Re: Quaternion problem

• To: mathgroup at smc.vnet.net
• Subject: [mg67463] Re: Quaternion problem
• From: Roger Bagula <rlbagula at sbcglobal.net>
• Date: Mon, 26 Jun 2006 00:13:15 -0400 (EDT)
• References: <e7ap5q\$952\$1@smc.vnet.net> <e7lf1d\$3l9\$1@smc.vnet.net>
• Sender: owner-wri-mathgroup at wolfram.com

```Roger Bagula wrote:

>Staffan Langin wrote:
>
>
>
>>Hello,
>>
>>I'm a total Mathematica newbee so please bear with me :-). I'm trying to let
>>Mathematica analytically evaluate a second order derivative for me. The
>>function has several parameters with some of them being of
>>"Quaternion-type". I want to instruct Mathematica that the commutative rule
>>doesn't apply when evaluation the function. Is that possible? Thanks in
>>
>>Staffan Langin
>>
>>
>>
>>
>>
>
>
>
>
I did a simple unitary quaternion differentiation using matrix quaternions
and tested the result: ( not a partial differentiation)

<< Algebra`Quaternions`
i = {{0, 1}, {-1, 0}};
j = {{0, I}, {I, 0}};
k = {{I, 0}, {0, -I}};
e = IdentityMatrix[2];
q[t_, x_, y_, z_] := e*t + x*i + j*y + k*z;
q[-1/2, 1/2, 1/2, 1/2].q[-1/2, -1/2, -1/2, -1/2]
qM[a_, b_, c_, d_] :=
Quaternion[Re[q[a, b, c, d][[1, 1]]], Re[
q[a, b, c, d][[1, 2]]], Re[q[a, b, c, d][[1, 2]]], Im[q[a, b, c, d][[1,
1]]]]
qM[-1/2, 1/2, 1/2, 1/2] ** qM[-1/2, -1/2, -1/2, -1/2]
(* Rodrigues unitary quaternion*)
qr = q[Cos[r], Sin[r]*Sin[p]*Cos[t], Sin[r]*Sin[p]*Sin[t], Sin[r]*Cos[p]]
qrs = q[Cos[r], -Sin[r]*Sin[p]*Cos[t], -Sin[r]*Sin[p]*Sin[t],
-Sin[r]*Cos[p]]
FullSimplify[ExpandAll[q[Cos[r], Sin[r]*Sin[p]*Cos[t],
Sin[r]*Sin[p]*Sin[t], Sin[r]*Cos[p]].q[Cos[r], -Sin[r]*Sin[p]*Cos[t], -
Sin[r]*Sin[p]*Sin[t], -Sin[r]*Cos[p]]]]
dqrs = D[qrs, {r, 2}]
FullSimplify[ExpandAll[
dqr.dqrs]]
q2_out = Quaternion[Re[dqr[[1, 1]]], Re[dqr[[1, 2]]], Re[dqr[[1, 2]]],
Im[dqr[[1, 1]]]]
(* if the angles are all real*)
Quaternion[-Cos[r], -Cos[t] Sin[p] Sin[r], -Cos[t] Sin[p]
Sin[r], -Cos[p] Sin[r]]
(* test*)
-dqr == qr
True

```

• Prev by Date: Re: Listing ALL Options for a mathematica Function
• Next by Date: Re: Integrate the Multivariate normal distribution
• Previous by thread: Re: Quaternion problem