Additional Mathematics Synthesis Guide 51: Local-to-Global Reasoning — Points, Intervals, Extrema and Whole-Graph Conclusions

Identify the scale of the evidence → identify the scale of the claim → build the bridge between them → check endpoints, boundaries and domain.

1. Point information is point information

If f(2)=5, we know one output value. We do not yet know whether 5 is a maximum, minimum, typical value or isolated point.

A single coordinate can support local reconstruction, but whole-graph conclusions require more structure.

2. Derivative at one point is local evidence

If f′(2)=3, the tangent gradient at x=2 is3. This describes the instantaneous rate of change there.

It does not prove f′(x)>0 for every x. To establish increasing behaviour on an interval, the derivative sign must be controlled throughout that interval.

3. Interval sign gives interval behaviour

If f′(x)>0 for all x in (a,b), then f is increasing on that interval. If f′(x)<0 throughout, f is decreasing there.

The logical bridge from local derivative information to interval behaviour is the sign statement holding across the whole interval.

4. Stationary point versus local extremum

f′(a)=0 says the tangent is horizontal. To classify the point, inspect nearby derivative signs or use the second derivative when its test is decisive.

• + to −: local maximum;
• − to +: local minimum;
• same sign through the point: stationary inflexion is possible.

5. Local maximum versus global maximum

A local maximum is larger than nearby values. A global maximum is at least as large as every allowed value in the full domain under consideration.

On a closed interval, global optimisation may require comparing all admissible stationary points with the endpoints.

6. Worked closed-interval example

Let f(x)=x³−3x on [−2,2].

f′(x)=3(x²−1), so stationary points occur at x=−1,1.

Evaluate:

• f(−2)=−2;
• f(−1)=2;
• f(1)=−2;
• f(2)=2.

The global maximum value is2, achieved at x=−1 and x=2. A local maximum at x=−1 is not the only point attaining the global maximum.

7. Quadratic structure gives global conclusions

For y=a(x−h)²+k, the sign of a gives whole-graph curvature and the vertex gives a global extremum over all real x.

• a>0: global minimum k at x=h;
• a<0: global maximum k at x=h.

Here the algebraic form itself supplies global structure, not merely local derivative evidence.

8. Range is a global statement over a domain

For y=3sinx+2 over all real x, the range is [−1,5]. This is a global conclusion derived from the whole sine range −1≤sinx≤1.

If the domain is restricted, the range may shrink. A global range claim must therefore name the domain it belongs to.

9. One period can support a global periodic conclusion

For a periodic function, analysing one complete period can determine behaviour on the whole real line once the period relation is established.

For sine and cosine, a complete cycle plus the identity f(x+2π)=f(x) extends the pattern globally.

The period relation is the bridge from local interval analysis to global repetition.

10. A graph window is not the whole graph

A calculator window may display one root while another lies outside the visible range. A local plot cannot by itself establish the total number of real roots.

Use algebraic structure, range, monotonicity or a justified domain-wide argument to make a global root-count claim.

11. Discriminant gives global root count for a quadratic

For a real quadratic, the discriminant gives the complete real-root count:

• Δ>0: two distinct real roots;
• Δ=0: one repeated real root;
• Δ<0: no real roots.

This is a whole-equation conclusion, stronger than inspecting a few graph points.

12. Monotonicity can prove uniqueness globally on an interval

If f is strictly increasing on an interval, then f(x)=k can have at most one solution there.

For eˣ, f′(x)=eˣ>0 for every real x. Hence eˣ is strictly increasing globally and each positive target k is attained exactly once because the range is y>0.

13. Tangency is local, intersection count can be global

Equal value and equal gradient establish local tangent contact at a point. But the curve and line may intersect again elsewhere unless the broader structure rules that out.

For a line and a quadratic, a repeated-root intersection equation with Δ=0 shows the line has exactly one real intersection x-value, so local tangency and global intersection count agree in that setting.

14. Local linear approximation is not global equality

A tangent line approximates a smooth curve near the contact point. It does not generally equal the curve elsewhere.

Confusing a local linear model with a global identity is a scale error.

15. Second derivative information has scale

f″(a)>0 at a stationary point supports local minimum at a when the usual test applies. It does not prove the function is convex everywhere.

To claim f″(x)>0 on an interval, one needs a sign argument across that interval.

16. Definite integrals are interval statements

∫_a^bf(x)dx describes signed accumulation over the interval [a,b]. It says nothing directly about accumulation outside that interval.

If f changes sign inside [a,b], total geometric area requires splitting at the relevant zeros. The local sign pattern controls the global interval total.

17. Kinematics converts local rates into interval motion

Velocity v(t) gives instantaneous motion direction. To understand motion over an interval, find where v=0, determine sign on each interval and then connect those intervals to displacement or total distance.

One velocity value cannot describe the whole journey.

18. Endpoint behaviour can decide global extrema

On a closed domain, an endpoint may be the global maximum or minimum even though it is not a stationary point.

This is why optimisation routines that inspect only f′=0 can miss the actual global answer.

19. Domain exclusions split global reasoning

A rational function such as 1/(x−2) has separate branches on (−∞,2) and (2,∞). Claims about monotonicity or range must respect the break at x=2.

A single formula does not imply one connected domain.

20. Modelling conclusions are global only over the modelled domain

An exponential or trigonometric model may fit a stated time interval. Extending its conclusion far beyond that interval is a new claim requiring new evidence.

A formula can be mathematically evaluable outside the observed range while its real-world interpretation becomes unreliable.

21. A local-to-global checklist

1. State the claim and its scale: point, neighbourhood, interval, full domain or modelled period.
2. State the evidence and its scale.
3. If the claim is larger than the evidence, identify the theorem or structure that extends it.
4. For calculus, control derivative signs over intervals rather than at isolated points.
5. For global extrema on closed intervals, include endpoints.
6. For periodic functions, establish the period before extending one cycle.
7. For root counts, use discriminant, monotonicity, range or other global structure.
8. Respect domain breaks and asymptotes.
9. For models, limit conclusions to the justified modelling domain unless extra evidence supports extrapolation.

22. Common failure patterns

• Turning one function value into a whole-graph conclusion.
• Using f′(a)>0 to claim increasing behaviour everywhere.
• Calling a local maximum the global maximum without comparing the domain.
• Ignoring endpoints in closed-interval optimisation.
• Reading a calculator graph window as the entire graph.
• Using one period without first establishing periodicity.
• Treating a tangent line as a global replacement for the curve.
• Calling f″(a)>0 evidence of global convexity.
• Ignoring domain splits at asymptotes.
• Extrapolating a model beyond its evidence without qualification.

23. Practice set

1. What does f(2)=5 tell you?
2. What does f′(2)=3 tell you?
3. What must be shown to prove f is increasing on (a,b)?
4. How can a stationary point be classified by derivative signs?
5. What is the difference between local and global maximum?
6. Why must endpoints be checked on a closed interval?
7. For y=2(x−3)²−5, state the global minimum.
8. What is the range of 3sinx+2 over all real x?
9. What identity extends one sine cycle globally?
10. Why can a graphing window miss roots?
11. What does Δ<0 prove for a real quadratic?
12. Why does strict monotonicity imply at most one solution to f(x)=k?
13. What two local conditions establish tangency?
14. Why can a tangent line not generally replace a curve globally?
15. Does f″(a)>0 prove f″(x)>0 for all x?
16. What does a definite integral describe?
17. Why must total distance split at direction changes?
18. Why is x=2 important for f(x)=1/(x−2)?
19. What is risky about extrapolating a fitted model?
20. What question should be asked before extending a local fact globally?

Answers

1. The function value at one point only.
2. The tangent gradient or instantaneous rate at x=2.
3. f′(x)>0 throughout the interval.
4. + to − gives local maximum; − to + gives local minimum.
5. Local compares nearby values; global compares every allowed value in the domain under consideration.
6. An endpoint can attain the largest or smallest allowed value without satisfying f′=0.
7. −5 at x=3.
8. [−1,5].
9. sin(x+2π)=sinx.
10. The chosen display range may exclude additional intersections.
11. No real roots.
12. A strictly increasing function cannot take the same output value twice on the interval.
13. Same point/value and same gradient.
14. It is only a local linear approximation unless the original function is itself that line.
15. No. The statement is only at x=a.
16. Signed accumulation over its stated interval.
17. Velocity sign determines motion direction; cancellation would turn total distance into displacement.
18. The function is undefined there, so the domain splits into two branches.
19. The relationship may cease to describe reality outside the calibrated or observed domain.
20. What theorem or structure justifies extending evidence from its current scale to the larger claim?

24. What mastery looks like

Mastery means the learner can name the scale of a fact, avoid overextending local evidence, use derivative signs and structural theorems to justify interval behaviour, and distinguish local extrema from global extrema or whole-domain claims.

The transfer test is to give a correct local statement and ask for a larger conclusion. If the learner first asks what bridge is needed rather than simply extending the statement, local-to-global reasoning has become controlled.

---

Continue through Batch 13

• Guide 49: Counterexamples and Theorem Boundaries
• Guide 50: Generalisation from Cases to Families
• Guide 52: Mathematical State Tracking — Variables, Conditions, Exactness, Branches and Intermediate Results
• Guide 31: Increasing and Decreasing Functions
• Guide 37: Domains, Ranges, Restrictions and Admissible Solutions
• Additional Mathematics Directory
• BTT Mathematics Hub