2

How to parametrize $x^2+y^2+z^2=4$ and $x+z=2$

$x$, $y$ and $z$ should be function of $t$

I have tried to eliminate $z$ but it doesn't work

mezzaluna
  • 319
  • 1
  • 5
  • This is not a curve, but a surface. Namely, a sphere. You need two parameters. I'd use spherical coordinates :) – ajotatxe Sep 10 '17 at 13:22
  • @ajotatxe:this is a curve in the exercise,but i have forgotten $x+z=2$,so i think it is maybe an intersection of two curves – mezzaluna Sep 10 '17 at 13:33
  • An intersection of two surfaces. Usually (but not always) that is a curve. And so it is here. – Jyrki Lahtonen Sep 10 '17 at 13:39

4 Answers4

3

Your curve $\gamma$ is the intersection of the sphere $x^2+y^2+z^2=4$ with the plane $x+z=2$, hence a circle. Looking at a figure one immediately sees that $[2{\bf e}_1,2{\bf e_3}]$ is a diameter of the circle, hence ${\bf m}:=(1,0,1)$ is its center, and $\rho:=\sqrt{2}$ its radius. We now need two orthogonal unit vectors spanning the plane of the circle. From the same figure we see that ${\bf a}:=(0,1,0)$, $\>{\bf b}:=\bigl({1\over\sqrt{2}},0,-{1\over\sqrt{2}}\bigr)$ do the job. It follows that a parametric representation of $\gamma$ is given by $$\gamma:\quad t\mapsto{\bf r}(t):={\bf m}+\rho\cos t\>{\bf a}+\rho \sin t\>{\bf b}\>\qquad(0\leq t\leq 2\pi)\ ,$$ or in coordinates: $$\gamma:\qquad t\mapsto\left\{\eqalign{x(t)&:=1+\sin t\cr y(t)&:=\sqrt{2}\cos t\cr z(t)&:=1-\sin t\cr}\right.\qquad(0\leq t\leq 2\pi)\ .$$

  • Good job looking at the circle of intersection without projections. I chose to do it the other way, because that was more in line with the OP's attempt to eliminate a variable. If the intersecting plane had been in general position, then I would have surely followed this route as well, because the projection would have been a tilted ellipse, and working out its semiaxes would have been an unwelcome chore. – Jyrki Lahtonen Sep 10 '17 at 15:05
2

From the second equation we get $z=2-x$. Plugging this into the equation of the sphere gives $$4=x^2+y^2+(2-x)^2=y^2+2x^2-4x+4,$$ or $$ 2=y^2+2x^2-4x+2=y^2+2(x-1)^2.\qquad(*) $$ In the $xy$-plane the equation $(*)$ defines an ellipse centered at $(x,y)=(1,0)$ (this is the projection of the curve of intersection into the $xy$-plane). We also see that the semiaxes of that ellipse are parallel to the coordinate axes. If their lengths are $a$ and $b$, then that ellipse has parametrization $$ x=1+a\cos t,\qquad y=b\sin t, $$ with $t$ ranging over the interval $[0,2\pi)$.

Leaving the tasks of finding $a$ and $b$ and writing $z$ as a function of $t$ to you.

Here's how Mathematica renders it.

enter image description here

Jyrki Lahtonen
  • 133,153
  • 1
    I think that there might be some un-clarity here (but it's early, so I could just be slow-witted). I think I'd say "the equation (*) defines an ellipse in the $xy$-plane, centered at ...", emphasizing that this thing is not the intersection curve itself, but something that gets used in finding the intersection curve in 3-space. – John Hughes Sep 10 '17 at 13:59
  • Thanks @JohnHughes. It was not badly written. Hopefully it is clearer now. – Jyrki Lahtonen Sep 10 '17 at 14:56
  • I know how tough it can be to see the lack of clarity in one's own (presumably perfect!) explanations. I usually really enjoy reading yours, and have learned a few useful presentation techniques (as well as lots of math) from them. – John Hughes Sep 10 '17 at 16:07
0

As ajotatxe says in the comments, $x^2+y^2+z^2=4$ describes a surface (a sphere in fact), not a curve. So we cannot parametrize it with one variable, we need two variables. The natural coordinate system is spherical coordinates in this case, as the surface is a sphere. The two parameters would be $\theta\in[0,\pi]$ and $\phi\in[0,2\pi)$, where $$\begin{cases}x=r\sin(\theta)\cos(\phi)\\y=r\sin(\theta)\sin(\phi)\\z=r\cos(\theta)\end{cases}$$ note that $r$ is the radius of the sphere, which is constant (and in this case is $\sqrt 4=2$). $\theta$ is the inclination angle (in the $z$-direction) and $\phi$ is the azimuth angle, which measure the angle in the $xy$-plane (with the positive $x$ axis being an angle of $0$).

Dave
  • 13,568
  • and if you make one variable a function of the other you will get a curve on that surface. – Paul Sep 10 '17 at 13:34
  • I would object only to the use of the word "natural"; instead I'd say "most often used", and perhaps not that the choice of which axis to distinguish by having it depend on only one variable -- $z$ in this case -- is a matter of considerable lack of consensus among individuals and fields. – John Hughes Sep 10 '17 at 13:54
0

1)$ x^2+y^2+z^2 = 4;$ A sphere center $(0,0,0)$, radius $= 2$.

2) $x+z = 2$ , a plane.

Intersection of the two yields a curve, a circle.

Let $\vec r(t)$ be the curve, parameter $t$.

Normalized normal to the plane:

$\vec n =(1/√2)(1,0,1).$

Centre of the circle $ (1,0,1)$. Radius of the circle: $r_0 := √2$

$\vec r(t)$ lies in the plane, with normal $n$, passing through $(1,0,1):$

$ \vec n \cdot (\vec r - (1,0,1)) = 0;$

$\vec r(t) = (1,0,1) + \vec d(t).$

Direction vector $\vec d(t)$ is perpendicular to $\vec n:$

Two perpendicular vectors are:

$ \vec a:=(1/√2)(-1,0,1)$, and $\vec b:= (0,1,0)$, normalized.

Note:$\vec a$ is perpendicular to $\vec b$.

In analogy to the 2D circle with

$Y -$ and $X -$ axes:

$\vec d(t) = r_0(\sin(t) \vec a + \cos(t) \vec b)$.

$\vec r(t) = (1/√2)r_0\sin(t)(-1,0,1) +r_0 \cos(t)(0,1,0) +(1,0,1)$.

Recall: $r_0= √2$ is the radius of a circle with centre (1,0,1).

Peter Szilas
  • 20,344
  • 2
  • 17
  • 28