I am faced with this problem:

Problem $\Gamma: z=f(x,y)$ is a surface in $\mathbb{R}^3$ and $f(x,y)$ has continuous derivatives on $\mathbb{R}^2$. All tangent planes of $\Gamma$ pass through a fixed point $(a,b,c)$. Prove that $$f(t(x-a)+a,t(y-b)+b)-c=t(f(x,y)-c), \forall t>0.$$ (Hint: What is the solution to $\frac{y'(t)}{y(t)}=\frac{1}{t}?$)

My Solution(unfinished) I have transform the aim formula into $$f(tx+(1-t)a,ty+(1-t)b)=tz+(1-t)c, \forall t>0,$$ which implies a line going through a random point $P\in \Gamma$ and the fixed point$(a,b,c)$ totally lies in $\Gamma$. Without losing generality, We take $(a,b,c)=(0,0,0)$ and the aim formula turns into $$f(tx,ty)=tf(x,y), \forall t>0.$$

The formula of tangent planes at a specific point $(x_0,y_0,z_0)\in \Gamma$ is $$ \frac{\partial f\left( x_0,y_0 \right)}{\partial x}\left( x-x_0 \right) +\frac{\partial f\left( x_0,y_0 \right)}{\partial y}\left( y-y_0 \right) -\left( z-z_0 \right) =0, $$ As $(0,0,0)$ lies in it, we have $$ \frac{\partial f\left( x_0,y_0 \right)}{\partial x}\left( 0-x_0 \right) +\frac{\partial f\left( x_0,y_0 \right)}{\partial y}\left( 0-y_0 \right) -\left( 0-z_0 \right) =0, $$ $$ \Rightarrow x\frac{\partial f}{\partial x}+y\frac{\partial f}{\partial y}=z=f\left( x,y \right) , \forall \left( x,y \right) \in \mathbb{R}^2. $$ This hints me at Euler Formula of homogeneous functions, but what we need to do is to prove the converse. We have $$ \frac{\mathrm{d}f\left( tx,ty \right)}{\mathrm{d}t}=\frac{\partial f}{\partial tx}\frac{\partial tx}{\partial t}+\frac{\partial f}{\partial ty}\frac{\partial ty}{\partial t}=x\frac{\partial f}{\partial x}+y\frac{\partial f}{\partial y}=f\left( x,y \right) , $$ Integrating on both sides of the equation, we have $$f(tx,ty)=tf(x,y)+C_0$$ where $C_0$ is an arbitrary constant in $\mathbb{R}$.

Here comes the tricky part: how to eliminate the constant $C_0$, i.e., to determine $C_0=0$?


I'm not sure I understand all your thoughts, but as I see it you already have all relevant formulas, just not in a straightforward order. Or perhaps completing your proof is as easy as inserting $t=1$ in your last formula, $f(x,y)=f(x,y)+C_0$.

The idea is that you restrict $f$ to the line from $(a,b)$ through some fixed $(x,y)$. $$ g(t)=f(a+t(x-a),b+t(y-b)). $$ The visual picture is that as a scalar function $g$ inherits the property that all of its tangents meet in $(0,c)$. This is only possible if $g$ itself is a linear function with $g(0)=c$. Then $g(t)=g(0)+t(g(1)-g(0))$ which is the claimed formula.

In detail, the time derivative of $g$ is $$ g'(t)=∂_xf(a+t(x-a),b+t(y-b))(x-a)+∂_yf(a+t(x-a),b+t(y-b))(y-b) $$ By the tangent plane condition you have $$ f(x(t),y(t))+∂_xf(x(t),y(t))(a-x(t))+∂_yf(x(t),y(t))(b-y(t))=c $$ Inserting $x(t)=a+t(x-a)$, $y(t)=b+t(y-b)$ results in the more compact formula $$ g(t)-tg'(t)=c. $$ This has an integrating factor $-t^{-2}$, so that $$ \frac{g(1)}{1}-\frac{g(t)}{t}=\frac{c}1-\frac{c}{t} \\ \implies f\bigl(a+t(x-a),b+t(y-b)\bigr)-c=t\,(f(x,y)-c) $$

  • $\begingroup$ You are right haha... t=1 really works! How did I miss that-_-#! And thaaaanks for your detailed answer! It actually makes the problem clearer. :) $\endgroup$ Oct 8 '20 at 9:46


From the tangency condition

$$ \frac{\partial f\left( x,y \right)}{\partial x}\left( x-a \right) +\frac{\partial f\left( x,y \right)}{\partial y}\left( y-b \right) -\left( f(x,y)-c \right) =0 $$

solving this PDE we get at

$$ f(x,y) = (x-a)\Phi\left(\frac{y-b}{x-a}\right)+c $$

now we can state that it is true

$$ f(a+t(x-a),b+t(y-b))-c - t(f(x,y)-c) = 0 $$

  • $\begingroup$ Thank you for your hint! Could you explain a bit about $\phi$ function, please? Is it a function in PDE course? I haven't learnt about PDE so far... $\endgroup$ Oct 8 '20 at 14:06
  • $\begingroup$ Well. It is an alternative approach to the question. The advantage is that you can get the functional structure to the ruled surface $f(x,y)$. Wait until you have enough knowledge in PDE's to enjoy the beauty of this result. $\endgroup$
    – Cesareo
    Oct 8 '20 at 16:54
  • $\begingroup$ I will keep it in mind. Many thanks! $\endgroup$ Oct 9 '20 at 6:31

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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