Scribd
Upload
12 views2 pages

PH - Algorithm (New)

The document describes a PH algorithm for solving multi-stage stochastic programs with recourse. The algorithm involves iteratively solving scenario subproblems and updating the solution and multipliers until satisfying nonanticipativity constraints. The example given involves 3 scenario subproblems that differ in their C2k matrices.

Uploaded by

mail.information0101
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
12 views2 pages

PH - Algorithm (New)

The document describes a PH algorithm for solving multi-stage stochastic programs with recourse. The algorithm involves iteratively solving scenario subproblems and updating the solution and multipliers until satisfying nonanticipativity constraints. The example given involves 3 scenario subproblems that differ in their C2k matrices.

Uploaded by

mail.information0101
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 2

PH Algorithm

1. Step 0: Set s0 = 0 and λ 0 = perpendicular to s0 . Set i = 0.


2. Step 1: For each k = 1, 2, 3, solve the scenario subproblems:
ρ i,k
min [(ci1 )T xi,k + (ci2 )T yi,k + (ci3 )T wi,k ] + λ i,k (xi,k − si ) + ∥x − si ∥2
xi,k ,yi,k ,wi,k 2
W1 xi,k ≤ r1i
C2k xi,k +W2 yi,k + R2 wi,k ≤ r2i,k
where i is for the iteration, k is for the scenario,
xk = (x1k , x2k , x3k ),
s = (s1 , s2 , s3 ),
yk = (yk1 , yk2 ),
wk = (wk1 , wk2 , wk3 , wk4 ).
The only thing that is changing is C2k for each scenario k.
3. Step 2: Stop if satisfy the nonanticipativity constraints (that is, xi,k = si ), other-
wise go to step 3.
4. Step 3: For stage 1, calculate the components of si+1 using xi,k as follows:
3
1
si+1 = 3 ∑ xi,k .
k=1

5. Step 4: Calculate the multipliers λ i+1,k by λ i+1,k = λ i,k + ρ(xi,k − si+1 ).


6. Step 5: Increase iteration i by 1 and go to step 1.
In this algorithm, I want to enter initial values of s0 = (0, 0, 0), λ 0 = (0, 0, 0), p = 1, and
the only changing values for each scenario which is in matrix C2k after running the code.

For your understanding, 3 sceanrio subproblems are as follows:

min 150x1 + 230x2 + 260x3 + 238y1 + 210y2 − 170w1 − 150w2 − 36w3 − 10w4
s. t. x1 + x2 + x3 ≤ 500,
− 3x1 − y1 + w1 ≤ −200,
− 3.6x2 − y2 + w2 ≤ −240,
− 24x3 + w3 + w4 ≤ 0,
w3 ≤ 6000.
(Scenario 1)

1
min 150x1 + 230x2 + 260x3 + 238y1 + 210y2 − 170w1 − 150w2 − 36w3 − 10w4
s. t. x1 + x2 + x3 ≤ 500,
− 2.5x1 − y1 + w1 ≤ −200,
− 3x2 − y2 + w2 ≤ −240,
− 20x3 + w3 + w4 ≤ 0,
w3 ≤ 6000.
(Scenario 2)

min 150x1 + 230x2 + 260x3 + 238y1 + 210y2 − 170w1 − 150w2 − 36w3 − 10w4
s. t. x1 + x2 + x3 ≤ 500,
− 2x1 − y1 + w1 ≤ −200,
− 2.4x2 − y2 + w2 ≤ −240,
− 16x3 + w3 + w4 ≤ 0,
w3 ≤ 6000.
(Scenario 3)

ci1 = 150
 i 
210 , ci3 = −170 −150
   
230  , c2 = 
160 238 −36 −10 ,
−1 0
   0 −1
W1 = 1 1 1 , W2 =  0
,
0
0 0
 
−200
−240
r1 = [500], r2 = 
 0 ,

6000
 
1 0 0 0
0 1 0 0
R2 = 0 0 1 1

0 0 1 0
 
−3 0 0
 0 −3.6 0 
The input values I want to give after run the code are: C21 =  ,
0 0 −24
0 0 0
   
−2.5 0 0 −2 0 0
 0 −3 , C3 =  0 −2.4
0   0 
C22 =  , ρ = 1, s0 = (0, 0, 0), λ = (0, 0, 0)
 0 0 −20 2  0 0 −16
0 0 0 0 0 0

You might also like