This site is supported by donations to The OEIS Foundation.

# Talk:Offsets

## Moving Examples of offsets to Offsets

Should we move Examples of offsets to Offsets, while keeping Examples of offsets as a redirect? (The page is semi-protected, so I can't do the move.) — Daniel Forgues 01:29, 16 August 2012 (UTC)

Maybe. I can't remember why it was I didn't just name it that way from the get go. Alonso del Arte 11:33, 16 August 2012 (UTC)
If you agree, I'll move Examples of offsets to Offsets#Examples of offsets, while keeping Examples of offsets as a redirect. — Daniel Forgues 03:49, 17 August 2012 (UTC)
Sounds good. Alonso del Arte 00:05, 18 August 2012 (UTC)
The page must be semi-protected, since I don't have a move tab for it. (Only an associate editor can do the move...) — Daniel Forgues 09:27, 18 August 2012 (UTC)
I did move the page, as requested by Daniel. Please edit the (sub-)headings to appropriate size. (= X = vs == X == ?) — Antti Karttunen 09:52, 18 August 2012 (UTC)
Done. (Top level <h1> ... </h1> or = ... = in wikitext happens to be used for the page title, so the sections start at level 2.) — Daniel Forgues 11:04, 18 August 2012 (UTC)

This article doesn't explain the fact that the offset is given as two numbers, not one. For example, A000290, the squares: a(n) = n^2 has its offset listed as 0, 3. What does the 3 mean?

Good point. When you contribute a new sequence, you need only concern yourself with the first number. If your sequence starts with f(0), the offset is 0. Then the machine adds a second number that gives the position of the first number greater than 1. So, if f(2) > 1 and 2 is the first number this is true for, this means the third entry (f(0) is the first entry) is the first one that is greater than 1, then the machine will add ", 3" to the offset field. This is the case with f(n) = n^2, as you've noticed.
Let me know if this explanation is clear. I admit that up to now I've just let the computer worry about that second number. Alonso del Arte 16:59, 30 October 2012 (UTC)
From Style Sheet#Offset:
There is a second part to the offset after a comma, which is the 1-based index of the first term which is greater than 1 in absolute value. If all terms are -1, 0, or 1 the second part should be 1.
You do not need to add this part; the server will calculate it for you.
Daniel Forgues 04:00, 31 October 2012 (UTC)
So Ira, please let us know if this makes it clear enough or if you still have questions on this topic. Alonso del Arte 23:56, 1 November 2012 (UTC)

Can we make this page clearer somehow? The way it is, I find it very confusing. In particular, the way it is currently written it sounds to me as if the only allowed offsets were 0, 1, 2 and 3. I would prefer if we could omit those specific examples and instead say something like that the offset should be 1, unless the sequence is generated by a function defined for all integers in which case offset should be 0 or the sequence is generated by a function that is defined only for some integers, in which case the offset should be equal to the least argument for which the function is defined. Felix Fröhlich 10:45, 17 July 2014 (UTC)

## Offset of arrays and triangles

What should it be? Should it be the index of the first row (when read by rows)? Li-yao Xia 17:35, 4 November 2014 (UTC)

Yes, I think it should be the first row index. Some helpful examples I found might be A030341, A154720 and A231473. Felix Fröhlich 22:41, 4 November 2014 (UTC)
The first row index is a bit insufficient piece of information. It would be great if OEIS allowed specifying two offsets for tabl and tabf sequences, the first row index and the first element in a row index. Another big problem is the b-file format. Look at the b-file of A231473, it looks awful. Changing the format of the offset field may be a too complicated task, but introducing a new b-file-like format seems possible. We may call it, say, t-file, and for A231473 it may look like:
```3 0 1
3 1 2
4 0 1
4 1 2
...
```
Andrey Zabolotskiy 07:22, 31 May 2016 (UTC)

What if I have a triangle T(n,k) of numbers? There is the offset of the index of the first row AND ( there is the offset of the index of the first column OR the sequence of offsets of the first columns, if not constant ). How do I record the offsets in the columns if constant and if not constant ? Christian Stump 08:37, 17 October 2015 (UTC)

In the case where the column offsets are constants, you could include this information in the definition of the sequence, like I did for example in A258787. I am not sure about the case where they are not constant, but maybe if there is a function or another simple rule for the column offsets, this rule could be given in the comments. If not, then maybe a comment could be given how the offsets for the columns shown in the entry were determined. That's just how I would do it, and I am not sure if there is some rule for it that I am not aware of. If you want some further input and no one else responds here you could also consider writing a message to SeqFan, which probably generates more responses. Best regards. Felix Fröhlich 13:29, 18 October 2015 (UTC)

## Why not use 0 for the secondary offset when all terms are -1, 0, or 1?

Currently there is no way from the secondary offset to distinguish between sequences where the first term is not -1, 0, or 1 and sequences where all the terms are -1, 0, or 1. Doug Bell 22:00, 6 June 2015 (UTC)