@Semiclassical I didn't do this on purpose. I thought that since $f(x,y,z)$ and $u(x,y,z)$ both depend on the same variables then one can define $g(x,y,z)=u+f$
if this is wrong then let me know, and i'll edit the question to avoid any issues
Well, I'm a know quite a bit about the straight undergrad math/physics mathematics subjects, i.e. I've already done the subjects. But, my question is a freshman one
It's about the concept of Subspace from linear algebra
So, I've already done exercises, tests and so on...
But I'm still do not grasp the very elementary notion about this structure, which means that I'm not grasp a few things:
I) It's ok for me that a vector subspace W must have maintain the operations defined on V (i.e. W is a subset of V), I mean we do not want to define other exterior (another) operations. But what isn't quite clear is that why we have some properties automatic enherited like commutative property?
II) Why neutral element and opposite aren't automatically enherited?
@M.N.Raia I) Commutativity and associativity are automatically inherited because, well, it's the same operation. If a + b = b + a for all elements of V, and W is in V, then a + b = b + a for all elements of W too.
You require further, on top of closure under addition and scalar mult., that the set be nonempty---and you might as well stipulate that 0 be in the subspace specifically, because the closure under scalar mult. will automatically mean 0 had better be in there anyway.
But you do need to check whether the set is nonempty, and 0 is the easiest element to check. And also must always lie in a subspace if it is indeed a subspace.
If you have established the closure properties, you don't need to look for 0 specifically: if something else (say, a) is in W, then by closure of scalar mult, 0 = 0a is also in W. But as I said, 0 is the easiest to check in pretty much all cases, so most people just say to look for that.
Conversely, if your set is nonempty, but doesn't contain 0, you can immediately eliminate it as a subspace, because it can't be closed under scalar mult!
Except, maybe they're pointing it out just because it is kind of special that closure under scalar mult buys you a lot less checking.
If you were just dealing with abelian groups, where there is no scalar multiplication, you would indeed have to check whether or not -a is in the subset for any a in the subset.
Well, it has to satisfy all of the vector space axioms. One of those, in particular, is that it must have a 0, yes.
It turns out that no matter what, it'll be the same 0 as in the big space.
The classic example is, say, a line through the (0,0) in the plane. It contains the zero vector; if you add any two vectors on the line, it stays on the line; if you scale a vector on the line, it stays on the line.
But, if we took a pair of different lines through (0,0): zero vector is still there, scaling still works, but addition doesn't---if you add a vector from one line to a vector from the other line, you'll land somewhere in between. The operation from the big space doesn't make the pair of lines into a vector space, because in any vector space, if you add two elements it needs to give you back an element of the vector space itself.
More briefly, if that pair of lines is W, then I just found a and b in W such that a + b is not in W.
But one of the axioms for vector spaces says, "if a and b are in V, then a + b is an element of V".
(Albeit, this is usually hidden behind "there is a binary operation + on V"...)
Similarly, you can work out what I meant by "closure under scalar multiplication".
Once you have those two things, most of the other properties are things you get for free.
Commutativity, associativity, distributivity.
Even opposites come for free with those two closure properties.
The only remaining axiom is the one that says "There exists an element 0 such that yadda yadda".
If you already know your set is nonempty, then by the closure of scalar mult, 0 must be in it, so you have no checking to do.
But usually you don't already know your set is nonempty, and so you check the zero vector, because that's probably easier than checking to see if (1,2,35,7) is in there.
Personally I like to check 0 first, because if it's not in there I don't even have to muck about with the closure stuff---I already know at that point that I have a non-subspace.
So, projections onto convex sets are unique. Somebody said that projections onto non-convex sets are not unique. This seems wrong, because a projection onto a non-convex set seems like projecting onto the convex hull of the non-convex set.
@AlessandroCodenotti you didn't understand what i meant
Here's an example: We know that if we convert binary number 111 to decimal we would get 3 which is a prime number. My question is that is it possible to check 111 is a prime number in base 10 without converting it to its decimal form and then checking for primality?
Being prime has nothing to do with decimal representation
If we used base 3, we’d still get the same set of prime numbers. The labels would look different, eg 12 =3*4 in base 10 would become 110=11*10 in base 3
To the extent that doing multiplication in a different base seems strange, it’s because we’ve committed to memory the multiplication table for decimal representation
what i mean is if you think of 111 as a base 10 number it is not prime but if you think of it as a base 2 number then it is prime as 111 in base 2 is equal to 7 in base 10
I don't see how is this different than just checking for primality in base $2$, $7$ is prime in base $10$ iff $111$ is prime in base $2$, run any primality testing algorithm on $111$ doing all arithmetic in binary and you'll find out whether it is prime (hence whether $7$ is prime in base $10$) without converting it
Hi. Can anyone direct to me to a detailed explanation of the urn problem in combinatorics? I want the general case with $p$ ball colors. Then we could count the number of ways to draw with/without replacement...
(PS, I found this weird subculture from some weird youtube recommendation which people came up all sorts of fictional cosmology beyond the observable universe)
@manooooh $A \subset A \cup B$ iff for every $x \in A$ we have $x \in A \cup B$ iff for every $x \in A$ we have $x \in A$ or $x \in B$ which is true, here it seems strange because we didn't need that B is a subset of A
@manooooh I don't know if something goes wrong if we don't have law of excluded middle - axiom of choice or something - but besides that, it seems like a trivial statement, yes
I have a quick question too: I am trying to visualize (S^1 smash product I) and it looks like S^2 but it must be homeomorphic to reduced suspension of S^1 which looks like S^1