临近点算法及其在最优化问题中的应用.pdfVIP

 Æ 2003 Æ Æ ℄ Æ Rn 0 ∈ T (x) Æ T : Rn → 2Rn ÆÆ℄ 2.1 1. x0 ∈ Rn k k k 2. x βk ∈ [β, ∞)(β 0), x˜ , e k k k k x + e ∈ x˜ + β T (˜x ), (1) k ek ek ≤ ηk xk − x˜k , (2) sup ηk = η 1, (3) k≥0 3. x˜k = xk , xk+1 = x˜k − ek (4) 2 k+1 k k [11] Æ [11] 3 x = PK (˜x − e ), K T PK () K [11] ℄Æ 2.1 I Æ2.1 (SV I (K, F )): K Rn x ∈ K , x∗ ∈ F (x) ∗ SV I (K, F ) x , y − x ≥ 0, ∀y ∈ K. F : Rn → 2Rn Æ Æ 2.1 Æ K Rn f : Rn → R ∪ {+∞} min f (x). x∈K ℄ Hilbert H ( V I (K, F )): K H x∗ ∈ K ∗ ∗ F (x ), x − x ≥ 0, ∀x ∈ K. F H F x∗ ∈ S , F (x), x − x∗ ≥ 0,∀x ∈ K , S (℄ (H2)). [8] ÆÆ℄ 3.1 1. x0 ∈ K 2. xk βk ∈ [β, ∞)(β 0), xk+1 ℄ xk + ek+1 ∈ xk+1 + βk [NK (xk+1) + F (xk+1)], (5) ek+1 ∞ k+1 k+1 k 2 e ≤ ηk x − x ηk +∞. (6) k=0 Æ H = Rn