Leverage Score Sampling
Opened this issue · 0 comments
nathanielpritchard commented
Write a function that computes a probability distribution over the rows or columns of a matrix,
- Computes an orthogonal matrix
$Q$ for the range of$A$ by performing qr - Computes the norm of each row of
$Q$ - Returns vector of each (row norm)/
$d$ , note $d = $ sum of row norms - If a distribution over the columns is desired it should be computed using the transpose of
$A$
See Page 2 for a description in the row case
- Add a folder in
linear_samplers
calledinduced_distributions
- Add file in
linear_samplers/induced_distributions
to that folder that Implements function that takes in a matrix,$A$ , as input and a boolean indicating if a row or column distribution should be used. Should return aWeights
data structure - add include statement for file with method in
linear_samplers.jl
- Export function in
RLinearAlgebra.jl
- Add tests for the function
- Add docstring reference to the created function under the section
Matrix Induced Probability Distributions
indocs/src/api/linear_samplers.md