This site is supported by donations to The OEIS Foundation.

User:Charles R Greathouse IV/Rule of thumb

From OeisWiki

Jump to: navigation, search

My rule of thumb is that a sequence should take at least one hour to submit, between researching, reading, writing programs, and typing it up.

This isn't a hard-and-fast rule, but if you're talking less than half an hour that's a red flag:

  • Are you making this sequence the best it could be? If not, maybe you should hold off submitting it until you can put in more work, or find a coauthor to finish it with you.
  • Is there really that little to say about it? Maybe it should be a comment on some other sequence rather than a sequence of its own.

Purpose

Sequences should be well-researched and professional. If someone comes across the sequence later, will the information they find be helpful? Spending a sufficient amount of time helps to ensure both.

Examples

Reading-heavy: Consider A226898. I spent about two hours reading the Hooley 1979 paper, then about half an hour searching (MathSciNet, Google, OEIS) for additional references, then about an hour skimming those papers and summarizing results in the sequence. While reading I had a program running in the background to compute terms. Total time: about 3.5 hours (3 hours reading, 30 minutes research, 5 minutes writing).

Exploration-heavy: I recently submitted three sequences: A227533, A227534, and A227535. I wrote a quick program to compute terms and then typed up the sequences (about 20 minutes). After 20 hours of computation my target had not been reached, so I studied the output to determine which values were taking a long time. I scribbled some basic formulas, experimented with different programs, and finally computed one additional term in a more clever fashion (1 hour). I then spent an hour and a half writing and debugging a program which generalized the solution from the single case it had handled earlier. This took longer than expected, since there were many special cases. I also found that I needed to compute two related sequences, so I wrote code for them (10 minutes, but not counted in the total since this was work on other sequences). I then spent a relatively long time formulating my thoughts coherently and typing them up in an email to SeqFan (about 45 minutes); this was worthwhile since it allowed me to sharpen my existing prose on the sequences (10 minutes) and may provide feedback on the sequences. Total time: about 3.75 hours (2.5 hours exploring, 1.25 hours writing), or about 75 minutes per sequence.

Exceptions

Not all sequences need a full hour to submit. Some sequences merely report the findings in existing papers; these sequences may not take as long to submit, since the research has already been done.

There are other exceptions. I spent only a few minutes submitting A182986, the possible characteristics of a field; I felt that it was important enough to have an A-number, even if there was not much to say beyond the wealth of information at A000040. These exceptional sequences should amount to no more than two or three in a year per contributor; most sequences need more detail.

Personal tools