@TedShifrin this is the part when you scold me hahaha As far as my knowledge goes I can write AB as the sum ( the definition of the multiplication ) Or multiply them and write them as the sum...
(Ctd from above) And that's when my students get a quiz back. I'll always frame it as "the grader chose to focus on this" even if I'm the one who graded that problem for the class
@TedShifrin I found this confusing footnote in Spivak, after he wrote down the definition of the distance between two points in the plane : "The fastidious reader might object to this definition on the grounds that nonnegative numbers are not yet known to have squre roots. This objection is really unanswerable atm - the definition will just have to be accepted with reservations, until the little point is settled.'' What on Earth did he mean ?
But yeah the problem was to show that if $x_1,\ldots,x_n$ are unit vectors in some normed linear space, assume that for some $\epsilon \in (0,\epsilon)$ such that $\|\sum_{j=1}^n \lambda_jx_j\| \le (1+\epsilon)\max_{1\le j\le n}|\lambda_j|$
@TedShifrin nothing concrete again. sounds either a politeness issue or an issue of expecting verb to follow first person pronoun; not a rule unfortunately. and yeah, I could not resist coming back for more :p
In the general case, yeah, and you just use $(1,0,0,\ldots)$ and $(0,1,0,\ldots)$
But the idea in the special case was that given any inner product $[\cdot,\cdot]$, you could express $\|x\|$ in that as $\langle Px, x\rangle$ for some symmetric, positive definite matrix $P$
Where $\langle \cdot,\cdot \rangle$ is the standard Euclidean inner product