Been trying to think of a way to interpret 0^-1 (12:58) as a multiplicative inverse to 0, i.e. 0 0^-1 = 1, rather than relying on the reciprocal. Unfortunately, he way I've been going about it usually ends up with expressions like the one at 8:35.
6:30 -- ε is a zero divisor (because ε*ε=0 and ε != 0). In theory, this means that it is not invertable (there exists no element x such that ε*x=1) **and** is not cancellable. (x*ε = y*ε does not imply x = y) A problematic example here is x = 1, y = 1+ε. This satisfies the un-cancelled equality, but not the cancelled equality. I *think* that restricting a and b to be real numbers would make ε cancellable, but I don't have a rigorous proof of this. (I bet someone else has either already proved or disproved that assertion) In short, feels like you need to be careful here.
There are systems that have positive and negative infinitesimals, that then have reciprocals of positive and negative infinities, however their sum still has to be either undefined, or an unsigned zero. I find find it easier to think of IEEE floating point -my arch-nemesis- as having positive and negative infinitesimals, and then lacking _any_ zero, rounding zero to one of the infinitesimals depending on the computer's mood (because IEEE floating point is not nearly as consistent as it should be).
There's another connection. As we saw here, -∞ = ∞, so unary negation does nothing, much like how -0 = 0. If you have an 8-bit signed integer, -128 is the minimum value. If you try to negate this, rather than getting 128, you instead get back -128. Much like with infinity and 0, unary negation does nothing. At the same, it kind of _does_ give +128, since -128 as an _unsigned_ 8-bit integer is 128. 8-bit 128 behaves like projective infinity under negation. It's precisely for this reason that posit numbers, an obscure alternative to IEEE floating point for representing fractional numbers, directly map 2's complement integers to the projective reals. The diagram with the circle and 0 on the bottom, + on the right, - on the left, and ∞ on the top, is exactly the same as the one used to demonstrate posit values, and making the + and - have a magnitude of 1 gives all 4 representable values for a 2-bit posit.
@8mr.y82 күн бұрын
0^-1=∞の証明、それあり!?
@8mr.y82 күн бұрын
これが無下限か
@匿名希望-j2f2 күн бұрын
0×∞=1 ?????
@yukkiee212 күн бұрын
無限を他の値で撮るのか...頭いいな
@piyashirikozo2 күн бұрын
符号の有るゼロを定義すれば良い。
@angeldude101Күн бұрын
If a system has any identity element, then the identity element is guaranteed to be unique. If you have 2 identity elements, then at least one of them is not a true identity. Distinguishing +0 from -0 deprives both of the ability to be true additive identities. You _can_ make infinitesimals +ε and -ε that _almost_ act like additive identities, but adding them together still needs to cancel out and give the true additive identity: unsigned zero. If they don't, then I'm sorry, but your math is fundamentally broken (or that particular sum is undefined just like ∞ - ∞).
@leht5882 күн бұрын
リーマン球面好き
@kiukiu19192 күн бұрын
無限大の彼方に発散するってかっこいい
@goblin50032 күн бұрын
I love their voice here, so I keep coming back even though there’s an English channel 😆
I've been waiting for a video like this keep up the great work ❤
@user-bp6mz2qw3j2 күн бұрын
3blue1brownの「四次元の見方」を思い出した
@Ny0s2 күн бұрын
By adding a point in the center of the circle corresponding to 0/0, this becomes a beautiful mathematical model of undefinedness! Thank you again for the subtitles, I simply love the japanese voices
@thechairguy2 күн бұрын
WHEEL THEORY!!!! unary operator division and x/x = 1 + 0x/x my beloved