Tossup

If the true parameter vector has this property, the Dantzig selector performs asymptotically as well as an oracle. Iterative methods for solving large positive semi-definite systems with this property first orthogonally project to a Krylov space, such as (-5[1])in the conjugate gradient method. (-5[1])The alpha (10[1])parameter measures the degree of this property of the coefficients estimated by LASSO. A type of PCA named for this property (10[1])sets a maximum on the l0 (10[1])(“little-L-zero”) pseudonorm (10[1])for each principal component and creates components from only a few input assets. The CSR format compresses a matrix with this property (10[1]-5[1])by storing it as three one-dimensional (10[1])arrays. (10[1])When creating matrices (10[1])with this property, (-5[1])thresholding sets entries below a certain value equal (10[1])to zero. (10[2]-5[1])For 10 points, name this property of matrices whose entries are mostly (10[2])zero. ■END■ (10[7]0[1])

ANSWER: sparse [or sparsity; accept sparse matrices or compressed sparse row format or sparse principal component analysis or sparse PCA; prompt on SPCA]
<Other Science>
= Average correct buzz position
Answerline and category may not exactly match the version played at all sites

Back to tossups