The set is compact because is closed and bounded, and the product
is continuous. By the extreme value theorem, has a maximizer. Since meets the interior of the nonnegative orthant , some feasible product is positive, so every maximizer lies in .
On , maximizing is equivalent to maximizing
The Hessian of is
which is negative definite. Thus is a strictly concave function, and it has at most one maximizer on the convex set . Denote this unique point by .
For any , the line segment
lies in . Since is maximal at , its right derivative there is nonpositive:
Therefore
This is the supporting inequality for the product maximizer on a compact convex subset of the positive orthant.
Suppose now that is invariant under cyclic coordinate permutation. That permutation preserves the product, so uniqueness forces it to fix . Hence
for some . More explicitly,
Indeed, cyclic symmetry averaging puts the diagonal point with coordinate equal to the mean of any back in , while the preceding inequality with gives .
Finally, given , define
This set is nonempty, closed, convex and bounded, and it contains . By the arithmetic-geometric mean inequality,
Equality holds only when all are equal and their sum is , namely only at . Thus this has
Let be continuous. If it were unbounded, one could choose with . The Bolzano-Weierstrass theorem gives a subsequence , but continuity would then give , contradicting unboundedness.
Now let . Choose with . A convergent subsequence and continuity give at its limit. Applying the same argument to shows that the infimum is attained. This proves the extreme value theorem on a closed bounded interval.
The function
is continuous and bounded, but attains neither its infimum nor its supremum . For the second example, enumerate the rationals in as and set
Every nondegenerate interval contains infinitely many rational numbers, hence some with arbitrarily large ; therefore is unbounded on every such interval.
For the running extrema, compactness lets us write
The function is uniformly continuous on . Given , choose such that whenever . If and , every new value with is at most . Since ,
Interchanging handles the other order. Applying this argument to proves continuity of as well.
Finally fix and put . For each positive integer , there must be some with . Otherwise, for some the continuous function would satisfy on the interval . By the intermediate value theorem, would have a constant sign there. The values
would then increase or decrease by at least at every step, contradicting boundedness of . Thus and
Rolle theorem: if is continuous on , differentiable on , and , then for some . Indeed, the extreme value theorem gives a maximum and minimum. If both occur only at the endpoints then is constant; otherwise an interior extremum satisfies by comparing the two-sided difference quotients.
Mean value theorem: if is continuous on and differentiable on , then some satisfies
Apply Rolle's theorem to
which has .
Set
The triangle inequality makes Lipschitz continuous and hence a continuous function. Since is an isometric embedding,
so as . The function is therefore coercive, and the extreme value theorem on a sufficiently large compact interval gives a minimizing parameter.
Suppose both minimize , put and , and let
The restriction of between the two parameters is a geodesic from to . Let be its midpoint. In the geodesic triangle with vertices , the thin geodesic triangle condition gives a point on one of the other two sides with . By symmetry suppose . Then
so
But lies on and is a closest point, so . Therefore , which is stronger than the required
This is the closest point on a geodesic line in a hyperbolic metric space estimate.
For , the triangle inequality and the Cauchy-Schwarz inequality give
Thus is a continuous function. The unit sphere in the finite-dimensional Euclidean normed vector space is compact. Moreover, implies that the polynomial vanishes identically, so every coefficient is zero. Hence is strictly positive on the unit sphere. By the extreme value theorem, it has a positive minimum
The supremum norm is homogeneous. For ,
It follows directly that whenever .
Write each polynomial of degree at most as and define
This is a continuous function, and the reverse triangle inequality together with part (a) gives
Consequently is a coercive function. Choose so large that outside the closed Euclidean ball of radius . That ball is compact, so the extreme value theorem supplies a minimizer there. Then is a best uniform approximation to among polynomials of degree at most .
Fix strictly between and , and set . The function is continuous on : interior differentiability gives interior continuity, and the finite one-sided derivatives give continuity at the endpoints. The extreme value theorem therefore gives a point where attains its minimum.
Since , for sufficiently small we have , so cannot be a minimizing point. Since , the quotient is positive for sufficiently small , giving . Thus cannot be a minimizing point either, and .
Fermat's theorem on stationary points states that a differentiable function at an interior local minimum or local maximum has derivative zero there. Applying it to gives
This proves the relevant case of Darboux's theorem; the argument uses continuity of , with no assumption of continuity of .
The continuous function attains a finite maximum on by the extreme value theorem. Additivity and the triangle inequality for the Riemann integral imply, for any ,
For reversed endpoints, reverse the orientation of the integral before taking its absolute value. Given , choose . Then implies . Thus is continuous on , including one-sided continuity at the endpoints; it is even Lipschitz continuous. If , the function is defined on one point and the conclusion is immediate.
Write and . By the extreme value theorem, attains a minimum and a maximum on . Because , the monotonicity of the Riemann integral gives
If , this forces , and every satisfies the required identity. This includes .
If , divide to obtain . The intermediate value theorem, applied between points where the continuous function attains and , supplies with . Hence the weighted mean value theorem for integrals gives
For , the gradient never vanishes on the circle. The Lagrange multiplier equations are
If , the two possibilities are . If , then , and the first equation gives . Combining this with yields
Put
Both and are positive, giving the other four points and . To classify all six, restrict to the circle by eliminating :
Its derivatives are and . In the interior, and . At either endpoint ; along the circle, has a strict local minimum at the left endpoint and a strict local maximum at the right endpoint. Consequently the six constrained extrema are
The circle is a compact set, so the extreme value theorem guarantees that its global minimum and global maximum occur among these six points. Their values give the corrected sharp bound
The printed lower bound is false. A particularly short counterexample is
The sole local minimum in part (a), , has value , larger than values on this circle; it is therefore not a global minimum. In fact , so has no global minimum on the plane.
Figure 1.
All six extrema on the constraint circle and the corrected minimum of the restricted polynomial
. Details.
We prove the symmetric derivative criterion for monotonicity using an antiderivative. Fix and define
The fundamental theorem of calculus gives because is continuous. For each interior ,
Since , the definition of symmetric derivative gives such that for . Therefore, for ,
We next show that is a convex function. On any closed interval let be the affine function joining to . If were positive anywhere, the extreme value theorem would give a positive maximum at an interior point . For every sufficiently small , that maximum would imply
The affine function contributes zero to this symmetric difference, contradicting the strict positive inequality above. Thus on every such interval, exactly convexity.
For a differentiable convex function, secant slopes bound endpoint derivatives. Explicitly, the chord inequality at and at , followed by , gives
So is nondecreasing. If for some , monotonicity would make constant on , forcing for , a contradiction. Consequently : is strictly increasing. This proof requires no continuity of and never applies the ordinary mean value theorem to a merely symmetrically differentiable function.
The extreme value theorem gives finite real numbers
Since these values are attained, the set of strict upper bounds in the definition of is precisely , and the set of strict lower bounds in the definition of is precisely . They are nonempty and bounded on the required sides. By the definitions of infimum and supremum,
The real spectral theorem states that a symmetric matrix has an orthonormal basis of real eigenvectors, and hence is a diagonalisable matrix. Here is a proof in this setting.
For , the quadratic form attains a maximum on the unit sphere, by compactness and the extreme value theorem. Choose a unit maximizer and put . For any , the Rayleigh quotient along has a maximum at , so
The last equality uses and . Thus is orthogonal to every , so .
The orthogonal complement is an invariant subspace, because for ,
The restriction of to is again symmetric with respect to the induced inner product. By mathematical induction on dimension, it has an orthonormal basis of eigenvectors; adjoining gives one for . If has those basis vectors as columns, then is a real diagonal matrix and .
For , the stationary points found in parts (a) and (b) have temperatures , , and . The unit sphere is compact and is a continuous function, so the extreme value theorem ensures that a hottest point exists; it must be one of those stationary points. Thus
For a direct bound, put . The identity gives , and
since . Equality forces , giving exactly the two points above.
Put and choose . For , the triangle inequality gives
The reverse triangle inequality therefore yields
If had no root, the function would be an entire function. The bound just proved makes as . On the remaining closed disk, continuity and the extreme value theorem make bounded. Thus is a bounded entire function, so Liouville's theorem makes it constant. Its limit at infinity forces that constant to be zero, impossible for a reciprocal.
Consequently has a root in . This proves the existence part of the fundamental theorem of algebra. The invoked Liouville theorem follows from the Cauchy integral formula: the Cauchy estimate on arbitrarily large circles gives when is bounded by .
We first show that a continuous injection must be surjective. Otherwise its image misses a point . Stereographic projection identifies with , giving a continuous injection . The extreme value theorem supplies distinct points where attains its minimum and maximum , with because is injective.
The two circular arcs from to each contain a point at which , by the intermediate value theorem. Those points are distinct because they lie in the disjoint interiors of the two arcs. This contradicts injectivity, proving the claim about .
Now suppose were a continuous injection. Its restriction to the equatorial circle would be a continuous injection from to itself, hence . The north pole is outside , but must equal for some . This contradicts injectivity. Therefore
Suppose were omitted by . Then
would be an entire function. The reverse triangle inequality gives as , so there. On a sufficiently large closed disk, is bounded by continuity and the extreme value theorem; outside it, the limit gives a bound. The Liouville theorem therefore makes constant, and the limit forces .
That is impossible, since is nonzero at every finite . No value can be omitted, and hence
The Liouville theorem states that every bounded entire function is constant.
By the Laurent theorem, has a unique Laurent series on . Split it into
Here is holomorphic on and is holomorphic on . To check their extended regions of convergence explicitly, fix and put . The Laurent coefficient formula gives
The first bound ensures convergence uniformly on every closed disk ; letting range up to covers . The second gives convergence uniformly on ; letting range down to covers . Locally uniform convergence of these power series, in and respectively, proves the claimed holomorphy. On the original annulus their sum is .
The second estimate also proves as : for any fixed and ,
For uniqueness under this condition, suppose . Define on and on . The two definitions agree on the overlap, and their domains cover , so they define an entire function. Moreover, at infinity. It is bounded outside a large disk by this limit, and inside by continuity and the extreme value theorem. The Liouville theorem makes constant, and its limit makes that constant zero. Hence and . Without the limiting condition, adding a constant to and subtracting it from would preserve the decomposition.
It suffices to assume . The polynomial interpolation error for the endpoint linear interpolant gives, for every interior , some with
Let , which exists by continuity of the second derivative and the extreme value theorem. Since ,
Integrating this pointwise inequality and using the triangle inequality for an integral yields
Thus
This derivation does not require choosing as a continuous function of ; the pointwise upper bound is sufficient.
The constant is sharp. Take the quadratic polynomial and write . Its exact integral is , its trapezoidal rule value is , and . Consequently
Equality holds, so no smaller universal constant than works under these assumptions. This is the sharp one-panel trapezoidal error bound.
Use constraints and . Both constraints and the objective are continuously differentiable. The feasible set is a nonempty closed subset of a bounded ellipsoid, hence compact, so the extreme value theorem guarantees both extrema.
The gradients and are independent everywhere feasible. If the first were times the second, then , , . The plane constraint would give , forcing the origin, which is not on the ellipsoid. Thus the Lagrange multiplier theorem applies at every feasible point.
Its equations are
Here , since otherwise the first two equations require . Subtract twice the first equation from the second, and use and ; the result is , so . Then and the ellipsoid gives . The only candidates are therefore , with matching signs. Consequently