Solving for unisolated variable
2 views (last 30 days)
Show older comments
Hello all, I'm having a bit of trouble finding a method to solve a nonlinear equation. I've looked into "fsolve" a little but couldn't figure out if it would be effective. I'm using MatLab 2009.
Here is the equation, for ease I've simplified the constants to A,B,C. Variable to solve for is x.
x = arctan( (A - C*cos(x)) / (B*cos(x)) )
Which was derived from
A = B*sin(x) + C*cos(x)
Any help would be greatly appreciated... Rather than relying on MatLab to solve I would like to have x = F(A,B,C) where A,B,C are known but change with each iteration.
Another consideration is that this value needs to be found 5-10 times per second.
Thank you in advance! - Jon
0 Comments
Answers (2)
Walter Roberson
on 22 Nov 2013
There are two solutions:
T0 = sqrt(C^2*B^2 - B^2*A^2 + B^4);
X = [(C*A + T0)/(C^2+B^2), (C*A - T0)/(C^2+B^2)]; %just sign difference between two
Y = (A - C*X) / B;
atan2(Y, X)
2 Comments
Roger Stafford
on 23 Nov 2013
Two solutions can be expressed as:
x = asin(A/sqrt(B^2+C^2)) - atan2(C,B);
and
x = pi - asin(A/sqrt(B^2+C^2)) - atan2(C,B)
Also either value with any integral multiple of 2*pi added or subtracted is a solution, thus giving infinitely many solutions.
3 Comments
Roger Stafford
on 23 Nov 2013
My reasoning went this way. Starting with
A = B*sin(x) + C*cos(x)
notice that any pair of numbers B and C can always be represented by
B = sqrt(B^2+C^2)*cos(t)
C = sqrt(B^2+C^2)*sin(t)
for an appropriate angle t, which can be evaluated by
t = atan2(C,B)
(That in fact is one way to define the function atan2.) This gives
A = sqrt(B^2+C^2)*(cos(t)*sin(x)+sin(t)*cos(x))
= sqrt(B^2+C^2)*sin(x+t)
sin(x+t) = A/sqrt(B^2+C^2)
x + t = asin(A/sqrt(B^2+C^2))
x = asin(A/sqrt(B^2+C^2)) - t = asin(A/sqrt(B^2+C^2)) - atan2(C,B)
See Also
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!