Slide 1: Having Fun in Research
Yuanyuan Zhou
06/25/07
Slide 2: Research is a Job!
You need to be responsible
You cannot cook your own results
Research should aim to make impact
Not just papers
Research pays
RAs ---- sorry, pay is a little too low Professor---pay reasonable Research scientist in a research lab Patent commercialized …
06/25/07
Slide 3: Research > a Job
What are the special characteristics demanded in research?
Hint: what is demanded in playing a computer game? Your curiosity Your passion Your focus Your patience
May take 2-3 years
Your obsess (your persistence)
Sometimes it does not work Rejections (10-20% acceptance rate)
Do you have all?
06/25/07
Slide 4: A table to relieve your stress
Enjoy yourself
Succeed (getting tenure, promotion, fame, etc) Not succeed
Not Enjoy yourself
Not too bad (at least you succeed!) BAD!!!!
Great!
Not too bad (at least you enjoy youself)
To avoid the only bad combination, you should just enjoy yourself! Because this is under your control, whether succeed or not in some degree also depends on others’ subjective evaluation!
06/25/07
Slide 5: Selecting Fun Research Areas
Match your interest/passion Common mistakes
Limit yourself only within your comfortable zone Something unknown must be fun Consider only the ultimate research goal but ignore
Research reality Research methodology You need to enjoy the journey
Consider only job opportunities Follow the trend --- hot areas
06/25/07
Slide 6: Adventurous in Your Research
Research is to find out the unknown Adventure in research
Propose new (maybe weird) solutions Jump into a new area unknown to you
Adventurous in research is fun
Not bored You can bring a fresh air into a direction Likely to yield unexpected results
Your start a new direction!
06/25/07
Slide 7: My Own Experience
1992-1993: Database (Peking Univ) 1993-1994: Mathematics & Internal medicine ( Univ. of Virginia) 1994-1995: Computational biology (Princeton) 1995-1999: Distributed systems (Princeton) 1999-2002: Storage Systems (NEC Lab) 2002-Now: Software reliability, Energy management (UIUC) Future: ?????
06/25/07
Slide 8: Problem-Driven Research
Identifying problems is more important than finding solutions Define the boundaries of your problem carefully
“Nobody will be impressed if you set the bar too low and jump over it. Nobody will be impressed if you set the bar too high and don’t jump over it.” (Dave Redell) Don’t try to “solve the world” or “boil the ocean”.
06/25/07
Slide 9: Picking Fun Research Problems
Criteria (agree with Lui & previous speakers)
Take time to understand the problem
Exciting and interesting area Important problems in area Activities suitable to you (theory vs systems)
Catch up background Get your own insights Hands-on: repeating the most recent, authoritative work in this direction to understand the limitations from YOUR own perspective!
06/25/07
Slide 10: Defining the Solution
Make the problem concrete
Start with particulars, then generalize
Know what makes the problem hard
“Why couldn’t you just...”
Identify the standard of success
How will you know when you are done? How to distinguish a good solution from a less-good solution?
06/25/07
Slide 11: Looking for Solutions
Don’t limit to your comfortable zone Broaden your eyes
Talk with researchers from other fields Attend talks in other fields
Don’t let fear of failure stop you
06/25/07
Slide 12: My Own “Gutsy” Solutions
Pushing bug detection support into hardware
Attended Ravi Iyer’s class at UIUC Attended Darko Marinov’s class at UIUC
Combining data mining into compiler’s program analysis
Attended Jiawei’s data mining class
Combining NLP into program analysis analysis
Reading some textbook
06/25/07
Slide 13: Conduct Research in a Fun Way
Feasibility analysis
Find out the potential in a quick way
Talk with other people
Get feedback
Divide and Conquer
See progress along the way
Be proud of your ideas!
06/25/07
Slide 14: Team Research is More Fun
Team discussion is stimulating You handle the up-and-downs together
Less pressure
Work party
Who likes to play games in a team?
Internet games Poker
06/25/07
Slide 15: Eight Ways to Destroy the Fun of Team Research?
Fight for credits/author order Not supportive Over-defensive Command each other No communication Blame each other Passive waiting for tasks No compromise
06/25/07
Slide 16: Tell Your Fun Findings to Others Write a paper Publish it at workshops/conferences Present it at a conference
06/25/07
Slide 17: Writing is Fun Writing helps you refine ideas
That’s why writing should be done in parallel with the design, implementation and experiments
Writing helps you communicate ideas
In most cases you will find out that you, your teammates and your adviser have totally different views of the project (the idea, etc)
06/25/07
Slide 18: Add Fun Analogies in your Presentation Explain boring technical things in an easy-to-understand manner
Demonstrate your key insights
Make your talk more lively
Throw jokes
06/25/07
Slide 19: Analogy Examples
The analogies used in my job talk The analogy used in my recent talks
06/25/07
Slide 20: Multi-level Server Cache Hierarchy
Database Clients File Clients
Networ k
Database Servers File Servers
Networ k
Storage Servers
…
Client Cache (64MB – 128MB)
06/25/07
Database Server Cache
Storage Server Cache
<<
(4GB – 32GB)
~
(1GB – 64GB)
No need for inclusion property
Slide 21: Multi-level Server Caching
accesses
in cache?
misses LRU?
Storage server Cache Storage Systems (Lower level)
hits
Least Recently Used (LRU)
06/25/07
Database or File server Cache Database Servers File Servers (Higher level)
Slide 22: Analogy: Storage Box (Basement)
Assumption for analogy: item = box Question: do you keep the box? If you have a basement, you can keep all the boxes
Living room (higher-level)
Basement (lower-level)
DELL
pizza
06/25/07
Traditional Client-Server Cache Hierarchy
Slide 23: Analogy: Storage Box (Closet)
If you just have a closet, you may keep only the box for your holiday decorations!
Living room (higher-level)
cold access hot access
Closet (lower-level)
hot miss
06/25/07
Database-Storage Server Cache Hierarchy
Slide 24: But If You Use LRU for Your Closet…
Your closet will be full of garbage!
Living room (higher-level)
Basement (lower-level)
pizza
06/25/07
Slide 25: Atomicity Violation Bugs
Programmers want atomicity
They assume some code regions are ‘atomic’ Lock/transaction are ways to ensure atomicity Incorrect implementation causes atomicity violations
06/25/07
Slide 26: Challenges of Atomicity Violation Detection
How to represent programmers atomicity intention? Infer programmer’s ? atomicity ow intention H Detect violation to the atomicity ? ow intention H
program
BUG!
06/25/07
Slide 27: An Analogy ---Don’t Disturb
Assumption:
I work on some tasks everyday and some are repetitive
Preparing lectures Writing proposal Playing computer games Thinking or day-dreaming…
Some tasks cannot be interrupted (e.g. playing games) I don’t tell students explicitly about what can or cannot be interrupted----too embarrassing…
Question:
Can my students automatically figure out what tasks can be interrupted?
06/25/07
Slide 28: Infer Professor’s “Atomicity” Desire
Finding Clues:
When a task is interrupted unwillingly, the professor is a little mad, dis-oriented and cannot wait to get back
Infer this task needs to be “atomic”
Otherwise, happy to be interrupted
06/25/07
Slide 29: Research Impacts
Papers on top conferences
Best paper award---even better
Visibility
Discussed in other institutes’ seminars Other people know about your work
Inspire other research projects
# of Citations
Start a new research direction Used in real systems
06/25/07
Slide 30: Final Words: Enjoy Your Research If you are not having fun in research now, ask yourself
Are you working on an area that you feel passionate? Are you working on a fun problem? Are you doing it alone? Have you seen any promising results? Do you see the impact of your research?
06/25/07
