Write , and . The law of total probability gives , a convex combination of the two fixed conditional values. HenceThese are the closed sharp bounds in the usual mixture-model interpretation, obtained by allowing to range from zero to one. The joint masses realize every such mixture.
There is an endpoint qualification if both conditionals must literally be defined by ratios: then . If , the marginal lies strictly between them, so the stated endpoints are an infimum and supremum rather than achieved extrema. If , every allowed mixture has exactly that common value. This is the second case of sharp bounds from binary conditional probabilities.
Put and . Decompose the marginal as , where . It follows thatBoth bounds are attainable: choose none or all of the mass outside to belong to . More explicitly, the four joint cell masses can be , all nonnegative. This is the first case of sharp bounds from binary conditional probabilities.
By Bayes' theorem,The denominator sums red outcomes from both possible jar choices, as calculated in (i).
The law of total probability weights each jar's red fraction by its selection probability:The large difference in jar sizes does not change their equal selection probabilities; pooling their contents would describe a different experiment.
The inverse appearing in the assumption requires . Taking determinants of giveswhere the minus sign comes from the odd matrix order. Multiplying by gives . Since is real, the second factor is positive. Thus and is a singular matrix. This is the singularity forced by an odd-dimensional transpose-inverse identity, with the roles of the two matrix letters exchanged.
The transpose rule gives . If is orthogonal, , and the identity becomesMultiplying on the left by yields . This is precisely the defining condition for an antisymmetric matrix.
The three columns of are , , and . Each has squared Euclidean norm eight. Their pairwise dot products areHence , so . Thus is an orthogonal matrix, with . The determinant from (i) places it in the determinant-one component.
Near zero, and , so . Replacing by similarly gives . ThereforeFor all sufficiently small nonzero , both cosines are positive and the denominator logarithm is nonzero, so this quotient and its two-sided real limit are well-defined.
Here , so grows exponentially rather than decaying. The positive terms do not tend to zero; indeed they tend to . Therefore the series diverges by the necessary condition that the terms of a convergent series tend to zero.
Let . The ratio test gives , so the series converges absolutely. In fact the exponential series identifies its value aswith the term subtracted.
At fixed , the argument is linear in , with . Each differentiation therefore contributes this same constant factor, while the outside factor remains unchanged. ThusHere denotes the third derivative of the single-variable profile, not a partial derivative of .
Put and write for differentiation with respect to . Work on a real branch where the powers are defined and differentiable; , suffices for arbitrary real , and the parameter choice below has . At fixed , . The product rule and chain rule therefore give
The number being squared has polar form of a complex number . Squaring doubles its argument, giving . Hence
The logarithm's argument is , with complex modulus one. The principal complex logarithm is therefore , givingIf denotes all logarithm values rather than the principal branch, they are for : the real part remains zero and the imaginary part changes by integer multiples of . The principal value is the conventional single-valued answer.
The adjugate matrix formula givesMultiplication by the right-side column yieldsMultiplying the inverse by checks the identity matrix, and substitution of this solution into the three original rows gives respectively.
Write the linear system as withExpansion along the first row gives . Since , is an invertible matrix, and is the unique solution for this right side.
There are unlisted articles, also show them or only show them.