Go plug that into a markdown cell of a jupyter notebook and get back to me
2 Likes
a kakoii person would have said here is the tofu you ordered
Can someone help me determine the gradient of this function??
f(x) = k(x, x) - k(x, Z) A^-1 k(Z, x).
x is a vector in R^n. Z has [p] R^n entries.
A^-1 is a pxp matrix (constant) of R.
k(Z, x) is a px1 matrix of R.
k(x, x) is a function that maps 2 vectors to R.
I am finding the value A^-1 k(Z, x) by finding Ar = k(A, x) through an iterative solution.
How the fuck do I find the gradient in respect to x?
1 Like
Do I take the derivative of k(Z, x), then find A^-1 through an iterative solution?
help
please help me
PLEASE
Bro its 3am
Sleep on it
Just going to use a low-rank approximation of A or something like a pseudo inverse/preconditioner and find K^-1 -- will solve my problems.
Looking into this
I solved it dude. Solving Ax = b with CG and never actually computing K^-1 isn't going to work; I would have to run CG in parallel and solve it M times and effectively monte carlo it.
I'm just going to randomized SVD it and then invert it and have K^-1 so I can calculate as many gradients as I want....
1 Like
Over my head. Sorry.
Don't forget that today is the last day to submit your research for SIAM 25
On a different note. I somehow corrupted the quadratic formula program in my ti84 and just only now realized it after losing my mind about getting nonsensical answers for probably a combined 4ish hours
1 Like
Including during lab and a quiz.
1 Like
I'm going to have to write an SVD algo from scratch? I don't have gesvd or anything??? lol??? Yeah, noo thanks... lmao
1 Like
The answer is 42!
1 Like
Matris E? Where’s A through D?