Total differential Created 2026-09-24 Updated 2026-10-14
The total differential is the linear map that combines the first-order responses to simultaneous changes in all coordinates. For a differentiable real-valued function at , it can be written
Here is an input vector; denotes the coordinate linear functional . The coordinate formula and the Fréchet derivative describe the same linear map. The defining approximation is
The little o notation says the remaining approximation error becomes small relative to the input norm, across every direction of approach.[ref]
For at , . A simultaneous change gives the linear approximation
The exact algebra is
Consequently the exact change is , and the approximation error is . If both input changes are halved, the linear prediction becomes and the error becomes : the first-order response halves, while the quadratic error falls by a factor of four. The right panel under directional derivative displays this comparison as the displacement is scaled.
For a quick check of differentiability, gives
Dividing by makes this bound tend to zero. The same calculation proves that the tangent plane to the graph of a function at is . It approximates the entire local surface, rather than only two coordinate slices.
The total differential is useful for estimating the combined effect of small input changes without reevaluating a complicated function. For example, if and , then . The exact remainder bound from the displayed polynomial is at most , so this example has throughout that input rectangle. These are deterministic bounds; a variance calculation additionally needs a joint probability distribution and its covariances. A first-order estimate is also local: large changes require control of the remainder, not merely knowledge of the partial derivatives at one point.
A length-, dimension- binary linear code has full support when no coordinate vanishes on every codeword, equivalently when the union of the supports of a vector of all codewords covers every coordinate. Every coordinate is then a nonzero linear functional on the message vector space . Translation by a vector on which it is one pairs the zero and one fibers, so exactly codewords contribute one in that coordinate. Summing these contributions proves the formula. If there are nonzero columns in a full-rank generator matrix, the same argument gives instead.
Trace is the unique normalized cyclic linear functional Created 2026-10-06 Updated 2026-10-07
Over or , every linear functional on all matrices satisfying is a scalar multiple of the trace. The matrix units show that the functional vanishes off the diagonal and has equal values on every diagonal unit. The normalization fixes the scalar to one.
Traceless matrix Created 2026-10-06 Updated 2026-11-05
A traceless matrix is a square matrix whose trace is zero. Such matrices form the kernel of the trace linear functional.
Transpose of a linear map 2026-10-14
For a linear map , the dual map sends a linear functional on the output space to the input measurement . It reverses direction because a measurement of the output can be evaluated on an input by first applying :
For and , the input measurement is . The output measurement ignores the part of this transformation contributed by .
If has matrix in specified bases, a row representing changes to . Storing the coefficients of a linear functional as a column in the corresponding dual basis gives , explaining why the transpose of a linear map is represented by the matrix transpose. This algebraic operation uses no inner product; it agrees with a matrix transpose even over the complex numbers, rather than automatically introducing complex conjugation. The adjoint operator is defined using an inner product and is a different construction. [ref].
In finite dimension, the kernel and image of the dual map are described by annihilators of vector subspaces:
Transvection Created 2026-10-14 Updated 2026-11-05
For a nonzero vector and nonzero linear functional vanishing on , this invertible map fixes the hyperplane pointwise and has determinant one. Its inverse is , since . A symplectic transvection is obtained by taking for a nondegenerate alternating bilinear form.
Variational problem Created 2026-10-06 Updated 2026-10-07
A variational problem seeks a stationary point or minimizer of a functional on a specified function space. For a symmetric coercive bilinear form and bounded linear functional on a real Hilbert space, minimizing is equivalent to the weak formulation for every test vector. The Lax-Milgram theorem gives a unique solution, and proves unique minimization. Specifying the admissible space and its boundary constraints is part of the problem.
Weak limit Created 2026-10-25 Updated 2026-10-31
The weak limit of a sequence in a normed vector space is an element at which every continuous linear functional has the limit of its values along the sequence. Thus the definition is precisely weak convergence to that element.
A map from an open subset of into a complex Banach space is weakly holomorphic when every composition with a bounded complex-linear functional is a scalar holomorphic function. Norm holomorphy requires its difference quotients to converge in the Banach norm. Weak holomorphy is initially a condition on scalar observations; the weak holomorphy implies norm holomorphy argument recovers an actual vector derivative. A closed target subspace containing values on a sequence with an interior accumulation point contains every value on the connected domain: annihilate the subspace, apply the scalar identity theorem, and use Hahn-Banach separation.
Weak mixed Poisson problem Created 2026-09-24 Updated 2026-11-05
Let be a bounded connected Lipschitz domain, and let a measurable boundary part have positive surface measure. Impose a homogeneous Dirichlet boundary condition there and a homogeneous Neumann boundary condition on the remainder. For , seek in the closed Sobolev space subspace
satisfying
The Trace operator makes the boundary restriction meaningful. The Poincare inequality with a partial Dirichlet boundary bounds the full norm by the gradient norm on . Thus the left side is a bounded coercive bilinear form, and the Lax-Milgram theorem gives a unique weak solution. The same proof allows any bounded linear functional on as the right side.
The Neumann condition is natural: integration by parts leaves the boundary term on the non-Dirichlet portion, and requiring the identity for all makes that term vanish. If is empty, constants remain in the kernel of a linear map; for homogeneous Neumann data, compatibility requires , and uniqueness requires a normalization such as zero mean. Positive Dirichlet surface measure is what removes this ambiguity.
Every bounded sequence in a Hilbert space has a weakly convergent subsequence. The closed span of a sequence is separable; diagonal convergence of coordinates along an orthonormal basis gives a bounded limiting linear functional, represented by a vector of the Hilbert space. Applied to a bounded family of mollifications, this proves that their strong Lebesgue limit inherits the weak derivative bound.