802 karma · joined September 16, 2013
f(x+dx) = f(x) + f'(x)dx + O(dx^2)
This should look familiar: taking a = f(x) and b = f'(x), this is just the line a + b.dx! In other words, Calculus is just a way transforming questions about (differentiable) functions into questions about lines.
So this is all well and good, but what if x is a vector? Or f(x) is a vector? Or both? Well, now we can approximate
f(x+dx) = f(x) + Df(x).dx + O(|dx|^2)
Here, Df(x) is nothing other than the matrix [(df_i/dx_j)(x)]. In other words, we still get a linear approximation. But now suppose we get greedy and want to take higher order derivatives: what is the derivative of Df(x)? It's a tensor!
There are a few ways to think about this (which gives rise to the different interpretations of a tensor):
1. It's just df_i/(dx_jdx_k). This is a multi-dimensional array.
2. In 1D, we have
f(x+dx) = f(x) + f'(x)dx + (1/2)f''(x)dx^2 + O(dx^3)
In higher dimensions, we get
f(x+dx) = f(x) + Df(x)*dx + dx'.D^2f(x).dx + O(|dx|^3)
Here, D^2f(x) is a multi-linear map: it takes in two copies of the vector dx, and returns a new vector.
More generalizations exist for various reasons, but I think this intuition captures why tensors show up so often.
[1] http://en.wikipedia.org/wiki/The_Market_for_Lemons [2] http://time.com/3592035/uber-taxi-history/
ϕ_mult = [[1 5] [1 1]]
This is because ϕ(x+y√5)=(x+5y)+(x+y)√5, and
[[1 5] [1 1]] [x y] = [x+5y x+y]
Now, to compute ϕ^N, we just have to compute [x y] = ϕ_mult^N[1 0] (since [1 0] is the representation of 1=1+0√5). Then ϕ^N=x+y√5. This is all exact, no floating point arithmetic!
It's actually just an example of an algebraic number field [1] -- the idea is that "algebraic numbers" (i.e., roots of polynomials) can be equivalently thought of as vectors and matrices. The simplest example, which everyone is familiar with, is the complex numbers, where i=√(-1). In this case, [x y] represents x+y*i, and multiplication by i corresponds to the matrix i_mult=[[0 1] [-1 0]]. Then i_mult^2= -[[1 0] [0 1]], in other words, multiplication by -1!
http://www.nature.com/nature/journal/v375/n6530/pdf/375367b0...
In particular, they have a description of why they find the traditional connection between uncertainty and wave-particle duality to be unsatisfactory. Unfortunately the technical details are beyond me, but hopefully someone else can sort them out?
Edit: Okay, some more digging suggests the following story: (i) We have wave-particle duality => Heisenberg uncertainty, but not the other way around (ii) With entropic uncertainty [1] (stronger than Heisenberg uncertainty), the authors show that we can obtain wave-particle duality
Again, can somebody more knowledgeable can clarify? :)
t = Bernoulli(T) (i.e., T = E[t])
The distribution is
m_n = M_1 t_1 ... t_n,
which has expected value
M_n = E[m_n] = E[M_1 t_1 ... t_n] = M_1 T^n
Here, each of the t_i is an i.i.d. sample of t. You can similarly compute the variance:
Var[m_n] = Var[M_1 t_1 ... t_n]
I don't know a closed form for this off the top of my head, but at this point simulations are probably the best way to go.
It turns out that if you can prove something is both true and false, then EVERYTHING is both true and false. For example, let's say you want to prove X. Note that Y is true. But then you obtain a contradiction (since Y is false)! Hence X is true.
"So there is a basic sense that whoever is providing that technology has to make sure it's secure."
I think he's just saying it's the job of the mail service to provide security, not something the end user should need to worry about.