Rotation Dynamics of a Rigid Body with Vector Calculus and Geometric Algebra
Transcript
rotation dynamics of a rigid body with vector calculus and geometric algebra I was told that geometric algebra is the more natural way to describe physics so I wonder whether that was true also for rotational dynamics of a rigid body since I haven't found a concrete derivation on youtube yet I decided to do it on my own in the first part of this video
we will first review intermediate dynamics we will go through the derivation of time derivative in a rotating moving frame and then we will use the results from the derivation to further derive the relations between angular momentum rotational inertia and torque in the second part we will go through the same topics as in the first but with geometric algebra and finally in the last part of the video
I will summarize and compare the two results so first we will recap some of the results from intermediate dynamics using the more common vector calculus suppose that we have a set of rotating orthonormal bases b and another set of fixed bases e in intermediate dynamics we are told that they are related by a rotational tensor r like this the rotation tensor r is symmetric and its bases are orthonormal to each other this means that the transpose
of r is also the inverse of r now if we take the time derivative of the rotating basis b we should get this expression db dt equals to r dot e since we can also write the basis e as the inverse of r multiplied by b we can also write the time derivative as db dt equals to r dot r transpose b we will call this r dot r transpose
tensor the angular velocity tensor omega now recall since r is both symmetric and also normal we had r r transpose equals the identity matrix if we take the time derivative of this equation we will get this r dot r transpose plus r r dot transpose equals to zero now the first term is just the transpose of the second term what this means is that the angular
velocity tensor omega is anti-symmetric so we can actually represent the tensor omega with the product between the three-dimensional level civita tensor and another vector which we will call it angular velocity vector small omega to show that this formulation of omega above really yields an anti-symmetric tensor recall that the levi civita tensor has the following property epsilon ijk equals to minus epsilon jik so if we swap the indices of omega
we get a minus sign now since we can express the derivative of a basis vector b as b i dot equals to omega i j b j we can also express it in terms of the three three-dimensional levi civita tensor and small omega b i dot equals to minus epsilon i j k b j omega k which equals to epsilon i j k omega j b k and this is just the outer product between small omega and b so in the compact form we can
write db dt equals to omega cross b now suppose we express the vector v with the rotating basis vector b so v equals to v i b i taking the time derivative of it we will get dv dt equals to vi dot bi plus vi bi dot for the second term in the derivative of v we know that db dt equals to omega cross b so we can also write the time derivative of v as dv dt equals to vi dot vi plus vi omega cross bi we can place vi just before of
bi so that the equations becomes dv dt equals to v i b i plus omega cross v i b i so the first term is the time derivative of the vector v with respect to the moving frame which we will express as delta over delta t in this video and the second term is just small omega cross v now that we are able to express the time derivative of a vector in the moving frame we can discuss the rotation
dynamics of a rigid body with it consider a rotating rigid body the angular momentum vector h of this body with respect to its center of mass is h equals to the integral r cross r dot where the vector r is the position vector with respect to the center of mass because this is a rigid body the positions of parts of the body are fixed and the time derivative of r with respect
to the rotating frame will be zero so the time derivative of r will only contain small omega cross r therefore h can be written as h equals to the integral r cross omega cross r let's dig into what this cross product actually is using the levi civita tensors we can write the cross product as epsilon i j k epsilon m n l r j omega n r l delta f k now the product of two levi civita
tensor is equal to the following so the cross product epsilon ijk epsilon mnl r j omega n r l delta m k will only have two non-zero terms which leads to omega i r j r j minus r i r j omega j or in compact form it is r dot r minus r tensor r omega therefore if we define the rotational inertia tensor as i equals to the integral of r
cross r identity minus r tensor r then we will get a nice equation for the angular momentum vector h h equals to i omega now let us consider a torque exerted onto a rigid body with respect to the center of mass of the body that would be t equals to integral r cross a here the vector a is the acceleration of parts of the rigid body with respect to the center of mass now then the vector for the acceleration is just the time derivative of vector v which can also be written as d dt integral r cross omega cross r since the first term of this derivative
dr dt cross small omega cross r is zero so with vector calculus we have the following rotation dynamics for a rigid body because the position vector r does not change on the rotating frame the rotational inertia tensor remains constant so the equation can be simplified to t equals to d h dt equals to i small omega dot plus small omega cross h let us now derive the same results from the previous part of the video using geometric algebra in geometric algebra the rotating basis b
and the fixed basis e has the following relation b equals to r minus theta over 2 e r theta over 2 where r theta is a geometric object called the rotor taking the time derivative of the equation we will have db dt equals to r minus theta over 2 dot e r theta over two plus r minus theta over two e r theta over two dot now since we can also write e as e equals to r theta over two b r minus theta over two replacing the e in the time derivative of b
we can also express b dot as db dt equals to r minus theta over two dot r theta over two b plus b r minus theta over two r theta over two dot this results from the property r minus theta over two r theta over 2 equals to 1. if we take the time derivative of this equation we get r minus theta over 2 dot r theta over 2 plus r
minus theta over 2 r theta over 2 dot equals to 0. so the time derivative of v can be written as db dt equals to b r minus theta over 2 r theta over 2 dot minus r minus theta over two r theta over two dot b now the anti-symmetric part of the geometric product between a vector and a bi-vector is the left contraction
between a vector and a bivector so we can write db dt equals to b left contract two r minus theta over two r theta over two dot if we define the angular velocity bivector omega theta as omega theta equals to 2 r minus theta over 2 r theta over 2 dot we will arrive at the nice equation dp dt equals to b left contract omega theta in geometric algebra the angular momentum of a rigid body with respect to its center of mass is a bivector h equals to integral r wedge r dot since parts of the rigid body are only
rotating with respect to the center of mass we can write the angular momentum as h equals to integral r wedge r left contract omega theta now since our left constraint omega theta must be perpendicular to r we can also write the equation as and we can tidy the formula up a little so that h equals to one half integral r dot r omega theta minus r omega theta r now if we decompose the vector r into a part
which is parallel to omega theta r parallel and the part which is perpendicular to the omega theta r perpendicular then we can write h equals to integral r r parallel omega theta so if we define the rotational inertia multi-vector as i equals to integral r parallel dot r parallel minus r parallel wedge r perpendicular we can express the angular momentum bivector as
h equals to i omega theta however unlike the rotational inertia tensor we derive with vector calculus the rotational inertia multi-vector we derive is a function of omega theta therefore it will not be as useful as the rotational inertia tensor nevertheless we can still find a relative simple equation for the rotational dynamics of a
rigid body with geometric algebra in geometric algebra torque is written as t equals to integral r wedge a as in the vector calculus case it is very easy to show that t equals to d h dt so what is the time derivative of h let us recall what h is it is h equals to one half integral r dot r omega theta minus r omega theta r since the position of parts of the
rigid body is fixed on the moving frame r is constant so the time derivative r dot r omega theta is just r dot r omega theta dot meanwhile for the second term we use the product rule to get r dot omega theta r plus r omega theta dot r plus r omega theta r dot which can be expressed as this very long equation we can cancel some of the terms out from the equation to get we can also add and subtract
r dot r omega theta omega theta inside the brackets to get so the time derivative of h is this long equation which we can tidy into dh dt equals to delta h delta t plus one half h omega theta minus omega theta h where delta h delta t is just integral r r left contract omega theta dot also the second part of the equation is the anti-symmetric component of the geometric product between h and omega theta which must be a bi vector so we can actually write the equation as
dh dt equals to delta h delta t plus grade two component of h omega theta when comparing the derivation of rotating dynamics of a rigid body with vector calculus and with geometric algebra there are three main points I found number one both ways result in very neat equations in vector calculus we have dh dt equals to delta h delta t plus omega cross h where in geometric algebra we have dh dt equals to delta h delta t plus the grade two component of h omega theta number two using rotational inertia in
vector calculus makes more sense in vector calculus we can derive a rotational inertia tensor which is independent of the angular velocity vector this is not the case in geometric algebra the rotational inertia multi vector is a function of the angular velocity bi-vector so it is more straightforward to carry out the calculations in vector calculus although in both ways it all breaks
down to a system of linear ordinary differential equations for the delta delta t term number three geometric algebra can describe rotation in higher dimensions the real advantage of expressing the rotation dynamics in geometric algebra is that it can be readily applied to rotations in dimensions higher than 3 since unlike the cross product products using geometric algebra can also be used in higher dimensions of course this extension has almost no
practical use in the fields of engineering but it might be useful in theoretical physics and mathematics