambda^rK_\lambda^mf.

Class III — Never globally continuous

For every finite n,

C_n\neq\mathbb R.

This third class contains both permanently discontinuous systems and systems with complicated partial healing.

32. Pointwise Versus Global Dynamics

The distinction between

\forall x\,\exists n(x)

and

\exists n\,\forall x

is fundamental.

Pointwise eventual continuity means

\forall x\in\mathbb R,\quad
\exists n=n(x)
\text{ such that }x\in C_n.

Global eventual continuity means

\exists n\in\mathbb N,\quad
C_n=\mathbb R.

The order of the quantifiers is reversed.

The infinite staircase demonstrates that these are genuinely different concepts.

33. Toward Kester Entropy

The itinerary

\omega(x)

suggests another possible invariant.

Instead of asking only how many times a point has been continuous, we could ask how complicated the collection of itineraries is across the domain.

For example, how many different strings of length n occur?

Define

\mathcal W_n
=
\{\omega_n(x):x\in\mathbb R\}.

One could investigate the growth of

|\mathcal W_n|.

If this grows rapidly, the Kester system has increasingly complicated continuity histories.

This suggests a possible analogue of symbolic complexity or topological entropy.

Whether such a theory is useful remains an open direction.

34. A Possible Kester Spectrum

For the generalized operator K_\lambda, one could also classify functions according to their long-term multiplier behaviour.

At a point,

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

If

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

then for |\lambda|\neq1,

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

Taking logarithms gives

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

If the limit exists,

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

This looks very much like a pointwise Lyapunov exponent.

One could therefore define a Kester growth exponent

\chi_f(x)
=
\rho(x)\log|\lambda|.

35. The Neutral Case Is Different

When

|\lambda|=1,

the growth exponent is zero.

But this does not mean the dynamics are trivial.

For

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

we have

K_\lambda^nf(x)
=
e^{2\pi i\alpha N_n(x)}f(x).

The magnitude is fixed while the phase may move forever.

Thus the neutral regime replaces exponential growth with rotational dynamics.

This resembles the distinction between expanding and elliptic behaviour in other dynamical systems.

36. The Geometry of Phase Orbits

Suppose

f(x)\neq0

and N_n(x) grows through all sufficiently large integers.

Then the orbit

\{K_\lambda^nf(x):n\ge0\}

lies on the circle

|z|=|f(x)|.

If \alpha is rational, only finitely many points are visited.

If \alpha is irrational, the orbit can be dense in that circle.

Thus a purely topological question—

When is x continuous?

—can generate a geometric orbit in the complex plane.

37. A New View of the Kester Operator

At this point, the Kester operator can be interpreted as combining three different mathematical structures:

Topology

Continuity determines the active set.

Algebra

The active set determines multiplication by \lambda.

Dynamics

The resulting function determines a new continuity set.

Thus the feedback loop is

\boxed{
\text{continuity}
\longrightarrow
\text{multiplication}
\longrightarrow
\text{new function}
\longrightarrow
\text{new continuity}.
}

That feedback loop is the defining feature of Kester dynamics.

38. What Remains Unknown?

Several important questions remain open.

Can a point genuinely oscillate

C_0\to D_1\to C_2\to D_3?

Can it have an infinite nonperiodic continuity itinerary?

Can the set of oscillating points be dense?

Can a Kester operator possess nontrivial cycles of functions?

Can the continuity-set dynamics itself have positive entropy?

Can one construct functions with prescribed healing-time functions

\tau_f(x)?

Can every sufficiently complicated binary itinerary be realised by some function?

These questions move the subject beyond elementary examples.

39. The Local Realisability Problem

A particularly natural next theorem would concern the question:

Given a desired finite continuity itinerary around a point, can we construct a function that realises it?

For example, can we construct

(1,0,1)?

at a prescribed point?

For a target point x, the condition at stage 2 is

f_2(t)
=
\lambda^{N_2(t)}f(t).

Continuity requires

\lim_{t\to x}
\lambda^{N_2(t)}f(t)
=
\lambda^{N_2(x)}f(x).

The challenge is therefore to construct neighbourhoods in which different itinerary classes carry precisely the compensating values required by this equation.

This is a promising route toward a Local Realisability Lemma.

40. A Possible Research Programme

The theory can now be developed systematically.

Problem 1: Itinerary realisability

Which binary sequences can occur as

\omega(x)?

Problem 2: Healing-time classification

Which functions

\tau:\mathbb R\to\mathbb N_0\cup\{\infty\}

can occur as Kester healing-time functions?

Problem 3: Oscillation

Can one construct

1,0,1,0,\ldots

or other infinite continuity itineraries?

Problem 4: Periodicity

Classify functions satisfying

K_\lambda^mf=f.

Problem 5: Phase dynamics

For |\lambda|=1, classify orbit closures.

Problem 6: Complexity

Study

|\mathcal W_n|

and possible entropy-like invariants.

Problem 7: General weights

Replace the constant multiplier \lambda with a function

g(x).

For example,

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

This creates an even larger class of self-referential operators.

41. The General Weighted Kester Operator

The most general version considered here is

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

The original Kester function corresponds to

g(x)=2.

The “three-times” version corresponds to

g(x)=3.

The complex phase version corresponds to

g(x)=e^{2\pi i\alpha}.

A variable weight allows the local geometry and local amplitude to interact simultaneously.

The resulting theory could be substantially richer.

42. Why the Kester Function Is Interesting

The formula itself is elementary.

That is part of what makes it interesting.

The complexity does not come from an elaborate equation.

It comes from allowing the function to examine its own continuity structure before deciding how to change.

The operator therefore creates a feedback system:

f
\rightarrow C_f
\rightarrow Kf
\rightarrow C_{Kf}
\rightarrow K^2f
\rightarrow\cdots

The function generates its own future rule.

43. A Working Terminology

For future development, the following terminology could be useful:

Kester operator

Kf=(1+\mathbf1_{C_f})f.

Kester orbit

f,Kf,K^2f,\ldots

Continuity itinerary

\omega(x).

Itinerary exponent

N_n(x).

Kester healing time

\tau_f(x).

Kester staircase

A finite sequence of strict continuity-set inclusions ending in \mathbb R.

Pointwise eventual continuity

Every point eventually becomes continuous.

Global eventual continuity

The entire function becomes continuous after finitely many iterations.

Kester fixed point

Kf=f.

Kester periodic point

K^mf=f

for some m>1.

Complex Kester operator

K_\lambda.

Kester phase orbit

The orbit generated by a unit-modulus complex multiplier.

44. Final Perspective

What began as the simple definition

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

quickly leads to a much broader mathematical framework.

The operator can:

• repair discontinuities,
• preserve discontinuities,
• create discontinuities,
• generate finite healing staircases,
• generate infinite pointwise healing,
• possess nonconstant fixed points,
• encode binary continuity histories,
• exhibit growth and decay,
• generate complex phase rotations,
• produce periodic phase behaviour,
• and potentially support complicated itinerary dynamics.

The fundamental formula

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

shows that the entire numerical evolution at a point is encoded by its continuity history.

This gives a particularly striking interpretation of the system:

In Kester dynamics, the history of continuity becomes a dynamical variable.

The function does not merely have a continuity set.

Its continuity set evolves.

And that evolving topology determines the future values of the function.

That is the central idea behind the Kester dynamical system.

A final conjectural picture

The long-term research programme can perhaps be summarised by the diagram

\boxed{
\text{Function}
\rightarrow
\text{Continuity Geometry}
\rightarrow
\text{Itinerary}
\rightarrow
\text{Multiplier}
\rightarrow
\text{New Function}
\rightarrow
\text{New Geometry}.
}

The original factor 2 is only the first member of the family.

Replacing it by 3, by a real parameter a, by a complex number \lambda, or ultimately by a variable weight g(x), turns the original construction into a much broader class of self-referential continuity operators.

The deepest unanswered question may therefore no longer be simply:

What does the Kester function do?

but rather:

Which dynamical systems can be generated when a function’s own topological regularity determines its evolution?

That question is potentially much larger than the original example.

This is an exploratory mathematical programme rather than a claim that all of the conjectures above have already been proved. In particular, the proposed oscillation and entropy questions should be treated as open problems until explicit constructions or rigorous impossibility proofs are established.