Take the 2-minute tour ×
Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

From my understanding, tangent planes in 3d space have the form ax+by+cz+d = 0.

If we have a function

$$f(x,y,z) = w$$

how would we represent a tangent plane in 4D space for 4D points of the form below? $$(x,y,z,f(x,y,z))$$

otherwise written:

$$(x,y,z,w)$$

(the gradient exists in 3 axis for the value w).

Thanks!

share|improve this question

3 Answers 3

up vote 2 down vote accepted

The formula for the tangent hyperplane in $\mathbb{R}^n$ always takes a simple form so long as the function $g(\vec x)$ is differentiable. At a point $\vec x_0$, the equation for the tangent hyperplane is

$$\nabla g(\vec x_0) \cdot (\vec x - \vec x_0) = 0.$$

To be explicit, this is the tangent to the set $g(\vec x ) = 0$ (or any other constant). The intuition is that the gradient should always be normal to the tangent plane. If you're interested in the tangent to some graph of a function $f$ (as you are), just take $g(\vec x) = x_n - f(x_1,\ldots,x_{n-1})).$ Writing this out in lower dimensions, and also in the case you're interested in $(n=4)$, we get the familiar formulas:

$$ a(x-x_0) + (y-f(x_0)) = 0, \qquad a= {\partial f \over \partial x}(x_0) $$ (the above is the tangent line to the point $(x_0,f(x_0))$ on the curve $f(x) = y$.) $$ a(x-x_0) + b(y-y_0) + (z-f(x_0,y_0)) = 0, \qquad a= {\partial f \over \partial x}(x_0,y_0), \qquad b= {\partial f \over \partial y}(x_0,y_0) $$ (the above is the tangent line to the point $(x_0,y_0,f(x_0,y_0))$ on the surface $f(x,y) = z$.) And finally, $$ a(x-x_0) + b(y-y_0) + c(z-z_0) +(w-f(x_0,y_0,z_0)) = 0, \qquad a= {\partial f \over \partial x}(x_0,y_0,z_0), \qquad b= {\partial f \over \partial y}(x_0,y_0,z_0),\qquad c= {\partial f \over \partial z}(x_0,y_0,z_0) $$ gives the tangent hyperplane to the hypersurface in $\mathbb{R}^4$ defined by $(x,y,z,f(x,y,z))$.

share|improve this answer
    
very cool answer. thanks for that! –  JayDee Jan 6 '13 at 3:23

How about $ax+by+cz+dw = e$?

I guess the only problem comes when you have 13 dimensions, then you run out of alphabets ... Hopefully by then, you let the coordinates be $x_i$, and the coefficients be $\alpha_i$.

share|improve this answer
    
could we thus rewrite it: ax+by+cz+dw+e = 0? –  JayDee Jan 6 '13 at 3:36
    
@JayDee Of course, since $a, b, c, d, e$ are variables, you could write it as $ax+by+cz+dw+(-e)=0$, and do a substitution $(-e)=E$. –  Calvin Lin Jan 6 '13 at 15:59
    
thanks Calvin, too bad i can't accept two answers. yours is equally right to the accepted one. I just give it to the others for the good explanatory build-up. thanks for your help though :) –  JayDee Jan 6 '13 at 20:08

Generally, if $F:M^k\to\mathbb{R}^n$ is an embedding, if $v_i$ span $T_pM$, then we can represent the tangent space to $p$ by the affine space $F(p) + \mbox{span}\{dFv_1,dFv_2,\ldots,dFv_k\}$.

In this case, $F:\mathbb{R}^3\to\mathbb{R}^4$ maps $\mathbb{R}^3$ to its graph. Particularly, $(x,y,z)\mapsto F(x,y,z) = (x,y,z,f(x,y,z))$. The differential is $$dF = \begin{pmatrix} I \\ \nabla f\end{pmatrix}$$ and so the tangent space to $F(x,y,z)$ is given by $$F(x,y,z) + \mbox{span}\{(1,0,0,f_x),(0,1,0,f_y),(0,0,1,f_z)\}$$

share|improve this answer

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.