CS 574, Spring 2021
Randomized Algorithms
Instructor
-
Timothy Chan
(tmc "at" illinois.edu, office hours: Fri 2:00pm-3:00pm on
zoom,
or if you want to meet at a different time, just email me)
Meeting Time
-
Mon & Wed 9:30am-10:45am on zoom
(if you are not officially registered but wants to attend, just email me)
Course Work
- 4 or 5 homeworks (problem sets), worth 70% (done in groups of up to 3)
- final exam, worth 30% (between May 10 Mon 2pm and May 12 Wed 2pm,
3 hours duration)
Tentative Topics/Outline
- Introduction: quicksort and quickselect, Rabin and Miller's
primality testing algorithm, complexity classes (ZPP, RP, BPP)
- Random re-ordering: examples from computational geometry,
AND/OR tree evaluation
- Random sampling: selection again (and
2-wise independence and 2-point sampling), Clarkson's linear programming algorithm
(and epsilon-nets),
Karger, Klein, and Tarjan's MST algorithm,
witness finding, color coding
- More sampling and concentration bounds:
Chernoff, Hoeffding, Azuma (and martingales), further applications
(balls and bins, approximate counting, ...)
- Algebraic methods:
fingerprinting (and Schwarz-Zippel lemma),
Rabin and Karp's string matching algorithm,
Mulmuley, Vazirani, and Vazirani's parallel
perfect matching algorithm (and isolation lemma)
- Randomized data structures:
skip-list, treap, hashing (2-universal, perfect, power of 2 choices, cuckoo, ...)
- Random projection and embeddings: locality-sensitive hashing,
dimension reduction (Johnson-Lindenstrass), ...
- Random walk and Markov chains: Schöning's k-SAT algorithm,
Goel, Kapralov, and Khanna's bipartite perfect matching algorithm for
regular graphs, approximate counting, expanders, ...
- Randomized rounding: MAX-SAT/MAX-CUT
- Other topics: derandomization (e.g., method of conditional probabilities),
Lovasz local lemma (and Moser and Tardos' algorithm),
randomized online algorithms, sublinear-time algorithms, ... (and maybe some of my recent research, if time permits :-)
Prerequisites
- Undergraduate algorithms, at the level of CS374
(CS473 is helpful but not required), and sufficient mathematical
maturity (including basic knowledge of probability theory)
Resources
- Motwani and Raghavan's book (1995) is highly recommended and
available online for UIUC students
-
Sariel's notes from past versions of the course are also recommended
- another book is Mitzenmacher and Upfal (2nd ed., 2017, but apparently not available online)
- other notes online can be found in similar courses elsewhere
(e.g., Tim Roughgarden (Stanford),
Eric Price (UT Austin),
James Aspnes (Yale), ...)
- Campuswire for discussion
(enrollment code: 1803)
Lectures
(Scribbles from class will be provided below.
Lectures will be also recorded, and for registered students, recordings may be accessed at mediaspace. [MR] refers to Motwani and Raghavan's book.
Many papers may be accessed via
UIUC library's proxy bookmarklet.)
- Jan 25: Introduction. Quicksort and quickselect. Primality testing (Miller-Rabin algorithm). Ref: [MR, Sec 14.6]
- Jan 27: Basic probability review. Randomized complexity classes (ZPP, RP). Ref: [MR, Sec 1.5]
- Feb 1: More randomized complexity classes (BPP).
Adleman's theorem (BPP in P/poly). Random re-ordering.
- Feb 3: Line segment intersection and trapezoidal decomposition (randomized incremental algorithm by Clarkson-Shor'89/Mulmuley'88, and Seidel's "backwards analysis"). Ref: [MR, Sec 9.6]
- Feb 8: AND/OR tree evaluation (Snir's algorithm). Yao's principle. Ref: [MR, Sec 2.1-2.2]
- Feb 10: Random sampling. Median finding (Floyd and Rivest's algorithm).
Ref: [MR, Sec 3.3]
- Feb 15: Median cont'd (pairwise independence, 2-point sampling, time/space tradeoffs).
Ref: [MR, Sec 3.3-3.4; my paper]
- Feb 17: no class (break).
- Feb 22: Min enclosing circle (Clarkson's sampling algorithm). Ref: [MR, Sec 9.10]
- Feb 24
and Mar 1: MST (Karger, Klein, and Tarjan's sampling-based algorithm). Ref: [MR, Sec 10.3] (and backwards analysis)
- Mar 3: Witness finding for Boolean matrix multiplication. Color coding for simple k-cycles.
Ref: [MR, Sec 10.1.2] and Alon, Yuster, and Zwick's paper
- Mar 8 and Mar 10: Concentration bounds. Chernoff, Hoeffding, and applications.
Ref: [MR, Sec 4.1] or [Sariel's notes, Ch 7]
- Mar 15
and Mar 17: Applications to approximate counting and approximate volume of unions (Karp and Luby's algorithm). Azuma and martingales (and method of bounded differences).
Ref: [MR, Sec 11.2 and 4.4]
- Mar 22: Algebraic techniques. Matrix multiplication verification (Freivalds), string matching (Rabin-Karp), Schwartz-Zippel lemma. Ref: [MR, Sec 7.1-7.6]
- Mar 24: no class (break).
- Mar 29: Perfect matching (Mulmuley, Vazirani, and Vazirani's parallel algorithm). Isolation lemma. Ref: [MR, Sec 7.3 and 12.4]
- Mar 31: Randomized data structures.
Skip lists. Treaps. Ref: [MR, Sec 8.3 and 8.2]
- Apr 5: Hashing (2-universal hash families, perfect hashing). Ref: [MR, Sec 8.4] or
Jeff's notes
- Apr 7: Approximate nearest neighbor search by
locality-sensitive hashing. Ref:
original paper by Indyk and Motwani
- Apr 12: Johnson-Lindenstrass lemma (dimension reduction, L2 to L1). Ref: Sec 2 of notes by Matousek
- Apr 14: Random walks.
Papadimitriou's 2SAT algorithm, Schöning's 3SAT algorithm.
Ref: [MR, Sec 6.1] and Schöning's paper
- Apr 19: Markov chains, stationary distribution, hitting time.
Ref: [MR, Sec 6.2]
- Apr 21: Perfect matching in regular bipartite graphs (Goel-Kapralov-Khanna). Undirected connectivity in randomized log space.
Ref: Goel et al.'s paper,
[MR, Sec 6.5-6.6]
- Apr 26: Cover time. Expanders and rapid mixing.
Ref: [MR, Sec 6.7], Hoory, Linial, and Wigderson's survey on expanders
(in particular, Sec 3.1)
- Apr 28: Expanders and probability amplification.
Ref: [MR, Sec 6.8] and Sec 3.2-3.3 of the above survey
- May 3: Miscellaneous topics. Lovasz Local Lemma and Moser's algorithm.
Ref: [MR, Sec 5.5] and
Lance Fortnow's blog post (or Terry Tao's)
- May 5: Derandomization via method of conditional probabilities (examples: discrepancy and MAX-SAT via randomized rounding).
Ref: [MR, Sec 5.6 and 5.2]