Hi all! I am a new phd student (started 4 months ago) and I think this forum is great!

I have 3 questions and hope someone can clarify my doubts.

1. When we do literature survey, do we need to understand totally other people's algorithms, or do we only need to know roughly what people have done?

2. Is using an existing algorithm for a new product considered "original"? (my research is more application based)

3. When we write papers, must we include every little detail? If i avoid writing something which gives better experimental results but does not have theoretical justification, will that be "cheating"?

Thank you very much!

My opinion is:
1) You need to be aware of other peoples algorithms, a grasp of the basics and be able to say why you are not using them. It is worth having a good grasp of what is currently considered the "best".
2) It isn't original but might get a conference publication and is worth doing if you can put some spin on it, you can do it as a tutorial or as a comparison.
3) I don't think better experimental results without theoretical justification will get you far unless you can explain why they are better, which would be theoretical justification, as it may be simply they work better for this experiment conditions but are not generally applicable. I was always told to provide enough detail to enable someone reading it to understand and reproduce it

Hi! Thanks for your advice!