The difference of the two eigenvalues is , hence . Its expectation is
For , put . The set of angles satisfying consists of , and , of total length . Thus the full cumulative distribution function is
Differentiating on gives the probability density function
This is the arcsine distribution, equivalently the beta distribution with both parameters . It has no atoms at either endpoint.
The characteristic polynomial of the rotation matrix is
Its roots are , so
For example is an eigenvector for and for . At or the two eigenvalues coincide; these events have probability zero under the uniform distribution.
The correlation function of a point process is even and tends to zero as . Its derivative factors as
Consequently its local maxima occur at , and its local minima at . The heights are
Figure 1.
The three peaks of the GUE one-point correlation function
. Original plot of the exact correlation function of a point process calculated above. Filled markers indicate local maxima at ; hollow markers indicate the local minima at . The total area is three. Details.
For an interval , is the expected value of the number of eigenvalues in . Thus equally short intervals have the highest expected counts near
with the central peak highest. Each exchangeably labelled eigenvalue has the same marginal probability density function . The three peaks should not be interpreted as three separate marginal densities for ordered eigenvalues.
In this case the most likely joint configuration also lies at these three positions. On the ordered chamber, the logarithm of the joint probability density function, up to a constant, is
Its stationary equations are , satisfied by . For every nonzero real vector , the second directional derivative is
Thus is strictly concave on this convex chamber, proving that the stationary configuration is its unique global maximum. Permuting the labels gives the other maxima on .
Evaluate the orthogonal polynomial projection kernel on its diagonal:
The terms cancel. Integrating against Lebesgue measure gives for a standard normal random variable , checking the correlation function of a point process normalization.
The correlation functions of a point process use factorial normalization:
Start with . By Gaudin's lemma, integrating once gives , and integrating once gives . The factorial factors cancel, so the eigenvalues form a determinantal point process with
and
The last equality uses the real symmetry of this kernel. In particular, and , as required for three points and six ordered pairs of distinct points.
Use the symmetric joint probability density function of exchangeably labelled eigenvalues, normalized over all of . The finite-rank projection kernel has
Successive applications of Gaudin's lemma therefore give
It follows that
As a normalization check, the weighted Vandermonde determinant yields
If instead one uses ordered eigenvalues on , the normalized density on that chamber is six times the labelled density. In that convention the determinant prefactor on the chamber is one.
The squared norm of in is , where . Choose positive normalizing constants
Then are an orthonormal set for Lebesgue measure. Expanding the finite sums gives
Thus this is an orthogonal polynomial projection kernel, explicitly
The positive constants above are forced by the reproducing identity: before normalization, its convolution multiplies each independent term by . Each multiplier must equal one. Changing the sign of a real leaves the kernel unchanged.
Gaudin's lemma is the projection kernel determinant integration identity. Suppose an integral kernel satisfies
with the relevant integrals absolutely convergent. Define and . Then
In particular, a finite-rank projection kernel of rank has .
One can see the factor directly from the permutation expansion of the determinant. Terms whose permutation fixes integrate to . Every permutation on the other indices admits insertions of into a cycle. Each insertion reverses its sign, and the reproducing identity contracts its two adjacent kernel factors into one. These terms integrate to , proving the formula.
Use the inner product for a standard normal random variable . The degree-zero monic polynomial is . Write and . The conditions
use and . Hence these first three Probabilists' Hermite polynomials are
The Gaussian fourth moment is , so their squared norms are
Let be the -by- evaluation matrix . The kernel evaluation matrix is the Gram matrix , so multiplicativity of the determinant gives
Because the polynomials are monic of consecutive degrees, their coefficient matrix relative to is a triangular matrix with unit diagonal. It has determinant one, so the Vandermonde determinant invariance under monic basis change gives
Squaring removes the ordering sign. Thus the Gaussian unitary ensemble joint probability density function has precisely the kernel form
No orthogonality of the polynomials is needed for this step.
Let be row of , viewed as a real vector. These are independent standard Gaussian random vectors, and . The trace expression for the empirical spectral measure gives
We need the norm moments
For completeness, has the chi-squared distribution with degrees of freedom. By independence and the Gaussian integral, its moment-generating function is for . Differentiating at zero once, twice and three times proves these three identities.
For the first moment,
For the second moment, if , conditional expectation given gives , hence an unconditional value . Separating equal and unequal row indices yields
For the third moment, the summand is . There are three disjoint cases.
Consequently
The three limiting expected values are therefore
First , so the spectral theorem for real symmetric matrices gives real eigenvalues and an orthonormal basis of eigenvectors. Moreover, for every real vector ,
Thus is a positive semidefinite matrix. For any nonzero eigenvector with eigenvalue ,
This matrix is a scaled example of the Wishart distribution.
Write the four cyclic factors as , , and . The Gaussian moment pairing theorem gives three terms, corresponding to , and . We sum each term over .
For , the entry covariance gives
Indeed the four products leave respectively three, two, two and one free indices. A cyclic relabelling gives exactly the same sum for .
For the crossing pairing ,
The first product allows and freely, whereas each other product forces all four indices to coincide. Combining the three pairings and the trace normalization,
The spectral theorem for real symmetric matrices identifies the sum of the th powers of the eigenvalues with the trace of the th matrix power. Therefore the moment of the empirical spectral measure is
For odd , every summand has zero expectation by part (a)(ii), so for every .
For the second moment, use . There are diagonal entries with variance two and ordered off-diagonal entries with variance one. Their means vanish, giving
Hence .
The probabilistic form of Wick's theorem is Isserlis's theorem. If is a centered random vector with a multivariate Gaussian distribution and covariance matrix , its odd product moments vanish, and, for even ,
Here is the set of perfect matchings of the labelled positions. The formula allows repeated coordinates and singular covariance matrices. To see the pairing rule, differentiate the moment-generating function once in each coordinate and set : only terms pairing all differentiated coordinates survive, each with the displayed product of covariances.
Take , with . These variables are linear combinations of the original independent normal random variables, hence are jointly Gaussian. They have zero means, so
Substituting the entry covariance from part (a)(i) identifies each parenthesis with , giving the requested identity.
The joint probability distribution of all entries of is unchanged by , so and have the same probability distribution. For the specified cyclic product , this transformation multiplies by . Every such product is an integrable random variable, since normal random variables have finite moments of all orders. Thus, for odd ,
Repeated indices cause no difficulty: all factors change sign simultaneously. It would be incorrect to assume that these factors are independent.
Taking the transpose gives , so is a symmetric matrix. For , has the standard normal distribution, whereas has the normal distribution . Distinct upper-triangular entries depend on disjoint sets of the original independent random variables, so they remain independent.
Consequently the joint probability density function, relative to , is
Here because is a symmetric matrix. This is the Gaussian orthogonal ensemble, with off-diagonal variance one and diagonal variance two; the rescaled matrix has the corresponding off-diagonal variance.
We will also use its entry covariance. Expanding in the independent entries of , for which , gives
The symbols are Kronecker deltas. This formula includes the diagonal case, where the two terms both contribute.
Diagonalizing the matrix gives the normalized eigenvectors
with eigenvalues , respectively. The first projective measurement prepares up to global phase, regardless of the earlier state, conditional on obtaining .
The unitary time evolution under gives, with ,
The probability amplitudes for the second projective measurement are
The Born rule therefore gives the general spin-one transverse measurement after axial precession:
At the specified times , , so
For even , the result is certainly ; for odd , it is certainly . Although zero belongs to the spectrum of the observable, it never occurs at these times. Indeed, for even the evolved vector is , while for odd it is , where the minus sign is a global phase.
The Hamiltonian operator is already diagonal in the angular momentum eigenstate basis of part (b). Its three distinct energy eigenvalues and stationary states are
Thus the three column eigenvectors are , and , with energies , zero and , respectively. Each is nondegenerate since . Any nonzero scalar multiple gives the same stationary state after normalization; a quantum superposition of different energies is generally not stationary.
Recover the angular momentum operators from and . The spin-one angular momentum matrices in the chosen descending- orthonormal basis are
They are self-adjoint operators, satisfy , and give , as required for .

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