Intermediate Value Theorem: Question bank — Maths deep dive

Reminder of the exact statement so every answer below is anchored to it:

Here is the geometry that statement describes: the level line $y = N$ (dashed) sits between the two endpoint heights, and a continuous curve is forced to touch it.

True or false — justify

Continuity on

[a, b]

alone guarantees

f

hits every value between

f (a)

and

f (b)

True. That is exactly IVT — no extra hypotheses (like differentiability or monotonicity) are needed; "no jumps" is the whole engine.

f (a)

and

f (b)

have the same sign, then

f

has no root in

(a, b)

False. IVT is only a sufficient condition for a root; same signs make it silent. E.g.

x^{2} - 1

[- 2, 2]

has

f (- 2) = f (2) = 3

yet two roots inside.

f

is continuous and has a root in

(a, b)

, then

f (a)

and

f (b)

must have opposite signs.

False. A root can exist with same-sign endpoints (the curve dips to zero and comes back), e.g.

x^{2}

touching

0

from a positive frame — opposite signs is a trigger, not a requirement.

IVT guarantees exactly one

c

with

f (c) = N

False. IVT is pure existence; it gives no count. There may be one, several, or infinitely many such

c

(think a wiggly curve crossing height

N

many times).

A function that satisfies the IVT conclusion (hits every intermediate value) must be continuous.

False. The converse fails.

sin (1/ x)

(extended by

0

0

) has the intermediate-value property near

0

but is not continuous there. Hitting all values ≠ continuity.

IVT works just as well on the open interval

(a, b)

as on

[a, b]

False. The hypothesis needs the closed

[a, b]

f (a)

and

f (b)

are actually defined and finite; on an open interval the endpoints (and thus the "in-between" values) may be missing or infinite.

The same statement of IVT holds if we replace

R

Q

(the rationals).

False.

f (x) = x^{2} - 2

[0, 2] \cap Q

changes sign but has no rational root. IVT relies on the Completeness Axiom of Real Numbers, which

Q

lacks.

f (a) = f (b)

, IVT tells us nothing.

True. IVT needs

N

strictly between

f (a)

and

f (b)

; if

f (a) = f (b)

there is no such

N

, so the theorem gives no conclusion (that gap is where Rolle's Theorem takes over for differentiable

f

IVT can prove a maximum of

f

[a, b]

exists.

False. That is the job of the Extreme Value Theorem. IVT is about values in between, not about largest/smallest values.

Spot the error

f (x) = 1/ x

[- 1, 1]

goes from

- 1

1

, so by IVT it hits

0

somewhere."

Error:

1/ x

is not continuous at

0

(it blows up to

\pm \infty

). The hypothesis fails, so IVT does not apply — and indeed

1/ x

never equals

0

f

is continuous on

(0, 1)

with

f \to - 1

x \to 0^{+}

and

f \to 1

x \to 1^{-}

, so IVT gives a root."

Error: the interval is open;

f (0)

and

f (1)

may not exist, so we have no genuine endpoint values to sandwich

N

. Limits at the ends are not the same as values at the ends.

f

jumps from

2

5

x = 3

, and

4

is between them, so IVT gives

c

with

f (c) = 4

Error: a jump is precisely a discontinuity. IVT forbids exactly this behaviour, so it cannot be invoked; the value

4

can be genuinely skipped.

"IVT says

\exists c

with

f (c) = N

, and since 'exists' means one thing,

c

is unique."

Error: "there exists at least one" is an existence quantifier, not a uniqueness claim. Uniqueness needs extra structure (e.g. strict monotonicity), never IVT alone.

f (0) = - 3

f (4) = 5

, so IVT gives a root in

(0, 4)

; therefore the root is at the midpoint

x = 2

Error: IVT locates no specific point. It gives no formula and no midpoint guarantee — you'd need the Bisection Method to actually pin the root down.

"The proof only handled

f (a) < 0 < f (b)

, so IVT is proved only for

N = 0

Error: the general

N

reduces to that case via

g (x) = f (x) - N

. A zero of

g

is a point where

f = N

, so the

N = 0

(Bolzano) case is the general theorem in disguise.

"Since IVT needs a supremum, it must also produce the maximum value of

f

Error: the supremum in the proof is of a set of $x$ -values (

S = {x : f (x) < 0}

), the location of the crossing — not the maximum output of

f

. Different object entirely.

Why questions

Why is continuity the single non-negotiable hypothesis?

Because the only way to skip a value

N

is to jump over it, and continuity is precisely the promise of "no jumps". Remove it and skipping becomes possible.

Why does the proof invoke the Completeness Axiom rather than just continuity?

Continuity keeps the curve from jumping, but we still need the crossing point

c = sup S

to exist as a real number. Completeness guarantees that supremum exists — over

Q

it might fall into a "gap".

Why do we build

g (x) = f (x) - N

instead of working with

f

and

N

directly?

It converts "does

f

hit the height

N

?" into "does

g

have a root?", i.e. reduces every case to the clean sign-change form

g (a) < 0 < g (b)

where Bolzano's argument runs.

Why must

N

be strictly between

f (a)

and

f (b)

N

equalled an endpoint value,

c

could be

a

b

itself, and the conclusion

c \in (a, b)

(open interior) might fail; strictness keeps the crossing genuinely inside.

Why does IVT prove

2

exists without ever computing it?

Applying IVT to

x^{2}

[0, 2]

shows some

c

with

c^{2} = 2

exists purely from continuity and completeness — existence is delivered before, and independently of, any decimal value.

Why is the fixed-point trick (

g (x) = f (x) - x

) a natural use of IVT?

A fixed point

f (x) = x

is a root of

g (x) = f (x) - x

; since

f

maps

[0, 1]

into

[0, 1]

g (0) \geq 0

and

g (1) \leq 0

, giving the sign change IVT needs (see Fixed Point Theorems).

Why does the antipodal-temperature argument work?

Define

h (θ) = T (θ) - T (θ + 18 0^{\circ})

; then

h (0) = - h (18 0^{\circ})

, so

h

has opposite signs (or is zero), and IVT forces some

θ

with

h (θ) = 0

— equal temperatures at opposite points.

Edge cases

f (a) = N

already, does IVT still apply?

The hypothesis wants

N

strictly between

f (a)

and

f (b)

, so with

f (a) = N

IVT is not invoked — but you already have the point

c = a

trivially (no theorem needed).

What if

f

is constant,

f (x) = k

, on

[a, b]

Then

f (a) = f (b) = k

and there is no

N

strictly between them, so IVT says nothing — consistent, since the only value achieved is

k

itself.

What if the two endpoints have opposite signs but

f

is continuous only at the endpoints, not inside?

IVT fails to apply — it demands continuity on the whole closed interval, not merely at the ends. A discontinuity inside can hide a skipped value.

What happens at a point where the curve merely touches the height

N

(tangent, no crossing)?

IVT still counts it:

f (c) = N

holds whether the curve crosses or just kisses the level. The theorem asks for equality, not a sign change of

f - N

around

c

Does IVT apply if

[a, b]

is a single point,

a = b

Degenerately,

[a, a] = {a}

has

f (a) = f (a)

, so no strictly-between

N

exists; the theorem is vacuous but not violated.

Can IVT be used on an unbounded "interval" like

[0, \infty)

Not directly — there is no finite right endpoint

f (b)

to sandwich

N

. You must first choose a finite

b

where the sign is known, then apply IVT on the closed

[0, b]

Connections

Continuity — the hypothesis every trap above secretly attacks.
Completeness Axiom of Real Numbers — why $Q$ counterexamples break IVT.
Bolzano's Theorem — the $N = 0$ special case powering the proof.
Bisection Method — what actually locates the root IVT only promises.
Extreme Value Theorem — the max/min sibling people confuse with IVT.
Rolle's Theorem and Mean Value Theorem — take over when $f (a) = f (b)$ .
Fixed Point Theorems — the $g (x) = f (x) - x$ application generalised.