[ODE] ODE's Jacobian matrices

Aaron Dwyer cubeleo at yahoo.com
Sun Sep 28 00:51:04 MST 2003


> "first create a position constraint equation C(q) 
> = 0, then take the time derivative of that" - 
> isn't the "time derivative" spoken of actually
> the partial derivatives wrt time of C(q)?

    I see what you mean, Gary.  That's where my head
starts hurting :)  Rigid bodies have 6 degrees of
freedom.  However, q is usually a position vector plus
an orientation quaternion (7 variables) or, as in the
case of the ball joint example in the joint creation
doc, a position vector plus an orientation matrix (12
variables).  Orientation expressed as quaternions and
matrices are examples of "redundant coordinates",
correct?

    Now, if I took the partial derivative of C wrt
these 12 variables (3 position vector and 9
orientation matrix), I'd end up with a much wider
matrix than I should have...but we need 6 lagrange
multipliers per body, not 12.  The position constraint
function C is a function of 12 variables, but dC/dt is
a function of 6.  That is where my confusion is
rooted, I think.

    Last, when you say "partial derivatives wrt time",
why would you say partial derivative when time is only
a single variable?  Or is it because time doesn't
appear explicitly in the constraint equation, but
instead, several variables which vary with time
appear?

-Aaron


More information about the ODE mailing list