Luigi's Mathpage

This is Luigi Leung's math.cmu page. This page is for:

21-765. Introduction to Parallel Computing and Scientific Computation

---

Project:

Last update: 5/21/2013
Project folder
or use git clone git://github.com/gitLouis/21-765.git ~/TARGET_DIR
and navigate to the "project/" directory

week7 - Design a modular implementation of the software

Design:
Option 1:
Modify exisiting C code in Alfy to become parallel, while having the parallel part to be modular so any new versions of Alfy can easily be converted.
Option 2:
Using tee to parallelize Alfy, and when new versions of Alfy exists, a change in a shell script varible (version number) can make the new version be parallelized.

More information on the exisiting software, Alfy:
Alignment-free detection of local similarity among viral and bacterial genomes. (2011)

week5 - Isolate class of algorithms, identify new research, pick algorithm

Isolate:
MSA:
progressive alignment (e.g. Clustal, MAFFT)
iterative method (e.g. MUSCLE)
Tree construction:
hierarchical clustering method (e.g. neighbor-joining, UPGMA)

New research:
MSA:
ClustalOmega is a parallel version of the Clustal alignment algorithm. (2010)
Tree construction:
Phylogenetic tree construction using stochastic optimization and clustering. (2006)

Pick:
MSA:
progressive alignment
Tree construction:
hierarchical clustering method

week3 - Clearly establish the problem

Problem:
In biology, running genetic sequence alignments and building phylogenetic trees take a long time.
For example, investigating variations of a particular gene (Pyl) between closely related organisms (20x species of Archaea) takes over one hour (on an Intel Core 2 Duo 2.4Ghz CPU) to build one phylogenetic tree per gene variation per tree construction "substitution model" method.

Solution/Attempt:
Parallel algorithm for multiple sequence alignment (MSA).
Parallel algorithm for phylogenetic tree building.

---

Homework:

hw8

5/18/2013:
q7. For a computer with an IP of 128.2.32.144 and netmask of 255.255.255.196,
the network address is 128.2.32.212 and the broadcast address is 128.2.32.255.

notes:
Network address:
convert IP and Netmask to binary, bitwise AND, convert to decimal.
Broadcast address:
bitwise OR the binary network address and inverted binary netmask.

q9. Using the option flag lsof -i can find the ports open by network services. The number under "name" after the ":" symbol is the port number.

hw7

5/18/2013:
q1. Pentium D CPU uses a LGA 775 socket, and a motherboard that can use this CPU is Intel S3120SHLC Server Motherboard. It is $225 on newegg.com.

q2. A CPU that will work with the motherboard is Intel E7500. It is LGA 775 and is $75 on newegg.com

q3. The motherboard uses DDR2 RAM, and can support up to 8GB ECC Fully Buffered. Crucial 2x4GB DDR2 GB-DIMM Server Memory Model can be used and is $255.

q4. The motherboard uses PCI-E x8 and x16. An Nvidia GPU can can work is Nvidia Tesla C2075. It has PCI-E 2.0 x16 and the monitor connector is DVI. It is $2000 on newegg.com

q5. A criteria for a SATA harddrive is to be reliable and fast. Crucial M500 is a good one. It uses SATA connectors.

q6. The motherboard is a ATX size. A case that would fit an ATX form factor is Lian Li PC-K65. It is $60 on newegg. A power supply to go with it can be SeaSonic X650. It is efficient with a gold "80 plus" certificate and is $110 on newegg.

q7. The difference between VGA and DVI is the shape of the connector and that VGA is analog and DVI is digital. A monitor for the Nvidia C2075 graphics card requires DVI and a Dell U2713HM can accept DVI connectors. It is a 27" monitor and is $700.

q8. The rest of the components like mouse and keyboard can use USB connectors and the CD drive can use SATA connectors.

hw6

5/16/2013: hw6's folder with Tcl/Tk files and question 6's number multiplying program
Finished on 5/20/2013.

hw5

2/25/2013: hw5's folder with C files

hw4

3/2/2013: hw4's folder with sh files
q9. It seem like there is no problem in creating 5000 files using the sh script?
q10. Need to solve "cat" error. Solved on 3/5/2013.
         Added q10_parallel_maketest_specificWait.sh on 3/8/2013.

hw3

2/17/2013: hw3's homework folder named "test1"
notes:
q6. Check apropos type | grep tree
q8. top greaterThanSymbol top_users ; grep -c 'top' top_users ?
       Better solution: pstree -p | grep -c 'top'
q11. ssh lleung@qwe1.math.cmu.edu top gives error
         TERM environment variable not set.
         But ssh lleung@qwe1.math.cmu.edu uname -a works.

hw2

2/2/2013: hw2's homework folder named "test"

hw1

2/2/2013: hw1
q13 (labsproject.org was inaccessible last week).
          One software project component described in class was the license.
          The labsproject.org project is licensed under LGPL.