91Ó°ÊÓ

In multivariable calculus, linearization is a method used to approximate a complicated function with a linear one. This is particularly useful for calculations and estimates around a certain point, where the function's behavior is predictable by its tangent plane. For a function of several variables, such as \(f(x, y, z)\), the linearization at a point \(P_0(x_0, y_0, z_0)\) is given by the formula:\[ L(x, y, z) = f(x_0, y_0, z_0) + f_x(x_0, y_0, z_0)(x - x_0) + f_y(x_0, y_0, z_0)(y - y_0) + f_z(x_0, y_0, z_0)(z - z_0) \]Where:

• \(L(x, y, z)\) is the linear approximation.
• \(f_x, f_y, f_z\) are the partial derivatives of \(f\) with respect to \(x\), \(y\), and \(z\) respectively.
• This approximation becomes accurate when \(x, y,\) and \(z\) are close to \(x_0, y_0,\) and \(z_0\).

Using linearization provides a simple linear equation that closely mirrors the performance of the function near \(P_0\). It's important, though, to evaluate its possible deviations from the original function.

Partial derivatives represent the rate at which a function changes with respect to one of its variables while keeping the others constant. For a function \(f(x, y, z) = x y + 2 y z - 3 x z\), the partial derivatives are calculated with respect to each variable by treating the other variables as constants:

• \(f_x = \frac{\partial}{\partial x}(x y + 2 y z - 3 x z) = y - 3z\)
• \(f_y = \frac{\partial}{\partial y}(x y + 2 y z - 3 x z) = x + 2z\)
• \(f_z = \frac{\partial}{\partial z}(x y + 2 y z - 3 x z) = 2y - 3x\)

These derivatives are fundamental tools in multivariable calculus, allowing for:

• Investigating changes along specific axes.
• Identifying the slope of the tangent plane to the surface at a point.
• Calculating the linear approximation of a function around a point.

Understanding the behavior of these derivatives at a certain point gives insights into how a function can be accurately linearized.

Estimating the error in linearization is essential because linear approximations aren't always perfect. To evaluate the approximation's accuracy, the second partial derivatives—or more specifically, the second-order partial derivatives—are used. These derivatives provide information about how much the linear approximation may deviate from the true function:

• The second derivative terms indicate the curvature's impact around the linearization point \(P_0\).
• The maximum values of these second derivatives are used to compute the error bound.

The error estimate is bounded by:\[ |E| \leq \frac{1}{2}(M_x \cdot |x - x_0| + M_y \cdot |y - y_0| + M_z \cdot |z - z_0|) \]Where:

• \(M_x, M_y, M_z\) are the bounds on the second-order partial derivatives.
• These values reflect how the function can curve away from its linear approximation.

By calculating these values, we obtain an upper bound for the potential deviation, ensuring predictions remain reasonably accurate within a specified region \(R\). This practice helps in real-world applications where precision is vital.