Some bounds for the eigenfunctions of the Stokes problem under Navier boundary conditions in a cube
Gianni ARIOLI - Alessio FALOCCHI - Filippo GAZZOLA · 2025
Read Some bounds for the eigenfunctions of the Stokes problem under Navier boundary conditions in a cube by Gianni ARIOLI - Alessio FALOCCHI - Filippo GAZZOLA (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2504-06679.txt.
Introduction
Gianni Arioli, Alessio Falocchi, and Filippo Gazzola prove bounds on the eigenfunctions of the Stokes problem in a cube, in a 2025 paper. The Stokes equations describe slow, viscous fluid flow, and their eigenfunctions are the natural vibration modes of such a flow in a box. The boundary condition is Navier's, which lets the fluid slip along the walls in proportion to the stress there, rather than sticking fast. The authors establish explicit bounds on those modes, feeding analysis of flow stability and numerical methods.
Catalog record
| CREATOR | Gianni ARIOLI - Alessio FALOCCHI - Filippo GAZZOLA |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 24 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2504-06679
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Some bounds for the eigenfunctions of the Stokes problem under Navier boundary conditions in a cube Gianni ARIOLI - Alessio FALOCCHI - Filippo GAZZOLA Abstract Dipartimento di Eccellenza MUR 2023-2027 (Italy) 1 Introduction In the cube \Omega=(0,\pi)^{3} we consider the following Stokes eigenvalue problem, studied in [ 2 ] : \left\{\begin{array}[]{ll}-\Delta\varphi=\lambda\varphi&\mbox{in }\Omega\,,\\ \nabla\cdot\varphi=0&\mbox{in }\Omega\,,\\ \varphi_{1}=\partial_{x}\varphi_{2}=\partial_{x}\varphi_{3}=0&\mbox{on }\{0,\pi\}\times(0,\pi)\times(0,\pi)\,,\\ \varphi_{2}=\partial_{y}\varphi_{1}=\partial_{y}\varphi_{3}=0&\mbox{on }(0,\pi)\times\{0,\pi\}\times(0,\pi)\,,\\ \varphi_{3}=\partial_{z}\varphi_{1}=\partial_{z}\varphi_{2}=0&\mbox{on }(0,\pi)\times(0,\pi)\times\{0,\pi\}\,.\end{array}\right. (1.1) We are interested in some properties of its eigenfunctions, that are computed explicitly in [ 2 ] . Proposition 1.1 . [ 2 ] All the eigenvalues of ( 1.1 ) have finite multiplicity and can be ordered in an increasing divergent sequence \{\lambda_{k}\}_{k\in\mathbb{N}_{+}} , in which the eigenvalues are repeated according to their multiplicity. For m,n,p\in\mathbb{N}_{+} , the eigenfunctions X_{0,n,p}(\xi):=\frac{2}{\sqrt{\pi^{3}(n^{2}+p^{2})}}\left(\begin{array}[]{c}0\\ p\sin(ny)\cos(pz)\\ -n\cos(ny)\sin(pz)\end{array}\right)\ , (1.2) Y_{m,0,p}(\xi):=\frac{2}{\sqrt{\pi^{3}(m^{2}+p^{2})}}\left(\begin{array}[]{c}-p\sin(mx)\cos(pz)\\ 0\\ m\cos(mx)\sin(pz)\end{array}\right)\ , (1.3) Z_{m,n,0}(\xi):=\frac{2}{\sqrt{\pi^{3}(m^{2}+n^{2})}}\left(\begin{array}[]{c}n\sin(mx)\cos(ny)\\ -m\cos(mx)\sin(ny)\\ 0\end{array}\right)\ , (1.4) V_{m,n,p}(\xi):=\frac{2\sqrt{2}\,\cos(mx)}{\sqrt{\pi^{3}(n^{2}+p^{2})}}\left(\begin{array}[]{c}0\\ p\sin(ny)\cos(pz)\\ -n\cos(ny)\sin(pz)\end{array}\right)\ , (1.5) W_{m,n,p}(\xi):=\frac{2\sqrt{2}}{\sqrt{\pi^{3}(m^{2}+n^{2}+p^{2})(n^{2}+p^{2})}}\left(\begin{array}[]{c}(n^{2}+p^{2})\sin(mx)\cos(ny)\cos(pz)\\ -mn\cos(mx)\sin(ny)\cos(pz)\\ -mp\cos(mx)\cos(ny)\sin(pz)\end{array}\right)\ , (1.6) form an orthogonal basis in U which is orthonormal in H . The statements proved in this paper will be used in [ 1 ] . We use the notation \xi:=(x,y,z)\in\overline{\Omega} . 2 Bounds for the eigenfunctions We first compute the squared L^{\infty}(\Omega) -norm of the eigenfunctions in Proposition 1.1 . Proposition 2.1 . Let m,n,p\in\mathbb{N}_{+} , then \begin{split}&\|X_{0,n,p}\|_{L^{\infty}(\Omega)}^{2}=\dfrac{4\max\{n^{2},p^{2}\}}{\pi^{3}(n^{2}+p^{2})},\quad\|Y_{m,0,p}\|_{L^{\infty}(\Omega)}^{2}=\dfrac{4\max\{m^{2},p^{2}\}}{\pi^{3}(m^{2}+p^{2})},\quad\|Z_{m,n,0}\|_{L^{\infty}(\Omega)}^{2}=\dfrac{4\max\{m^{2},n^{2}\}}{\pi^{3}(m^{2}+n^{2})},\\ &\|V_{m,n,p}\|_{L^{\infty}(\Omega)}^{2}=\dfrac{8\max\{n^{2},p^{2}\}}{\pi^{3}(n^{2}+p^{2})},\quad\|W_{m,n,p}\|_{L^{\infty}(\Omega)}^{2}=\dfrac{8\max\{(n^{2}+p^{2})^{2},m^{2}n^{2},m^{2}p^{2}\}}{\pi^{3}(n^{2}+p^{2})(m^{2}+n^{2}+p^{2})}.\end{split} Let {\bf e}=(a,b,c)\in\mathbb{R}^{3} be a unit vector, we now determine the maximum of the projection of the eigenfunctions in Proposition 1.1 onto {\bf e} . Proposition 2.2 . Let m,n,p\in\mathbb{N}_{+} , {\bf e}=(a,b,c)\in\mathbb{R}^{3} be such that a^{2}+b^{2}+c^{2}=1 . Then \begin{split}&\|{\bf e}\cdot X_{0,n,p}\|_{L^{\infty}(\Omega)}^{2}=\dfrac{4\max\{b^{2}p^{2},c^{2}n^{2}\}}{\pi^{3}(n^{2}+p^{2})},\\ &\|{\bf e}\cdot Y_{m,0,p}\|_{L^{\infty}(\Omega)}^{2}=\dfrac{4\max\{a^{2}p^{2},c^{2}m^{2}\}}{\pi^{3}(m^{2}+p^{2})},\\ &\|{\bf e}\cdot Z_{m,n,0}\|_{L^{\infty}(\Omega)}^{2}=\dfrac{4\max\{a^{2}n^{2},b^{2}m^{2}\}}{\pi^{3}(m^{2}+n^{2})},\\ &\|{\bf e}\cdot V_{m,n,p}\|_{L^{\infty}(\Omega)}^{2}=\dfrac{8\max\{b^{2}p^{2},c^{2}n^{2}\}}{\pi^{3}(n^{2}+p^{2})},\\ &\|{\bf e}\cdot W_{m,n,p}\|_{L^{\infty}(\Omega)}^{2}=\dfrac{8\max\{a^{2}(n^{2}+p^{2})^{2},b^{2}m^{2}n^{2},c^{2}m^{2}p^{2}\}}{\pi^{3}(n^{2}+p^{2})(m^{2}+n^{2}+p^{2})}.\end{split} (2.1) In particular, ( 2.1 ) 1 is equivalent to \begin{split}&\max_{(y,z)\in[0,\pi]^{2}}\big[bp\sin(ny)\cos(pz)-cn\cos(ny)\sin(pz)\big]^{2}=\max\{b^{2}p^{2},c^{2}n^{2}\},\end{split} whereas ( 2.1 ) 5 is equivalent to \begin{split}&\displaystyle\max_{(x,y,z)\in\overline{\Omega}}\big[a(n^{2}\!+\!p^{2})\sin(mx)\cos(ny)\cos(pz)\!-\!bmn\cos(mx)\sin(ny)\cos(pz)\!-\!cmp\cos(mx)\cos(ny)\sin(pz)\big]^{2}\\ &=\max\{a^{2}(n^{2}+p^{2})^{2},b^{2}m^{2}n^{2},c^{2}m^{2}p^{2}\}\,.\end{split} Hence, the proof of ( 2.1 ) 1,2,3,4 is almost immediate, while for ( 2.1 ) 5 it is lengthy and tedious. Next, we bound sums of the squared projections of the eigenfunctions along {\bf e} . For instance, fix \xi\in\overline{\Omega} , \mu>0 , and consider \displaystyle\sum_{i=1}^{h}\dfrac{\big({\bf e}\cdot X_{0,n_{i},p_{i}}(\xi)\big)^{2}}{(\lambda_{i}+\mu)^{2}} \displaystyle= \displaystyle\sum_{i=1}^{h}\dfrac{4}{\pi^{3}(n_{i}^{2}+p_{i}^{2})}\dfrac{\big(bp_{i}\sin(n_{i}y)\cos(p_{i}z)-cn_{i}\cos(n_{i}y)\sin(p_{i}z)\big)^{2}}{(n_{i}^{2}+p_{i}^{2}+\mu)^{2}} \displaystyle\leq \displaystyle\dfrac{4}{\pi^{3}}\sum_{i=1}^{h}\dfrac{\max\{b^{2}p_{i}^{2},c^{2}n_{i}^{2}\}}{(n_{i}^{2}+p_{i}^{2})(n_{i}^{2}+p_{i}^{2}+\mu)^{2}}\leq\dfrac{4}{\pi^{3}}\int_{0}^{\infty}\int_{0}^{\infty}\dfrac{\max\{b^{2}y^{2},c^{2}x^{2}\}}{(x^{2}+y^{2})(x^{2}+y^{2}+\mu)^{2}}dxdy, where we applied Proposition 2.2 . Similarly, for the sums related to the other four families of eigenfunctions in Proposition 1.1 . Then we compute the integral bounding the sum. To do this, with the convention that \arctan(+\infty)=\tfrac{\pi}{2} , for (b,c)\in\mathbb{R}^{2} such that b^{2}+c^{2}\leq 1 we introduce the function \displaystyle\Upsilon(b,c) \displaystyle:= \displaystyle b^{2}\arctan\Big|\frac{c}{b}\Big|+c^{2}\arctan\Big|\frac{b}{c}\Big|+|bc|, (2.2) and we state the next result. Proposition 2.3 . Let \mu>0 , m,n,p\in\mathbb{N}_{+} and (a,b,c)\in\mathbb{R}^{3} be such that a^{2}+b^{2}+c^{2}=1 . Then \begin{split}&\int_{0}^{\infty}\int_{0}^{\infty}\dfrac{\max\{b^{2}y^{2},c^{2}x^{2}\}}{(x^{2}+y^{2})(x^{2}+y^{2}+\mu)^{2}}dxdy=\dfrac{1}{4\mu}\,\Upsilon(b,c)\\ &\int_{0}^{\infty}\int_{0}^{\infty}\int_{0}^{\infty}\dfrac{\max\{b^{2}y^{2},c^{2}x^{2}\}}{(x^{2}+y^{2})(x^{2}+y^{2}+z^{2}+\mu)^{2}}dxdydz=\dfrac{\pi}{8\sqrt{\mu}}\,\Upsilon(b,c)\\ &\int_{0}^{\infty}\int_{0}^{\infty}\int_{0}^{\infty}\dfrac{\max\{a^{2}(y^{2}+z^{2})^{2},b^{2}x^{2}y^{2},c^{2}x^{2}z^{2}\}}{(y^{2}+z^{2})(x^{2}+y^{2}+z^{2})(x^{2}+y^{2}+z^{2}+\mu)^{2}}\,dxdydz\leq\frac{\pi}{12\sqrt{\mu}}\bigg[\pi\,a^{2}+\dfrac{\Upsilon(b,c)}{2}\bigg].\end{split} (2.3) Our last computation deals with a suitable combination of the integrals in Proposition 2.3 in order to maximize it with respect to the components of the unit vector {\bf e} . Proposition 2.4 . Let k_{0}:=1+4\sqrt{2}\frac{\pi}{3} , k_{1}:=\frac{2}{3}\sqrt{2}\pi^{2} and \Gamma(a,b,c):=k_{0}\Upsilon(b,c)+\Upsilon(a,c)+\Upsilon(a,b)+k_{1}a^{2}, where the function \Upsilon(\cdot,\cdot) is defined in ( 2.2 ). Then it holds \max_{a^{2}+b^{2}+c^{2}=1}\ \Gamma(a,b,c)\approx 10.91\,. (2.4) 3 Proofs of the bounds Proof of Proposition 2.1 . . We maximize \xi\mapsto|X_{0,n,p}(\xi)|^{2}=\dfrac{4}{\pi^{3}(n^{2}+p^{2})}[p^{2}\sin^{2}(ny)\cos^{2}(pz)+n^{2}\cos^{2}(ny)\sin^{2}(pz)]\qquad\xi\in\overline{\Omega}\,. By setting Y=\sin^{2}(ny) and Z=\sin^{2}(pz) , this is equivalent to maximize \Gamma(Y,Z):=p^{2}Y(1-Z)+n^{2}(1-Y)Z over the square (Y,Z)\in[0,1]^{2} . The function \Gamma is harmonic and strictly positive in (0,1)^{2} : therefore, it achieves the (positive) maximum on the boundary of the square [0,1]^{2} . Hence, \max_{(Y,Z)\in[0,1]^{2}}\ \Gamma(Y,Z)=\max\{n^{2},p^{2}\}, (3.1) giving the first norm in Proposition 2.1 . We proceed similarly for the eigenfunctions Y_{m,0,p}(\xi) and Z_{m,n,0}(\xi) . On the other hand, if we allow nonconstant coefficients in the linear combinations, we find the following relations between eigenfunctions: V_{m,n,p}(\xi)=\sqrt{2}\,\cos(mx)\,X_{0,n,p}(\xi)\qquad\forall\xi\in\Omega\,, (3.2) W_{m,n,p}(\xi)=\tfrac{n\sqrt{2}\sqrt{m^{2}+n^{2}}}{\sqrt{(m^{2}+n^{2}+p^{2})(n^{2}+p^{2})}}\,\cos(pz)\,Z_{m,n,0}(\xi)-\tfrac{p\sqrt{2}\sqrt{m^{2}+p^{2}}}{\sqrt{(m^{2}+n^{2}+p^{2})(n^{2}+p^{2})}}\,\cos(ny)\,Y_{m,0,p}(\xi)\quad\forall\xi\in\Omega\,. Then, for the eigenfunctions ( 1.5 ), from ( 3.2 ) we infer that \|V_{m,n,p}\|_{L^{\infty}(\Omega)}^{2}=2\|X_{0,n,p}\|_{L^{\infty}(\Omega)}^{2}=\dfrac{8\max\{n^{2},p^{2}\}}{\pi^{3}(n^{2}+p^{2})}. Finally, for the eigenfunctions ( 1.6 ) we notice that |W_{m,n,p}(\xi)|^{2}=\tfrac{8[(n^{2}+p^{2})^{2}\sin^{2}(mx)\cos^{2}(ny)\cos^{2}(pz)+m^{2}n^{2}\cos^{2}(mx)\sin^{2}(ny)\cos^{2}(pz)+m^{2}p^{2}\cos^{2}(mx)\cos^{2}(ny)\sin^{2}(pz)]}{\pi^{3}(n^{2}+p^{2})(m^{2}+n^{2}+p^{2})} and we have to maximize this function over \overline{\Omega} . By setting X=\sin^{2}(mx) , Y=\sin^{2}(ny) , Z=\sin^{2}(pz) , and \Gamma(\xi):=(n^{2}+p^{2})^{2}X(1-Y)(1-Z)+m^{2}n^{2}(1-X)Y(1-Z)+m^{2}p^{2}(1-X)(1-Y)Z\,, we are led to maximize \Gamma over the cube [0,1]^{3} . Since \Gamma is positive and harmonic, it achieves its maximum on the boundary of this cube and \max_{\xi\in[0,1]^{3}}\Gamma(\xi)=\max\{(n^{2}+p^{2})^{2},m^{2}n^{2},m^{2}p^{2}\}\, completing the proof of Proposition 2.1 . \Box Proof of Proposition 2.2 . . We begin with the proof of ( 2.1 ) 1 . Let us introduce the function M(y,z)=\big[bp\sin(ny)\cos(pz)-cn\cos(ny)\sin(pz)\big]^{2} with b^{2}+c^{2}\leq 1 . The gradient of M vanishes either when M=0 (and M attains its minimum 0) or when (y,z)\in[0,\pi]^{2} solve the homogeneous trigonometric system \begin{pmatrix}bp&cn\\ cn&bp\\ \end{pmatrix}\begin{pmatrix}\cos(ny)\cos(pz)\\ \sin(ny)\sin(pz)\\ \end{pmatrix}=\begin{pmatrix}0\\ 0\\ \end{pmatrix}\,. If (bp)^{2}\neq(cn)^{2} we find as possible extremal points \begin{cases}\cos(ny)\cos(pz)=0\\ \sin(ny)\sin(pz)=0\end{cases}\qquad\Longrightarrow\qquad\cos(ny)=\sin(pz)=0\mbox{ or }\sin(ny)=\cos(pz)=0\,, which proves the proposition. If (bp)^{2}=(cn)^{2} , then M(y,z)=c^{2}n^{2}\sin^{2}(ny\mp pz)\leq c^{2}n^{2} , proving again the proposition. The proof of ( 2.1 ) 2,3 follows similarly. To prove ( 2.1 ) 4 we recall ( 3.2 ), so that ({\bf e}\cdot V_{m,n,p}(\xi))^{2}=\frac{8\cos^{2}(mx)}{\pi^{3}(n^{2}+p^{2})}M(y,z); arguing on the function M(y,z) as before we obtain ( 2.1 ) 4 . The remaining part of the proof is devoted to ( 2.1 ) 5 . We put A:=a(n^{2}+p^{2}) , B:=-bmn , C:=-cmp and, over \overline{\Omega} , define the function K(\xi)=A\sin(mx)\cos(ny)\cos(pz)+B\cos(mx)\sin(ny)\cos(pz)+C\cos(mx)\cos(ny)\sin(pz)\,. The proof follows if we show that \|K\|_{L^{\infty}(\Omega)}=\max\{|A|,|B|,|C|\}\,. (3.3) By the Weierstrass Theorem, K achieves a maximum over \overline{\Omega} and two different situations may occur. \bullet The maximum is achieved on \partial\Omega . Here, we notice that |K(0,y,z)|=|K(\pi,y,z)|=|B\sin(ny)\cos(pz)+C\cos(ny)\sin(pz)|\quad\forall(y,z)\in[0,\pi]^{2}\,, |K(x,0,z)|=|K(x,\pi,z)|=|A\sin(mx)\cos(pz)+C\cos(mx)\sin(pz)|\quad\forall(x,z)\in[0,\pi]^{2}\,, |K(x,y,0)|=|K(x,y,\pi)|=|A\sin(mx)\cos(ny)+B\cos(mx)\sin(ny)|\quad\forall(x,y)\in[0,\pi]^{2}\,. Then, by Proposition 2.2 , we know that \max_{(y,z)\in[0,\pi]^{2}}|K(0,y,z)|=\max_{(y,z)\in[0,\pi]^{2}}|K(\pi,y,z)|=\max\{|B|,|C|\}\,, \max_{(x,z)\in[0,\pi]^{2}}|K(x,0,z)|=\max_{(x,z)\in[0,\pi]^{2}}|K(x,\pi,z)|=\max\{|A|,|C|\}\,, \max_{(x,y)\in[0,\pi]^{2}}|K(x,y,0)|=\max_{(x,y)\in[0,\pi]^{2}}|K(x,y,\pi)|=\max\{|A|,|B|\}\,. Therefore, if the maximum of |K| is achieved on \partial\Omega , then ( 3.3 ) is proved. \bullet The maximum is achieved in \Omega . Let \overline{\xi}\in\Omega be a maximum point for |K(\xi)| , then \nabla K(\overline{\xi})=0 , i.e., \left\{\begin{array}[]{lll}A\cos(mx)\cos(ny)\cos(pz)-B\sin(mx)\sin(ny)\cos(pz)-C\sin(mx)\cos(ny)\sin(pz)=0\,,\\ -A\sin(mx)\sin(ny)\cos(pz)+B\cos(mx)\cos(ny)\cos(pz)-C\cos(mx)\sin(ny)\sin(pz)=0\,,\\ -A\sin(mx)\cos(ny)\sin(pz)-B\cos(mx)\sin(ny)\sin(pz)+C\cos(mx)\cos(ny)\cos(pz)=0\,.\end{array}\right. Putting X=mx , Y=ny , Z=pz , the above system becomes \left\{\begin{array}[]{lll}A\cos(X)\cos(Y)\cos(Z)-B\sin(X)\sin(Y)\cos(Z)-C\sin(X)\cos(Y)\sin(Z)=0\,,\\ A\sin(X)\sin(Y)\cos(Z)-B\cos(X)\cos(Y)\cos(Z)+C\cos(X)\sin(Y)\sin(Z)=0\,,\\ A\sin(X)\cos(Y)\sin(Z)+B\cos(X)\sin(Y)\sin(Z)-C\cos(X)\cos(Y)\cos(Z)=0\end{array}\right. (3.4) and, from now on, the original parameters ( a,b,c,x,y,z,m,n,p ) will not appear any more. We compute the values of K=K(X,Y,Z) for any triplet (X,Y,Z)\in\mathbb{R}^{3} satisfying ( 3.4 ). To this end, we distinguish several cases, with some subcases. Case 1: \cos(X)\cos(Y)\cos(Z)=0 . If \cos(X)=0 and if the critical point \overline{\xi}\in\Omega is a maximum for |K| , then |K(\overline{\xi})|=|A\cos(Y)\cos(Z)|\leq|A| , with equality for suitable (Y,Z) , so that ( 3.3 ) is satisfied. Similarly, if \cos(Y)=0 we find that |K(\overline{\xi})|\leq|B| , while if \cos(Z)=0 we find that |K(\overline{\xi})|\leq|C| . Therefore, in Case 1, ( 3.3 ) is satisfied. Case 2: \cos(X)\cos(Y)\cos(Z)\neq 0 . After dividing by \cos(X)\cos(Y)\cos(Z) each of the equations in ( 3.4 ), we obtain the system \left\{\begin{array}[]{lll}B\tan(X)\tan(Y)+C\tan(X)\tan(Z)=A\,,\\ A\tan(X)\tan(Y)+C\tan(Y)\tan(Z)=B\,,\\ A\tan(X)\tan(Z)+B\tan(Y)\tan(Z)=C\,.\end{array}\right. (3.5) In order to solve ( 3.5 ) we distinguish two subcases. Case 2.1: ABC=0 . If A=0 , then ( 3.5 ) 2 -( 3.5 ) 3 yield B^{2}=C^{2}>0 so that C=\pm B and K may be written as K(\xi)=B\cos(X)[\sin(Y)\cos(Z)\pm\cos(Y)\sin(Z)]=B\cos(X)\sin(Y\pm Z)\,, implying that |K(\xi)|\leq|B| for all \xi\in\overline{\Omega} and showing that ( 3.3 ) holds. Similarly, if B=0 then \|K\|_{L^{\infty}(\Omega)}\leq|C| , while if C=0 then \|K\|_{L^{\infty}(\Omega)}\leq|A| . Hence, if ABC=0 , then ( 3.3 ) holds. Case 2.2: ABC\neq 0 . In this case, the coefficient matrix of the (linear) system ( 3.5 ) is nonsingular and ( 3.5 ) is uniquely solvable in terms of the three unknowns \tan(X)\tan(Y) , \tan(X)\tan(Z) , \tan(Y)\tan(Z) . We find \tan(X)\tan(Y)=\frac{A^{2}\!+\!B^{2}\!-\!C^{2}}{2AB},\ \tan(X)\tan(Z)=\frac{A^{2}\!+\!C^{2}\!-B^{2}}{2AC},\ \tan(Y)\tan(Z)=\frac{B^{2}\!+\!C^{2}\!-\!A^{2}}{2BC}. (3.6) We now claim that \tan(X)\tan(Y)\tan(Z)=0\ \Longrightarrow\ \mbox{(\ref{claimK}) holds}. Indeed, if \tan(X)=0 from ( 3.6 ) 1 -( 3.6 ) 2 we infer that A^{2}+B^{2}-C^{2}=A^{2}+C^{2}-B^{2}=0 , which yields B^{2}=C^{2} and, at the corresponding critical point \overline{\xi} , we have that |K(\overline{\xi})|=|B|\,|\sin(Y)\cos(Z)\pm\cos(Y)\sin(Z)|=|B|\,|\sin(Y\pm Z)|\leq|B|\,, with equality