Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

So I have two sets of equations:

$\mathcal{A}$ = \begin{equation} \{ f(y_{0},x), \, f(y_{1},x) , \;... \;, f(y_{n},x) \} \end{equation}

$\mathcal{B}$ = \begin{equation} \{ g(y,x_{0}), \, g(y,x_{1}) , \;... \;, g(y,x_{n}) \} \end{equation}

And I created a surface out of these equations in a plot. I know I can find an equation for this surface by fitting the plot data to known surfaces, but that is not what I want.

I want to be able to directly derive an equation for the surface from just these equations. Is this possible? How can I do this?

share|cite|improve this question
Just a fleeting though, perhaps this could be considered as the (possibly overdetermined) method of lines solution of a PDE, and then one could run an inverse method approach to reconstruct the original surface. – Emily Nov 2 '12 at 16:03
The equations in the sets are determined experimentally themselves and don't represent differential equations, if that helps. – The Dude Nov 2 '12 at 16:09

You are not given "equations", but $2n+2$ curves in three-space, namely the curves $$\gamma_j:\quad x\mapsto\bigl(x,y_j,f(x,y_j)\bigr)\qquad(0\leq j\leq n)$$ in the vertical planes $y=y_j$, and the curves $$\gamma_k'\quad y\mapsto\bigl(x_,y,g(x_k,y)\bigr)\qquad(0\leq k\leq n)$$ in the vertical planes $x=x_k$. These $2n+2$ curves form a wire mesh over the $(x,y)$-plane which from above looks like an orthogonal grid. Now you are in search of a function $$\phi: \ (x,y)\mapsto z=\phi(x,y)\ ,$$ such that the graph of $\phi$ (a surface over the $(x,y)$-plane) contains all these wires $\gamma_j$ and $\gamma_k'$.

In the first place such a $\phi$ is not uniquely determined at all. There are various principles to choose an "optimal" $\phi$ corresponding to the given data. Which of these principles suits best in your situation has to be judged by an "expert". In any case the problem formulated here is a problem of utmost interest in practical applications (e.g., sculpting of car fenders), and there is a large variety of numerical techniques (and software) available to deal with it.

share|cite|improve this answer
Yes, thank you for the formalism. Is there a way to restrict the family of functions I need to fit by looking at the form of these maps? – The Dude Nov 2 '12 at 16:28
@The Dude: If the given input consists only of the values $z_{kj}=\phi(x_k,y_j)$ then the method of choice is called spline interpolation. Look it up in a suitable text book, e.g., Gerald Farin: Curves and surfaces for computer aided geometric design; Academic press 1988. I'm sure there are newer books on this topic available, but that's the book on my shelf. – Christian Blatter Nov 2 '12 at 20:51
Okay. I'll give it a shot. I've used b-splines for 2D fitting before, but this is my first real attempt at 3D fitting. Thanks! – The Dude Nov 2 '12 at 22:00

Your Answer


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.