From ofirgeri at stanford.edu Mon Oct 1 11:33:22 2018
From: ofirgeri at stanford.edu (Ofir Geri)
Date: Mon, 1 Oct 2018 18:33:22 +0000
Subject: [theory-seminar] Theory Seminar (10/5): Madhu Sudan
Message-ID:
Hi all,
This week in theory seminar Madhu Sudan from Harvard will be giving a talk on General Strong Polarization (abstract below). It will take place on Friday 10/5 at 3:00 PM in Gates 463A.
A reminder: Next week we will have two theory seminar talks at non-standard times. Shahar Dobzinski will be speaking on Monday 10/8 and Juba Ziani on Thursday 10/11 (and there will be no talk on Friday).
Hope to see you there!
Ofir
General Strong Polarization
Speaker: Madhu Sudan (Harvard)
A martingale is a sequence of random variables that maintain their future expected value conditioned on the past. A [0,1]-bounded martingale is said to polarize if it converges in the limit to either 0 or 1 with probability 1. A martingale is said to polarize strongly, if in t steps it is sub-exponentially close to its limit with all but exponentially small probability. In 2008, Arikan built a powerful class of error-correcting codes called Polar codes. The essence of his theory associates a martingale with every invertible square matrix over a field (and a channel) and showed that polarization of the martingale leads to a construction of codes that converge to Shannon capacity. In 2013, Guruswami and Xia, and independently Hassani et al. showed that strong polarization of the Arikan martingale leads to codes that converge to Shannon capacity at finite block lengths, specifically at lengths that are inverse polynomial in the gap to capacity, thereby resolving a major mathematical challenge associated with the attainment of Shannon capacity.
We show that a simple necessary condition for an invertible matrix to polarize over any non-trivial channel is also sufficient for strong polarization over all symmetric channels over all prime fields. Previously the only matrix which was known to polarize strongly was the 2x2 Hadamard matrix. In addition to the generality of our result, it also leads to arguably simpler proofs. The essence of our proof is a local definition" of polarization which only restricts the evolution of the martingale in a single step, and a general theorem showing the local polarization suffices for strong polarization.
In this talk I will introduce polarization and polar codes and, time permitting, present a full proof of our main theorem. No prior background on polar codes will be assumed.
Based on joint work with Jaroslaw Blasiok, Venkatesan Guruswami, Preetum Nakkiran and Atri Rudra.
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
From rmhulett at stanford.edu Tue Oct 2 09:18:40 2018
From: rmhulett at stanford.edu (Reyna Marie Hulett)
Date: Tue, 2 Oct 2018 16:18:40 +0000
Subject: [theory-seminar] Theory Lunch -- Don Knuth
Message-ID:
Hi all,
This week's theory lunch will be given by Don Knuth! The title is "The First Greedy Algorithm," which was published in 1825. As always, please join us on Thursday from noon to 1 pm in Gates 463A!
Best,
-Reyna
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
From rmhulett at stanford.edu Tue Oct 2 09:18:40 2018
From: rmhulett at stanford.edu (Reyna Marie Hulett)
Date: Tue, 2 Oct 2018 16:18:40 +0000
Subject: [theory-seminar] Theory Lunch -- Don Knuth
Message-ID:
Hi all,
This week's theory lunch will be given by Don Knuth! The title is "The First Greedy Algorithm," which was published in 1825. As always, please join us on Thursday from noon to 1 pm in Gates 463A!
Best,
-Reyna
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
From ofirgeri at stanford.edu Tue Oct 2 14:44:08 2018
From: ofirgeri at stanford.edu (Ofir Geri)
Date: Tue, 2 Oct 2018 21:44:08 +0000
Subject: [theory-seminar] Additional Theory Talk This Thursday (10/4):
Jelani Nelson on Sketching Algorithms (3pm at Gates 104)
Message-ID:
Hi all,
There will be an additional theory talk this week on Thursday (10/4), 3-4pm in Gates 104. Jelani Nelson (Harvard) will be speaking on Sketching Algorithms (see details below).
Hope to see you there!
Ofir
Sketching Algorithms
Jelani Nelson
Harvard University
ABSTRACT:
A "sketch" is a data structure supporting some pre-specified set of queries and updates to a database while consuming space substantially (often exponentially) less than the information theoretic minimum required to store everything seen. Thus, sketching can be seen as some form of functional compression. The advantages of sketching include reduced memory consumption, faster algorithms, and reduced bandwidth requirements in distributed computing environments.
Sketching has been a core technique in several domains, including processing massive data streams with low memory footprint, 'compressed sensing' for lossy compression of signals with few linear measurements, and dimensionality reduction or 'random projection' methods for speedups in large-scale linear algebra algorithms, and high-dimensional computational geometry.
This talk will provide a glimpse into some recent progress on core problems in the theory of sketching algorithms.
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
From moses at cs.stanford.edu Wed Oct 3 07:15:33 2018
From: moses at cs.stanford.edu (Moses Charikar)
Date: Wed, 3 Oct 2018 07:15:33 -0700
Subject: [theory-seminar] Fwd: Open positions in Duke CS
In-Reply-To:
References:
Message-ID:
Hi folks,
Kamesh Munagala (a Stanford grad) writes that they have many open faculty
positions in the Duke CS department this year, and some of the focus areas
are algorithms, optimization, and crypto (see below). If you are graduating
this year, I encourage you to seriously consider Duke. They have a strong
theory group:
http://theorywiki.cs.duke.edu/index.php/Home
Cheers,
Moses
==
----
*Tenure-Track Faculty Positions in Computer Science*
Duke University invites applications and nominations for four tenure-track
or tenured faculty positions in all areas of computer science, and for two
additional joint positions between Computer Science and other departments.
Areas include artificial intelligence, machine learning, computer systems,
security and privacy, database systems, computer vision, algorithms,
optimization, as well as interdisciplinary work that relates to the social
sciences or biological sciences. Positions are at all ranks and to begin
July 2019.
Candidates are expected to have a doctoral degree in computer science or
related areas. A successful candidate must have a solid disciplinary
foundation and demonstrate promise of outstanding scholarship in every
regard, including research and teaching.
The Duke faculty and student communities are currently very diverse and are
strongly committed to further enhancing this diversity. We seek faculty
members who are committed to building a diverse and inclusive community,
which fosters excellence in research and teaching. We strongly encourage
applications from women and underrepresented minorities in computing.
Please see http://www.cs.duke.edu and http://www.provost.duke.edu/faculty for
information about the Department and advantages that Duke offers to faculty.
Applicants should submit their materials (cover letter, research statement,
teaching statement, contacts for at least three references) electronically
through AcademicJobsOnline: https://academicjobsonline.org/ajo/jobs/12079,
and solicit letters of reference. For full consideration, applications and
references should be received by December 15, 2018.
Duke University is an Affirmative Action/Equal Opportunity Employer
committed to providing employment opportunity without regard to an
individual's age, color, disability, genetic information, gender, gender
identity, national origin, race, religion, sexual orientation, or veteran
status.
Durham, Chapel Hill, and the Research Triangle of North Carolina are
frequently ranked among the best places in the country to live and work.
Duke and the many other universities in the area offer a wealth of
education and employment opportunities for spouses and families.
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
From baxelrod at stanford.edu Thu Oct 4 11:41:59 2018
From: baxelrod at stanford.edu (Brian Axelrod)
Date: Thu, 4 Oct 2018 18:41:59 +0000
Subject: [theory-seminar] Books in Theory Seminar Room (463A)
Message-ID:
Hi,
When I was travelling this summer my office got accidentally cleaned out and all my textbooks got moved into the theory seminar room (463A) or were lost.
Sadly, I've been unable to recover all my textbooks. If you've borrowed a textbook from the room recently, or otherwise obtained a textbook that looks like it's mine, can you please let me check if it was mine or drop it off in one of the locations listed below (no questions asked)? They're mostly math textbooks and a couple of CS books.
Some of the books are of sentimental value to me as they were gifts from a professor or are from classes I worked really hard on. It will be very sad if I'm not able to recover some of them. Most of the missing ones were bound at the MIT printing press as opposed to a mass market printer and look the part (but still nicer than a ring binding).
Some of the book include:
Nonlinear Optimization
Introduction to Representation Theory by Etingof
Probability Theory by Durret
A bound copy of my MIT master's thesis (author is Brian Axelrod)
Some of the titles I don't remember off the top of my head but it includes:
2x Statistics Books (both by Stanford Profs if I remember correctly. One is a classical text, the other is modern)
Convex Geometry
A Small algebraic geometry paperback (mass market bound)
[Nonlinear] Optimization
A bound set of notes by Terence Tao (I think these have my scribbles in them making it easy to identify)
I never wrote down a complete inventory, so if you borrowed another mathy book from the room recently, I'd really appreciate you getting in touch with me to let me check if it was mine.
If you don't want to reach out to me (I won't be angry at you!) and want to return a book anonymously you can do so at by visiting one of the admin listed below:
Megan Harris's office, rm 479
Lashonda Eagels, rm 217 (if she's not around Tin Tin can open the office, she works in rm 152)
Thanks for your time!
Brian
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
From ofirgeri at stanford.edu Fri Oct 5 08:51:57 2018
From: ofirgeri at stanford.edu (Ofir Geri)
Date: Fri, 5 Oct 2018 15:51:57 +0000
Subject: [theory-seminar] Theory Seminar (10/5): Madhu Sudan
In-Reply-To:
References:
Message-ID:
Reminder: Madhu Sudan is giving a talk at theory seminar today at 3pm!
Hope to see you there!
Ofir
________________________________
From: Ofir Geri
Sent: Monday, October 1, 2018 11:33:22 AM
To: thseminar at cs.stanford.edu
Subject: Theory Seminar (10/5): Madhu Sudan
Hi all,
This week in theory seminar Madhu Sudan from Harvard will be giving a talk on General Strong Polarization (abstract below). It will take place on Friday 10/5 at 3:00 PM in Gates 463A.
A reminder: Next week we will have two theory seminar talks at non-standard times. Shahar Dobzinski will be speaking on Monday 10/8 and Juba Ziani on Thursday 10/11 (and there will be no talk on Friday).
Hope to see you there!
Ofir
General Strong Polarization
Speaker: Madhu Sudan (Harvard)
A martingale is a sequence of random variables that maintain their future expected value conditioned on the past. A [0,1]-bounded martingale is said to polarize if it converges in the limit to either 0 or 1 with probability 1. A martingale is said to polarize strongly, if in t steps it is sub-exponentially close to its limit with all but exponentially small probability. In 2008, Arikan built a powerful class of error-correcting codes called Polar codes. The essence of his theory associates a martingale with every invertible square matrix over a field (and a channel) and showed that polarization of the martingale leads to a construction of codes that converge to Shannon capacity. In 2013, Guruswami and Xia, and independently Hassani et al. showed that strong polarization of the Arikan martingale leads to codes that converge to Shannon capacity at finite block lengths, specifically at lengths that are inverse polynomial in the gap to capacity, thereby resolving a major mathematical challenge associated with the attainment of Shannon capacity.
We show that a simple necessary condition for an invertible matrix to polarize over any non-trivial channel is also sufficient for strong polarization over all symmetric channels over all prime fields. Previously the only matrix which was known to polarize strongly was the 2x2 Hadamard matrix. In addition to the generality of our result, it also leads to arguably simpler proofs. The essence of our proof is a local definition" of polarization which only restricts the evolution of the martingale in a single step, and a general theorem showing the local polarization suffices for strong polarization.
In this talk I will introduce polarization and polar codes and, time permitting, present a full proof of our main theorem. No prior background on polar codes will be assumed.
Based on joint work with Jaroslaw Blasiok, Venkatesan Guruswami, Preetum Nakkiran and Atri Rudra.
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
From ofirgeri at stanford.edu Fri Oct 5 18:18:23 2018
From: ofirgeri at stanford.edu (Ofir Geri)
Date: Sat, 6 Oct 2018 01:18:23 +0000
Subject: [theory-seminar] Theory Seminar (10/8): Shahar Dobzinski
Message-ID:
Hi all,
On Monday (10/8) Shahar Dobzinski from Weizmann will give a theory seminar talk: From Cognitive Biases to the Communication Complexity of Local Search (abstract below). The talk will be at 3:00 PM in Gates 463A.
Please note the non-standard day for the talk. We will have another theory seminar at a non-standard time next week: Juba Ziani will be speaking on Thursday (10/11) at 4:15 PM (and there will be no talk on Friday).
For the snacks, if anyone has any restrictions/allergies (or requests), please let me know.
Hope to see you there!
Ofir
>From Cognitive Biases to the Communication Complexity of Local Search
Speaker: Shahar Dobzinski (Weizmann)
In this talk I will tell you how analyzing economic markets where agents have cognitive biases has led to better understanding of the communication complexity of local search procedures.
We begin the talk with studying combinatorial auctions with bidders that exhibit endowment effect. In most of the previous work on cognitive biases in algorithmic game theory (e.g., [Kleinberg and Oren, EC'14] and its follow-ups) the focus was on analyzing the implications and mitigating their negative consequences. In contrast, we show how cognitive biases can sometimes be harnessed to improve the outcome.
Specifically, we study Walrasian equilibria in combinatorial markets. It is well known that a Walrasian equilibrium exists only in limited settings, e.g., when all valuations are gross substitutes, but fails to exist in more general settings, e.g., when the valuations are submodular. We consider combinatorial settings in which bidders exhibit the endowment effect, that is, their value for items increases with ownership. Our main result here shows that when the valuations are submodular even a mild level of endowment effect suffices to guarantee the existence of Walrasian equilibrium. In fact, we show that in contrast to Walrsian equilibria with standard utility maximizers bidders -- in which the equilibrium allocation must be a global optimum -- when bidders exhibit endowment effect any local optimum can be an equilibrium allocation.
This raises the natural question of understanding the complexity of computing a local maximum in combinatorial markets. We reduce it to the following communication variant of local search: there is some fixed, commonly known graph G. Alice holds f_A and Bob holds f_B, both are functions that specify a value for each vertex. The goal is to find a local maximum of f_A+f_B, i.e., a vertex v for which f_A(v)+f_B(v) >= f_A(u)+f_B(u) for every neighbor u of v. We prove that finding a local maximum requires polynomial (in the number of vertices) bits of communication.
Based on joint works with Moshe Babaioff, Yakov Babichenko, Noam Nisan, and Sigal Oren.
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
From ofirgeri at stanford.edu Mon Oct 8 10:12:56 2018
From: ofirgeri at stanford.edu (Ofir Geri)
Date: Mon, 8 Oct 2018 17:12:56 +0000
Subject: [theory-seminar] Theory Seminar (10/8): Shahar Dobzinski
In-Reply-To:
References:
Message-ID:
Reminder: Shahar Dobzinski is giving a talk at theory seminar today at 3pm.
________________________________
From: Ofir Geri
Sent: Friday, October 5, 2018 6:18:23 PM
To: thseminar at cs.stanford.edu
Subject: Theory Seminar (10/8): Shahar Dobzinski
Hi all,
On Monday (10/8) Shahar Dobzinski from Weizmann will give a theory seminar talk: From Cognitive Biases to the Communication Complexity of Local Search (abstract below). The talk will be at 3:00 PM in Gates 463A.
Please note the non-standard day for the talk. We will have another theory seminar at a non-standard time next week: Juba Ziani will be speaking on Thursday (10/11) at 4:15 PM (and there will be no talk on Friday).
For the snacks, if anyone has any restrictions/allergies (or requests), please let me know.
Hope to see you there!
Ofir
>From Cognitive Biases to the Communication Complexity of Local Search
Speaker: Shahar Dobzinski (Weizmann)
In this talk I will tell you how analyzing economic markets where agents have cognitive biases has led to better understanding of the communication complexity of local search procedures.
We begin the talk with studying combinatorial auctions with bidders that exhibit endowment effect. In most of the previous work on cognitive biases in algorithmic game theory (e.g., [Kleinberg and Oren, EC'14] and its follow-ups) the focus was on analyzing the implications and mitigating their negative consequences. In contrast, we show how cognitive biases can sometimes be harnessed to improve the outcome.
Specifically, we study Walrasian equilibria in combinatorial markets. It is well known that a Walrasian equilibrium exists only in limited settings, e.g., when all valuations are gross substitutes, but fails to exist in more general settings, e.g., when the valuations are submodular. We consider combinatorial settings in which bidders exhibit the endowment effect, that is, their value for items increases with ownership. Our main result here shows that when the valuations are submodular even a mild level of endowment effect suffices to guarantee the existence of Walrasian equilibrium. In fact, we show that in contrast to Walrsian equilibria with standard utility maximizers bidders -- in which the equilibrium allocation must be a global optimum -- when bidders exhibit endowment effect any local optimum can be an equilibrium allocation.
This raises the natural question of understanding the complexity of computing a local maximum in combinatorial markets. We reduce it to the following communication variant of local search: there is some fixed, commonly known graph G. Alice holds f_A and Bob holds f_B, both are functions that specify a value for each vertex. The goal is to find a local maximum of f_A+f_B, i.e., a vertex v for which f_A(v)+f_B(v) >= f_A(u)+f_B(u) for every neighbor u of v. We prove that finding a local maximum requires polynomial (in the number of vertices) bits of communication.
Based on joint works with Moshe Babaioff, Yakov Babichenko, Noam Nisan, and Sigal Oren.
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
From ofirgeri at stanford.edu Mon Oct 8 16:25:38 2018
From: ofirgeri at stanford.edu (Ofir Geri)
Date: Mon, 8 Oct 2018 23:25:38 +0000
Subject: [theory-seminar] Theory Seminar (10/11): Juba Ziani
Message-ID:
Hi all,
On Thursday (10/11) Juba Ziani from Caltech will give a theory seminar talk on Optimal Data Acquisition for Statistical Estimation (abstract below).
The talk will be at 4:15 PM in Gates 463A. Please notice the non-standard time.
Hope to see you there!
Ofir
Optimal Data Acquisition for Statistical Estimation
Speaker: Juba Ziani (Caltech)
We consider a data analyst's problem of purchasing data from strategic agents to compute an unbiased estimate of a statistic of interest. Agents incur private costs to reveal their data and the costs can be arbitrarily correlated with their data. Once revealed, data are verifiable. This paper focuses on linear unbiased estimators. We design an individually rational and incentive compatible mechanism that optimizes the worst-case mean-squared error of the estimation, where the worst-case is over the unknown correlation between costs and data, subject to a budget constraint in expectation. We characterize the form of the optimal mechanism in closed-form. We further extend our results to acquiring data for estimating a parameter in regression analysis, where private costs can correlate with the values of the dependent variable but not with the values of the independent variables.
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
From rmhulett at stanford.edu Tue Oct 9 13:07:43 2018
From: rmhulett at stanford.edu (Reyna Marie Hulett)
Date: Tue, 9 Oct 2018 20:07:43 +0000
Subject: [theory-seminar] Theory Lunch -- Nima Anari
Message-ID:
Hi everyone,
This week's theory lunch will be given by Nima Anari, on ?Log-Concave Polynomials: Combinatorics and Algorithms?--see abstract below. As always, please join us Thursday from noon to 1 pm in Gates 463A!
----------------------------------------
Abstract:
I will discuss an analytic property of multivariate polynomials, which we call complete log-concavity, and its surprising uses to attack several problems in combinatorics, discrete optimization, and discrete sampling. This property defines a large class of discrete distributions that should be thought of as the analog of the well-studied continuous log-concave distributions. This class includes uniform distributions over bases or independent sets of matroids, determinantal point processes, the random cluster model for some regimes of parameters, and several other generalizations.
I will then discuss an application of this theory in which we resolve a combinatorial conjecture of Mason on the ultra-log-concavity of the number of independent sets of varying sizes in matroids.
Based on joint work with Kuikui Liu, Shayan Oveis Gharan, and Cynthia Vinzant.
----------------------------------------
Cheers,
-Reyna
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
From rmhulett at stanford.edu Tue Oct 9 13:07:43 2018
From: rmhulett at stanford.edu (Reyna Marie Hulett)
Date: Tue, 9 Oct 2018 20:07:43 +0000
Subject: [theory-seminar] Theory Lunch -- Nima Anari
Message-ID:
Hi everyone,
This week's theory lunch will be given by Nima Anari, on ?Log-Concave Polynomials: Combinatorics and Algorithms?--see abstract below. As always, please join us Thursday from noon to 1 pm in Gates 463A!
----------------------------------------
Abstract:
I will discuss an analytic property of multivariate polynomials, which we call complete log-concavity, and its surprising uses to attack several problems in combinatorics, discrete optimization, and discrete sampling. This property defines a large class of discrete distributions that should be thought of as the analog of the well-studied continuous log-concave distributions. This class includes uniform distributions over bases or independent sets of matroids, determinantal point processes, the random cluster model for some regimes of parameters, and several other generalizations.
I will then discuss an application of this theory in which we resolve a combinatorial conjecture of Mason on the ultra-log-concavity of the number of independent sets of varying sizes in matroids.
Based on joint work with Kuikui Liu, Shayan Oveis Gharan, and Cynthia Vinzant.
----------------------------------------
Cheers,
-Reyna
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
From ofirgeri at stanford.edu Wed Oct 10 18:13:46 2018
From: ofirgeri at stanford.edu (Ofir Geri)
Date: Thu, 11 Oct 2018 01:13:46 +0000
Subject: [theory-seminar] Combinatorics seminar talk next Thursday (Oct. 11)
at 3pm by Persi Diaconis (Stanford) in 384-H
In-Reply-To:
References: ,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
Message-ID:
Hi all,
There is a talk of interest tomorrow at the combinatorics seminar - see details below.
Best,
Ofir
________________________________
From: Jacob Fox
Sent: Wednesday, October 10, 2018 4:45 PM
To: Ofir Geri
Subject: Combinatorics seminar talk next Thursday (Oct. 11) at 3pm by Persi Diaconis (Stanford) in 384-H
Dear Ofir,
I think the combinatorics seminar talk on Thursday by Persi Diaconis will be of interest to many in theoretical computer science. Can you advertise it on the theory seminar mailing list?
Best wishes,
Jacob
When: Thursday, October 11, 3-4pm
Room: 384-H
Speaker: Persi Diaconis (Stanford)
Title: PERMANENTAL GENERATING FUNCTIONS AND SEQUENTIAL IMPORTANCE SAMPLING
Abstract: Counting permutations with restricted positions is a classical activity (menage problem 1708). Many approaches will be reviewed (evaluating permanents,Markov chain Monte Carlo, ...). A new 'sequential importance sampling algorithm' is proposed. This seems to work well and I have been able to prove some things about it using Bregman's inequality (solution of Minc conjecture). Working it out in special cases (eg, count permutations that move i at most k places) needs delicate analysis of permanents with variables within used to keep track of things). these yield elegant generating functions. this is joint work with Fan Chung, Brett Kolasnick, Ron Graham and Andy Tsao.
The seminar webpage is:
http://mathematics.stanford.edu/combinatorics-seminar/
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
From ofirgeri at stanford.edu Thu Oct 11 10:20:52 2018
From: ofirgeri at stanford.edu (Ofir Geri)
Date: Thu, 11 Oct 2018 17:20:52 +0000
Subject: [theory-seminar] Theory Seminar (10/11): Juba Ziani
In-Reply-To:
References:
Message-ID:
Reminder: today at 4:15pm, Juba Ziani is giving a talk at theory seminar.
________________________________
From: Ofir Geri
Sent: Monday, October 8, 2018 4:25:38 PM
To: thseminar at cs.stanford.edu
Subject: Theory Seminar (10/11): Juba Ziani
Hi all,
On Thursday (10/11) Juba Ziani from Caltech will give a theory seminar talk on Optimal Data Acquisition for Statistical Estimation (abstract below).
The talk will be at 4:15 PM in Gates 463A. Please notice the non-standard time.
Hope to see you there!
Ofir
Optimal Data Acquisition for Statistical Estimation
Speaker: Juba Ziani (Caltech)
We consider a data analyst's problem of purchasing data from strategic agents to compute an unbiased estimate of a statistic of interest. Agents incur private costs to reveal their data and the costs can be arbitrarily correlated with their data. Once revealed, data are verifiable. This paper focuses on linear unbiased estimators. We design an individually rational and incentive compatible mechanism that optimizes the worst-case mean-squared error of the estimation, where the worst-case is over the unknown correlation between costs and data, subject to a budget constraint in expectation. We characterize the form of the optimal mechanism in closed-form. We further extend our results to acquiring data for estimating a parameter in regression analysis, where private costs can correlate with the values of the dependent variable but not with the values of the independent variables.
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
From ccanonne at cs.stanford.edu Thu Oct 11 13:25:25 2018
From: ccanonne at cs.stanford.edu (=?UTF-8?Q?Cl=c3=a9ment_Canonne?=)
Date: Thu, 11 Oct 2018 13:25:25 -0700
Subject: [theory-seminar] TCS+ talk: Wednesday, October 17, C. Seshadhri,
UC Santa Cruz
Message-ID: <059c2418-855f-ee6c-2378-f44c529ee865@cs.stanford.edu>
Hi everyone,
Next Wednesday, we'll have a TCS+ talk in Gates 463A: we'll be watching
Sesh (UCSC) talk about a recent result on efficiently finding forbidden
minors through random walks in graphs (title and abstract below).
it's as usual at 9:55am: come for the breakfast, stay for the
K_{3,3}-minor finding!
Best,
-- Cl?ment
-------------------------------
Speaker: C. Seshadhri (UC Santa Cruz)
Title: Finding forbidden minors through random walks: (almost) n^{1/2}
query one-sided testers for minor closed properties
Abstract: Let G be an undirected, bounded degree graph with n vertices.
Fix a finite graph H, and suppose
one must remove eps*n edges from G to make it H-minor free (for some
small constant eps > 0). We give a nearly n^{1/2} time algorithm that,
with high probability, finds an H-minor in such a graph.
As an application, consider a graph G that requires eps*n edge removals
to make it planar. This result implies an algorithm, with the same
running time, that produces a K_{3,3} or K_5 minor in G. No prior
sublinear time bound was known for this problem. By the graph minor
theorem, we get an analogous result for any minor-closed property.
Up to n^{o(1)} factors, this result resolves a conjecture of
Benjamini-Schramm-Shapira (STOC 2008) on the existence of one-sided
property testers for minor-closed properties. Furthermore, our algorithm
is nearly optimal, by lower bounds of Czumaj et al (RSA 2014).
Joint work with Akash Kumar and Andrew Stolman
From ccanonne at cs.stanford.edu Sun Oct 14 17:02:16 2018
From: ccanonne at cs.stanford.edu (=?UTF-8?Q?Cl=c3=a9ment_Canonne?=)
Date: Sun, 14 Oct 2018 17:02:16 -0700
Subject: [theory-seminar] Theory Happy Hour: S02E01 ("The Meadterms")
Message-ID: <3399679e-c887-d4b4-e72d-931d1a40e0ad@cs.stanford.edu>
Hi everybody,
Since the semester has by now long-resumed, and the palm trees around
are looking at each other pretty confused about this whole "Fall" thing,
it's more than time to have our first *happy hour* of the quarter.
It'll be on *Wednesday, October 17th* at *17h30* (5:30pm for the
12h-clock people), in Gates 463A. Migration to the AT&T patio in case of
nice weather is a clear possibility, so stay tuned.
For those joining from afar (or a-close), the concept is pretty simple:
we bring food/snacks, drinks (wine, cold beer, and the occasional soda),
and possibly-good company. You bring you.
See you on Wednesday!
PS: if you have preferences or restrictions on the food/drinks, please
send me an email and I'll try to accommodate them.
-- Cl?ment
From ofirgeri at stanford.edu Mon Oct 15 11:11:16 2018
From: ofirgeri at stanford.edu (Ofir Geri)
Date: Mon, 15 Oct 2018 18:11:16 +0000
Subject: [theory-seminar] Theory/ML Seminar (TOMORROW 10/16): Elad Hazan
Message-ID:
Hi all,
Tomorrow (Tuesday 10/16) we will have a joint theory/machine learning seminar, where Elad Hazan (Princeton) will be speaking on Taking Control by Convex Optimization (abstract below). The talk will be at 3:00 PM in Gates 463A. There will be no seminar on Friday.
Hope to see you there!
Ofir
Taking Control by Convex Optimization
Speaker: Elad Hazan (Princeton)
Linear dynamical systems, a.k.a. Kalman filtering, are a class of time-series models widely used in robotics, finance, engineering, and meteorology. In it's general form (unknown system), learning LDS is a classic non-convex problem, typically tackled with heuristics like gradient descent ("backpropagation through time") or the EM algorithm.
I will present our new "spectral filtering" approach to the identification and control of discrete-time general linear dynamical systems with multi-dimensional inputs, outputs, and a latent state. This approach yields a simple and efficient algorithms for low-regret prediction (i.e. asymptotically vanishing MSE) as well as finite-time control.
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
From ofirgeri at stanford.edu Tue Oct 16 08:32:10 2018
From: ofirgeri at stanford.edu (Ofir Geri)
Date: Tue, 16 Oct 2018 15:32:10 +0000
Subject: [theory-seminar] Theory/ML Seminar (TOMORROW 10/16): Elad Hazan
In-Reply-To:
References:
Message-ID:
Reminder: today at 3pm, Elad Hazan is giving a talk on Taking Control by Convex Optimization.
________________________________
From: Ofir Geri
Sent: Monday, October 15, 2018 11:11:16 AM
To: thseminar at cs.stanford.edu
Subject: Theory/ML Seminar (TOMORROW 10/16): Elad Hazan
Hi all,
Tomorrow (Tuesday 10/16) we will have a joint theory/machine learning seminar, where Elad Hazan (Princeton) will be speaking on Taking Control by Convex Optimization (abstract below). The talk will be at 3:00 PM in Gates 463A. There will be no seminar on Friday.
Hope to see you there!
Ofir
Taking Control by Convex Optimization
Speaker: Elad Hazan (Princeton)
Linear dynamical systems, a.k.a. Kalman filtering, are a class of time-series models widely used in robotics, finance, engineering, and meteorology. In it's general form (unknown system), learning LDS is a classic non-convex problem, typically tackled with heuristics like gradient descent ("backpropagation through time") or the EM algorithm.
I will present our new "spectral filtering" approach to the identification and control of discrete-time general linear dynamical systems with multi-dimensional inputs, outputs, and a latent state. This approach yields a simple and efficient algorithms for low-regret prediction (i.e. asymptotically vanishing MSE) as well as finite-time control.
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
From ccanonne at cs.stanford.edu Tue Oct 16 11:18:54 2018
From: ccanonne at cs.stanford.edu (=?UTF-8?Q?Cl=c3=a9ment_Canonne?=)
Date: Tue, 16 Oct 2018 11:18:54 -0700
Subject: [theory-seminar] TCS+ talk: Wednesday, October 17, C. Seshadhri,
UC Santa Cruz
In-Reply-To: <059c2418-855f-ee6c-2378-f44c529ee865@cs.stanford.edu>
References: <059c2418-855f-ee6c-2378-f44c529ee865@cs.stanford.edu>
Message-ID: <8bbca2d3-86b7-a728-605c-18bfbe468b97@cs.stanford.edu>
Reminder: this is *tomorrow morning*. With pastries, and resolution of
an eight-year old conjecture.
-- Cl?ment
On 10/11/18 1:25 PM, Cl?ment Canonne wrote:
> Hi everyone,
>
> Next Wednesday, we'll have a TCS+ talk in Gates 463A: we'll be watching
> Sesh (UCSC) talk about a recent result on efficiently finding forbidden
> minors through random walks in graphs (title and abstract below).
>
> it's as usual at 9:55am: come for the breakfast, stay for the
> K_{3,3}-minor finding!
>
> Best,
>
> -- Cl?ment
>
> -------------------------------
> Speaker: C. Seshadhri (UC Santa Cruz)
> Title: Finding forbidden minors through random walks: (almost) n^{1/2}
> query one-sided testers for minor closed properties
>
> Abstract: Let G be an undirected, bounded degree graph with n vertices.
> Fix a finite graph H, and suppose
> one must remove eps*n edges from G to make it H-minor free (for some
> small constant eps > 0). We give a nearly n^{1/2} time algorithm that,
> with high probability, finds an H-minor in such a graph.
>
> As an application, consider a graph G that requires eps*n edge removals
> to make it planar. This result implies an algorithm, with the same
> running time, that produces a K_{3,3} or K_5 minor in G. No prior
> sublinear time bound was known for this problem. By the graph minor
> theorem, we get an analogous result for any minor-closed property.
>
> Up to n^{o(1)} factors, this result resolves a conjecture of
> Benjamini-Schramm-Shapira (STOC 2008) on the existence of one-sided
> property testers for minor-closed properties. Furthermore, our algorithm
> is nearly optimal, by lower bounds of Czumaj et al (RSA 2014).
>
> Joint work with Akash Kumar and Andrew Stolman
>
> _______________________________________________
> theory-seminar mailing list
> theory-seminar at lists.stanford.edu
> https://mailman.stanford.edu/mailman/listinfo/theory-seminar
From gvaliant at cs.stanford.edu Tue Oct 16 12:58:42 2018
From: gvaliant at cs.stanford.edu (Gregory Valiant)
Date: Tue, 16 Oct 2018 12:58:42 -0700
Subject: [theory-seminar] Elad Hazan talk today @3pm
Message-ID:
Hi Friends,
Elad Hazan will be visiting from Princeton, and will be giving a talk 3-4pm
in Gates 463a. Hope to see you there,
-Greg
Time/Location: Tues Oct 16, 3-4pm Gates 463a.
Speaker: Elad Hazan (Princeton University)
Title: Taking Control by Convex Optimization
Abstract:
Linear dynamical systems, a.k.a. Kalman filtering, are a class of
time-series models widely used in robotics, finance, engineering, and
meteorology. In it's general form (unknown system), learning LDS is a
classic non-convex problem, typically tackled with heuristics like gradient
descent ("backpropagation through time") or the EM algorithm.
I will present our new "spectral filtering" approach to the identification
and control of discrete-time general linear dynamical systems with
multi-dimensional inputs, outputs, and a latent state. This approach yields
a simple & efficient algorithms for low-regret prediction (i.e.
asymptotically vanishing MSE) as well as finite-time control.
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
From rmhulett at stanford.edu Tue Oct 16 14:27:53 2018
From: rmhulett at stanford.edu (Reyna Marie Hulett)
Date: Tue, 16 Oct 2018 21:27:53 +0000
Subject: [theory-seminar] Theory Lunch -- Bruce Spang
Message-ID:
Hi all,
This week in theory lunch, Bruce will talk to us about "Unconstraining Graph-Constrained Group Testing." As always, please join us from noon to 1 pm in Gates 463A!
===========================================================
Abstract:
In network tomography, one goal is to find a small set of failed links in a network by sending a few packets through the network and seeing which reach their destination. This can be seen as a variant of combinatorial group testing, which has been studied before under the moniker "graph-constrained group testing."
I will discuss some of our recent work on this problem, in which we show that for many graphs, the ?constraints? imposed by the underlying graph are no constraint at all. That is, for many graphs it is possible to identify the failed links in the ?graph-constrained? group testing problem with a number of tests which is near-optimal even for the group testing problem with no graph constraints.
Based on joint work with Mary Wootters.
===========================================================
Cheers,
-Reyna
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
From rmhulett at stanford.edu Tue Oct 16 14:27:53 2018
From: rmhulett at stanford.edu (Reyna Marie Hulett)
Date: Tue, 16 Oct 2018 21:27:53 +0000
Subject: [theory-seminar] Theory Lunch -- Bruce Spang
Message-ID:
Hi all,
This week in theory lunch, Bruce will talk to us about "Unconstraining Graph-Constrained Group Testing." As always, please join us from noon to 1 pm in Gates 463A!
===========================================================
Abstract:
In network tomography, one goal is to find a small set of failed links in a network by sending a few packets through the network and seeing which reach their destination. This can be seen as a variant of combinatorial group testing, which has been studied before under the moniker "graph-constrained group testing."
I will discuss some of our recent work on this problem, in which we show that for many graphs, the ?constraints? imposed by the underlying graph are no constraint at all. That is, for many graphs it is possible to identify the failed links in the ?graph-constrained? group testing problem with a number of tests which is near-optimal even for the group testing problem with no graph constraints.
Based on joint work with Mary Wootters.
===========================================================
Cheers,
-Reyna
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
From rmhulett at stanford.edu Tue Oct 16 15:30:13 2018
From: rmhulett at stanford.edu (Reyna Marie Hulett)
Date: Tue, 16 Oct 2018 22:30:13 +0000
Subject: [theory-seminar] Theory Lunch -- Bruce Spang
In-Reply-To:
References:
Message-ID:
P.S. Theory lunch is on Thursday, oops.
________________________________
From: theory-seminar on behalf of Reyna Marie Hulett
Sent: Tuesday, October 16, 2018 2:27:53 PM
To: theory-seminar at lists.stanford.edu; thseminar at cs.stanford.edu
Subject: [theory-seminar] Theory Lunch -- Bruce Spang
Hi all,
This week in theory lunch, Bruce will talk to us about "Unconstraining Graph-Constrained Group Testing." As always, please join us from noon to 1 pm in Gates 463A!
===========================================================
Abstract:
In network tomography, one goal is to find a small set of failed links in a network by sending a few packets through the network and seeing which reach their destination. This can be seen as a variant of combinatorial group testing, which has been studied before under the moniker "graph-constrained group testing."
I will discuss some of our recent work on this problem, in which we show that for many graphs, the ?constraints? imposed by the underlying graph are no constraint at all. That is, for many graphs it is possible to identify the failed links in the ?graph-constrained? group testing problem with a number of tests which is near-optimal even for the group testing problem with no graph constraints.
Based on joint work with Mary Wootters.
===========================================================
Cheers,
-Reyna
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
From rmhulett at stanford.edu Tue Oct 16 15:30:13 2018
From: rmhulett at stanford.edu (Reyna Marie Hulett)
Date: Tue, 16 Oct 2018 22:30:13 +0000
Subject: [theory-seminar] Theory Lunch -- Bruce Spang
In-Reply-To:
References:
Message-ID:
P.S. Theory lunch is on Thursday, oops.
________________________________
From: theory-seminar on behalf of Reyna Marie Hulett
Sent: Tuesday, October 16, 2018 2:27:53 PM
To: theory-seminar at lists.stanford.edu; thseminar at cs.stanford.edu
Subject: [theory-seminar] Theory Lunch -- Bruce Spang
Hi all,
This week in theory lunch, Bruce will talk to us about "Unconstraining Graph-Constrained Group Testing." As always, please join us from noon to 1 pm in Gates 463A!
===========================================================
Abstract:
In network tomography, one goal is to find a small set of failed links in a network by sending a few packets through the network and seeing which reach their destination. This can be seen as a variant of combinatorial group testing, which has been studied before under the moniker "graph-constrained group testing."
I will discuss some of our recent work on this problem, in which we show that for many graphs, the ?constraints? imposed by the underlying graph are no constraint at all. That is, for many graphs it is possible to identify the failed links in the ?graph-constrained? group testing problem with a number of tests which is near-optimal even for the group testing problem with no graph constraints.
Based on joint work with Mary Wootters.
===========================================================
Cheers,
-Reyna
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
From ccanonne at cs.stanford.edu Tue Oct 16 16:19:58 2018
From: ccanonne at cs.stanford.edu (=?UTF-8?Q?Cl=c3=a9ment_Canonne?=)
Date: Tue, 16 Oct 2018 16:19:58 -0700
Subject: [theory-seminar] Theory Happy Hour: S02E01 ("The Meadterms")
In-Reply-To: <3399679e-c887-d4b4-e72d-931d1a40e0ad@cs.stanford.edu>
References: <3399679e-c887-d4b4-e72d-931d1a40e0ad@cs.stanford.edu>
Message-ID: <92561345-fc0e-d0a6-898f-aecbd1ea2f4e@cs.stanford.edu>
Reminder: this is tomorrow!
-- Cl?ment
On 10/14/18 5:02 PM, Cl?ment Canonne wrote:
> Hi everybody,
>
> Since the semester has by now long-resumed, and the palm trees around
> are looking at each other pretty confused about this whole "Fall" thing,
> it's more than time to have our first *happy hour* of the quarter.
>
> It'll be on *Wednesday, October 17th* at *17h30* (5:30pm for the
> 12h-clock people), in Gates 463A. Migration to the AT&T patio in case of
> nice weather is a clear possibility, so stay tuned.
>
> For those joining from afar (or a-close), the concept is pretty simple:
> we bring food/snacks, drinks (wine, cold beer, and the occasional soda),
> and possibly-good company. You bring you.
>
> See you on Wednesday!
>
> PS: if you have preferences or restrictions on the food/drinks, please
> send me an email and I'll try to accommodate them.
> -- Cl?ment
>
> _______________________________________________
> theory-seminar mailing list
> theory-seminar at lists.stanford.edu
> https://mailman.stanford.edu/mailman/listinfo/theory-seminar
From ccanonne at cs.stanford.edu Wed Oct 17 16:23:41 2018
From: ccanonne at cs.stanford.edu (=?UTF-8?Q?Cl=c3=a9ment_Canonne?=)
Date: Wed, 17 Oct 2018 16:23:41 -0700
Subject: [theory-seminar] Theory Happy Hour: S02E01 ("The Meadterms")
In-Reply-To: <92561345-fc0e-d0a6-898f-aecbd1ea2f4e@cs.stanford.edu>
References: <3399679e-c887-d4b4-e72d-931d1a40e0ad@cs.stanford.edu>
<92561345-fc0e-d0a6-898f-aecbd1ea2f4e@cs.stanford.edu>
Message-ID: <3b5fa102-ca1a-9c51-886e-f6d88bc80afd@cs.stanford.edu>
Update: this will take place on the AT&T patio, because apparently rain
is really not a thing around here and it is wonderful.
Also, if you're only coming after 5:50 or so because of a class -- fear
not! There will still be happihourness aplenty.
Best,
-- Cl?ment
On 10/16/18 4:19 PM, Cl?ment Canonne wrote:
> Reminder: this is tomorrow!
>
> -- Cl?ment
>
> On 10/14/18 5:02 PM, Cl?ment Canonne wrote:
>> Hi everybody,
>>
>> Since the semester has by now long-resumed, and the palm trees around
>> are looking at each other pretty confused about this whole "Fall"
>> thing, it's more than time to have our first *happy hour* of the quarter.
>>
>> It'll be on *Wednesday, October 17th* at *17h30* (5:30pm for the
>> 12h-clock people), in Gates 463A. Migration to the AT&T patio in case
>> of nice weather is a clear possibility, so stay tuned.
>>
>> For those joining from afar (or a-close), the concept is pretty
>> simple: we bring food/snacks, drinks (wine, cold beer, and the
>> occasional soda), and possibly-good company. You bring you.
>>
>> See you on Wednesday!
>>
>> PS: if you have preferences or restrictions on the food/drinks, please
>> send me an email and I'll try to accommodate them.
>> -- Cl?ment
>>
>> _______________________________________________
>> theory-seminar mailing list
>> theory-seminar at lists.stanford.edu
>> https://mailman.stanford.edu/mailman/listinfo/theory-seminar
>
> _______________________________________________
> theory-seminar mailing list
> theory-seminar at lists.stanford.edu
> https://mailman.stanford.edu/mailman/listinfo/theory-seminar
From ofirgeri at stanford.edu Thu Oct 18 20:12:01 2018
From: ofirgeri at stanford.edu (Ofir Geri)
Date: Fri, 19 Oct 2018 03:12:01 +0000
Subject: [theory-seminar] Bonus Theory Seminar on Tuesday (10/23): Srikanth
Srinivasan
Message-ID:
Hi all,
On Tuesday (10/23) we will have a bonus theory seminar, where Srikanth Srinivasan (IIT Bombay) will be speaking on constant-depth circuits for the coin problem (abstract below). The talk will be at 4:15 PM in Gates 463A. Please note the non-standard day and time. We will also have theory seminar as usual next Friday (but not tomorrow).
Hope to see you there!
Ofir
On Constant-Depth Circuits for the Coin Problem
Speaker: Srikanth Srinivasan (IIT Bombay)
We come up with explicit functions for which we can prove tight (up to polynomial factors) upper and lower bounds in the AC^0[2] circuit model. In particular, this implies the first Fixed-Depth Size Hierarchy theorem for this model.
The explicit functions are obtained by constructing explicit AC^0[2] circuits for solving the coin problem, which is defined as follows. For a parameter delta, we have to decide whether a given coin has bias (1+delta)/2 or (1-delta)/2.
Our upper bounds are proved by derandomizing a circuit construction of O'Donnell-Wimmer (2007) and Amano (2009) to reduce the number of samples. Our lower bounds follow from a modification of the Razborov-Smolensky polynomial method.
Joint work with Nutan Limaye (IITB CSE), Karteek Sreenivasiah (IITH CSE), Utkarsh Tripathi (IITB Math) and S Venkitesh (IITB Math).
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
From saba at cs.stanford.edu Sun Oct 21 18:08:16 2018
From: saba at cs.stanford.edu (Saba Eskandarian)
Date: Sun, 21 Oct 2018 18:08:16 -0700
Subject: [theory-seminar] Fwd: Friday,
October 26 -- Dana Dachman-Soled: Non-Malleable Codes from
Average-Case Hardness
In-Reply-To:
References:
Message-ID:
This week's security seminar talk will likely be of interest to the theory
group as well.
Thanks,
~saba
---------- Forwarded message ---------
From: Saba Eskandarian
Date: Sun, Oct 21, 2018 at 6:05 PM
Subject: Friday, October 26 -- Dana Dachman-Soled: Non-Malleable Codes from
Average-Case Hardness
To:
Non-Malleable Codes from Average-Case Hardness
Dana Dachman-Soled
Friday, October 26, 2018
Talk at 11:00am
Gates 392
Abstract:
We show a general framework for constructing non-malleable codes
against tampering families with average-case hardness bounds. Our
framework adapts ideas from the Naor-Yung double encryption paradigm
such that to protect against tampering in a class F, it suffices to
have average-case hard distributions for the class, and underlying
primitives satisfying certain properties with respect to the class. We
then present instantiations of the framework to achieve non-malleable
codes for various classes F.
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
From ofirgeri at stanford.edu Mon Oct 22 10:24:02 2018
From: ofirgeri at stanford.edu (Ofir Geri)
Date: Mon, 22 Oct 2018 17:24:02 +0000
Subject: [theory-seminar] TODAY Probability Seminar with Moses Charikar:
"Efficient Profile Maximum Likelihood for Universal Symmetric Property
Estimation"
In-Reply-To:
References: <64229855-29cb-a896-1306-d5c345d882ba@stanford.edu>
,
Message-ID:
Hi all,
Moses is giving a talk today at 4pm in the probability seminar -- see details below.
Best,
Ofir
---------- Forwarded message ---------
From: Cindy R Kirby >
Date: Mon, Oct 22, 2018 at 9:30 AM
Subject: [probability] Seminars: Week of October 22, 2018
To: Statistics Seminars >, Probability Seminars >
Monday, October 22, 4pm: Probability Seminar, Sequoia Hall Room 200
Moses Charikar, Stanford Computer Science
"Efficient profile maximum likelihood for universal symmetric property estimation"
Abstracts from: https://statistics.stanford.edu/events/probability-seminar
Tuesday, October 23, 4:30pm: Statistics Seminar, Sloan Mathematics Room 380C
Art Owen, Stanford Statistics
"Optimizing the tie-breaker regression discontinuity design"
Abstracts from: https://statistics.stanford.edu/events/statistics-seminar
Thursday, October 25, 1:30pm: Workshop in Biostatistics, Medical School Office Building Room x303
Neil Risch, UC San Francisco
"Genetic Epidemiology Research Based in Electronic Health Records"
Abstracts from: http://med.stanford.edu/biostatistics/workshop
If you have questions about this email, please reply to the sender.
Inquiries about the Seminars should be sent to seminars at stat.stanford.edu.
--
Cindy R Kirby Administrative Associate Department of Statistics
Sequoia Hall, 390 Serra Mall Room 218 Stanford, CA 94305-4065
phone 650 725 2229 fax 650 725 8977 statistics.stanford.edu
_______________________________________________
probability-seminar mailing list
probability-seminar at lists.stanford.edu
https://mailman.stanford.edu/mailman/listinfo/probability-seminar
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
From gvaliant at cs.stanford.edu Mon Oct 22 11:44:29 2018
From: gvaliant at cs.stanford.edu (Gregory Valiant)
Date: Mon, 22 Oct 2018 11:44:29 -0700
Subject: [theory-seminar] Fwd: Michael O. Rabin Postdoctoral Fellowship in
Theoretical Computer Science
In-Reply-To:
References:
Message-ID:
Hi Friends,
For those of you graduating, below is a solicitation for postdoc
applications for the Rabin postdoctoral fellowship at Harvard. Aviad was
one of the first Rabin postdocs, so I'm sure you could get the inside scoop
from him : )
Cheers,
-g
----------------------
*Michael O. Rabin Postdoctoral Fellowship in Theoretical Computer Science*
Deadline for full consideration: *December 3, 2018*. Applications can be
submitted here::
https://academicpositions.harvard.edu/postings/8512
The Harvard John A. Paulson School of Engineering and Applied Sciences at
Harvard University seeks applicants for the Michael O. Rabin Postdoctoral
Fellowship in Theoretical Computer Science. The standard duration of the
Rabin Fellowship is two years. Rabin Fellows will receive a generous salary
as well as an annual allocation for research and travel expenses.
Past fellows are Mika Goose and Aviad Rubinstein and the current fellow is
Alexander Golovnev.
We are looking for exceptional junior scientists in theoretical computer
science, broadly construed. Rabin Fellows will be provided with the
opportunity to pursue their research agenda in an intellectually vibrant
environment with ample mentorship. While interaction with Harvard faculty,
students, and visitors is encouraged, Rabin Fellows are free to pursue
their own interests. Candidates are required to have a doctorate or
terminal degree in Computer Science or a related area by the expected start
date.
Required application documents include a cover letter, research statement,
CV (including a list of publications), and names and contact information
for three references. We recommend that papers mentioned in the CV be
available online on your homepage or on electronic archives. Applicants
will apply on-line at the above address. We encourage candidates to apply
by December 3, 2018, but applications will be accepted until the position
is filled.
Harvard is an equal opportunity employer and all qualified applicants will
receive consideration for employment without regard to race, color,
religion, sex, sexual orientation, gender identity, national origin,
disability status, protected veteran status, or any other characteristic
protected by law.
*The fellowship is named after Michael O. Rabin, pioneer in Computer
Science research and winner of numerous awards including the A. M. Turing
award in 1976. Michael Rabin has been on the faculty at Harvard since 1981,
and currently is the Thomas J. Watson, Sr. Research Professor of Computer
Science in the Harvard Paulson School. The fellowship is aimed at
researchers in all areas of theoretical computer science, including fellows
that, like Rabin, might create new areas that do not yet exist.
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
From ofirgeri at stanford.edu Tue Oct 23 13:15:06 2018
From: ofirgeri at stanford.edu (Ofir Geri)
Date: Tue, 23 Oct 2018 20:15:06 +0000
Subject: [theory-seminar] Bonus Theory Seminar on Tuesday (10/23):
Srikanth Srinivasan
In-Reply-To:
References:
Message-ID:
Reminder: Srikanth Srinivasan will be speaking at theory seminar today at 4:15pm.
________________________________
From: theory-seminar on behalf of Ofir Geri
Sent: Thursday, October 18, 2018 8:12:01 PM
To: thseminar at cs.stanford.edu
Subject: [theory-seminar] Bonus Theory Seminar on Tuesday (10/23): Srikanth Srinivasan
Hi all,
On Tuesday (10/23) we will have a bonus theory seminar, where Srikanth Srinivasan (IIT Bombay) will be speaking on constant-depth circuits for the coin problem (abstract below). The talk will be at 4:15 PM in Gates 463A. Please note the non-standard day and time. We will also have theory seminar as usual next Friday (but not tomorrow).
Hope to see you there!
Ofir
On Constant-Depth Circuits for the Coin Problem
Speaker: Srikanth Srinivasan (IIT Bombay)
We come up with explicit functions for which we can prove tight (up to polynomial factors) upper and lower bounds in the AC^0[2] circuit model. In particular, this implies the first Fixed-Depth Size Hierarchy theorem for this model.
The explicit functions are obtained by constructing explicit AC^0[2] circuits for solving the coin problem, which is defined as follows. For a parameter delta, we have to decide whether a given coin has bias (1+delta)/2 or (1-delta)/2.
Our upper bounds are proved by derandomizing a circuit construction of O'Donnell-Wimmer (2007) and Amano (2009) to reduce the number of samples. Our lower bounds follow from a modification of the Razborov-Smolensky polynomial method.
Joint work with Nutan Limaye (IITB CSE), Karteek Sreenivasiah (IITH CSE), Utkarsh Tripathi (IITB Math) and S Venkitesh (IITB Math).
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
From ofirgeri at stanford.edu Tue Oct 23 13:23:40 2018
From: ofirgeri at stanford.edu (Ofir Geri)
Date: Tue, 23 Oct 2018 20:23:40 +0000
Subject: [theory-seminar] Theory Seminar (10/26): Erik Waingarten
Message-ID:
Hi all,
This Friday at theory seminar, Erik Waingarten from Columbia will be giving a talk on Approximate Nearest Neighbors via Non-Linear Spectral Gaps (see abstract below). The talk will be as usual on Friday 3pm in Gates 463A.
Please note that we also have a theory seminar today.
Hope to see you there!
Ofir
Approximate Nearest Neighbors via Non-Linear Spectral Gaps
Speaker: Erik Waingarten (Columbia)
I will present recent advances in approximate nearest neighbor search data structures for general normed spaces. I will explain what non-linear spectral gaps are, and how to use estimates on non-linear spectral gaps to partition large graphs whose vertices lie in a normed space. As a result, I will present the first sub-linear time data structure for approximate nearest neighbor search in high-dimensional normed spaces with approximation which is sub-polynomial in the dimension.
Based on joint work with Alex Andoni, Assaf Naor, Sasho Nikolov, and Ilya Razenshteyn.
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
From rmhulett at stanford.edu Tue Oct 23 15:55:11 2018
From: rmhulett at stanford.edu (Reyna Marie Hulett)
Date: Tue, 23 Oct 2018 22:55:11 +0000
Subject: [theory-seminar] Theory Lunch -- Paris Syminelakis
Message-ID:
Hi folks,
This week's theory lunch will be given by Paris Syminelakis, on ?Efficient Density Evaluation for Smooth Kernels?--see abstract below. As always, please join us Thursday from noon to 1 pm in Gates 463A!
----------------------------------------
Abstract:
A fundamental problem in Scientific Computing and Machine Learning is that of Interpolation: "given measurements (y_i, x_i) where y_i is a real number and x_i is a d-dimensional vector for i \in [n], construct a continuous/smooth function f: \R^d \to \R such that f(x_i) = y_i for all i \in [n]". In order to solve such problems starting with work in Cartography in the 60's and adopted widely in Machine Learning in the 90's, scientists have used functions of the form: f(x) = \sum_{i} w_i \phi( \| x - x_i \|) by picking coefficients w_i appropriately (linear system solving/optimization). The functions \phi(r) as known as Radial Basis Functions (RBF) and the method is known as "RBF Interpolation".
In this talk, I will discuss a set of techniques that show that such functions can be approximately evaluated in *poly-logarithmic time in the number of points* but having an *exponential dependence on a smoothness parameter*. For many functions used in the literature, this smoothness parameter is small (constant), giving overall fast algorithms for evaluating such "kernel densities" in high dimensions. For decreasing and smooth kernels, we also show a black-box reduction of this problem to the problem of approximate nearest neighbor search, that might be of independent interest.
This is based on joint work with Arturs Backurs (MIT), Moses Charikar and Piotr Indyk (MIT) presented at FOCS 2018.
----------------------------------------
Cheers,
-Reyna
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
From rmhulett at stanford.edu Tue Oct 23 15:55:11 2018
From: rmhulett at stanford.edu (Reyna Marie Hulett)
Date: Tue, 23 Oct 2018 22:55:11 +0000
Subject: [theory-seminar] Theory Lunch -- Paris Syminelakis
Message-ID:
Hi folks,
This week's theory lunch will be given by Paris Syminelakis, on ?Efficient Density Evaluation for Smooth Kernels?--see abstract below. As always, please join us Thursday from noon to 1 pm in Gates 463A!
----------------------------------------
Abstract:
A fundamental problem in Scientific Computing and Machine Learning is that of Interpolation: "given measurements (y_i, x_i) where y_i is a real number and x_i is a d-dimensional vector for i \in [n], construct a continuous/smooth function f: \R^d \to \R such that f(x_i) = y_i for all i \in [n]". In order to solve such problems starting with work in Cartography in the 60's and adopted widely in Machine Learning in the 90's, scientists have used functions of the form: f(x) = \sum_{i} w_i \phi( \| x - x_i \|) by picking coefficients w_i appropriately (linear system solving/optimization). The functions \phi(r) as known as Radial Basis Functions (RBF) and the method is known as "RBF Interpolation".
In this talk, I will discuss a set of techniques that show that such functions can be approximately evaluated in *poly-logarithmic time in the number of points* but having an *exponential dependence on a smoothness parameter*. For many functions used in the literature, this smoothness parameter is small (constant), giving overall fast algorithms for evaluating such "kernel densities" in high dimensions. For decreasing and smooth kernels, we also show a black-box reduction of this problem to the problem of approximate nearest neighbor search, that might be of independent interest.
This is based on joint work with Arturs Backurs (MIT), Moses Charikar and Piotr Indyk (MIT) presented at FOCS 2018.
----------------------------------------
Cheers,
-Reyna
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
From ccanonne at cs.stanford.edu Wed Oct 24 16:58:38 2018
From: ccanonne at cs.stanford.edu (=?UTF-8?Q?Cl=c3=a9ment_Canonne?=)
Date: Wed, 24 Oct 2018 16:58:38 -0700
Subject: [theory-seminar] Math and Biology-- BaMBA Day Nov.3rd 2018 @
Stanford (w/ FREE breakfast and lunch)
In-Reply-To:
References:
Message-ID:
Hi everyone,
This may be a long shot, but if you or people you know are working on a
mathematical or computational take on biology questions, this may be
relevant to you.
Best,
-- Cl?ment
-------- Forwarded Message --------
Subject: [EVENT] Math and Biology-- BaMBA Day Nov.3rd 2018 @ Stanford
(w/ FREE breakfast and lunch)
Date: Tue, 23 Oct 2018 16:03:09 -0700
From: Massa Shoura
To: postdoc-exchange at lists.stanford.edu
Dear all,
On behalf of the organization committee, This note is to invite you to
participate in BaMBA!
BaMBA 2018 November 3rd, 2018
9:00 AM-6:00 PM
James H. Clark Center
318 Campus Drive, Stanford , CA 94305
BaMBA is a one-day meeting aimed at exploring the role of mathematics in
biology in an informal atmosphere. Going beyond traditional applied
mathematics, the topics include applications of algebraic, topological,
statistical and computational methods. Our goal is to encourage dialogue
between researchers and students from different disciplines in an
atmosphere that promotes the open exchange of ideas and viewpoints. We
expect a day full of enticing discussions!!
Participation in BaMBA is free and open to everyone, but?registration?is
required.?Undergraduates, graduates, and postdocs involved in
mathematical and computational investigations of biological systems are
invited to submit an abstract for a poster presentation.
*BaMBA Day is Nov.3rd 2018 @ Stanford (w/ FREE breakfast and lunch) --
See link for more info regarding the program. *https://bambameeting.org/
Best,
Massa
Massa Shoura, Ph.D.
Arnold and Mabel Beckman Postdoctoral Fellow
Fire Laboratory
Stanford University School of Medicine
300 Pasteur Dr.
Lane Building L302
Department of Pathology
Stanford, CA 94305
-------------- next part --------------
--+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
* postdoc-exchange info and options: https://mailman.stanford.edu/mailman/listinfo/postdoc-exchange
* SURPAS website: http://www.surpas.org
* OPA Website: http://postdocs.stanford.edu
* To unsubscribe: Send e-mail to postdoc-exchange-unsubscribe at lists.stanford.edu (no subject or content required)
All e-mails must contain one of the following tags in the subject line: [ADVICE] [BIO] [CHEM] [CLIPSS] [COMP] [EVENT] [HOUSING] [JOB] [MISC] [PETS] [SALE] [TALK] [VOLUNTEER]
From ofirgeri at stanford.edu Fri Oct 26 11:08:44 2018
From: ofirgeri at stanford.edu (Ofir Geri)
Date: Fri, 26 Oct 2018 18:08:44 +0000
Subject: [theory-seminar] Theory Seminar (10/26): Erik Waingarten
In-Reply-To:
References:
Message-ID:
Reminder: theory seminar is today at 3pm.
________________________________
From: Ofir Geri
Sent: Tuesday, October 23, 2018 1:23:40 PM
To: thseminar at cs.stanford.edu
Subject: Theory Seminar (10/26): Erik Waingarten
Hi all,
This Friday at theory seminar, Erik Waingarten from Columbia will be giving a talk on Approximate Nearest Neighbors via Non-Linear Spectral Gaps (see abstract below). The talk will be as usual on Friday 3pm in Gates 463A.
Please note that we also have a theory seminar today.
Hope to see you there!
Ofir
Approximate Nearest Neighbors via Non-Linear Spectral Gaps
Speaker: Erik Waingarten (Columbia)
I will present recent advances in approximate nearest neighbor search data structures for general normed spaces. I will explain what non-linear spectral gaps are, and how to use estimates on non-linear spectral gaps to partition large graphs whose vertices lie in a normed space. As a result, I will present the first sub-linear time data structure for approximate nearest neighbor search in high-dimensional normed spaces with approximation which is sub-polynomial in the dimension.
Based on joint work with Alex Andoni, Assaf Naor, Sasho Nikolov, and Ilya Razenshteyn.
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
From ccanonne at cs.stanford.edu Fri Oct 26 13:11:21 2018
From: ccanonne at cs.stanford.edu (=?UTF-8?Q?Cl=c3=a9ment_Canonne?=)
Date: Fri, 26 Oct 2018 13:11:21 -0700
Subject: [theory-seminar] STOC Paper swap
Message-ID: <70b5d567-9dcd-c0c6-2bb7-3fa8e3e5a03c@cs.stanford.edu>
Hi everyone,
As briefly evoked at the theory lunch, there is a deadline* coming, and
with it a *paper swap* for those eager to get feedback on their submission.
You do not have to /have/ a submission to participate. You simply have
to be willing to read 10 pages or so (or more if you want) of a draft,
and provide feedback.
Concept:
* if you are interested, sign up here before *Sunday, noon*:
https://docs.google.com/forms/d/e/1FAIpQLScUP9HuCJdft0SNLWmeRFj5_0TvELGhYe7yq4kZnjo_i4QQoA/viewform
* Ill match people for the swap, and send you names by Sunday evening.
* Send the corresponding person(s) your draft(s) by Monday, *noon*
* Come on Tuesday, *6pm*, for giving and receiving feedback over dinner.
* STOC: next Friday, 5:59pm PDT.
Best,
--
-- Cl?ment
From ofirgeri at stanford.edu Fri Oct 26 15:01:02 2018
From: ofirgeri at stanford.edu (Ofir Geri)
Date: Fri, 26 Oct 2018 22:01:02 +0000
Subject: [theory-seminar] Theory Seminar (10/26): Erik Waingarten
In-Reply-To:
References: ,
Message-ID:
Correction: The talk will take place in Gates 104.
________________________________
From: Ofir Geri
Sent: Friday, October 26, 2018 11:08:44 AM
To: thseminar at cs.stanford.edu
Subject: Re: Theory Seminar (10/26): Erik Waingarten
Reminder: theory seminar is today at 3pm.
________________________________
From: Ofir Geri
Sent: Tuesday, October 23, 2018 1:23:40 PM
To: thseminar at cs.stanford.edu
Subject: Theory Seminar (10/26): Erik Waingarten
Hi all,
This Friday at theory seminar, Erik Waingarten from Columbia will be giving a talk on Approximate Nearest Neighbors via Non-Linear Spectral Gaps (see abstract below). The talk will be as usual on Friday 3pm in Gates 463A.
Please note that we also have a theory seminar today.
Hope to see you there!
Ofir
Approximate Nearest Neighbors via Non-Linear Spectral Gaps
Speaker: Erik Waingarten (Columbia)
I will present recent advances in approximate nearest neighbor search data structures for general normed spaces. I will explain what non-linear spectral gaps are, and how to use estimates on non-linear spectral gaps to partition large graphs whose vertices lie in a normed space. As a result, I will present the first sub-linear time data structure for approximate nearest neighbor search in high-dimensional normed spaces with approximation which is sub-polynomial in the dimension.
Based on joint work with Alex Andoni, Assaf Naor, Sasho Nikolov, and Ilya Razenshteyn.
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
From reingold at stanford.edu Sun Oct 28 09:51:47 2018
From: reingold at stanford.edu (Omer Reingold)
Date: Sun, 28 Oct 2018 09:51:47 -0700
Subject: [theory-seminar] Research methods class + playback 2019
Message-ID:
Hi all,
1. I am contemplating a new course about ?research methods for
theoretical CS.? In the social sciences, students are expected to take
quantitative and qualitative research methods courses before starting
research. In CS, we learn on the job. Are there things that can be taught
which would make undergrad and master research more effective? If you have
any thoughts, please email me: what would you wished someone taught you
before you started research? Let me emphasize that the purpose of the
course will not be useful knowledge but rather useful skills.
2. In the winter quarter, I?ll be teaching ?CS 83: Playback Theater
for Research? (
https://omereingold.wordpress.com/cs-83-playback-theater-for-research/) for
the second time. The first time, the students (including PhD students) felt
extremely rewarded. If you think of a student that would gain from such a
class (or if it is a good fit for you), please spread the word.
Thanks,
Omer
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
From ofirgeri at stanford.edu Mon Oct 29 21:40:00 2018
From: ofirgeri at stanford.edu (Ofir Geri)
Date: Tue, 30 Oct 2018 04:40:00 +0000
Subject: [theory-seminar] Two Theory Seminars This Week: Toniann Pitassi on
10/31 and Rotem Oshman on 11/2
Message-ID:
Hi all,
This week we will have two theory seminars. On Wednesday (10/31), Toniann Pitassi (University of Toronto and Institute for Advanced Study) will give a talk on Lifting in Communication Complexity and Beyond (see abstract below). The talk will take place in Gates 463A at 3:00 PM.
Later this week on Friday, Rotem Oshman (Tel-Aviv University) will give a talk on Two Open Problems in Distributed Graph Algorithms. Another announcement will be sent closer to Friday.
The seminar schedule and abstracts are also available at:
http://theory.stanford.edu/seminar/
Hope to see you there!
Ofir
Lifting in Communication Complexity and Beyond
Speaker: Toniann Pitassi (University of Toronto and Institute for Advanced Study)
Hardness escalation or query-to-communication lifting is a method of proving tight upper and lower bounds on the complexity of a composed function or relation by a reduction to the query complexity of the base function. This talk will primarily be a tutorial. We will give many applications of lifting, including new and improved circuit lower bounds, as well as lower bounds in game theory, proof complexity and extended formulation complexity. We will sketch the main ideas in the proofs of lifting for randomized communication complexity, and conclude with new directions. This is joint work with Mika Goos and Thomas Watson.
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
From rmhulett at stanford.edu Tue Oct 30 13:28:17 2018
From: rmhulett at stanford.edu (Reyna Marie Hulett)
Date: Tue, 30 Oct 2018 20:28:17 +0000
Subject: [theory-seminar] Theory Lunch -- Andrew Stolman (UCSC)
Message-ID:
Hi everybody,
This week's theory lunch will be given by Andrew Stolman (UCSC) on ?Finding forbidden minors through random walks: an almost optimal one-sided tester for minor-closed properties?--see abstract below. As always, please join us Thursday from noon to 1 pm in Gates 463A!
----------------------------------------
Abstract:
Let [G] be an undirected, bounded degree graph with vertices. Fix a finite graph [H] , and suppose one must remove edges from [G] to make it [H] -minor-free (for some small constant [\varepsilon > 0] ). We give an n1/2 + o(1)-time randomized algorithm that, with high probability, finds an [H] -minor in such a graph. As an application, suppose one must remove edges from a bounded degree graph [G] to make it planar. This result implies an algorithm, with the same running time, that produces a [K_{3,3}] or [K_5] minor in [G] . No prior sublinear time bound was known for this problem.
By the graph minor theorem, we get an analogous result for any minor-closed property. Up to [n^{o(1)}] factors, this resolves a conjecture of Benjamini-Schramm-Shapira (STOC 2008) on the existence of one-sided property testers for minor-closed properties. Furthermore, our algorithm is nearly optimal, by an [\Omega(\sqrt{n})] lower bound of Czumaj et al (RSA 2014).
This result appeared in FOCS 2018 and the paper can be found here https://arxiv.org/abs/1805.08187.
Joint work with C. Seshadhri of UCSC and Akash Kumar of Purdue University.
----------------------------------------
Cheers,
-Reyna
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
From rmhulett at stanford.edu Tue Oct 30 13:28:17 2018
From: rmhulett at stanford.edu (Reyna Marie Hulett)
Date: Tue, 30 Oct 2018 20:28:17 +0000
Subject: [theory-seminar] Theory Lunch -- Andrew Stolman (UCSC)
Message-ID:
Hi everybody,
This week's theory lunch will be given by Andrew Stolman (UCSC) on ?Finding forbidden minors through random walks: an almost optimal one-sided tester for minor-closed properties?--see abstract below. As always, please join us Thursday from noon to 1 pm in Gates 463A!
----------------------------------------
Abstract:
Let [G] be an undirected, bounded degree graph with vertices. Fix a finite graph [H] , and suppose one must remove edges from [G] to make it [H] -minor-free (for some small constant [\varepsilon > 0] ). We give an n1/2 + o(1)-time randomized algorithm that, with high probability, finds an [H] -minor in such a graph. As an application, suppose one must remove edges from a bounded degree graph [G] to make it planar. This result implies an algorithm, with the same running time, that produces a [K_{3,3}] or [K_5] minor in [G] . No prior sublinear time bound was known for this problem.
By the graph minor theorem, we get an analogous result for any minor-closed property. Up to [n^{o(1)}] factors, this resolves a conjecture of Benjamini-Schramm-Shapira (STOC 2008) on the existence of one-sided property testers for minor-closed properties. Furthermore, our algorithm is nearly optimal, by an [\Omega(\sqrt{n})] lower bound of Czumaj et al (RSA 2014).
This result appeared in FOCS 2018 and the paper can be found here https://arxiv.org/abs/1805.08187.
Joint work with C. Seshadhri of UCSC and Akash Kumar of Purdue University.
----------------------------------------
Cheers,
-Reyna
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
From ofirgeri at stanford.edu Wed Oct 31 11:42:39 2018
From: ofirgeri at stanford.edu (Ofir Geri)
Date: Wed, 31 Oct 2018 18:42:39 +0000
Subject: [theory-seminar] Two Theory Seminars This Week: Toniann Pitassi
on 10/31 and Rotem Oshman on 11/2
In-Reply-To:
References:
Message-ID:
Reminder: Toniann Pitassi's talk is today at 3pm in Gates 463A.
________________________________
From: Ofir Geri
Sent: Monday, October 29, 2018 9:40:00 PM
To: thseminar at cs.stanford.edu
Subject: Two Theory Seminars This Week: Toniann Pitassi on 10/31 and Rotem Oshman on 11/2
Hi all,
This week we will have two theory seminars. On Wednesday (10/31), Toniann Pitassi (University of Toronto and Institute for Advanced Study) will give a talk on Lifting in Communication Complexity and Beyond (see abstract below). The talk will take place in Gates 463A at 3:00 PM.
Later this week on Friday, Rotem Oshman (Tel-Aviv University) will give a talk on Two Open Problems in Distributed Graph Algorithms. Another announcement will be sent closer to Friday.
The seminar schedule and abstracts are also available at:
http://theory.stanford.edu/seminar/
Hope to see you there!
Ofir
Lifting in Communication Complexity and Beyond
Speaker: Toniann Pitassi (University of Toronto and Institute for Advanced Study)
Hardness escalation or query-to-communication lifting is a method of proving tight upper and lower bounds on the complexity of a composed function or relation by a reduction to the query complexity of the base function. This talk will primarily be a tutorial. We will give many applications of lifting, including new and improved circuit lower bounds, as well as lower bounds in game theory, proof complexity and extended formulation complexity. We will sketch the main ideas in the proofs of lifting for randomized communication complexity, and conclude with new directions. This is joint work with Mika Goos and Thomas Watson.
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
From ofirgeri at stanford.edu Wed Oct 31 16:19:35 2018
From: ofirgeri at stanford.edu (Ofir Geri)
Date: Wed, 31 Oct 2018 23:19:35 +0000
Subject: [theory-seminar] Theory Seminar (11/2): Rotem Oshman
Message-ID:
Hi all,
This Friday in theory seminar, Rotem Oshman (Tel-Aviv University) will give a talk on Two Open Problems in Distributed Graph Algorithms (see abstract below). The talk will be at 3pm in Gates 392 - please note the unusual location!
Hope to see you there!
Ofir
Two Open Problems in Distributed Graph Algorithms
Speaker: Rotem Oshman (Tel-Aviv University)
In a distributed graph algorithm, we have a network of computing nodes, where each node initially knows only its own local neighborhood; the nodes communicate over the network edges in order to solve some problem on the network graph. We are interested in algorithms that are fast, but also do not require a lot of communication between the network nodes.
In this talk I will describe recent algorithms and lower bounds for two graph problems: exact maximum bipartite matching, and testing whether the network contains an even-length cycle of a specific length. Both problems do not have matching upper and lower bounds, so their complexity remains open. The talk will not assume any prior knowledge about distributed computing.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: