Search papers, labs, and topics across Lattice.
L,ηγ‖C‖22<1.\begin{aligned} x_{k+1}&=x_{k}-\gamma(Ax_{k}+b+C^{\top}\lambda_{k}),\\ \lambda_{k+1}&=\big[\lambda_{k}+\eta(Cx_{k+1}-d)\big]_{+},\end{aligned}\qquad\begin{gathered}0<\gamma<\frac{2}{L},\\ \eta\gamma\|C\|_{2}^{2}<1.\end{gathered} (6) ISTA (R) Iterative Shrinkage-Thresholding Algorithm for regularization: yk=xk−γ(Axk+b),xk+1=𝒮γλ(yk)=proxγλ∥⋅∥1(yk),0<γ≤
1
0
2
Transformers can directly solve quadratic programs and leverage covariance matrices for superior decision-making, outperforming traditional "predict-then-optimize" methods in portfolio construction.
