![]() |
Máximos y Mínimos de Funciones de Varias VariablesAngel Carrillo Hoyo, Elena de Oteyza de Oteyza\(^2\), Carlos Hernández Garciadiego\(^1\), Emma Lam Osnaya\(^2\) | ![]() |
Solución:
Tenemos la restricción \begin{equation*} x+y+z=24. \end{equation*} Queremos que su media geométrica \begin{equation*} \sqrt[3]{xyz} \end{equation*} sea máxima.
Consideramos las funciones \begin{equation*} f\left( x,y,z\right) =\sqrt[3]{xyz} \end{equation*} y \begin{equation*} g\left( x,y,z\right) =x+y+z-24, \end{equation*} con lo cual la restricción se escribe como \begin{equation*} g\left( x,y,z\right) =0. \end{equation*} Calculamos los gradientes de \(f\) y \(g\): \begin{eqnarray*} \nabla f\left( x,y,z\right) & = & \left( \dfrac{\partial f}{\partial x},\dfrac{ \partial f}{\partial y},\dfrac{\partial f}{\partial z}\right) =\left( \dfrac{ 1}{3}\left( xyz\right) ^{-\frac{2}{3}}\left( yz\right) ,\dfrac{1}{3}\left( xyz\right) ^{-\frac{2}{3}}\left( xz\right) ,\dfrac{1}{3}\left( xyz\right) ^{- \frac{2}{3}}\left( xy\right) \right) \\ \nabla g\left( x,y,z\right) & = & \left( \dfrac{\partial g}{\partial x},\dfrac{ \partial g}{\partial y},\dfrac{\partial g}{\partial z}\right) =\left( 1,1,1\right) . \end{eqnarray*} De acuerdo con el método de Lagrange, debemos resolver el sistema de ecuaciones \begin{eqnarray*} \nabla f\left( x,y,z\right) & = & \lambda \nabla g\left( x,y,z\right) \\ x+y+z & = & 24, \end{eqnarray*} es decir \begin{eqnarray*} \dfrac{1}{3}\left( xyz\right) ^{-\frac{2}{3}}\left( yz\right) & = & \lambda \\ \dfrac{1}{3}\left( xyz\right) ^{-\frac{2}{3}}\left( xz\right) & = & \lambda \\ \dfrac{1}{3}\left( xyz\right) ^{-\frac{2}{3}}\left( xy\right) & = & \lambda \\ x+y+z & = & 24. \end{eqnarray*} De las dos primeras, obtenemos \begin{equation*} x=y, \end{equation*} de la segunda y la tercera tenemos \begin{equation*} y=z. \end{equation*} Por tanto, \begin{equation*} x=y=z. \end{equation*} Sustituyendo en la cuarta obtenemos \begin{equation*} x=y=z=8 \end{equation*} y sustituyendo en cualquiera de las tres primeras, obtenemos: \begin{equation*} \lambda =\dfrac{1}{3}\left( 8^{3}\right) ^{-\frac{2}{3}}\left( 8\right) ^{2}=\dfrac{1}{3}. \end{equation*} Definimos la función auxiliar \begin{eqnarray*} h\left( x,y,z\right) & = & f\left( x,y,z\right) -\lambda g\left( x,y,z\right) \\ & = & \sqrt[3]{xyz}-\lambda \left( x+y+z-24\right) . \end{eqnarray*} Calculamos las derivadas de \(h\) de orden \(1\). \begin{eqnarray*} \dfrac{\partial h }{\partial x}\left( x,y,z\right) & = & \dfrac{1}{3}\left( xyz\right) ^{-\frac{2}{3}}\left( yz\right) -\lambda \\ \dfrac{\partial h }{\partial y}\left( x,y,z\right) & = & \dfrac{1}{3}\left( xyz\right) ^{-\frac{2}{3}}\left( xz\right) -\lambda \\ \dfrac{\partial h }{\partial z}\left( x,y,z\right) & = & \dfrac{1}{3}\left( xyz\right) ^{-\frac{2}{3}}\left( xy\right) -\lambda . \end{eqnarray*} Las derivadas parciales de \(h\) de orden \(2\) son: \begin{eqnarray*} \dfrac{\partial ^{2}h }{\partial x^{2}}\left( x,y,z\right) & = & \dfrac{1}{3} \left( -\dfrac{2}{3}\right) \left( xyz\right) ^{-\frac{5}{3}}\left( yz\right) ^{2}=-\dfrac{2}{9}\left( xyz\right) ^{-\frac{5}{3}}\left( yz\right) ^{2} \\ \dfrac{\partial ^{2}h }{\partial y^{2}}\left( x,y,z\right) & = & -\dfrac{2}{9} \left( xyz\right) ^{-\frac{5}{3}}\left( xz\right) ^{2} \\ \dfrac{\partial ^{2}h }{\partial z^{2}}\left( x,y,z\right) & = & -\dfrac{2}{9} \left( xyz\right) ^{-\frac{5}{3}}\left( xy\right) ^{2} \end{eqnarray*} y \begin{eqnarray*} \dfrac{\partial ^{2}h }{\partial y\partial x}\left( x,y,z\right) & = & -\dfrac{2 }{9}\left( xyz\right) ^{-\frac{5}{3}}\left( yz\right) \left( xz\right) + \dfrac{1}{3}z\left( xyz\right) ^{-\frac{2}{3}}= \dfrac{\partial ^{2}h }{\partial x\partial y}\left(x,y,z\right) \\ \dfrac{\partial ^{2}h }{\partial z\partial y}\left( x,y,z\right) & = & -\dfrac{2 }{9}\left( xyz\right) ^{-\frac{5}{3}}\left( xz\right) \left( xy\right) + \dfrac{1}{3}x\left( xyz\right) ^{-\frac{2}{3}}= \dfrac{\partial ^{2}h }{\partial y\partial z}\left(x,y,z\right) \\ \dfrac{\partial ^{2}h }{\partial z\partial x}\left( x,y,z\right) & = & -\dfrac{2 }{9}\left( xyz\right) ^{-\frac{5}{3}}\left( yz\right) \left( xy\right) + \dfrac{1}{3}y\left( xyz\right) ^{-\frac{2}{3}}= \dfrac{\partial ^{2}h }{\partial x\partial z}\left(x,y,z\right). \end{eqnarray*} Evaluamos las derivadas parciales de orden \(2\) en el punto \(\left( 8,8,8\right) \) \begin{equation*} \dfrac{\partial ^{2}h }{\partial x^{2}}\left( 8,8,8\right)=-\dfrac{2}{9} \left( 8^{3}\right) ^{-\frac{5}{3}}\left( 8^{2}\right) ^{2}=-\dfrac{2}{9} \left( 8\right) ^{-5}\left( 8^{4}\right) =-\dfrac{1}{36}, \end{equation*} de la misma manera \begin{equation*} \dfrac{\partial ^{2}h }{\partial y^{2}}\left( 8,8,8\right)=-\dfrac{1}{36}= \dfrac{\partial ^{2}h }{\partial z^{2}}\left( 8,8,8\right) \end{equation*} y \begin{eqnarray*} \dfrac{\partial ^{2}h }{\partial y\partial x}\left( 8,8,8\right) & = & \dfrac{ \partial ^{2}h }{\partial x\partial y}\left( 8,8,8\right)=-\dfrac{2}{9} \left( 8^{3}\right) ^{-\frac{5}{3}}\left( 8^{4}\right) +\dfrac{1}{3}\left( 8\right) \left( 8^{3}\right) ^{-\frac{2}{3}}=-\dfrac{2}{9}\left( \dfrac{1}{8} \right) +\dfrac{1}{3}\left( \dfrac{1}{8}\right) =\dfrac{1}{72} \\ \dfrac{\partial ^{2}h }{\partial z\partial y}\left( 8,8,8\right) & = & \dfrac{ \partial ^{2}h }{\partial y\partial z}\left( 8,8,8\right)=\dfrac{1}{72} \\ \dfrac{\partial ^{2}h }{\partial z\partial x}\left( 8,8,8\right) & = & \dfrac{ \partial ^{2}h }{\partial x\partial z}\left( 8,8,8\right)=\dfrac{1}{72}. \end{eqnarray*} Evaluamos el determinante hessiano limitado en \(\left( 8,8,8\right) \) \begin{eqnarray*} \left\vert \overline{H}_{4}\left( 8,8,8\right) \right\vert & = & \left\vert \begin{array}{cccc} 0 & \dfrac{\partial g}{\partial x}\left( 8,8,8\right) & \dfrac{\partial g}{ \partial y}\left( 8,8,8\right) & \dfrac{\partial g}{\partial z}\left( 8,8,8\right) \\ & & & \\ \dfrac{\partial g}{\partial x}\left( 8,8,8\right) & \dfrac{\partial ^{2}h}{ \partial x^{2}}\left( 8,8,8\right) & \dfrac{\partial ^{2}h}{\partial y\partial x}\left( 8,8,8\right) & \dfrac{\partial ^{2}h}{\partial z\partial x }\left( 8,8,8\right) \\ & & & \\ \dfrac{\partial g}{\partial y}\left( 8,8,8\right) & \dfrac{\partial ^{2}h}{ \partial x\partial y}\left( 8,8,8\right) & \dfrac{\partial ^{2}h}{\partial y^{2}}\left( 8,8,8\right) & \dfrac{\partial ^{2}h}{\partial z\partial y} \left( 8,8,8\right) \\ & & & \\ \dfrac{\partial g}{\partial z}\left( 8,8,8\right) & \dfrac{\partial ^{2}h}{ \partial x\partial z}\left( 8,8,8\right) & \dfrac{\partial ^{2}h}{\partial y\partial z}\left( 8,8,8\right) & \dfrac{\partial ^{2}h}{\partial z^{2}} \left( 8,8,8\right) \end{array} \right\vert \\ & & \\ & = & \left\vert \begin{array}{cccc} 0 & 1 & 1 & 1 \\ 1 & -\dfrac{1}{36} & \dfrac{1}{72} & \dfrac{1}{72} \\ 1 & \dfrac{1}{72} & -\dfrac{1}{36} & \dfrac{1}{72} \\ 1 & \dfrac{1}{72} & \dfrac{1}{72} & -\dfrac{1}{36} \end{array} \right\vert \end{eqnarray*} Ahora debemos calcular los determinantes de orden \(3\) y \(4\) y analizar sus signos. \begin{equation*} \left\vert \overline{H}_{3}\left( 8,8,8\right) \right\vert =\left\vert \begin{array}{ccc} 0 & 1 & 1 \\ 1 & -\dfrac{1}{36} & \dfrac{1}{72} \\ 1 & \dfrac{1}{72} & -\dfrac{1}{36} \end{array} \right\vert =\dfrac{1}{12}>0 \end{equation*} y \begin{equation*} \left\vert \overline{H}_{4}\left( 8,8,8\right) \right\vert =\left\vert \begin{array}{cccc} 0 & 1 & 1 & 1 \\ 1 & -\dfrac{1}{36} & \dfrac{1}{72} & \dfrac{1}{72} \\ 1 & \dfrac{1}{72} & -\dfrac{1}{36} & \dfrac{1}{72} \\ 1 & \dfrac{1}{72} & \dfrac{1}{72} & -\dfrac{1}{36} \end{array} \right\vert =-\dfrac{1}{192} < 0. \end{equation*}
Por tanto, puesto que el signo del determinante de la submatriz de orden \(3\) es positivo y el de orden \(4\) es negativo, entonces, por el inciso 1 del Teorema 2, la función \(h\) alcanza un máximo. Por tanto, la media geométrica es máxima cuando \( x=y=z=8\) y el valor máximo para la media geométrica es \begin{equation*} \sqrt[3]{xyz}=\sqrt[3]{8^{3}}=8. \end{equation*}