Put , so . The power series coefficients of the hyperbolic sine preprocessing are
Thus for every , and the series converge on the full interval. Applying coefficient-dominated entrywise positivity gives .
Every diagonal entry belongs to one of the diagonal blocks and equals . Because and , this is . Therefore
The matrix lies in the elliptope and admits a Gram matrix representation by unit vectors. This preprocessing is the Krivine rounding scheme; the equality of absolute coefficients is what preserves positive semidefiniteness even though the cross-block sine coefficients alternate in sign.
The three fixed- estimates below cease to be uniform as the minimum of the phase approaches the endpoint. For the cubic endpoint-to-saddle transition, write
The Taylor series of the hyperbolic sine gives, for bounded and fixed ,
The cubic term controls the tail, so localization of the Laplace integral gives the uniform leading formula
Here is the cubic Laplace transition integral, equal to in terms of the Scorer Hi function.
To recover the endpoint regime, let . Scale in ; the cubic term becomes negligible and . Therefore
which agrees with part (i) in the overlap . At , the Gamma integral gives , recovering part (iii).
For , set . The exponent has its maximum at , with second derivative . Thus Laplace's method gives
and the transition formula becomes
For with , part (ii) has
Putting reproduces both the exponential and its prefactor. For relative agreement of these leading exponentials, one may use the overlap , which makes . Thus the same transition integral connects all three regimes.
Figure 1.
The cubic endpoint-to-saddle transition
. Direct numerical integration of the original phase approaches the same cubic transition function as the large parameter increases. Negative transition parameter places the minimum inside the interval; positive parameter leaves an ordinary endpoint minimum. Details.
Write
The first column has unit length for the indefinite quadratic form:
Because , there is a unique such that
The two columns are orthogonal for the same form and the second has squared length :
The vectors with these properties are
Using the hyperbolic cosine and hyperbolic sine, the two possible families are therefore
When , set . The reduced ordinary differential equation is . Enforcing the same initial conditions gives
The corresponding overdamped normal mode can also be written
Both rates in the exponential functions are negative because , so the hyperbolic sine factor does not prevent decay.
Set in Laplace's equation. Separation of variables gives
The Sturm-Liouville problem in has the constant mode with , and the eigenfunctions with , . Indeed, integration by parts makes nonnegative; for the first Neumann boundary condition removes the sine term and the second requires to be an integer. For , both derivative conditions make constant.
Expand the top data using a Fourier cosine series:
The constant coefficient is
For , the constant part has zero integral against , and two integrations by parts give
Since the bottom value is zero, a convenient factor depending on for the nonconstant modes is , which vanishes at . Its value at is . The zero mode is linear in , so matching both constant boundary values gives
To obtain exactly the paper's form, use the hyperbolic-function identity
Thus
Each product of hyperbolic sine or hyperbolic cosine with has . The zero mode is also a harmonic function. On , the factor lies between zero and one. Since converges, the Weierstrass M-test gives uniform convergence on the closed rectangle. The series therefore gives zero at the bottom and the continuous Fourier cosine series of at the top, including its endpoints. On compact subsets with , the factors decay exponentially with , justifying termwise partial derivatives. In particular on both open vertical edges because every differentiated cosine factor vanishes there. The solution has finite Dirichlet energy: Fourier orthogonality gives the th nonconstant mode's energy as , whose sum is finite. Here is the hyperbolic cotangent. Thus a proves uniqueness in that class.
There is a corner compatibility condition for mixed boundary data in the printed wording. The top data have , but the paper writes even at . The series solves the interior equation and all the open-edge boundary conditions; the literal derivative condition at is incompatible with the top data. A function differentiable in along the top edge would necessarily have . Hence it cannot simultaneously satisfy the printed zero derivative at that corner, and there is no solution with all those literal corner requirements. The standard interpretation of this mixed boundary condition imposes the Neumann boundary condition on the open sides and the Dirichlet boundary condition continuously on the horizontal edges. Differentiating the series term by term on the top edge would be invalid at this corner; the uniform convergence of the undifferentiated series does not license that operation.
Substitute the exponential definitions of hyperbolic cosine and hyperbolic sine. The cross terms cancel:
This proves the identity for real arguments, and the same algebra proves it for complex arguments.
Separate one factor of the hyperbolic sine and use the hyperbolic Pythagorean identity . With and , integration by substitution gives
For real , the hyperbolic cosine never vanishes. Differentiating the result gives , checking the sign of the reciprocal term.
The exponential definition of hyperbolic sine gives , or . Taking every branch of the complex logarithm gives
For hyperbolic cosine, multiply by to obtain . Thus , giving
The exponential definitions of hyperbolic sine and hyperbolic cosine, together with Euler's formula, give and . The argument of the complex logarithm therefore reduces to .
Using the principal logarithm, , so
If no branch of the complex logarithm is selected, its complete set of values is , . The real part remains zero, while the imaginary part is . The original paper does not explicitly state a branch convention.
The definition of hyperbolic sine simplifies the composition exactly for :
Set . The geometric series , valid for , then gives
At , all odd-indexed cosine terms vanish, while . Thus the Fourier series from part (c) gives
Divide by and rearrange. The hyperbolic sine identity yields
The evaluation is legitimate by the convergence established in part (c).
The Cauchy-Riemann equations are and . A holomorphic function is smooth, so equality of mixed partial derivatives gives
Thus its real and imaginary parts are harmonic functions.
An entire function is holomorphic at every finite point of the complex plane. The globally convergent power series for the exponential implies that and are entire. Their difference therefore gives
at every complex , proving that the hyperbolic sine is entire. For powers, the base case has derivative one everywhere. If is entire with derivative , the product rule gives everywhere. Induction proves that every positive integer power is entire.
A function has a pole of order at when it is holomorphic in a punctured neighbourhood and extends holomorphically to with . An isolated essential singularity is neither a removable singularity nor a pole; equivalently its Laurent series has infinitely many nonzero negative-power terms. In particular
has an essential singularity at zero.
For , the Laurent series about infinity is
To classify the point correctly, use . Then , and is holomorphic with value one at zero. Therefore
The pole at infinity is compatible with an essential singularity at the finite point zero. Negative powers of are regular powers of the local coordinate near infinity, so they do not make that point essential.
Start with the exponential function definition of the hyperbolic sine and use Euler's formula:
Identifying the hyperbolic sine and hyperbolic cosine in the final line gives
Put and . Traverse the rectangle with positive orientation of a contour, taking . Part (i) shows that the only enclosed poles of are
They are simple poles because
The residue formula for a quotient at a simple denominator zero gives
Their imaginary terms cancel and their residues sum to . Thus Cauchy's residue theorem gives
Figure 1.
The positively oriented contour encloses the two poles at
. The top edge is traversed from right to left. Its reversed orientation of a contour and the imaginary shift in the numerator produce times the desired real integral. Details.
The choice of numerator makes the horizontal sides useful. The hyperbolic sine satisfies , so . Consequently the lower and upper contour integrals, with their actual orientations, add to
It remains to justify discarding both vertical sides. On either side , , the reverse triangle inequality and part (b)(ii) imply
This lower bound is positive for sufficiently large . Also and each vertical side has length . The estimation lemma bounds the absolute value of the sum of the vertical contributions by
The real improper integral converges absolutely, since its integrand is continuous on the real axis and decays like at infinity. Combining all four sides and taking gives
Therefore
An independent real-variable check uses the hyperbolic tangent substitution : the integral becomes . Its antiderivative uses the inverse hyperbolic tangent, giving . As , this tends to , agreeing with and checking the factor of two.
The zeros of the hyperbolic sine are exactly , , since is equivalent to . Each zero is simple because . The quotient is therefore holomorphic precisely on
At zero, the Taylor expansion gives , so zero is a simple pole. At , the denominator has a double zero and the numerator is nonzero, giving a pole of order two. Thus none of the omitted points admits a holomorphic extension.
The hyperbolic sine satisfies . The complex exponential function is entire, and sums and compositions of entire functions are entire. Therefore