See you in Kyoto!

Map of Kyoto around the Miyako Hotel

Faculty Positions in Engineering and Computer Science

The School of Engineering at the University of New Mexico invites applications for multiple full-time probationary tenure-track faculty appointments at the rank of Assistant Professor to begin August 1, 2012. Appointment at the associate professor level will be considered for individuals with appropriate qualifications. The appointments will reside in the appropriate SOE Department, one of: Civil Engineering, Chemical and Nuclear Engineering, Computer Science, Electrical and Computer Engineering, Mechanical Engineering.
We seek applicants from across the engineering and computer science spectrum with expertise in CyberSecurity, Sensing Technologies and Advanced Materials. The successful applicant should have exceptional research credentials and the ability to contribute to the School’s interdisciplinary approach to a wide variety of national challenges relating to the safety and security of physical, social, and digital infrastructures.
Applicants are required to have a Ph.D. in engineering, computer science or a related field by the start date of the appointment. A demonstrated record of research and scholarly achievements; interest and ability to engage in interdisciplinary research and education; a strong commitment to teaching excellence at both the undergraduate and graduate levels; and an expressed interest in mentoring students with diverse backgrounds, are preferred. We seek individuals with clear potential for developing an internationally acclaimed research program and passion for interdisciplinary collaborations.
The School of Engineering (SOE) has over 100 faculty and plans a significant expansion during the next five years across all its academic departments: Chemical and Nuclear Engineering, Civil Engineering, Computer Science, Electrical and Computer Engineering, and Mechanical Engineering. Among the internationally recognized centers associated with the school are the Center for High Technology Materials, the Center for Microengineered Materials, the Center for Advanced Research Computing, the Mind Research Network, the Center for Biomedical Engineering, and the Center for Emerging Energy Technologies. The SOE academic budget is nearly $13M and annual research and contract funding is $31M. The School has close relations with industry and with three federal laboratories in New Mexico – Sandia National Laboratories, Los Alamos National Laboratory, and the Air Force Research Laboratory Phillips Research Site. The SOE, with enrollments of about 1,100 undergraduates and 600 graduate students, is committed to diversity with several externally funded programs encouraging women and minority students in science and engineering. The SOE is ranked nationally for both its undergraduate and graduate programs, and all undergraduate programs are either ABET or ASCE accredited. For more information about the School of Engineering see http://www.soe.unm.edu
The University of New Mexico (UNM) is the premier research university in the state of New Mexico. UNM is a Carnegie Very High Research Activity Institution and a federally designated Hispanic Serving Institution, with nearly 35,000 students on the main and branch campuses. The University is located in Albuquerque, the largest city in New Mexico. The Albuquerque metropolitan area is ethnically diverse with over 900,000 residents, and has been listed among the smartest U.S. cities and best places to relocate in America. The city has a rich culture and a location offering unparalleled opportunities for outdoor adventure. The University is located within one hour of Santa Fe and within minutes of the Sandia and Manzano mountain ranges, which offer great opportunities for hiking, biking, rock climbing and skiing.

For best consideration, complete applications must be received by February 6, 2012. The positions will remain open until filled. A complete application consists of 1) a letter of interest; 2) a comprehensive CV; 3) a research statement; 4) a teaching statement; and 5) the names and contact information (address, phone number and email address) of four professional references.
Applications must be submitted online through UNMJobs.unm.edu, by referencing posting #0813676. Academic couples are encouraged to apply.
Inquiries should be sent to: SOE-Faculty-Search@cs.unm.edu
The University of New Mexico has a progressive spousal hiring policy and is an equal opportunity/affirmative action employer and educator. We especially encourage members of underrepresented groups to apply.


The last column of the year is dedicated, as always, to a review of distributed computing awards and conferences in 2011. It includes:

* “Sharing Memories, Robustly” by Hagit Attiya
* “Prize for Innovation in Distributed Computing”
* “Review of PODC 2011″ by Maryam Helmi
* “Review of DISC 2011″ by Michael Hakimi and Adam Morrison
* “Review of SIROCCO 2011″ by Adrian Kosowski, Dominik Pajak and Zuzanna Stamirowska

The column is online on the column’s archive, both at the Technion:

http://www.ee.technion.ac.il/~idish/sigactNews

and at MIT:

http://people.csail.mit.edu/idish/sigactNews

You are given a large array in the form:

You want to output an array in the form:

The catch is that you have very little external memory (only O(log n)), so you need to change the array in place.

It’s not too hard to get a solution to this problem that runs in O(n log n) time (think recursion).  But surprisingly, you can do better.  Can you get a linear time solution? Beware: this is harder than it seems!

Below is a picture of the protocol originally proposed by Yao.  Alice and Bob are the two millionaires and i and j are the private net worths of Alice and Bob respectively.  The protocol makes use of public key cryptography: Alice is assumed to be able to generate a public encryption key E_a() along with a private decryption key D_a().  To see that the protocol reveals who is richer, note that y_j = D_a(k) = x.  Thus w_j = x if j<= i and w_j = x+1 if j>i.  Showing that the protocol doesn’t reveal any information beyond who is richer is more challenging and is presented in detail in  the paper.

I should point out that this protocol takes exponential time and that this run time has been improved in subsequent papers.  However, there is a question about this problem I haven’t been able to answer.  In the case where there are n players, do we need private channels or cryptographic assumptions to solve the problem?  Are private channels needed even if we’re happy with a Monte Carlo solution?  I’ve seen several papers that remove cryptographic assumptions, but none that seem to remove the need for private channels.  Conversely, I’ve seen no papers that prove that private channels are necessary…

midterm problem

Recall from last time that we can build a MUX over 2^x signals with O(2^x) noisy gates.  Using a similar technique, for any set of x inputs, we can build an “All Functions” circuit that computes all 2^(2^x) possible functions of these inputs with O(2^(2^x)) gates. Our result today will make use of both 1) an “All Functions” circuit and 2) a MUX.

The technique is illustrated in the figure below.  Assume we’re trying to compute a function f of n inputs.  Let a = log (n-log n) Basically, we send a of the inputs the “All Functions” circuit and we send n-a of the inputs to the MUX.  The MUX selects, based on the first n-a inputs, the exact function of the remaining a inputs that determines the function f over all n inputs.  The MUX selects 1 of 2^(n-a) such “completion functions”.

The beautiful thing here is that both the MUX and the “All Functions” circuit require only O(2^n/n) gates.  Thus, this clever combination of both types of circuits has saved us a factor of n in the total number of noisy gates required.

In previous posts, we showed that n \log n gates are always sufficient and we showed that for the exclusive-or function, n \log n gates are necessary.  This implies that in the worst case, there is a log n blowup in the number of gates required.

There is another result by Pippenger that shows that for *most* boolean functions, only a *constant* blowup is required.  Today I want to focus on the first part of this result, which is to show that a multiplexer (MUX) over 2^r signals can be computed with only O(2^r) noisy gates.  Recall that a MUX over 2^r signals is a circuit that takes as input both 1) 2^r possible signals; and 2) an r bit selector.  The MUX outputs exactly one of the 2^r possible signals, specifically the one that is specified by the r bit selector.

For example, a MUX for r=1 has as input 1) 2 signal bits; and 2) a single selector bit.  If the selector bit is 0, the MUX outputs the first signal bit and if the selector bit is 1, the MUX outputs the second selector bit.  Let G be a gate that computes the MUX for r=1.  Then it’s easy to create a MUX for any r by wiring up O(2^r) copies of G (try it).  In particular, a MUX for r requires at least 2^r noise free gates.

Pippenger shows that you can create a MUX for any r using only O(2^r) noisy gates.  Clearly this is just a constant blowup over the number of noiseless gates needed to create such a MUX.  The construction used to prove this result is sketched in the figure below. See also this pdf: pipp-mux

Note that any boolean function over x inputs can be computed using a MUX, where r=x (the computation is done essentially by table lookup).  This means that we can now compute any boolean function over x inputs with O(2^x) noisy gates.  It turns out that most boolean functions over x variables require 2^x/x noise-free gates.   In my next post, I’ll show how we can shave off this factor of x in the noisy gate world.

In some peer-to-peer applications, this censorship is done on the client side, so there is a subroutine in, e.g. Skype chat, that checks if an outgoing message contains a keyword on the blacklist.  If you are the censor and you don’t care about revealing the blacklist, then there are techniques for doing this in an efficient manner (hint: FSA’s).  However, it’s an interesting (but evil) theoretical problem to think about how to efficiently do keyword censorship if you are also trying to also hide the blacklist.  In particular, if you want to hide it from someone who may be running your executable in a debugger.  Hint: the Chinese Skype program did it incorrectly and that’s why we were able to decrypt their blacklist!