0-1 Knapsack problemKnapsack Problem You are a mischievous child camping in the woods. You would like to steal items from other campers, but you can only carry so much mass in your knapsack. You see seven items worth stealing. Each item has a certain mass and monetary value. How do you maximize your profit so you can buy more video games later?Knapsack Problem > Value and mass of each item is given > Maximize profit > Subject to mass constraint of knapsack: 15 kg > Being a smart kid, you apply dynamic programming.Solution oo 8Sensitivity Analysis > Valuables are not very valuable when camping, pick richer people to steal from. > Suddenly a valuable item of $15 that weighs 11kg is available, will the optimal solution change? Yes, the maximum profit becomes $23Knapsack problem Given some items, pack the knapsack to get the maximum total value. Each item has some weight and some value. Total weight that we can carry is no more than some fixed number W. So we must consider weights of items as well as their values. Item #(i) Weight(w;) Value(b;) 1 1 8 2 3 6 3 5 §0-1 Knapsack problem Weight —_ Benefit value Items Wi b, mm} 3 This is a knapsack || 3 4 Max weight: W = 20 Hil 4 5 5 8 ve»| L] ° °Knapsack problem There are two versions of the problem: 1. “0-1 knapsack problem” + Items are indivisible; you either take an item or not. Some special instances can be solved with dynamic programming 2. “Fractional knapsack problem” + Items are divisible: you can take any fraction of an item0-1 Knapsack problem * Given a knapsack with maximum capacity W, and a set S consisting of nitems * Each item / has some weight w, and benefit value 6, (all w, b,and W are integer values) Problem: How to pack the knapsack to achieve TEPC Me TLR cyte0-1 Knapsack problem * Problem, in other words, is to find max )"b, subject to ieT The problem is called a “0-7” problem, because each item must be entirely accepted or rejected.Brute-force approach Let’s first solve this problem with a straightforward algorithm * Since there are n items, there are 2” possible combinations of items. * We go through all combinations and find the one with maximum value and with total weight less or equal to W * Running time will be O/2")Dynamic programming approach * We can do better with an algorithm based on dynamic programming * We need to carefully identify the subproblemsDefining a Subproblem If items are labeled 1..n, then a subproblem would be to find an optimal solution for S, = {items labeled 1, 2, .. k} * This is a reasonable subproblem definition. * The question is: Can we describe the final solution (S,,, ) in terms of subproblems (S,)? * Unfortunately, we can’t do that. ically, th i mization problem for S,,, might NOT contain the optimal solution from problem S,.Knapsack 0-1 Problem * Let’s illustrate that point with an example: item Weight Value ly 3 10 1 8 4 in 9 9 \s Is 8 a * The maximum weight the knapsack can hold is 20. * The best set of items from {Io, !1, |y} iS {lo, I, Io} * BUT the best set of items from {lo, |, I5, Ig} iS {lo, Io, Is}. * In this example, note that this optimal solution, {lp, !,, |5}, does NOT build upon the previous optimal solution, {lo, !;, ls}: * (Instead it builds upon the solution, (|, |,}, which is really the optimal subset of {lg |,,|,} with weight 12 or less.)Defining a Subproblem * Let’s add another parameter: w, which will represent the maximum weight for each subset of items * The subproblem then will be to compute V/k,wJ, i.e., to find an optimal solution for S, = {items labeled 1, 2, .. k} in a knapsack of size w. * Assuming knowing V{I, j], where i=0,1, 2, ... k-1, j=0,1,2, ...w, how to derive V[k,w]?Recursive Formula * The best subset of 5; that has the total weight < w, either contains item / or not. * First case: w;>w. Item / can’t be part of the solution, since if it was, the total weight would be > w, which is unacceptable. * Second case: w, < w. Then the item j can be in the solution, and we choose the case with greater value.0-1 Knapsack Algorithm forw=0toW V[0,w] = 0 fori=1ton Vii fori=1ton forw=1toW if w, <= w// item i can be part of the solution if b, + V[i-1,w-w]] > V[i-1,w] Vii,w] = b, + V[i-2,w- wi) else V[i.w] = Vii-1.w] else V[i,w] = V[i-1,w] // w,>wRunning Time — What is the running time of this algorithm? O(n*W) forw=0toW V[0,w] = 0 oe fori=1ton O(n) V[i,0] =0 ferl = dton Repeat n times forw=1toW if w, <= w// item ican be part of the solution o(w) if b, + V[i-1,w-w] > V[i-1,w]) V{iw] = b, + V[i-1,w- wi] else Saat Remember that the i,w] = V[i-1,w) Meer atTee alGGri else V[i,w] = V[i-1,w] //w,>w or inExample Let’s run our algorithm on the following data: n=4 (# of elements) W =5 (max weight) Elements (weight, benefit): (2,3), (3,4), (4,5), (5,6)Example iW oO - 2 lo ls ln 0; 0}| 0 0 0 Ree for w= 0 to W V[0,w] = 0Example Wo 1 2 3) 4 5 0 | 0 0 1 0 2 0 3 | 0 4| 0 fori=1ton V[i,0] = 0Example Woo 1 2) 3) 4 of o[o]olfolfo 1 {0 |*o 2/0 3 0 4[Lo fori=1ton for w=1toW if w, <= w // item i can be part of the solution if bj + VLi-l.w-wjl > Viielw) Viinw] = b, + Vli-l.w- wal else Viinw] = Viel w) else V[i, Vii-l,w] fw, >w Vf1,1] = V10,1) oeExample Woo 1 2) 3 4S 0 oto | 0 ofojo 1| 0 | 0 3 2/0 3 0 4[o fori=1ton for w=1toW if w, <= w // item i can be part of the solu if by + Vii-l,w-wjl > Vii-lw] 34V10,0] > V{0,2] V[i,w] = b, + Vfi-L.w- wi] 3. >0 else . Vii.w] = V[i-1.w] else V[i,w] = Vii-I,w] // w, > wExample Wo 1 2 23 4 5 0 | 0 oo [0 00 rio] o | 3 %3 2/0 3 0 4[o fori=1ton for w= 1toW if w, <= w // item i can be part of the solut if b, + V[i-I.w-w,] > Viielw] 34V10,11 > Viisw] = b, + Vli-Lw- wi) 340 > else Viiwl = Viel wh else V[iww] = Vii-Lw] // w,>wExample Woo 1 2) 3 4S 0 [0 0 to fo 0 1} o]o]3 | 3 3 2| 0 a) 4 [0 fori=1ton if w Vi-l.w) 34¥(0,2] > VI0,4] Viiw] =b, + Vii-Lw- wi] 340 >0 else = vi else VIi,w] = Li-Lw] Lw] fw, >Example Woo 1 2) 3 4S 0 [0 0 oo [0 1{o][o/3 [3 | 3 %3 2 [0 3 0 4 [0 fori=1ton for w=1toW if w; <= w/// item i can be part of the solution if, + Vii-l,w-w]> Vii-lw] — [3syjo3} > vios) Viiwl = b, + Vli-Lw- wi) 340 > 0 else = Vfiw = Vii-lw] else V[i,w] = Vii ww] fiw,>wExample Wo 1 2 3) 4 5 ol o oj|/ofolo r/o], | 3 )3 [3/3 2 [0 | 3 0 4[Lo fori=1ton for w=1toW if w, <= w // item i can be part of the solution if b, + Vli-L,w-w, > View] Viiw] = b, + Vii-Lw- w,] else Viiww] = Vii-L.w] else V[i,w] = V[i-l,w] //w, > w V2.1) = ViL4Example i\W oO 1 2 3 4 5 of o] 0 0 [0 1fofo] 3 [3 ]3 [3 2{ 0] 0 [% 3 0 4[o fori=1ton forw=1toW if w, <= w //item i can be part of the solution. if b, + Vli-lw-w,l > Viel wl Viiww] = b, + Vii-L.w- wi] else View] = Vii-lw) Viz2)=vina) else V[i,w] = V[i-lw] //w,> w oeExample Woo 1 2) 3 4S ofolofolofolj|o 1 Ono | 3 | 3 3.13 2}/o0|0/3 74 3 0 4[Lo fori=1ton for w=1toW if w, <= w // item i can be part of the solution b, + VLi-l.w-w] > VIi-lw] 44V[1, 1) VIi.w] = b; + V[i-L,w- wi) 4 else Vii.w] = Vli-Lw] else V[i,w] = Vli-l,w] // w,> wExample Woo 1 2) 3 4S ofolo]fojfofojo 1/o0;/ot3 ] 3/3 /3 foto ls pata 3 0 4 [0 fori=1ton for w=1toW if w, <= w // item i can be part of the solution. if by + Vii-lw-w)> Vii-lw) — [4*vita) > via) V[i.w] = b; + V[i-l.w- wi] 4. * else V[i.w] = Vfi-1.w] else V[i,w] = Vii-l,w] // w, > wExample i\W 0 1 2 3 4 5 0 | 0 0 0 0 0 0 1 0 0 a ee 3 2 0 0 3 4 4 7 3 0 4|0 fori=1ton for w=1toW if w, <= w // item i can be part of the solution if b, + Vii-l,w-w,] > Vii-Lw] Reena Vli,w] = b, + VLi-l.w- Wj] else Vii,w] = Vii-Lw] else Vii,w] = Vii-L.w] /w, > wExample Woo 1 2) 3 4S ofoflo]lofofolo 1fo[o/3/3 {3 | 3 2[o0f[ 0/3 [4 [4/7 3[ 0 ]* [%3 |*4 4 [0 fori=1ton forw=1toW if w, <= w // item i can be part of the solution if b; + Vli-L,w-w,l > Vii-Lw] Viiww] = by + Vii-Lw- wi) else Viiw] = View] else V[i,w] = V{i-l,w] “/ w, > w Vi3,31= Vi2,3)Example oO 1 2 3 4 5 o|o|o]o/}o0]o0]|o0 1}/o0/o0/3 )3 |3 )3 2 ono ts fs | 417 3|0]|0 {3 [4s 4| 0 fori=1ton for w=1toW if w, <= w // item i can be part of the solution if b; + V[i-1,w-w,] > V[i-l.w] 5+V(2,0] > V[2,4] Viiew] =, + VIi-Lw- wi] 510 > 4 else = Viiw] = Viielw] else Vfi,w] = V[i-l,w] /w,> wExample Woo 1 2) 3 4S ofolofolofolj|o 1}o]o]3 )3 [3/3 2 0 0 3 4 4 rt 3 0 0 3 4 5 7 4[Lo fori=1ton for w=1toW if w, <= w///item i can be part of the solution if b; + Vli-L.w-w,l > Vli-Lw] Viinw] = b; + Viel w- wi] else Viiw] = Vii-Lw] else V[i,w] = V{i- w] fw, > w 54VI2,1] > VI2,5] 540 > 7 Vi3.4] = 7i_ wb, Example 1 (2,3) 2: 3,4) Wo 1 2 3 4 5 jeg 2% 4S) ofofo]fTolololo|y5 #& 69 1fofol3 [3/3 /3 | ys 2fofo][s]f4fa4[7}] vig af of] o [3 [4 [5 [7 4[o [40 ['3 [*4 |¥s fori=1ton forw=1toW if w, <= w//item i can be part of the solution if b; + VLielw-wjl > Vii-lw] Viiaw] = b, + Vii-Lw- wi else v else V[i,w (i-1.w] Iw] fw, > wItems: i__w, b, Example I: (2,3) 2: (3,4) Wo 1 2 3 4 5 jeg 2% 45) ofofo]folololo|ye5 #& 69 i rfolo [3/3 ]3 [3 | V5 2,o/o]3|4/4 [7] \5 3 0 0 x 4 5 ¢ w- w.=0 4fofo[3 [4 [5 [% fori=1ton for w=1toW if w, <= w// item i can be part of the solution if, + VEi-l.w-w,] > VEi-lw] 6+V13,1] > VI3,5) , + Vii-lw- wi) 610 >7 ‘|= Vii-Lw] i-Lw] dw, > wExercise 1. a, Apply the bottom-up dynamic programming algorithm to the following instance of the knapsack problem: 3 oe oe item | weight value $25 sx S15 $40 $50 , capacity W = 6. How to find out which items are in the optimal subset?Comments * This algorithm only finds the max possible value that can be carried in the knapsack + i.e., the value in V[n,W] that ata) aT) Solty tates layHow to find actual Knapsack Items * All of the information we need is in the table. * Vin,W] is the maximal value of items that can be placed in the Knapsack. * Let i=n and k=W if V{i,k] # V[i-1,k] then mark the /* item as in the knapsack i=i-l, k= kw, else i=i-1 // Assume the i item is not in the knapsack // Could it be in the optimally packed knapsack?Items: iw, b Finding the Items I: (2,3) 2: (3,4) Wo 1 2 3 4 5 ing 3 5) ofofo]oflofo!o jes # G,6) 1{o0]o|3 | 3 [3 [3 | ba 2fo0[o0/3 |4 [4/7 |» 3, o0];o0 [3 [4 ]s5 [7 | afolo[3[4[s [7] WtItems: - iw by Finding the Items I: (2,3) 2: G4) Wo 1 2 3 4 5 ing 3 GS) ofofo]ojlo]o!o jes *# G,6) 1{o]o/3 /3 | 3 | 3 | b= 2/0] 0 | 3 4 [4 | 7 | wes 3/ofo ]3)4]s5 Gy Vii Kl = VI4,5]=7 4[o[o]3 [4 [5 [\7)] Wi-+4-vB5)=7Items: - iw by Finding the Items I: (2,3) 2: (3,4) iW 9 1 2 3 4 5 3: (4,5) ofolofolofolj|o 4: (5,6) 1}o]o]3 )3 /3 )3 2{/o0[o0 [3 |4]|4 |f/7 3| 0] 0 | 3 | 4 | 5 |\7/| WiM=Vi35]-7 alolo 3 4 5 7 | VWi-LK=VI2,5] =7Items: iw, b Finding the Items I: (2,3) 2: (3,4) wo 1 2 3 4 5 i > 45) ofofolofo]o]o ls # 66 1; o]o eH) b= 0 0 3 4 4 we. @ ofo [3 [4 [5 | 7_| VéK-vesb7 V{i-1,k]=V[1,5] =3 4fo lo 13 1415 17 Pes i=n, k=W while i,k > 0 if V/i.k) # V/i-Z,k] then: mark the # item as in the knapsack isi kekw; else isitFinding the Items i\W 0 Items: iw, b 1? (2,3) 2: (3,4) 3: 4: slolclolo clolelol- @ isn, keW while i,k > 0 if V/i.k) # V/i-Z,k] then: mark the # item as in the knapsack isi kekw; else isitItems: iw, b Finding the Items I: (2,3) 2: G4) Wo 1 Py ae ee ee) of o]o 19 | 0 | 0 | xa2 # 6,6) 0 | 0 3 [3 | 3 | b= 0 | 0 4 [4 [7 | 3/o0lo13 1415 17 | Vdl=viL2ie3 alolo 3 4 5 7 V[i-1,k]=V(0,2] =0 k-w,=0Items: iw by Finding the Items I: (2,3) 2: G4) Wo 1 Py ae ee ee) of o]o 19 | 0 | 0 | xa2 # 6,6) 0 | 0 3 [3 | 3 | b= 0 | 0 4 [4 [7 | 3/o0lo13 1415 17 | Védl=VviL2ie3 alolo 3 4 5 7 V[i-1,k]=V(0,2] =0 k-w,=0Items: iw, b Finding the Items I: (2,3) 2: G4) Wo 1 2 3 4 5 io 3% 5) ofolo]folo]fo]o|kro # 6,6) ofo {3 [3/3 ]3 ofo/3 [4 [4/7 3 0 0 3 4 5 7 The optimal knapsack # o i 3 4: 5 7 should contain isn, k=W (1,2) while ik > 0 if Vfi.k] # V[iL.K] then mark the n™ fsihk else tsi mas in the knapsackItems: iw by Finding the Items I: (2,3) 2: (3,4) Wo 1 3: (4,5) 0 0 0 0 4: (5,6) 0 0 3 0 0 2 3/0] 0 } 3 The optimal afolols3 knapsack else mark the n™ isik isit mas in the knapsack should contain (1,2)Memorization (Memory Function Method) * Goal: + Solve only subproblems that are necessary and solve it only once + Memorization is another way to deal with overlapping subproblems in dynamic programming * With memorization, we implement the algorithm recursively: * If we encounter a new subproblem, we compute and store the solution. + If we encounter a subproblem we have seen, we look up the answer * Most useful when the algorithm is easiest to implement recursively + Especially if we do not need solutions to all subproblems.Conclusion * Dynamic programming is a useful technique of solving certain kind of problems * When the solution can be recursively described in terms of partial solutions, we can store these partial solutions and re-use them as necessary (memorization) * Running time of dynamic programming algorithm vs. naive algorithm: * 0-1 Knapsack problem: O(W*n) vs. O(2")