I want to transform smoothly (and if possible by rotation) $(2,0,0,0)$ to $(1,1,1,1)$ in Euclidean space isometrically. How to find explicit formula for this transformation?
Thanks.
I want to transform smoothly (and if possible by rotation) $(2,0,0,0)$ to $(1,1,1,1)$ in Euclidean space isometrically. How to find explicit formula for this transformation?
Thanks.
On
There are many such rotations. Let's first make an orthonormal basis so that one of the rotations only touches the first two coordinates. One such basis is $$ (1,0,0,0)\\ \frac{1}{\sqrt3}(0,1,1,1)\\ \frac{1}{\sqrt2}(0,1,-1,0)\\ \frac{1}{\sqrt6}(0,1,1,-2) $$ In this basis, your vectors become $(2,0,0,0)$ and $\left(1, \sqrt3,0,0\right)$. That's a rotation through $\pi/3$ radians. So in this basis, the rotation has matrix $$ \begin{bmatrix} \cos(\pi/3) & -\sin(\pi/3)&0&0\\ \sin(\pi/3) & \cos(\pi/3)&0&0\\ 0&0&1&0\\ 0&0&0&1 \end{bmatrix} $$ (If you want the whole rotation as an actual movement, not just an instantaneous transformation, just insert a $t$ next to all the $\pi$'s and let it vary from $0$ to $1$.) Finally, to get your rotation in the standard basis, we have to transform this matrix to your basis. That means that the final result is $$ \begin{bmatrix} 1&0&0&\vphantom{\frac1{\sqrt2}}0\\ 0&\frac{1}{\sqrt3}&\frac{1}{\sqrt2}&\frac1{\sqrt6}\\ 0&\frac{1}{\sqrt3}&\frac{-1}{\sqrt2}&\frac1{\sqrt6}\\ 0&\frac{1}{\sqrt3}&0&\frac{-2}{\sqrt6} \end{bmatrix} \times \begin{bmatrix} \cos(\pi/3) & -\sin(\pi/3)&\vphantom{\frac1{\sqrt2}}0&0\\ \sin(\pi/3) & \cos(\pi/3)&0\vphantom{\frac1{\sqrt2}}&0\\ 0&\vphantom{\frac1{\sqrt2}}0&1&0\\ 0&0\vphantom{\frac1{\sqrt2}}&0&1 \end{bmatrix} \times \begin{bmatrix} 1&\vphantom{\frac1{\sqrt2}}0&0&0\\ 0&\frac{1}{\sqrt3}&\frac{1}{\sqrt3}&\frac1{\sqrt3}\\ 0&\frac{1}{\sqrt2}&\frac{-1}{\sqrt2}&0\\ 0&\frac{1}{\sqrt6}&\frac{1}{\sqrt6}&\frac{-2}{\sqrt6} \end{bmatrix} $$ where I've used that the inverse of an othogonal matrix is its transpose.
Note on how I came up with the basis: I wanted the first basis vector to be a unit vector in the direction of $(2,0,0,0)$, because that's easy and means that as long as all other vectors have $0$ as first component they will be orthogonal to that one. I wanted the second unit vector to be one that makes the $(1,1,1,1)$ be in the span of the first two basis vectors. So those two pretty much found themselves.
I just picked the third one because it was a vector obviously orthogonal to the previous two. By then, the last one was fixed: I wanted it to be orthogonal to the third basis vector, so the second and third component had to be equal, and I wanted it to be orthogonal to the second basis vector, so that forced the final component. Finally, I scaled them to be unit length.
Let $b=\{e_1,e_2,e_3,e_4\}$ be the standard basis of $\mathbb{R}^4$. Let
Apply Gram-Schmidt to $\{u_1,u_2,u_3,u_4\}$, thereby obtaining an orthonormal basis $b'=\{f_1,f_2,f_3,f_4\}$ of $\mathbb{R}^4$. Of course, $f_1=\frac12(1,1,1,1)$. Finally, consider the map $T\colon\mathbb{R}^4\longrightarrow\mathbb{R}^4$ such that $(\forall j\in\{1,2,3,4\}):f(e_j)=f_j$. Then $T$ is an isometry and $T(2,0,0,0)=(1,1,1,1)$.