Concave up / concave down. A graph is concave up (convex down) at a point if the tangent line there lies below the graph nearby; it is concave down (convex up) if the tangent lies above the graph nearby.
Definition via f′. On an open interval I where f′′ exists: f is concave up on I if f′ is strictly increasing on I; concave down if f′ is strictly decreasing on I.
Test of Concavity (Theorem 7.11 — the working tool).
- f′′(x)>0 on an open interval I ⇒ f is concave up on I.
- f′′(x)<0 on an open interval I ⇒ f is concave down on I.
Remarks. A local maximum of a function that is convex-up (concave-down) on the whole interval [a,b] is automatically its absolute maximum there (and symmetrically for a concave-up function's local minimum). There is only ever one absolute maximum (and one absolute minimum) on an interval, but there can be several local maxima/minima.
Points of inflection. A point where the graph switches from concave-up to concave-down (or vice versa) is a point of inflection.
Test for Points of Inflection (Theorem 7.12). If f′′(c) exists and changes sign as x passes through c, then (c,f(c)) is a point of inflection; and if f′′ exists at a genuine point of inflection, then necessarily f′′(c)=0 there.
Three cautions (all illustrated in the textbook remarks) that keep this test from being mechanically misapplied:
- f′′(c) may fail to exist at a genuine inflection point (e.g. f(x)=x1/3 at x=0) — so "f′′ undefined" is not automatically "no inflection point," it needs separate checking. …