Equation of a SphereDate: 09/17/97 at 09:20:03 From: Matt Ream Subject: Equation of a sphere given 3 points I am helping write a geometry computer program and we are stuck. We are trying to find an equation or group of equations simplified that would solve for the center point and radius of a sphere given three points in 3D space. We have already tried using the equation of a sphere and substituting to isolate one variable, but the algebra is far too complex and tedious. How would you use the equation of a plane to solve the problem? Thank you, Matt Date: 09/17/97 at 11:16:11 From: Doctor Rob Subject: Re: Equation of a sphere given 3 points Three points are not enough to determine a sphere, even if they are not colinear (lying on the same line). You need four points, not coplanar (lying in the same plane) to uniquely determine a sphere. Nevertheless, three noncolinear points do determine a family of spheres. The points determine a plane, and within that plane, they determine a circle whose center and radius can be determined, but which are really unnecessary to the calculation. The line perpendicular to that plane at the center of the circle contains the centers of this family of spheres. For each point on the line, there is a sphere whose center is there, and which intersects the plane in the circle. Its radius is the distance from that center to each of the three original points. Let the three points be P, Q, and R, whose x-, y- and z-coordinates are given. The equation of a plane can be written in the form A*x + B*y + C*z = D, with A^2 + B^2 + C^2 = 1. You can substitute the coordinates of P, Q, and R into that equation to get three linear equations in the coefficients A, B, C, and D, which, together with the quadratic equation above, determine A, B, C, and D (up to sign, which is irrelevant). Next you can find the equations of the lines PQ, PR, and QR, and find the perpendicular bisectors of those line segments. Their intersection is the line of centers of the sphere family. The line connecting P(xP,yP,zP) to Q(xQ,yQ,zQ) is given by: (x-xP)/(xQ-xP) = (y-yP)/(yQ-yP) = (z-zP)/(zQ-zP), provided none of the denominators is zero. (If one of the denominators is zero, say xQ = xP, drop that part of the equation and replace it by the numerator set equal to zero, i.e. x - xP = 0. If two of the denominators are zero, drop all of the above and replace it with the two numerators set equal to zero.) In parametric form, with parameter t, in all cases, the line can take the form x = t*(xQ-xP) + xP, y = t*(yQ-yP) + yP, z = t*(zQ-zP) + zP. The parameter value t = 0 corrsponds to P, and the parameter value t = 1 corresponds to Q. The midpoint of PQ is ([xP+xQ]/2,[yP+yQ]/2,[zP+zQ]/2), correponding to parameter value t = 1/2. The equation of the perpendicular bisecting plane of PQ is then (xQ-xP)*(x-[xP+xQ]/2) + (yQ-yP)*(y-[yP+yQ]/2) + (zQ-zP)*(z-[zP+zQ]/2) = 0. Create the equations of the perpendicular bisecting planes of PR and QR in an analogous fashion. Then solve the three equations simultaneously for x, y, and z. Actually this system of equations is redundant, since the three planes must share a common line, so any two of them will suffice. The result of this solving will be the equation of a line, which can be put into the parametric form x = A*t + a, y = B*t + b, z = C*t + c. If all of A, B, and C are nonzero, this can be written as (x-a)/A = (y-b)/B = (z-c)/C = t, and if not, a related form results. This is the line which consists of the centers of the family of spheres. Then the equation of the sphere corresponding to parameter value t is (x - [A*t+a])^2 + (y - [B*t+b])^2 + (z - [C*t+c])^2 = (xP - [A*t+a])^2 + (yP - [B*t+b])^2 + (zP - [C*t+c])^2. The right-hand side is the square of the radius, and the center is O(A*t+a, B*t+b, C*t+c). Testing to see if the three points are colinear is a separate issue! -Doctor Rob, The Math Forum Check out our web site! http://mathforum.org/dr.math/ |
Search the Dr. Math Library: |
[Privacy Policy] [Terms of Use]
Ask Dr. Math^{TM}
© 1994- The Math Forum at NCTM. All rights reserved.
http://mathforum.org/dr.math/