XGBoost-Style Leaf Weight

Compute the optimal leaf weight in XGBoost-style gradient boosting from the sum of gradients G, the sum of Hessians H of the samples in the leaf, and the L2 regularisation strength lam.

Implement solve(G,H,lam).

Returns. Return the weight as a float, $w^{\ast }=-G\mathrm{/}(H+\lambda )w^*=-G/(H+\backslash lambda)$.

Examples

Example 1

Input:

solve(-4.0, 3.0, 1.0)

Output:

1.0

Example 2

Input:

solve(2.0, 1.0, 3.0)

Output:

-0.5