Compare L1 and L2 solutions

Problem

Minimize

J1(w)=12(w−3)2+∣w∣andJ2(w)=12(w−3)2+12w2.J_1(w)=\frac{1}{2}(w-3)^2+|w|\qquad\text{and}\qquad J_2(w)=\frac{1}{2}(w-3)^2+\frac{1}{2}w^2.

Find both minimizers and explain their shrinkage behavior.

Reveal answer or reference solution

For J1J_1, the derivative in the positive region is w−3+1w-3+1, which gives w∗=2w^*=2; the subgradient condition confirms the global minimum. For J2J_2, J2′(w)=2w−3J_2'(w)=2w-3, giving w∗=32w^*=\frac{3}{2}. Both objectives are convex. L1L_1 soft-thresholding can produce exact zeros; L2L_2 gives smooth proportional shrinkage and usually does not produce exact zeros.

Outcome

Previous attempts

Loading attempts saved in this browser…

    Review Screening Results