convex hull formation for programmable matternavlakha/bda2017/material/talks/daymudecon… ·...
TRANSCRIPT
![Page 1: Convex Hull Formation for Programmable Matternavlakha/BDA2017/Material/Talks/DaymudeCon… · Current Programmable Matter Convex Hull Formation for Programmable Matter Biological](https://reader033.vdocuments.us/reader033/viewer/2022050323/5f7c5b9548272338f369a3a6/html5/thumbnails/1.jpg)
Convex Hull Formation for Programmable MatterJOSHUA J. DAYMUDE AND ANDRÉA W. R ICHA – ARIZONA STATE UNIVERSITY
ROBERT GMYR, CHRISTIAN SCHEIDELER, AND THIM STROTHMANN –UNIVERSITY OF PADERBORN
Biological Distributed Algorithms 2017
![Page 2: Convex Hull Formation for Programmable Matternavlakha/BDA2017/Material/Talks/DaymudeCon… · Current Programmable Matter Convex Hull Formation for Programmable Matter Biological](https://reader033.vdocuments.us/reader033/viewer/2022050323/5f7c5b9548272338f369a3a6/html5/thumbnails/2.jpg)
Current Programmable Matter
Convex Hull Formation for Programmable Matter Biological Distributed Algorithms 2017
Introduction Background & Model Problem Statement Algorithm Description Runtime Analysis Conclusion
[2] RCN 2014: “Programmable self-assembly in a thousand-robot swarm”[1] RGR 2013: "M-blocks: Momentum driven, magnetic modular robots"
![Page 3: Convex Hull Formation for Programmable Matternavlakha/BDA2017/Material/Talks/DaymudeCon… · Current Programmable Matter Convex Hull Formation for Programmable Matter Biological](https://reader033.vdocuments.us/reader033/viewer/2022050323/5f7c5b9548272338f369a3a6/html5/thumbnails/3.jpg)
Inspirations & Applications
Convex Hull Formation for Programmable Matter Biological Distributed Algorithms 2017
Introduction Background & Model Problem Statement Algorithm Description Runtime Analysis Conclusion
![Page 4: Convex Hull Formation for Programmable Matternavlakha/BDA2017/Material/Talks/DaymudeCon… · Current Programmable Matter Convex Hull Formation for Programmable Matter Biological](https://reader033.vdocuments.us/reader033/viewer/2022050323/5f7c5b9548272338f369a3a6/html5/thumbnails/4.jpg)
The Amoebot Model
Convex Hull Formation for Programmable Matter Biological Distributed Algorithms 2017
Introduction Background & Model Problem Statement Algorithm Description Runtime Analysis Conclusion
Particles move by expanding and contracting, and are:
• Anonymous (no unique identifiers)
• Without global orientation or compass (no shared sense of “north”)
• Limited in memory (constant size)
• Activated asynchronously
![Page 5: Convex Hull Formation for Programmable Matternavlakha/BDA2017/Material/Talks/DaymudeCon… · Current Programmable Matter Convex Hull Formation for Programmable Matter Biological](https://reader033.vdocuments.us/reader033/viewer/2022050323/5f7c5b9548272338f369a3a6/html5/thumbnails/5.jpg)
Our Past Work
Convex Hull Formation for Programmable Matter Biological Distributed Algorithms 2017
Introduction Background & Model Problem Statement Algorithm Description Runtime Analysis Conclusion
• Leader Election [DNA21, ALGOSENSORS ‘17]
• Shape Formation [NANOCOM ’15, SPAA ‘16]
• Object Coating [Theoretical Computer Science, Natural Computing]
• Full list of publications can be found at: sops.engineering.asu.edu/publications-press/.
![Page 6: Convex Hull Formation for Programmable Matternavlakha/BDA2017/Material/Talks/DaymudeCon… · Current Programmable Matter Convex Hull Formation for Programmable Matter Biological](https://reader033.vdocuments.us/reader033/viewer/2022050323/5f7c5b9548272338f369a3a6/html5/thumbnails/6.jpg)
Convex Hull: Definitions
Convex Hull Formation for Programmable Matter Biological Distributed Algorithms 2017
Introduction Background & Model Problem Statement Algorithm Description Runtime Analysis Conclusion
• We begin with an object O, which is a connected set of nodes in our graph G = (V, E).
![Page 7: Convex Hull Formation for Programmable Matternavlakha/BDA2017/Material/Talks/DaymudeCon… · Current Programmable Matter Convex Hull Formation for Programmable Matter Biological](https://reader033.vdocuments.us/reader033/viewer/2022050323/5f7c5b9548272338f369a3a6/html5/thumbnails/7.jpg)
Convex Hull: Definitions
Convex Hull Formation for Programmable Matter Biological Distributed Algorithms 2017
Introduction Background & Model Problem Statement Algorithm Description Runtime Analysis Conclusion
• We begin with an object O, which is a connected set of nodes in our graph G = (V, E).
• Let O* be the minimal convex set of nodes containing O.
![Page 8: Convex Hull Formation for Programmable Matternavlakha/BDA2017/Material/Talks/DaymudeCon… · Current Programmable Matter Convex Hull Formation for Programmable Matter Biological](https://reader033.vdocuments.us/reader033/viewer/2022050323/5f7c5b9548272338f369a3a6/html5/thumbnails/8.jpg)
Convex Hull: Definitions
Convex Hull Formation for Programmable Matter Biological Distributed Algorithms 2017
Introduction Background & Model Problem Statement Algorithm Description Runtime Analysis Conclusion
• We begin with an object O, which is a connected set of nodes in our graph G = (V, E).
• Let O* be the minimal convex set of nodes containing O.
• The convex hull of O, denoted C(O), is the set of nodes in V \ O* adjacent to some node(s) of O*. (Essentially the “external boundary” of O*).
![Page 9: Convex Hull Formation for Programmable Matternavlakha/BDA2017/Material/Talks/DaymudeCon… · Current Programmable Matter Convex Hull Formation for Programmable Matter Biological](https://reader033.vdocuments.us/reader033/viewer/2022050323/5f7c5b9548272338f369a3a6/html5/thumbnails/9.jpg)
Why Convex Hulls?
Convex Hull Formation for Programmable Matter Biological Distributed Algorithms 2017
Introduction Background & Model Problem Statement Algorithm Description Runtime Analysis Conclusion
• Interesting problem in computational geometry, especially in distributed settings.
• Can be viewed as a relaxation of object coating.
Incomplete Coating Convex Hull
![Page 10: Convex Hull Formation for Programmable Matternavlakha/BDA2017/Material/Talks/DaymudeCon… · Current Programmable Matter Convex Hull Formation for Programmable Matter Biological](https://reader033.vdocuments.us/reader033/viewer/2022050323/5f7c5b9548272338f369a3a6/html5/thumbnails/10.jpg)
Our Goal
Convex Hull Formation for Programmable Matter Biological Distributed Algorithms 2017
Introduction Background & Model Problem Statement Algorithm Description Runtime Analysis Conclusion
Given: a connected object O with no holes, a connected particle system P such that |P| ≥ |C(O)|, and a unique seed particle s which is adjacent to O.
Goal: reconfigure P so that every node of C(O) is occupied by a contracted particle.
![Page 11: Convex Hull Formation for Programmable Matternavlakha/BDA2017/Material/Talks/DaymudeCon… · Current Programmable Matter Convex Hull Formation for Programmable Matter Biological](https://reader033.vdocuments.us/reader033/viewer/2022050323/5f7c5b9548272338f369a3a6/html5/thumbnails/11.jpg)
Our Goal
Convex Hull Formation for Programmable Matter Biological Distributed Algorithms 2017
Introduction Background & Model Problem Statement Algorithm Description Runtime Analysis Conclusion
Given: a connected object O with no holes, a connected particle system P such that |P| ≥ |C(O)|, and a unique seed particle s which is adjacent to O.
Goal: reconfigure P so that every node of C(O) is occupied by a contracted particle.
![Page 12: Convex Hull Formation for Programmable Matternavlakha/BDA2017/Material/Talks/DaymudeCon… · Current Programmable Matter Convex Hull Formation for Programmable Matter Biological](https://reader033.vdocuments.us/reader033/viewer/2022050323/5f7c5b9548272338f369a3a6/html5/thumbnails/12.jpg)
Algorithm: High Level
Convex Hull Formation for Programmable Matter Biological Distributed Algorithms 2017
Introduction Background & Model Problem Statement Algorithm Description Runtime Analysis Conclusion
Our algorithm is broken up into two main phases:
![Page 13: Convex Hull Formation for Programmable Matternavlakha/BDA2017/Material/Talks/DaymudeCon… · Current Programmable Matter Convex Hull Formation for Programmable Matter Biological](https://reader033.vdocuments.us/reader033/viewer/2022050323/5f7c5b9548272338f369a3a6/html5/thumbnails/13.jpg)
Algorithm: High Level
Convex Hull Formation for Programmable Matter Biological Distributed Algorithms 2017
Introduction Background & Model Problem Statement Algorithm Description Runtime Analysis Conclusion
Our algorithm is broken up into two main phases:
1. Phase I: Escaping the Object
![Page 14: Convex Hull Formation for Programmable Matternavlakha/BDA2017/Material/Talks/DaymudeCon… · Current Programmable Matter Convex Hull Formation for Programmable Matter Biological](https://reader033.vdocuments.us/reader033/viewer/2022050323/5f7c5b9548272338f369a3a6/html5/thumbnails/14.jpg)
Algorithm: High Level
Convex Hull Formation for Programmable Matter Biological Distributed Algorithms 2017
Introduction Background & Model Problem Statement Algorithm Description Runtime Analysis Conclusion
Our algorithm is broken up into two main phases:
1. Phase I: Escaping the Object
2. Phase II: Constructing the Convex Hull
![Page 15: Convex Hull Formation for Programmable Matternavlakha/BDA2017/Material/Talks/DaymudeCon… · Current Programmable Matter Convex Hull Formation for Programmable Matter Biological](https://reader033.vdocuments.us/reader033/viewer/2022050323/5f7c5b9548272338f369a3a6/html5/thumbnails/15.jpg)
Phase I: Escaping the Object
Convex Hull Formation for Programmable Matter Biological Distributed Algorithms 2017
Introduction Background & Model Problem Statement Algorithm Description Runtime Analysis Conclusion
Phase I is responsible for reorganizing P into a straight line of particles, which must necessarily reach outside O* (recall: |P| ≥ |C(O)|).
![Page 16: Convex Hull Formation for Programmable Matternavlakha/BDA2017/Material/Talks/DaymudeCon… · Current Programmable Matter Convex Hull Formation for Programmable Matter Biological](https://reader033.vdocuments.us/reader033/viewer/2022050323/5f7c5b9548272338f369a3a6/html5/thumbnails/16.jpg)
Phase I: Escaping the Object
Convex Hull Formation for Programmable Matter Biological Distributed Algorithms 2017
Introduction Background & Model Problem Statement Algorithm Description Runtime Analysis Conclusion
Phase I is responsible for reorganizing P into a straight line of particles, which must necessarily reach outside O* (recall: |P| ≥ |C(O)|).
First, organize P using the “spanning forest primitive”.
![Page 17: Convex Hull Formation for Programmable Matternavlakha/BDA2017/Material/Talks/DaymudeCon… · Current Programmable Matter Convex Hull Formation for Programmable Matter Biological](https://reader033.vdocuments.us/reader033/viewer/2022050323/5f7c5b9548272338f369a3a6/html5/thumbnails/17.jpg)
Phase I: Escaping the Object
Convex Hull Formation for Programmable Matter Biological Distributed Algorithms 2017
Introduction Background & Model Problem Statement Algorithm Description Runtime Analysis Conclusion
Phase I is responsible for reorganizing P into a straight line of particles, which must necessarily reach outside O* (recall: |P| ≥ |C(O)|).
First, organize P using the “spanning forest primitive”.
![Page 18: Convex Hull Formation for Programmable Matternavlakha/BDA2017/Material/Talks/DaymudeCon… · Current Programmable Matter Convex Hull Formation for Programmable Matter Biological](https://reader033.vdocuments.us/reader033/viewer/2022050323/5f7c5b9548272338f369a3a6/html5/thumbnails/18.jpg)
Phase I: Escaping the Object
Convex Hull Formation for Programmable Matter Biological Distributed Algorithms 2017
Introduction Background & Model Problem Statement Algorithm Description Runtime Analysis Conclusion
Phase I is responsible for reorganizing P into a straight line of particles, which must necessarily reach outside O* (recall: |P| ≥ |C(O)|).
First, organize P using the “spanning forest primitive”.
![Page 19: Convex Hull Formation for Programmable Matternavlakha/BDA2017/Material/Talks/DaymudeCon… · Current Programmable Matter Convex Hull Formation for Programmable Matter Biological](https://reader033.vdocuments.us/reader033/viewer/2022050323/5f7c5b9548272338f369a3a6/html5/thumbnails/19.jpg)
Phase I: Escaping the Object
Convex Hull Formation for Programmable Matter Biological Distributed Algorithms 2017
Introduction Background & Model Problem Statement Algorithm Description Runtime Analysis Conclusion
Next, alternate between the following subphases until Phase I is complete:
1. Wall Following: Follow the object using right-hand-rule until finding a “concave turn”.
![Page 20: Convex Hull Formation for Programmable Matternavlakha/BDA2017/Material/Talks/DaymudeCon… · Current Programmable Matter Convex Hull Formation for Programmable Matter Biological](https://reader033.vdocuments.us/reader033/viewer/2022050323/5f7c5b9548272338f369a3a6/html5/thumbnails/20.jpg)
Phase I: Escaping the Object
Convex Hull Formation for Programmable Matter Biological Distributed Algorithms 2017
Introduction Background & Model Problem Statement Algorithm Description Runtime Analysis Conclusion
Next, alternate between the following subphases until Phase I is complete:
1. Wall Following: Follow the object using right-hand-rule until finding a “concave turn”.
![Page 21: Convex Hull Formation for Programmable Matternavlakha/BDA2017/Material/Talks/DaymudeCon… · Current Programmable Matter Convex Hull Formation for Programmable Matter Biological](https://reader033.vdocuments.us/reader033/viewer/2022050323/5f7c5b9548272338f369a3a6/html5/thumbnails/21.jpg)
Phase I: Escaping the Object
Convex Hull Formation for Programmable Matter Biological Distributed Algorithms 2017
Introduction Background & Model Problem Statement Algorithm Description Runtime Analysis Conclusion
Next, alternate between the following subphases until Phase I is complete:
1. Wall Following: Follow the object using right-hand-rule until finding a “concave turn”.
![Page 22: Convex Hull Formation for Programmable Matternavlakha/BDA2017/Material/Talks/DaymudeCon… · Current Programmable Matter Convex Hull Formation for Programmable Matter Biological](https://reader033.vdocuments.us/reader033/viewer/2022050323/5f7c5b9548272338f369a3a6/html5/thumbnails/22.jpg)
Phase I: Escaping the Object
Convex Hull Formation for Programmable Matter Biological Distributed Algorithms 2017
Introduction Background & Model Problem Statement Algorithm Description Runtime Analysis Conclusion
Next, alternate between the following subphases until Phase I is complete:
1. Wall Following: Follow the object using right-hand-rule until finding a “concave turn”.
2. Line Probing: Attempt to build the desired line, and backtrack on failure.
![Page 23: Convex Hull Formation for Programmable Matternavlakha/BDA2017/Material/Talks/DaymudeCon… · Current Programmable Matter Convex Hull Formation for Programmable Matter Biological](https://reader033.vdocuments.us/reader033/viewer/2022050323/5f7c5b9548272338f369a3a6/html5/thumbnails/23.jpg)
Phase I: Escaping the Object
Convex Hull Formation for Programmable Matter Biological Distributed Algorithms 2017
Introduction Background & Model Problem Statement Algorithm Description Runtime Analysis Conclusion
Next, alternate between the following subphases until Phase I is complete:
1. Wall Following: Follow the object using right-hand-rule until finding a “concave turn”.
2. Line Probing: Attempt to build the desired line, and backtrack on failure.
![Page 24: Convex Hull Formation for Programmable Matternavlakha/BDA2017/Material/Talks/DaymudeCon… · Current Programmable Matter Convex Hull Formation for Programmable Matter Biological](https://reader033.vdocuments.us/reader033/viewer/2022050323/5f7c5b9548272338f369a3a6/html5/thumbnails/24.jpg)
Phase I: Escaping the Object
Convex Hull Formation for Programmable Matter Biological Distributed Algorithms 2017
Introduction Background & Model Problem Statement Algorithm Description Runtime Analysis Conclusion
Next, alternate between the following subphases until Phase I is complete:
1. Wall Following: Follow the object using right-hand-rule until finding a “concave turn”.
2. Line Probing: Attempt to build the desired line, and backtrack on failure.
![Page 25: Convex Hull Formation for Programmable Matternavlakha/BDA2017/Material/Talks/DaymudeCon… · Current Programmable Matter Convex Hull Formation for Programmable Matter Biological](https://reader033.vdocuments.us/reader033/viewer/2022050323/5f7c5b9548272338f369a3a6/html5/thumbnails/25.jpg)
Line Bending: A Helpful Tool for Phase II
Convex Hull Formation for Programmable Matter Biological Distributed Algorithms 2017
Introduction Background & Model Problem Statement Algorithm Description Runtime Analysis Conclusion
Bending a straight line by some angle is easy in a synchronous setting, but we have asynchronous activations.
![Page 26: Convex Hull Formation for Programmable Matternavlakha/BDA2017/Material/Talks/DaymudeCon… · Current Programmable Matter Convex Hull Formation for Programmable Matter Biological](https://reader033.vdocuments.us/reader033/viewer/2022050323/5f7c5b9548272338f369a3a6/html5/thumbnails/26.jpg)
Line Bending: A Helpful Tool for Phase II
Convex Hull Formation for Programmable Matter Biological Distributed Algorithms 2017
Introduction Background & Model Problem Statement Algorithm Description Runtime Analysis Conclusion
Bending a straight line by some angle is easy in a synchronous setting, but we have asynchronous activations.
Synchronous Asynchronous
![Page 27: Convex Hull Formation for Programmable Matternavlakha/BDA2017/Material/Talks/DaymudeCon… · Current Programmable Matter Convex Hull Formation for Programmable Matter Biological](https://reader033.vdocuments.us/reader033/viewer/2022050323/5f7c5b9548272338f369a3a6/html5/thumbnails/27.jpg)
Line Bending: A Helpful Tool for Phase II
Convex Hull Formation for Programmable Matter Biological Distributed Algorithms 2017
Introduction Background & Model Problem Statement Algorithm Description Runtime Analysis Conclusion
Bending a straight line by some angle is easy in a synchronous setting, but we have asynchronous activations.
Synchronous Asynchronous
![Page 28: Convex Hull Formation for Programmable Matternavlakha/BDA2017/Material/Talks/DaymudeCon… · Current Programmable Matter Convex Hull Formation for Programmable Matter Biological](https://reader033.vdocuments.us/reader033/viewer/2022050323/5f7c5b9548272338f369a3a6/html5/thumbnails/28.jpg)
Line Bending: A Helpful Tool for Phase II
Convex Hull Formation for Programmable Matter Biological Distributed Algorithms 2017
Introduction Background & Model Problem Statement Algorithm Description Runtime Analysis Conclusion
Bending a straight line by some angle is easy in a synchronous setting, but we have asynchronous activations.
Synchronous Asynchronous
![Page 29: Convex Hull Formation for Programmable Matternavlakha/BDA2017/Material/Talks/DaymudeCon… · Current Programmable Matter Convex Hull Formation for Programmable Matter Biological](https://reader033.vdocuments.us/reader033/viewer/2022050323/5f7c5b9548272338f369a3a6/html5/thumbnails/29.jpg)
Line Bending: A Helpful Tool for Phase II
Convex Hull Formation for Programmable Matter Biological Distributed Algorithms 2017
Introduction Background & Model Problem Statement Algorithm Description Runtime Analysis Conclusion
Bending a straight line by some angle is easy in a synchronous setting, but we have asynchronous activations.
Synchronous Asynchronous
![Page 30: Convex Hull Formation for Programmable Matternavlakha/BDA2017/Material/Talks/DaymudeCon… · Current Programmable Matter Convex Hull Formation for Programmable Matter Biological](https://reader033.vdocuments.us/reader033/viewer/2022050323/5f7c5b9548272338f369a3a6/html5/thumbnails/30.jpg)
Phase II: Constructing the Convex Hull
Convex Hull Formation for Programmable Matter Biological Distributed Algorithms 2017
Introduction Background & Model Problem Statement Algorithm Description Runtime Analysis Conclusion
Phase II begins from the straight line of particles obtained in Phase I. This phase is also divided into two subphases (but these don’t alternate):
1. Moving the root of the particle line to some position in C(O) by bending + forwarding.
![Page 31: Convex Hull Formation for Programmable Matternavlakha/BDA2017/Material/Talks/DaymudeCon… · Current Programmable Matter Convex Hull Formation for Programmable Matter Biological](https://reader033.vdocuments.us/reader033/viewer/2022050323/5f7c5b9548272338f369a3a6/html5/thumbnails/31.jpg)
Phase II: Constructing the Convex Hull
Convex Hull Formation for Programmable Matter Biological Distributed Algorithms 2017
Introduction Background & Model Problem Statement Algorithm Description Runtime Analysis Conclusion
Phase II begins from the straight line of particles obtained in Phase I. This phase is also divided into two subphases (but these don’t alternate):
1. Moving the root of the particle line to some position in C(O) by bending + forwarding.
![Page 32: Convex Hull Formation for Programmable Matternavlakha/BDA2017/Material/Talks/DaymudeCon… · Current Programmable Matter Convex Hull Formation for Programmable Matter Biological](https://reader033.vdocuments.us/reader033/viewer/2022050323/5f7c5b9548272338f369a3a6/html5/thumbnails/32.jpg)
Phase II: Constructing the Convex Hull
Convex Hull Formation for Programmable Matter Biological Distributed Algorithms 2017
Introduction Background & Model Problem Statement Algorithm Description Runtime Analysis Conclusion
Phase II begins from the straight line of particles obtained in Phase I. This phase is also divided into two subphases (but these don’t alternate):
1. Moving the root of the particle line to some position in C(O) by bending + forwarding.
![Page 33: Convex Hull Formation for Programmable Matternavlakha/BDA2017/Material/Talks/DaymudeCon… · Current Programmable Matter Convex Hull Formation for Programmable Matter Biological](https://reader033.vdocuments.us/reader033/viewer/2022050323/5f7c5b9548272338f369a3a6/html5/thumbnails/33.jpg)
Phase II: Constructing the Convex Hull
Convex Hull Formation for Programmable Matter Biological Distributed Algorithms 2017
Introduction Background & Model Problem Statement Algorithm Description Runtime Analysis Conclusion
Phase II begins from the straight line of particles obtained in Phase I. This phase is also divided into two subphases (but these don’t alternate):
1. Moving the root of the particle line to some position in C(O) by bending + forwarding.
![Page 34: Convex Hull Formation for Programmable Matternavlakha/BDA2017/Material/Talks/DaymudeCon… · Current Programmable Matter Convex Hull Formation for Programmable Matter Biological](https://reader033.vdocuments.us/reader033/viewer/2022050323/5f7c5b9548272338f369a3a6/html5/thumbnails/34.jpg)
Phase II: Constructing the Convex Hull
Convex Hull Formation for Programmable Matter Biological Distributed Algorithms 2017
Introduction Background & Model Problem Statement Algorithm Description Runtime Analysis Conclusion
Phase II begins from the straight line of particles obtained in Phase I. This phase is also divided into two subphases (but these don’t alternate):
1. Moving the root of the particle line to some position in C(O) by bending + forwarding.
2. Bending the line around the object, occupying the rest of C(O).
![Page 35: Convex Hull Formation for Programmable Matternavlakha/BDA2017/Material/Talks/DaymudeCon… · Current Programmable Matter Convex Hull Formation for Programmable Matter Biological](https://reader033.vdocuments.us/reader033/viewer/2022050323/5f7c5b9548272338f369a3a6/html5/thumbnails/35.jpg)
Phase II: Constructing the Convex Hull
Convex Hull Formation for Programmable Matter Biological Distributed Algorithms 2017
Introduction Background & Model Problem Statement Algorithm Description Runtime Analysis Conclusion
Phase II begins from the straight line of particles obtained in Phase I. This phase is also divided into two subphases (but these don’t alternate):
1. Moving the root of the particle line to some position in C(O) by bending + forwarding.
2. Bending the line around the object, occupying the rest of C(O).
![Page 36: Convex Hull Formation for Programmable Matternavlakha/BDA2017/Material/Talks/DaymudeCon… · Current Programmable Matter Convex Hull Formation for Programmable Matter Biological](https://reader033.vdocuments.us/reader033/viewer/2022050323/5f7c5b9548272338f369a3a6/html5/thumbnails/36.jpg)
Phase II: Constructing the Convex Hull
Convex Hull Formation for Programmable Matter Biological Distributed Algorithms 2017
Introduction Background & Model Problem Statement Algorithm Description Runtime Analysis Conclusion
Phase II begins from the straight line of particles obtained in Phase I. This phase is also divided into two subphases (but these don’t alternate):
1. Moving the root of the particle line to some position in C(O) by bending + forwarding.
2. Bending the line around the object, occupying the rest of C(O).
![Page 37: Convex Hull Formation for Programmable Matternavlakha/BDA2017/Material/Talks/DaymudeCon… · Current Programmable Matter Convex Hull Formation for Programmable Matter Biological](https://reader033.vdocuments.us/reader033/viewer/2022050323/5f7c5b9548272338f369a3a6/html5/thumbnails/37.jpg)
Phase II: Constructing the Convex Hull
Convex Hull Formation for Programmable Matter Biological Distributed Algorithms 2017
Introduction Background & Model Problem Statement Algorithm Description Runtime Analysis Conclusion
Phase II begins from the straight line of particles obtained in Phase I. This phase is also divided into two subphases (but these don’t alternate):
1. Moving the root of the particle line to some position in C(O) by bending + forwarding.
2. Bending the line around the object, occupying the rest of C(O).
![Page 38: Convex Hull Formation for Programmable Matternavlakha/BDA2017/Material/Talks/DaymudeCon… · Current Programmable Matter Convex Hull Formation for Programmable Matter Biological](https://reader033.vdocuments.us/reader033/viewer/2022050323/5f7c5b9548272338f369a3a6/html5/thumbnails/38.jpg)
Phase II: Constructing the Convex Hull
Convex Hull Formation for Programmable Matter Biological Distributed Algorithms 2017
Introduction Background & Model Problem Statement Algorithm Description Runtime Analysis Conclusion
Phase II begins from the straight line of particles obtained in Phase I. This phase is also divided into two subphases (but these don’t alternate):
1. Moving the root of the particle line to some position in C(O) by bending + forwarding.
2. Bending the line around the object, occupying the rest of C(O).
![Page 39: Convex Hull Formation for Programmable Matternavlakha/BDA2017/Material/Talks/DaymudeCon… · Current Programmable Matter Convex Hull Formation for Programmable Matter Biological](https://reader033.vdocuments.us/reader033/viewer/2022050323/5f7c5b9548272338f369a3a6/html5/thumbnails/39.jpg)
Phase II: Constructing the Convex Hull
Convex Hull Formation for Programmable Matter Biological Distributed Algorithms 2017
Introduction Background & Model Problem Statement Algorithm Description Runtime Analysis Conclusion
Phase II begins from the straight line of particles obtained in Phase I. This phase is also divided into two subphases (but these don’t alternate):
1. Moving the root of the particle line to some position in C(O) by bending + forwarding.
2. Bending the line around the object, occupying the rest of C(O).
![Page 40: Convex Hull Formation for Programmable Matternavlakha/BDA2017/Material/Talks/DaymudeCon… · Current Programmable Matter Convex Hull Formation for Programmable Matter Biological](https://reader033.vdocuments.us/reader033/viewer/2022050323/5f7c5b9548272338f369a3a6/html5/thumbnails/40.jpg)
Phase II: Constructing the Convex Hull
Convex Hull Formation for Programmable Matter Biological Distributed Algorithms 2017
Introduction Background & Model Problem Statement Algorithm Description Runtime Analysis Conclusion
Phase II begins from the straight line of particles obtained in Phase I. This phase is also divided into two subphases (but these don’t alternate):
1. Moving the root of the particle line to some position in C(O) by bending + forwarding.
2. Bending the line around the object, occupying the rest of C(O).
![Page 41: Convex Hull Formation for Programmable Matternavlakha/BDA2017/Material/Talks/DaymudeCon… · Current Programmable Matter Convex Hull Formation for Programmable Matter Biological](https://reader033.vdocuments.us/reader033/viewer/2022050323/5f7c5b9548272338f369a3a6/html5/thumbnails/41.jpg)
Preliminary Worst-Case Runtime Analysis
Convex Hull Formation for Programmable Matter Biological Distributed Algorithms 2017
Introduction Background & Model Problem Statement Algorithm Description Runtime Analysis Conclusion
Let n = |P| and m be the area occupied by O.
Phase I: O(n + m) rounds. ?
• Spanning forest primitive: O(n) rounds. ✓
• Wall following subphase: O(m) rounds. ✓
• Line probing subphase: O(m) rounds. ?
All together: O(n + m) rounds…?
We measure runtime in asynchronous rounds.
Phase II: O(n) rounds. ?
• Each line bending: O(n) rounds. ?
• Move the root to the hull: ≤ 6 line bends. ✓
• Wrap the rest of the line: 6 line bends. ✓
![Page 42: Convex Hull Formation for Programmable Matternavlakha/BDA2017/Material/Talks/DaymudeCon… · Current Programmable Matter Convex Hull Formation for Programmable Matter Biological](https://reader033.vdocuments.us/reader033/viewer/2022050323/5f7c5b9548272338f369a3a6/html5/thumbnails/42.jpg)
Future Work
Convex Hull Formation for Programmable Matter Biological Distributed Algorithms 2017
Introduction Background & Model Problem Statement Algorithm Description Runtime Analysis Conclusion
• For convex hull formation (work-in-progress):• Formalize the ideas outlined here into a fully developed distributed algorithm.
• Theoretical results: work out the details of correctness and runtime proofs.
• For Self-Organizing Particle Systems in general:• Pushing towards applications: bridging/filling gaps, etc.
• Investigate more fault tolerant algorithms.
• Generalize the existing model and algorithms to 3-dimensional space, if possible.
![Page 43: Convex Hull Formation for Programmable Matternavlakha/BDA2017/Material/Talks/DaymudeCon… · Current Programmable Matter Convex Hull Formation for Programmable Matter Biological](https://reader033.vdocuments.us/reader033/viewer/2022050323/5f7c5b9548272338f369a3a6/html5/thumbnails/43.jpg)
Collaborators
Convex Hull Formation for Programmable Matter Biological Distributed Algorithms 2017
Introduction Background & Model Problem Statement Algorithm Description Runtime Analysis Conclusion
Andréa W. Richa Joshua J. Daymude Christian Scheideler Robert Gmyr Thim Strothmann
![Page 44: Convex Hull Formation for Programmable Matternavlakha/BDA2017/Material/Talks/DaymudeCon… · Current Programmable Matter Convex Hull Formation for Programmable Matter Biological](https://reader033.vdocuments.us/reader033/viewer/2022050323/5f7c5b9548272338f369a3a6/html5/thumbnails/44.jpg)
Convex Hull Formation for Programmable Matter Biological Distributed Algorithms 2017
Introduction Background & Model Problem Statement Algorithm Description Runtime Analysis Conclusion
Thank you!sops.engineering.asu.edu