Comments on: The Knuth-Morris-Pratt Algorithm in my own words

By: Noname

Noname — Fri, 30 Jul 2010 17:16:17 +0000

You should add a link to this on wikipedia. Much better than most explainations.

Thanks!

By: Jake Boxer

Jake Boxer — Wed, 07 Jul 2010 22:19:47 +0000

Glad I could help :) If I run across anything else that I feel isn’t generally described well, I’ll do another.

By: helped

helped — Wed, 07 Jul 2010 21:33:05 +0000

you are a GENIUS!!!!

you don’t know how much this has helped me!!!

love to see more of these explanations ;)

thanks alot

By: Denis

Denis — Wed, 16 Jun 2010 19:43:42 +0000

Thank you for a very good explanation! tnx again..

By: Jake Boxer

Jake Boxer — Mon, 14 Jun 2010 14:28:39 +0000

Tim: Thanks! I felt the same way about KMP when I was doing my exams a few years ago. Glad I could help!

By: Tim

Tim — Mon, 14 Jun 2010 13:26:55 +0000

Fantastic explanation mate. I coded it using pseudocode a few weeks ago but I’m going into the exam in a couple of days and I had no idea. Still so much to learn but this has been a great help!

By: efemers

efemers — Sun, 13 Jun 2010 14:43:18 +0000

u sir just made my day. Thank you very much :)

By: Tri

Tri — Fri, 07 May 2010 20:46:15 +0000

Sounds good! :)

By: Jake Boxer

Jake Boxer — Thu, 06 May 2010 23:21:55 +0000

Hey Tri,

If my table’s values are shifted to the left by one column in comparison to your table’s values, it must mean you’re prepending an additional empty cell (which you could argue represents a partial match of length zero). This is totally valid, and may make some areas of the code look nicer. However, doing length – 1 is a very common pattern in programming (to compensate for 0-based arrays), while prepending an empty cell to an array isn’t.

Thus, I think doing index arithmetic is better for an algorithm explanation than adding cells. Like I said though, both are totally valid in practice, so feel free to use either one :) I appreciate the comment, and I’m glad I was able to help.

By: Tri

Tri — Thu, 06 May 2010 22:06:12 +0000

Hey Jake,

This was great! It definitely helped me understand the Partial Matching Table for KMP. However, I was wondering if you’re generating the correct Table. When I’m comparing the table that I would get from your method, the values are shifted to the left by one column. I see that your algorithm will still work though, since you account for this by doing table[partial_match_length - 1], or taking the left value. I think typically you would just subtract table[partial_match_length]. I’m just concerned because some people might read this and unknowingly compute the wrong table. This was a great explanation though, and please tell me if I’m misunderstanding something! Thanks again for this!