Now the inverse integrand can be written
For , the exponent has a positive real time parameter. Close the Bromwich contour to the left; the residue at is . For , close to the right, where there are no poles. The reversed orientation there still gives zero. Consequently
The same result follows from the Laplace transform time-shift rule: corresponds to , where is the Heaviside step function.
The original transform cannot fix a value at the single point . The Bromwich inversion formula chooses the average of the limiting values and zero, giving under that convention. Indeed,
regardless of the assigned value at the jump.
Choose . The inverse integrand is . Because , the exponential factor decays towards the left. Closing the Bromwich contour to the left and using the residue theorem picks up just the simple pole at , with residue . Hence
Its defining integral checks both the answer and its convergence condition:
Assume the regular real Sturm-Liouville problem, with , ordered eigenvalues , and real orthonormal eigenfunctions in the weighted inner product
The Rayleigh quotient for an admissible trial function is
The second expression follows by integration by parts. It is the quantity used in the Rayleigh-Ritz method.
For , orthonormality eliminates every cross-term in the denominator:
Applying to the eigenfunction expansion and using the same orthogonality in the numerator gives
Hence, whenever the trial function has finite energy,
Put and . The exact error and its expansion are
In particular, under a common small scaling with fixed finite sums and ,
The fourth-order remainder includes mixed products , not just individual fourth powers.
Error comparison and sign. Since , the relative weighted norm error of the supplied trial function is
For , the relative eigenvalue error is . Thus the Rayleigh quotient has a second-order error for a first-order error in the trial eigenfunction, provided the energy-weighted coefficient sum is controlled. Small weighted norm error alone does not control arbitrarily high-frequency contamination, because can grow without bound. If , an absolute error must replace the undefined relative one.
Finally every summand of is nonnegative. Thus
with strict inequality whenever some . This is the upper-bound property of the Rayleigh-Ritz variational principle for the lowest eigenvalue.
For the real Sturm-Liouville operator , two integrations by parts give the Lagrange identity for a Sturm-Liouville operator:
Set . Both vanish at the endpoints, so the boundary term is zero. Since ,
Consequently
This is self-adjointness with the homogeneous Dirichlet boundary conditions, and expresses the Fredholm solvability condition at the eigenvalue . In this unweighted integral already contains its factor of ; adding another weight would be incorrect.
By superposition, . Split the integral at : the portion uses the first branch of the Green's function, and the portion uses the second. This yields
For , put . The two integrals are
Substitution and cancellation give
Indeed . For a direct differential check,
and therefore
The homogeneous uniqueness shown in part (a) makes this the unique solution.
Choose the homogeneous solutions , which vanishes at , and , which vanishes at . Their Wronskian in the order is . For a source point , seek the Green's function as
It is continuous at ; otherwise its second distributional derivative would contain a derivative of the Dirac delta function, which the source does not contain. Integrating the differential equation across the source, using continuity and the leading coefficient one, gives the jump condition
Continuity and this derivative jump require
Solving with the Wronskian gives and . Therefore
The two values at are identical. The derivative jump is explicitly
Both endpoint values vanish and the differential expression is zero off the diagonal, completing the verification of the Green's function.
Use reduction of order by setting . Then and , so the differential expression simplifies to
In the homogeneous case is constant. Hence , and a basis of the solution space is
Their Wronskian is on , proving linear independence. No nonzero homogeneous solution satisfies both endpoint conditions: sets , and then sets .
Mixed endpoint spectrum. For , integration against and integration by parts, using , gives for every nonzero real eigenfunction. Thus and the first condition makes . The second requires , so , . Since
the normalized eigenfunctions and eigenvalues are
The orthogonality also follows by applying integration by parts to two different eigenfunctions of the same self-adjoint differential operator.
Forced expansion. Let and write , . The Fourier coefficients of the forcing are
because the boundary cosine vanishes and . Applying to the eigenfunction expansion gives . Since is not an eigenvalue,
The coefficients are , so the series and its first two differentiated series converge absolutely and uniformly on the interval. Each term satisfies the two endpoint conditions, and equals the mixed-endpoint Fourier series of . The forcing coefficients are , giving uniform convergence of that series as well. Uniqueness follows because the homogeneous problem has a nonzero solution only at one of the excluded eigenvalues.
Odd fourth-power sum. Set . Direct integration of , with the same endpoint conditions, gives
At , the eigenfunction expansion instead gives . Equating the two values yields
Absolute and uniform convergence justify evaluation at the endpoint.
Write a nonterminating parity series as , where or . Its recurrence relation is
The ratio test applied to its nonzero successive terms has limiting absolute ratio . Hence the series converges absolutely for and diverges for , with
At , the limiting ratio is one, so the basic ratio test alone is inconclusive. Retain the next term of the ratio instead:
For large the ratio is positive. Taking logarithms and summing gives ; the remainders have a finite sum. Thus eventually. Comparison with the convergent p-series proves absolute convergence at both endpoints as well:
The radius of convergence one is the distance from the expansion point to the nearest singular points of a differential equation, . Convergence of function values at those points does not make the solutions analytic there; endpoint derivatives can diverge. The terminating Chebyshev polynomial is an exception with infinite radius of convergence. For example, at the other solution is , whose endpoint values are finite while its first derivative diverges.
Let for integer . In the even or odd power series whose parity matches , the numerator in the recurrence relation vanishes at the coefficient of . All coefficients of lower powers in that parity are nonzero, and all higher ones vanish. Therefore a polynomial solution exists, of polynomial degree , containing
For nonnegative these are the Chebyshev polynomials . The normalized low-degree cases follow directly from the recurrence relation:
Enforcing gives
To see that normalization at one is always possible, set in the Chebyshev differential equation. For , the differential expression becomes . Thus gives on and has value one at . It is a polynomial because the cosine addition formula gives . Negative integer produces the same solution as .
Divide the Chebyshev differential equation by . Its coefficients and are analytic at zero, so The finite singular points of a differential equation are .
Insert , . The coefficient of the lowest power is , hence or . The next power gives , and the remaining powers yield
For , is free and
Choosing selects the even power series, with . For , the next-power equation forces and
This selects the odd power series, with . Their different initial values prove linear independence. Although the Frobenius method produces two possible leading exponents, zero is an ordinary point, and a general series already includes both initial constants.
The original PDF uses the conjugate transpose , rather than the transpose found in the transcription. The printed diagonalisation criterion is correct when diagonalisation means unitary diagonalization of a normal matrix. Since ,
Consequently if and only if . By part (ii), this gives a common unitary matrix and
which is diagonal. Conversely, if for a diagonal and a unitary matrix , then
Thus the precise conclusion is
If arbitrary invertible changes of basis are allowed, the printed assertion is false. For example,
satisfy . Both Hermitian matrices have distinct eigenvalues: for and for . The distinct eigenvalues make diagonalizable by similarity, but and differ. This distinguishes ordinary diagonalization from the required unitary version.
Suppose and . Then
Because the eigenvalues of are distinct, the eigenspace for is one-dimensional. It follows that , including the possibility . Thus an orthonormal eigenbasis of is also an eigenbasis of . The simplicity of both spectra means their eigenvectors coincide as directions, up to ordering, phases and nonzero scalar multiples.
Conversely, if they share a complete eigenbasis, choose the unitary matrix whose columns are these common normalized eigenvectors. Then and are diagonal, so and . Hence
This is simultaneous diagonalization of the two Hermitian matrices.
The conjugate transpose reverses products and conjugates scalars. Since are Hermitian matrices,
Therefore In particular the commutator of two Hermitian matrices is anti-Hermitian, and multiplication by converts it to a Hermitian matrix.
As is a unitary matrix, . Let . Its conjugate transpose is , so and every positive integer power are Hermitian matrices and have real eigenvalues.
More explicitly, if with real, then form an eigenbasis and
Therefore
where denotes the matrix size, to distinguish it from the power .
The first three output components equal , while the fourth is . Hence
Its outputs have component sum zero, and applying to any such output leaves it fixed. Thus
The first three columns give a basis of the image of a linear map:
They are linearly independent because their first three coordinates are the standard basis. Their span is precisely the three-dimensional hyperplane .
Apply the Gram-Schmidt process. First . Subtracting the projection from gives , of length . Subtracting both preceding projections from gives , of length . Therefore an orthonormal basis is
Their pairwise dot products vanish and each has length one. Although this image admits an orthonormal basis, itself is an oblique projection: its kernel is the fourth coordinate axis, which is not perpendicular to the image, and .
If for a nonzero eigenvector, then . Since , : every eigenvalue is zero or one. Let be the number of ones, counted with algebraic multiplicity. The trace is and the determinant is the product of these eigenvalues. Therefore
Every output is fixed by a second application:
The image of a linear map is thus the eigenspace for the eigenvalue one. To establish its dimension without assuming diagonalizability, decompose every input as
The first term lies in the image of a linear map and the second in its kernel. Their intersection is zero, because a vector in the image is fixed by whereas a vector in the kernel is killed by . Thus the space is their direct sum, and an adapted basis puts into the form , where . Comparing traces gives :
This is an idempotent linear map, and trace of an idempotent equals its rank. If , no nonzero output exists. The dimension in the question refers to the image subspace, including its zero vector.
By the Cartesian formula for the curl,

There are unlisted articles, also show them or only show them.