On the Sign Change of a Continuous Function

Suzy Manuela Prajea

November 8, 2021

Abstract
The behavior of the functions is the main goal when study real-valued functions. It is known
that continuity does not imply monotony as well as monotony does not imply continuity. Any-
way, if a continuous function changes from negative to positive values, a glimpse of increasing
behavior might be questioned. This article provides a few answers to this concern and estab-
lish the existence of an interval of ’local’ monotony that holds in certain plain conditions of
differentiability. From a didactic perspective, the narrative intends to support the AP Calculus
students with a better understanding of the continuity and differentiability of the functions.

1
I dedicate this paper to my beloved students from the AP Calculus classes at J. L. Chambers High
School, as a sign of deep appreciation for their engagement to decipher the mystery of Calculus in
spite of all the odds.
On the Sign Change of a Continuous Function Suzy Manuela Prajea

1 Preliminaries
The introductory section presents the theme of the study and the terminology used throughout the
article. If nothing else is specified, real-valued functions defined on an non-empty open interval as
well as open intervals will be considered in the following sections.
Definition 1.1. Let be f a real-valued function defined on the interval (a, b).

(1) The function f changes from negative to positive values on the interval (a, b) if there is a
number c ∈ (a, b) s.t.:

(i) f (x) < 0 for x ∈ (a, c)

(ii) f (c) = 0

(ii) f (x) > 0 for any x ∈ (c, b)

(see Figure 1)

(2) The function f changes from positive to negative values on the interval (a, b) if there is a
number c ∈ (a, b) s.t.:

(i) f (x) < 0 for x ∈ (a, c)

(ii) f (c) = 0

(ii) f (x) > 0 for any x ∈ (c, b)

(3) The function f changes the sign at x = c on the interval (a, b) if the function changes from
positive to negative values or from negative to positive values.
Example:

Figure 1: Example of function that changes from negative to positive values on the interval (a, b)

Definition 1.2. Let be c a real number. A neighborhood of the point c is an open interval containing
the point c.

Ph.D. in Mathematics, J.L.Chambers, Charlotte, NC, USA 3

On the Sign Change of a Continuous Function Suzy Manuela Prajea

Figure 2: The interval (−2, 1) is neighborhood of any of its points

Example: The interval (-2, 1) is a neighborhood of 0.

*Note: It can be proved that an open interval is a neighborhood of any of its points.
Definition 1.3. Let be f a real-valued function defined on the interval (a, b). The function f is
called

(i) increasing on the interval (a, b) iff for any x, y ∈ (a, b), x < y, it results f (x) ≤ f (y)

(ii) strictly increasing on the interval (a, b) iff for any x, y ∈ I, x < y, it results f (x) < f (y)

(iii) decreasing on the interval (a, b) iff for any x, y ∈ (a, b), x < y. it results f (x) ≥ f (y)

(iv) strictly decreasing on the interval (a, b) iff for any x, y ∈ (a, b), x < y, it results f (x) > f (y)

(v) monotonic on the interval (a, b) iff the function f is increasing on the interval (a, b) or
decreasing on decreasing on the interval (a, b)

(vi) strictly monotonic on the interval (a, b) iff the function f is strictly increasing on the in-
terval (a, b) or strictly decreasing on the interval (a, b)

Figure 3: Example of function that is strictly increasing on the interval (5, 10)

Ph.D. in Mathematics, J.L.Chambers, Charlotte, NC, USA 4

On the Sign Change of a Continuous Function Suzy Manuela Prajea

Definition 1.4. Let be f a real-valued function defined on the open interval I. Define the difference
quotient of the function f on the non-empty interval (x, y) ⊆ I as:

f (y) − f (x)
R(x, y) =
y−x

*Note that R(x, y) represents the slope of the secant line determined by the points (x, f (x)) and
(y, f (y)) of the graph of the function f (see Figure 4).

Figure 4: The difference quotient R(x, y) is the slope of the secant line AB to the graph of the
function f and R(x, y) = tan α

Lemma 1.1. Let be f a real-valued function defined on the open interval I. The following holds:

(i) the function f is increasing on the interval I iff for any x, y ∈ I it results R(x, y) ≥ 0

(ii) the function f is strictly increasing on the interval I iff for any x, y ∈ I it results R(x, y) > 0

(iii) the function f is decreasing on the interval I iff for any x, y ∈ I it results R(x, y) ≤ 0

(iv) the function f is strictly decreasing on the interval I iff for any x, y ∈ I it results R(x, y) < 0

Proof. The proof is immediate from the Definition 1.3.

Ph.D. in Mathematics, J.L.Chambers, Charlotte, NC, USA 5

On the Sign Change of a Continuous Function Suzy Manuela Prajea

Remark 1.1. In the context of Definition 1.1, the following holds:

If a function f changes from negative to positive values then a certain increase tendency in the
values of the function can be described by:

(∀)x, y ∈ (a, b), x < c < y ⇒ f (x) < 0 < f (y) (1)

Proof. (1) is a reformulation of the conditions (i)-(iii) from Definition 1.1

*Note: The characterization (1) does not describe the increasing behavior of the function f on
the interval (a, b) in the sense of Definition 1.3 (i). In fact, the characterization (1) is a weaker
condition than the condition (i) from Definition 1.3 because it establishes the inequality f (x) < f (y)
for any x < c < y and not for any x < y of the interval (a, b) as in Definition 1.3 (i).

The following definition will enable a specific terminology for the behavior described in (1).
Definition 1.5. Let be f a real-valued function defined on the interval I and c a point of this
interval. The function f is called:

(i) locally increasing at the point c iff there is a neighborhood (a, b) of c, (a, b) ⊆ I s.t.

(∀)x, y ∈ (a, b), x < c < y ⇒ f (x) ≤ f (c) ≤ f (y)

(ii) locally strictly increasing at the point c iff there is a neighborhood (a, b) of c, (a, b) ⊆ I s.t.

(∀)x, y ∈ (a, b), x < c < y ⇒ f (x) < f (c) < f (y)

(iii) locally decreasing at the point c iff there is a neighborhood (a, b) of c, (a, b) ⊆ I s.t.

(∀)x, y ∈ (a, b), x < c < y ⇒ f (x) ≥ f (c) ≥ f (y)

(iv) locally strictly decreasing at the point c iff there is a neighborhood (a, b) of c, (a, b) ⊆ I s.t.

(∀)x, y ∈ (a, b), x < c < y ⇒ f (x) > f (c) > f (y)

Definition 1.6. Let be f a real-valued function defined on the interval I and c a point of this
interval. The function f is called:

(i) locally monotonic on the interval (a, b) iff the function f is locally increasing on the interval
(a, b) or the function f is locally decreasing on the interval (a, b)

(ii) locally strictly monotonic on the interval (a, b) iff the function f is locally strictly increasing
on the interval (a, b) or the function f is locally strictly decreasing on the interval (a, b)

Proposition 1.1. Let be f a function defined on the interval (a, b). The following holds:

(i) if f is increasing on the interval (a, b) then f is locally increasing at every point c of the
interval (a, b)

Ph.D. in Mathematics, J.L.Chambers, Charlotte, NC, USA 6

On the Sign Change of a Continuous Function Suzy Manuela Prajea

Figure 5: Example of function that is locally strictly increasing at x = c and not increasing on the
interval (a, b)

(ii) if f is strictly increasing on the interval (a, b) then f is locally strictly increasing at every
point c of the interval (a, b)

(iii) if f is decreasing on the interval (a, b) then f is locally decreasing at every point c of the
interval (a, b)

(iv) if f is strictly decreasing on the interval (a, b) then f is locally strictly decreasing at every
point c of the interval (a, b)

Proof. (i) Let be c an arbitrary point from (a, b). It is required to prove that f is locally increasing
at the point c.
From f increasing on the interval (a, b), for any x from (a, b), x < c, it results f (x) ≤ f (c).
Also, again from f increasing on the interval (a, b), for any y from (a, b), c < y, it results f (c) ≤ f (y).
In conclusion, for any x, y from (a, b) with x < c < y it results f (x) ≤ f (c) <≤ f (y) i.e. the function
f is locally increasing at the point c.
(ii)-(iv) analogously with (i)

It seems natural to question an eventual potential of increasing in the behavior of the function
f in the hypothesis of Definition 1.3 (1) i.e. the function f changes from negative to positive values
on the interval (a, b):

(∗) There is a neighborhood of the point c on that the function f is increasing?

(∗∗) There is an interval in the neighborhood of c, not necessarily containing the point c, on
that the function f is increasing?

where c is the zero of the function f as in the Definition 1.3 (1).

Finally, what will be the answers to the questions above, if specific conditions of continuity or
differentiability are imposed to function f ?

Ph.D. in Mathematics, J.L.Chambers, Charlotte, NC, USA 7

On the Sign Change of a Continuous Function Suzy Manuela Prajea

2 Exploratory Examples
If f is a function that changes from negative to positive values, the answers to the questions (∗)
and (∗∗) are both negative, as provided by the next example.
Example 2.1. Consider f the piecewise function defined by:

−x − 1 0 < x < 2

f (x) = 0 x=2

−x + 4 2<x<4


Figure 6: f changes from negatives to positives and f is not increasing on any interval that contains
x=2
Indeed, the function f has the following properties:
→ f is strictly decreasing on each of the intervals (0, 2) and (2, 4)
→ f changes from negatives to positives values in the sense of Definition 1.1 i.e.
f (x) < 0, 0<x<2
f (2) = 0
f (x) > 0, 2<x<4
→ f is not increasing on any interval included in its domain of definition.
→ f is locally strictly increasing at x = 2
→ f is locally strictly decreasing at each point of the intervals (0, 2) and (2, 4)

The properties of the function from Example 2.1 justify why the answers to the questions:

(∗) There is a neighborhood of the point c on that the function f is increasing?

(∗∗) There is an interval in the neighborhood of c, not necessarily containing the point c, on
that the function f is increasing?

Ph.D. in Mathematics, J.L.Chambers, Charlotte, NC, USA 8

On the Sign Change of a Continuous Function Suzy Manuela Prajea

are both negative.

Conclusion 2.1.1. A function f that changes from negative to positive values on the interval (a, b)
does not necessarily have an interval on that it is increasing. (see Figure 6)
Examples 1.1 and 2.1 intuitively support the idea that a function changing from negative to
positive values might involve an eventual potential of increasing behavior on a neighborhood of
the point where it changes the sign only if its "jumping" behavior is canceled. Most precisely, any
removable or non-removable point of discontinuity is not welcomed in the search of an interval of
increasing for the function due to the fact that an eventual increasing behavior of the function can
be easily troubled in a point of discontinuity. The condition of continuity seems to be essential
in the searching of an interval of increase for the functions that changes from negative to positive
values i.e. for functions locally increasing at a point.

Back to the intuitive image of a continuous function that changes from negative to positive val-
ues as in the Figures 1 and 7, at the first glance it seems obvious that the answers to the questions:

(∗) There is a neighborhood of the point c on that the function f is increasing?

(∗∗) There is an interval in the neighborhood of c, not necessarily containing the point c, on
that the function f is increasing?

are YES (Figure 1) and respectively YES. (see Figure 7).

Figure 7: f changes from positives to negatives and f is increasing on an interval that contains the
point c

Motto: What seems obvious can be misleading ...

Example 2.2. Consider the piecewise function f defined by:
  
sin 1

x ̸= 0
f (x) = x

0 x=0

(see Figure 8)
This well-known example (see proofs in [2], [3], [4] or [7]) enables the following:

Ph.D. in Mathematics, J.L.Chambers, Charlotte, NC, USA 9

On the Sign Change of a Continuous Function Suzy Manuela Prajea

Conclusion 2.2.1. The following are true:

1 1 1
(i) the points where the function f changes from negative to positive values are x = ± , ± , ± , ...
π 3π 5π
(ii) the function f is continuous on (−∞, 0) ∪ (0, ∞)

(iii) for any integer k, there is an interval that contain the point x = (2k + 1)πon that the func- 
2 2
tion f is strictly increasing. More precisely, f is strictly increasing on the interval , .
(4k + 3)π (4k + 1)π
1 1 1
In consequence, the answers to the questions (∗) and (∗∗) for the points x = ± , ± , ± , ...
π 3π 5π
are both positive. (see Figure 8)
Proof. see [2], [3], [4] or [7]

*Note: What about the answers to the questions (∗) and (∗∗) when c = 0?

Motto: The most interesting part of the investigation it’s just about to start ...

Figure 8: f discontinuous and does NOT change from positives to negatives or from negatives to
positives at 0, in the sense of Definition 1.1

Also, Figure 8 suggests that the pattern from Conclusion 2.2.1. continues to appear for any
integer k. One might say that for k −→ ∞, the claim referring to the existence of a neighborhood
on that the function is increasing in the sense of the questions (∗) and (∗∗) seems to remain valid
at c = 0, i. e. there is a neighborhood of c = 0 on that the function f is increasing.

Motto: When technology fails to see...

1
It is true that when k −→ ∞, it results −→ 0 and this implies sin ((2k + 1)π) −→ 0.
(2k + 1)π  
2 2
On the other hand, the corresponding neighborhood of increasing , −→
(2k + 5)π (2k + 3)π
[0, 0] = 0 is not an open interval, so basically the expected neighborhood of c = 0 on that the
function would be increasing does not exist.

Ph.D. in Mathematics, J.L.Chambers, Charlotte, NC, USA 10

On the Sign Change of a Continuous Function Suzy Manuela Prajea

Finally, the answers at the questions (∗) and (∗∗) at the point c = 0 are given by:

Conclusion 2.2.2. At x = 0, the following are true:

(i) the function f does NOT change from negatives to positives values or from positive to nega-
1 1 1
tive values at x = 0 as it changes at x = ± , ± , ± , ...
π 3π 5π
(ii) the function f is NOT continuous at x = 0

(iii) there is no open interval that contains x = 0 on that the function f is increasing

(iv) any interval that contains x = 0 has an subinterval on that the function f is increasing as
well a subinterval on that the function f is decreasing. (see Example 2.2, Figure 8)
*Note: Due to Conclusion 2.2.2.(i), the case c = 0 does not correspond to the initial declared
investigation, so the answer to the questions (∗) and (∗∗) is not an interest at this moment. For
this reason, (iii) and (iv) will not be proved here but a detailed proof can be found in [2], [3] or [4].

The question that arises after Conclusion 2.2.2. is referring to continuity: if the function f from
Example 2.2 would be continuous overall its domain, then Conclusion 2.2.1.(iii) can be extended
at the point x = 0? The answer suggested by the limits-considerations above seems to be negative.
Indeed ...

The next Example 2.3 will confirm the fact that even the function is continuous overall, then
it does not necessary exists a neighborhood of x = 0 where the function changes from negatives to
positives or positives to negatives.
Example 2.3. Consider the piecewise function f defined by:
  
x sin 1

x ̸= 0
f (x) = x

0 x=0
(see Figure 9)
Intuitive, the monotonic patterns in the behavior of the function f can be noticed in any open
interval centered at x = 0 (Figure 9).
 
2 2
Indeed, it is not difficult to observe that the function f is increasing on any interval ,
  (4k + 3)π (4k + 1)π
2 2
and decreasing on any interval , , for any integer k.
(4k + 1)π (4k − 1)π
Example 2.3 provides reason to formulate the following:
Conclusion 2.3.1. If a non-constant continuous function defined on an open interval has a zero at
x = c and in any interval that contains x = c the function takes positive and negatives values, then
the function NOT necessarily changes the sign from negatives to positives values or from positive
to negatives values at x = c in the sense of Definition 1.1. (see Example 2.3, Figure 9)

Ph.D. in Mathematics, J.L.Chambers, Charlotte, NC, USA 11

On the Sign Change of a Continuous Function Suzy Manuela Prajea

Moral: What we see (Figure 1) is not all what it is (Figure 9) ...

... and Conclusion 2.3.1 justify more clearly now the reason of introducing Definition 1.1.

* Note
In other terms said, if the Figure 1 would represent the graph of any continuous function that
changes from negative to positive values in an appropriate neighborhood of its zero, what sense
could one find in the introduction of the Definition 1.1? Obviously, it would be no sense to
introduce the Definition 1.1 if the scenarios described in the Definition 1.1 would be the only
options for a continuous function to change the sign in an appropriate neighborhood of its zero.
Examples 2.3.1 shows that the Definition 1.1 makes sense for continuous functions because
continuous functions may have zeros not necessarily described by a neighborhood where the
function changes from negative to positive values or from positive to negative values as in the
scenarios described by Definition 1.1.

Examples 2.2 and 2.3 prove that if a continuous or discontinuous function has a zero then the
function does not necessarily change from negative to positive values or from positive to negative
values on a neighborhood of its zero. This unpredictable behavior surprises the mind because one
of the first insights that one gets when starting the learning of continuous functions is the
intuitive presentation of a continuous function over an interval: a function whose graph can be
drawn without lifting the pen from the paper. In this context, it seems obvious that the graph of
a continuous function that changes from negative to positive values can be drawn without lifting
the pen from the paper in an approximate manner as in Figure 1. The intuitive drawing pen
model might trouble seriously the perception of continuity if the learning of the continuity will not
be backed up by accurate mathematical definitions.

Figure 9: f continuous and f does NOT change from positives to negatives or from negatives to
positives at the point x = 0 (in the sense of Definition 1.1)

Mathematical analysis taught us that beyond intuition, an unexplored universe might be hidden.
For instance, the next conclusion. Before that, a fascinating animation provided by the
magnification of the Koch curve: Koch Snow Flake reminds of the self-similarity of the fractals
when ’zooming in’ the graph of the function from Example 2.3 in the neighborhood of x = 0.

Ph.D. in Mathematics, J.L.Chambers, Charlotte, NC, USA 12

On the Sign Change of a Continuous Function Suzy Manuela Prajea

If Weierstrass Function is considered the very first beginning of fractals, then a function as one
from Figure 9 might be considered the ancestor of Weierstrass function.
In consequence, the following conclusion can be enabled:
Conclusion 2.3.2. Example 2.3 shows that even though the function is continuous, its behavior
in the neighborhood of one of its zeros can be:

⋄ unpredictable (referring at the graph from Figure 9 in the neighborhood of x = 0 versus the
’foreseeing’ graphs, Figure 1 in the neighborhood of x = c)

⋄ erratic (referring at the intervals of monotony that appears in any neighborhood of x = 0, Figure
9)

⋄ fascinating (referring at the self-similarity of the graph in the neighborhood of x = 0; imagine

the perfect ’zoom in’ animation technology of the function represented in Figure 9)
Consider a non-continuous function that has zeros, not necessarily changing the sign in a
neighborhood of its zeros. The weirdness of the behavior of this function from the perspective of
the existence of an interval of monotony in the neighborhood of those zeros, it does not end with
the above examples and the erratic behavior as it is described in Conclusion 2.3.2.
There are continuous functions that have such an erratic behavior in the neighborhoods of its
zeros (and not only there) that it does not admit not even a single interval of monotony in these
neighborhoods. Here it comes the famous Weierstrass function, an intricate real-valued function
that disrupt the terms in that the continuity of functions was seen in the 19th century. 1 . It is the
very first example of fractal curve and it marked a new milestone in the further development of
the mathematical analysis.

Example 2.4. (The Weierstrass Function)

Let be the function
∞
X
f (x) = an cos (bn πx)
n=0

3π
for 0 < a < 1 and b an odd positive integer greater than 1 s.t. ab > 1 + .
2
This function is everywhere continuous but nowhere differentiable.

See the graph here: the Weierstrass Function on the interval [−2, 2].

More than this, Weierstrass function has a ’fickle’ behavior at any point c, i.e.

there are two sequences ym < c < zm with ym −→ c, zm −→ c as m −→ ∞ s. t. the difference

quotients R(ym , c) and R(zm , c):

1 "deplorable evil", "lamentable scourge", "pathological function"

Ph.D. in Mathematics, J.L.Chambers, Charlotte, NC, USA 13

On the Sign Change of a Continuous Function Suzy Manuela Prajea

(i) have oscillating opposite signs for any m

(ii) do not approach 0 as m −→ ∞ (2)

The sequences described in (2) (i)-(ii) are the core of the proof that Weierstrass function is
nowhere differentiable. The proof can be found in [5].

From the perspective of the interest of this paper, at a closer look, the (2) (i)-(ii) reveals that
Weierstrass function is nowhere locally monotonic in the sense of Definition 1.5 introduced in
section 1.Indeed...

Proposition 2.1. The Weierstrass function

∞
X
f (x) = an cos (bn πx)
n=0

3π
for 0 < a < 1 and b an odd positive integer greater than 1 s.t. ab > 1 + ,
2
has the following properties:

(i) f is not monotonic on any interval

(ii) f is not locally monotonic at any point

Proof. (i) Assuming by contradiction that there is an interval (a, b) on that f is monotonic.
To fix the ideas, consider a point c in the interval (a, b). Notice that the interval (a, b) is a
neighborhood of the point c in the sense of Definition 1.2.
According to (2) (i)-(ii), there are two sequences ym < c < zm with ym −→ c, zm −→ c as
m −→ ∞ s. t. the difference quotients R(ym , c) and R(zm , c) have oscillating opposite signs for
any m i.e.
(I) : R(ym , c) < 0, R(zm , c) > 0
or
(II) : R(ym , c) < 0, R(zm , c) > 0
Let’s prove case (I), case (II) being similar.
From ym −→ c when m −→ ∞, it follows that exists a rank m1 from where ym is in the interval
(a, b) for any m ≥ m1 . Analogously, from zm −→ c when m −→ ∞, it follows that exists a rank
m2 from where zm is in the interval (a, b) for any m ≥ m2 .
Define m0 = max {m1 , m2 }, so

(∀) m ≥ m0 =⇒ ym < c < zm

Fix n > m0 and it results that yn < c < zn .

From case (I) : R(ym , c) < 0, R(zm ) > 0, so f (ym ) < f (c) and f (zm ) > f (c) for all m ≥ m0 .

Ph.D. in Mathematics, J.L.Chambers, Charlotte, NC, USA 14

On the Sign Change of a Continuous Function Suzy Manuela Prajea

In particular, n > m0 implies f (yn ) < f (c) and f (zn ) > f (c).
Finally,
yn < c < zn , f (yn ) < f (c) > f (zn )
The last relationships prove that f is neither increasing nor decreasing on the interval (a, b).
This is a contradiction with the assumption that f is monotonic on the interval (a, b).
Analogously can be proved the case (II).
In conclusion, f is not monotonic on any interval (a, b).
(ii) analogously to (i) (refer to Definition 1.5)

Conclusion 2.4.1. Example 2.4 shows that even though a function is continuous, its behavior at
any of its zeros can be NOT monotonic in any neighborhood of its zeros nor locally monotonic in
any of its zeros .

3 On the Sign Change of a Continuous Function

In spite of the unpredictable, erratic, or ’fickle’ behavior of a continuous function as it was
revealed in the previous section, if the function changes from negative to positive values and it is
differentiable with the derivative continuous, then a certain degree of predictability in its
increasing behavior in a neighborhood of its zero, can be proved.

The following definitions and lemmas are needed.

Definition 3.1. (ε − δ definition)

Let be f a real-valued function defined on the open interval (a, b) and α a point of the interval
(a, b) .

The function f is continuous at the point x = α iff for any ε > 0, there exists δ > 0 s.t.

for any x of the interval (a, b) with α − δ < x < α + δ, it results f (α) − ε < f (x) < f (α) + ε.

*Note that the ε − δ definition of continuity at a point x = α can be rewritten equivalently:

The function f is continuous at the point x = α iff for any ε > 0, there is δ > 0 s.t.

for any x of the interval (a, b) with |x − α| < δ, it results |f (x) − f (α)| < ε

or

f is continuous at the point x = α iff

(∀) ε > 0, (∃) δ > 0 s.t. (∀) x ∈ (a, b), |x − α| < δ ⇒ |f (x) − f (α)| < ε (3)
One of the most powerful piece of information about the values of a continuous function that ε − δ
definition provides is that if the function is continuous and strictly positive at a point then the

Ph.D. in Mathematics, J.L.Chambers, Charlotte, NC, USA 15

On the Sign Change of a Continuous Function Suzy Manuela Prajea

Figure 10: ε − δ definition of continuity at the point x = α

function is strictly positive on a neighborhood of that point. In other words, for a function defined
on an interval, the characteristic of a function to be strictly positive at a point of continuity is not
isolated, it is induced on a whole neighborhood of that point. (4)

The following lemma will prove the result mentioned above.

Lemma 3.1. (The inertia property of a continuous function)

Let be f a real-valued function defined on the open interval (a, b) and α a point of the interval
(a, b). The following hold:
(i) if the function f is continuous and strictly positive at the point α, then there is a neighborhood
of the point α included in (a, b) on that function f is strictly positive.

(ii) if the function f is continuous and strictly negative at the point α, then there is a
neighborhood of the point α included in (a, b) on that function f is strictly negative.

Shortly,

(i) if f is continuous at α and f (α) > 0, then

(∃) (c, d) ⊆ (a, b) s.t. (∀) x, c < x < d ⇒ f (x) > 0 (5)

(ii) if f is continuous at α and f (α) < 0, then

(∃) (c, d) ⊆ (a, b) s.t. (∀) x, c < x < d ⇒ f (x) < 0 (5′ )

Proof. (i) The function f is continuous at x = α and f (α) > 0

Ph.D. in Mathematics, J.L.Chambers, Charlotte, NC, USA 16

On the Sign Change of a Continuous Function Suzy Manuela Prajea

The ε − δ definition offers the opportunity to choose ε > 0 on purpose. Because f (α) > 0 and the
intention is to get a neighborhood of f (α) on that the function f remains positive, the ε-value will
be chosen s.t. ε < f (α). In this way, the desired interval (f (α) − ε, f (α) + ε) will contain only
positive values.

f (α)
To fix the ideas, choose for instance ε = .
2
According to the ε − δ definition, there is δ > 0 s.t. for any x in the interval (α − δ, α + δ) ∩ (a, b)
it results f (x) in the interval (f (α) − ε, f (α) + ε), so f (x) will be positive. (6)

The intersection (α − δ, α + δ) ∩ (a, b) is also an open interval and this interval can be determined
by considering c = max (α − δ, a) , d = min (α + δ, b). In consequence, the interval
(α − δ, α + δ) ∩ (a, b) = (c, d) (7).

From (6) and (7), it follows that for any x in the interval (c, d) ⊆ (a, b) it results f (x) in the
interval (f (α) − ε, f (α) + ε), so f (x) is positive.
(ii) can be proved analogously.
Lemma 3.2. If the function f defined on the interval (a, b) is differentiable, then the following
implications hold:

(i) if f increasing on (a, b) then f ′ (x) ≥ 0, (∀) x, a < x < b

(ii) if f strictly increasing on (a, b) then f ′ (x) ≥ 0, (∀) x, a < x < b

(iii) if f decreasing on (a, b) then f ′ (x) ≤ 0, (∀) x, a < x < b

(iv) if f strictly decreasing on (a, b) then f ′ (x) ≤ 0, (∀) x, a < x < b.

Proof. see [2],[3], or [4]
Lemma 3.3. If the function f defined on the interval (a, b) is differentiable, then the following
implications hold:

(i) if f ′ (x) ≥ 0, (∀) x, a < x < b, then f is increasing on (a, b)

(ii) if f ′ (x) > 0, (∀) x, a < x < b, then f is strictly increasing on (a, b)

(iii) if f ′ (x) ≤ 0, (∀) x, a < x < b, then f is decreasing on (a, b)

(iv) if f ′ (x) < 0, (∀) x, a < x < b, then f is strictly decreasing on (a, b)

Proof. see [2],[3], or [4]

Conclusion 2.4.1 revealed the fact that even though a function is continuous, its behavior in any
neighborhood of its zeros can NOT be locally monotonic.

Ph.D. in Mathematics, J.L.Chambers, Charlotte, NC, USA 17

On the Sign Change of a Continuous Function Suzy Manuela Prajea

Back to Definition 1.1 and the terms of the initial investigative questions (∗) and (∗∗), a certain
expected behavior in terms of monotonicity can be proved in conditions of differentiability.
More specifically, the following will be proved:
Theorem 1. Let be f a function defined on the interval (a, b) that changes from negative to
positive values on the interval (a, b).

If the function f is differentiable with the derivative continuous, then there is an subinterval (g, h)
of (a, b) on that the function is strictly increasing i.e.

(∃) (g, h) ⊆ (a, b) s.t. f is strictly increasing on the interval (g, h).

Proof. Notice that the derivative f ′ cannot be negative on the interval (a, b). (8)

Indeed, if f ′ (x) ≤ 0 for any x from (a, b) then f will be decreasing on (a, b) (Lemmas 3.3 and 1.1)
and the difference quotient R(x, y) ≤ 0 for any x, y from (a, b). (9)

From f changes from positive to negative values on (a, b), it results that there is a point c of (a, b)
s.t.

(i) f (x) < 0 for x ∈ (a, c)

(ii) f (c) = 0

(ii) f (x) > 0 for any x ∈ (c, b).

Choosing x, y from (a, b) s.t. x < c < y, it results f (x) < 0 < f (y) and from here it follows
f (y) − f (x)
immediately that > 0 or equivalently R(x, y) > 0. The last inequality is in
y−x
contradiction with the inequality (9).
Therefore f ′ cannot be negative on (a, b), so the statement (8) is proved.
It follows that there exists a point α of (a, b) where the f ′ is strictly positive i.e. f ′ (α) > 0.
The derivative f ′ is continuous and therefore the property of inertia holds for the continuous
function f ′ (Lemma 3.1 (i)): there exists a neighborhood (g, h) of α, (g, h) ⊆ (a, b) s.t. f ′ (x) > 0
for all x with g < x < h. According to Lemma 3.3 (ii) it results f strictly increasing on the
interval (g, h).
Theorem 3.3 and its proof allows to formulate the following:
Conclusion 3.0.1. If the function f defined on the interval (a, b) is differentiable with the
derivative continuous and it changes from negative to positive values at the point c of the interval
(a, b), then the answers to the initial questions:

(∗) There is a neighborhood of the point c on that the function f is strictly increasing?

(∗∗) There is an interval in the neighborhood of c, not necessarily containing the point c, on that
the function f is strictly increasing?

Ph.D. in Mathematics, J.L.Chambers, Charlotte, NC, USA 18

On the Sign Change of a Continuous Function Suzy Manuela Prajea

are NO and respectively, YES.

4 Locally Monotonic Functions

Proposition 4.1. Let be f a function defined on the interval (a, b) and c a point of this interval.
The following holds:
(i) if f is strictly increasing on both intervals (a, c) and (c, b) and locally strictly increasing at the
point c, then f is strictly increasing on (a, b).

(ii) if f is increasing on both intervals (a, c) and (c, b) and locally increasing at the point c, then f
is increasing on (a, b).

(iii) if f is strictly decreasing on both intervals (a, c) and (c, b) and locally strictly decreasing at
the point c, then f is strictly decreasing on (a, b).

(iv) if f is decreasing on both intervals (a, c) and (c, b) and locally decreasing at the point c, then
f is decreasing on (a, b).
Proof. (i) It is given f strictly increasing on each of the intervals (a, c) and (c, b), and locally
strictly increasing at c. To prove the conclusion, it is needed to prove the following:

(∀) x, y ∈ (a, b), x < y =⇒ f (x) < f (y) (10)

Let be x, y ∈ (a, b), x < y. The following cases are considered:

(j) if a < x < y < c, from f strictly increasing on the interval (a, c) it results immediately
f (x) < f (y)

(jj) if c < x < y < b, from f strictly increasing on the interval (a, c) it results immediately
f (x) < f (y)

(jjj) if a < x < c < y < b, from f locally increasing at c it results f (x) < f (c) < f (y), so
f (x) < f (y).

In consequence, (10) is proved. (ii), (iii), (iv), the proofs are analogously with the proof given at
(i)

Theorem 2. Let be f a function defined on the interval (a, b) and c a point of the interval (a, b).
If the function f is:

(i) differentiable with the derivative continuous

(ii) locally strictly increasing at the point c

then there is a subinterval (g, h) of the interval (a, b) on that the function f is strictly increasing.

Ph.D. in Mathematics, J.L.Chambers, Charlotte, NC, USA 19

On the Sign Change of a Continuous Function Suzy Manuela Prajea

Proof. It results from applying Theorem 1.1 to the function k(x) = f (x) − f (c).
Indeed, from f locally increasing at c it follows that function k changes from negative to positive
values (Definition 1.1 and 1.5). Obviously, the function k is differentiable and its derivative is
continuous so Theorem 1 can be applied to the function k.
Theorem 3. If the function f defined on the interval (a, b) is differentiable and locally increasing
on the interval (a, b), then the function f is increasing on (a, b).
Proof. From f locally increasing on the interval (a, b) (Definition 1.6) it follows that f is locally
increasing at each point c of the interval (a, b) i.e. it exists the interval (g, h) ⊆ (a, b) s.t. for any
x, y ∈ (g, h) with x < c < y, it results that the difference quotients R(x, c) and R(c, y) are strictly
positive. Because f is differentiable at the point c, the both limits of R(x, c) and R(c, y) exists
when x −→ c and y −→ c, and they should be equal.
It follows that f ′ (c) ≥ 0. (12)
Due to the fact that the relationship (12) holds for every c from (a, b), it results f ′ (x) ≥ 0 for any
a < x < b. From Lemma 3.3 it follows that f is increasing on the interval (a, b).
Corollary 3.1. If the function f defined on the interval (a, b) is differentiable and locally strictly
increasing at every point c from (a, b), then f is increasing on (a, b).

Proof. Although f is locally strictly increasing and not only locally increasing as in the Theorem
4, the conclusion remains the same due to the fact that even if R(x, c) > 0 for any c from (a, b),
the limit of R(x, c) when x −→ c will be positive and not strictly positive.

5 Instead of Epilogue
There is no magic more captivating than mathematics: where one sees a path, one can find soon,
or later a way. Definition 1.1 framed the intent: if a continuous function has a zero while changing
from negative to positive values, it must definitely show a form of increase in its values. Then, the
questions rose naturally:

(∗) if the function is continuous, an inertia property of increase might hold in a sufficiently small
neighborhood of the zero?
or maybe...

(∗∗) if the function is continuous, an eventual inertia property of increase might hold in a
sufficiently small neighborhood of the zero but maybe it does not contain the zero?
or maybe...

(⋄) a weaker pattern of increase can define better this behavior?

and not ultimately ...

(⋄⋄) imagining the graph of the continuous non-differentiable function while changing from
positive to negative in a ’sufficiently small neighborhood’ of its zero is more complex than it
would be if the function would be differentiable. The usual pattern based on ’zooming in’ a
’sufficiently small neighborhood’ to see if the function changes from negative to positive values is

Ph.D. in Mathematics, J.L.Chambers, Charlotte, NC, USA 20

On the Sign Change of a Continuous Function Suzy Manuela Prajea

worthless when the fractals are hidden.

Motto: The thinking is magic while the intuition is not magic ...

... the beauty of the real analysis is yet not exhausted in the 21st century, centuries after the
childhood of Calculus.

...and yes, why not write stories about the most beautiful ideas instead of the implacable abstract
math literature only for those initiated?

References
[1] Garling, D.J.H., A Course In Mathematical Analysis, Volume 1, Foundations and Elementary
Real Analysis
[2] Megan, M., Bus, e, C., Lat, cu, Analiză Matematică, 1995
[3] Megan M., Popa, C., Hiris. V., Introducere în Analiza Matematică prin Exercit, ii s, i Probleme,
1976

[4] Siretchi, Gh., Calcul diferential si integral, Ed. Ştiinţifică şi Enciclopedică, Bucureşti 1985.
[5] McLaughlin, M.E. An introduction to everywhere continuous, nowhere differentiable functions,
2013
[6] Vesnesque, S., Continuous, nowhere differentiable functions, 2019

[7] Talman, L.A., On Functions "Increasing at a Point", 2005

November 8, 2021

Dr. Suzy Manuela Prajea,

J.L. Chambers High School,
Charlotte, North Carolina, USA
[email protected]

Ph.D. in Mathematics, J.L.Chambers, Charlotte, NC, USA 21