1 / 21

Restricted Track Assignment with Applications

Restricted Track Assignment with Applications. 報告人:林添進. Channel Assignment Problem. Given a set of intervals S = {I i | 1≦i ≦ n} 1. To find the degree of overlap S, k, where k = max {|S’| | S’  S and S’ is overlapping}

taryn
Download Presentation

Restricted Track Assignment with Applications

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Restricted Track Assignment with Applications 報告人:林添進

  2. Channel Assignment Problem • Given a set of intervals S = {Ii| 1≦i ≦ n} • 1. To find the degree of overlap S, k, where k = max {|S’| | S’ S and S’ is overlapping} • 2. To partition S into k subsets such that the intervals in each subsets are pairwise non-overlapping

  3. k = 1 temp = 1 k = 2 temp = 2 k = 3 temp = 3 k = 3 temp = 2 k = 3 temp = 1 Algorithm for finding degree k • To sort 2n endpoints in the increasing order of their x-coordinates O(n log n) • To scan the 2n endpoints from left to right. O(n) • Why this algorithm works?

  4. Algorithm for finding degree k is optimal Ω(n log n) lower bound on the time required to determine whether n intervals on the line are pairwise disjoint.

  5. Algorithm for partitioning S into k subsets • To sort 2n endpoints in the increasing order of their x-coordinates • To find degree k • To scan the intervals form left to right • If a left endpoint is encountered, we assign the corresponding interval to the next available channel. • If a right endpoint is encountered, we release the channel occupied by the corresponding interval and make the released channel available again. • Why the algorithm works ?

  6. c1 c2 c3 c4 c5

  7. k channel is necessary, but why k channels is also sufficient ? • 當使得degree變成 k 的 interval I 進來以前,若 k 個 channel 被用光了,則在 I 進來以前必有某一個 interval 要離開,否則 degree 就大於 k了,這是不可能的。 • 為何對channel的長度不固定的Track Assignment Problem或RTAP,k channels is not sufficient ?

  8. Definitions • Given a set of intervals I = {Ii| 1≦i ≦ n}, where Ii = (li, ri). • Two intervals Ii = (li, ri) and Ij = (lj, rj) are • independent: if ri <lj or rj <li ; • dependent: otherwise. Two dependent intervals are • crossing: if li <lj< ri <rj or lj <li< rj<ri . • Ii contains Ij: If li <lj< rj <ri.

  9. Definitions • Track Assignment Problem (TAP) is to assign the intervals into tracks 1 to t such that in each track, intervals are pairwise independent. The goal is to minimize the number of tracks t. Whether density of I is sufficient or not? • Thedensity of a column c, denoted dc, is the number of interval in I that contain c. • The density of the set I is d = maxc dc,

  10. Definitions • Consider a set of points C = { c1, c2, …, cm}, called a restricted set. • The maximal subset of I that contains a point of C is denoted by Ic; intervals in Ic are called restricted intervals.

  11. Restricted Track Assignment Problem Given I and C, the restricted track assignment problem (RTAP) is the problem of assigning intervals into tracks 1 to t such that: p1) In each track, intervals are pairwise independent. p2) If a restricted interval Ii contains another restricted interval Ij then τi > τj , where τa is the track to which Ia is assigned. The goal is, as before, to minimize the number of tracks t. Note that when when C = { c1} then RTAP is equivalent to TAP. Here, we will show RTAP is NP-hard for |C|=2 (and |C|>2). Then we propose an approximation algorithm for solving an arbitrary instance (I, C) of RTAP.

  12. Why density is not sufficient? c1 c2 d = 3 but It need at least 4 tracks

  13. NP-hardness of RTAP Consider an arbitrary instance (I, C) of TRAP , where I = { I1, I2, …, In} and C = { c1, c2, …, cm}. Certainly, the number of tracks is lower-bounded by the density d. We will show that it is NP-complete to decide if d tracks are sufficient of assignment of intervals, when m=2. We shall transform the problem of coloring a circular-arc graph to RTAP.

  14. NP-hardness of RTAP Consider a set A = {A1, A2,…, As } of arcs, where = (ai, bi), 1≦ ai, bi ≦2s. The intersection graph of A is called a circular-arc graph. Consider a point P on the circle. The set of arcs cut by a line segment starting at the center of the circle and passing through P is denoted by AP. The arcs in AP are said to be P-equivalent. The circular-arc coloring problem (CACP) is to assign the minimum number of colors to arcs of A such that every two P-equivalent arcs, for all P, are assigned distinct colors. The maximum of |AP|, over all P, is called the density of A and is denoted byδ. It is NP-complete to decide ifδcolors are sufficient to color an arbitrary instance A of CACP.

  15. NP-hardness of RTAP We will show that CACP can be reduced to RTAP in polynomial-time. Lemma 1 An arbitrary instance A of CACP is δ-colorable if and only if (IA, CA) can be assigned into δ tracks of RTAP.

  16. An Approximation Algorithm In this section, first we show there are instances of RTAP with t≧ 2d-1, where d is the problem density. Then, we propose an approximation algorithm for an arbitrary instance of RTAP achieving t ≦ 2d-1. 為何與C無關?

  17. An Approximation Algorithm We define an instance (I, C) of TRAP that requires at least 2d-1 tracks in a recursive manner. It consists of a collection of blocks. B1 has a single unit-length interval. We obtain Bi, i > 1 , by combining two copies of Bi-1 and two intervals, called top-intervals. The instance Bd with C being the set of right points of all unit intervals is called a difficult example. Lemma 2 In an arbitrary instance of difficult example Bd, t≧ 2d-1.

  18. An Approximation Algorithm • Consider an arbitrary instance (I, C) of the problem with density d. Let {k1, k2,…, ks} denote the set of max-density columns with k1< k2<…< ks. Among all intervals containing k1 we select the one with the rightmost right point. This interval is denoted by I1,1. Assume I1,1 contains columns k1,…, kj , does not contain column kj+1. Then among all intervals containing kj+1 we select the one with the rightmost right point. This interval is denoted by I1,2. This task is repested until all columns {k1, k2,…, ks} is processed. The set of intervals selected is denoted by I1 = {I1,1, I1,2, …} • The set I1 is removed and the process is repeated for the rest of intervals ─ we obtain I1, I2, … respectively until no interval is remained. • We assign Ii, 1≦ i ≦d-1 to tracks 2d-2i+1 and 2d-2i. Id has density 1, and thus we assign it to track 1. • Why the algorithm works ?

  19. Three Properties of the Algorithm Prop 1. Density of each Ii is at most two. Prop 2. After removing each Ii the density of I is reduced by at least one. Prop 3. Consider any interval Ii,a Ii and any interval Ij,b Ij for i < j. Then Ii,ais not contained in Ij,b.

  20. k1 k2 k3 k4 k5 k1 k2 k3 k4 k1 k2 k3 k4 k1 k2 k3 k4 k5 k6 k1 k2 k3 k4 k5 k6

More Related