Harish Chandran
Contact
N 021 North Building
Voice:
(919) 724-8205
Information
Department of Computer Science
Fax:
(919) 660-6519
Duke University
email:
harish@cs.duke.edu
Durham, NC 27708 USA
URL:
www.cs.duke.edu/~harish
Research
Nanoscience, Algorithms, Computer Architecture, Mathematical Modeling and Scientific Simula-
Interests
tions
Education
Duke University, Durham, North Carolina, USA
Doctoral Candidate, Computer Science (expected graduation date: 2012)
Certificate in Nanoscience, Graduate Program in Nanoscience (expected graduation date: 2012)
Anna University, Chennai, Tamil Nadu, India
Bachelor of Engineering, Computer Science and Engineering (May 2007)
Honors and
Nanoscience Fellowship, Duke University (Aug 2007 - May 2008)
Awards
Graduated First Class with Distinction, Anna University (May 2007)
Research
Waran Research Foundation, Chennai, Tamil Nadu India
Experience
Research Trainee (July 2005 - July 2007)
Microprocessor Design Automation, Brain Modeling, High Performance Computing, Benchmarking, VLSI Routing
Publications
Microprocessor Design Automation: A DNA Based Evolutionary Approach, BICS 06
DNA Based Evolutionary Approach for Microprocessor Design Automation, ICANNGA 07
Projects
Computer Architecture - Dynamic Heterogeneous Core Scheduling
We investigated the benefits of a multicore processor design with cores of varied pipeline depth and frequency.
Through intelligent speculative context swapping between cores, concurrent threads could be optimally allocated
among the various heterogeneous cores based on their respective likelihood of branch misprediction. To test this,
we considered a dynamic analysis method by which we heuristically determined when such swaps should occur.
In order to evaluate our design, we compared our simulation results with those achieved by both an equivalent
homogeneous multi-core processor and an idealized heterogeneous design using an oracle to make optimal swapping
decisions.
Linear Programming - Geometric Duality and Linear Programming
We used geometric duality to convert a set of constraints C to a point set S in the dual space and then apply the
idea of approximate point set of S. In particular we studied the following problem: Let for any set of constraints
C and an objective function obj, L(C, obj) represent the the maximal value of obj constrained by C. Then, for
given set of constraints C with a non null feasible region, a parameter ǫ ∈ (0, 1) and any objective function obj is
there a sparse set of constraints C′ such that L(C, obj) ≤ L(C′, obj) ≤ (1 + ǫ)L(C, obj).
Nanoscience - Biomolecular Detection using DNA Based Nanoparticle Arrays
Disease diagnosis often requires multiple time-consuming, labor-intensive methods. A quick and simple detection
system for multiplexed analysis, using self-assembly of nanoparticle functionalized DNA tiles, was proposed. Two
potential approaches to biomolecular detection were examined: SERS assays and a computational DNA tiling
system. Spectra of the different systems and enhancements of nanoparticle arrays were determined analytically. It
was found that both detection schemes exhibit promise, and further experimental exploration was recommended.
Computational Neuroscience - Pro ject MMINiDASS
The MMINi-DASS aims at predicting the neuronal interconnectivity of various cortices of the human brain. We
start out with a random interconnectivity of a cortex and then refine it iteratively based on actual fMRI imaging.
In each iteration, input stimuli are injected as spike trains into few neurons and the overall activity of the cortex is
recorded based on the hemodynamic of the neurons. This simulated fMRI is then compared with the actual fMRI
obtained for the same input. The difference between the two is reduced by altering the structure using Simulated
Annealing.
Benchmarking - Pro ject BENSIM
BENSIM is an effort to create a synthetic benchmark for use on supercomputing clusters. Key metrics deter-
mining the goodness of a system: power, performance and scalability, are determined when the cluster runs a
synthetic mixture of algorithms created based on inputs from the user from a pre-existing bank of algorithms. A
kernel extracts performance and power figures from the cluster-specific simulator based inputs and also calculates
scalability by varying cluster size for the same benchmark.
Microprocessor Design Automation - DNA Evo Based Design Automation
This project aimed at developing a methodology to automate the design process of a microprocessor by using a
DNA based evolutionary approach. Parameters that are defined a microprocessor are encoded onto DNA sequences
which then undergo recombinations with other sequences along with mutations. These offsprings are then decoded
into microprocessors and evaluated. Over a period of time a Gene-pool is built up from which processors according
to user specifications can be evolved.
VLSI Routing - Channel Routing using PSO
The project aimed at using the methods of Particle Swarm Optimization for solving the channel routing problem
in VLSI. Based on the requirements of the PSO algorithm,a set of parameters were identified, which affect the
routing decisions. These parameters are used in turn, to develop the fitness functions, which are used to gauge
the effectiveness of the solution produced. The project analyzed the results obtained from this algorithm for
routing and provided insights that could result in further improvements with respect to algorithm fine tuning and
parameters used.
Presentations/
Programmable Biomolecular Self-Assembly Pathways
Talks
Biomolecular Computation Journal Club, Duke University, Feb 2008
Professional
Organizing Committee
Services
Secretary, Dhi Yantra 07, Workshop on Brain Modeling and Supercomputing, July 2007
Computer
Languages: C, C++, Java and HTML
Skills
Applications: Photoshop, Dreamweaver, L TEX, Office Suites, Oracle and MySQL
A
Operating Systems: UNIX, Linux and Windows
Work
AIMS Education, Chennai, Tamil Nadu, India
Experience
Content Developer (May 2004 - May 2005)
I created educational content for grades eleven and twelve in forms of quizzes, examinations and other academic
materials, both electronic as well as paper-based. I also organized a city wide science talent test organized by
AIMS Educations which was attended by around sixteen thousand students.
Relevant
Graduate
Coursework
Advanced Algorithms, Advanced Computer Architecture, Mathematical Methods in System Analysis, Randomized
Algorithms, Linear & Integer Programming, Bio Nanotechnology, Foundations of Nanoscience.
