This site is supported by donations to The OEIS Foundation.

OEIS Help Page

From OeisWiki
Jump to: navigation, search

Have a question about the OEIS? Enter it here! (You need to Register first)

Someone will enter the answer here. (Click the "watch" tab so you are notified of changes)

Note that there is also an FAQ page, and a page about Trouble registering, logging in, changing password

Memo to sequence fans, associate editors, editors-in-chief: please click the "watch" tab so that you can help answer these questions.

Q: I registered, but I don't see how to add a comment to a sequence.

(Question and answer posted by N. J. A. Sloane, May 7 2013, based on an email received that day.)


  • Go to the OEIS database (this is not the wiki!), and in the top-right corner, click Login, then enter your name and password.
  • Go to the sequence you want to edit, A021754, for example.
  • Click the "Edit" link, as shown in this screenshot:


  • Enter your comment or formula, etc.
  • Click Save.
  • Click "These changes are ready for review by an OEIS editor".
  • You will be able to check your proposal via the (new) link "draft edits", but it will not be displayed immediately in the "published" record: This will happen only after review by Editors and approval by an Editor-in-Chief.

Q: Should "doubled" sequences be added to the OEIS?

(Posted by Jess Tauber, May 9, 2013.)

I've been looking at atomic numerical relations- in the nucleus, apparently because of nucleon pairing, relevant sequences of numerical differences between filled shells and subshells tend to be strongly related to double Pascal Triangle diagonal numbers. The same sort of doubling also appears in similar Pascal-based sequences in atomic clusters. Electronic relations in single atoms use single Pascal numbers.

The question is, when such relevant sequences are simply doubled variants of singular ones already existing in the OEIS, should comments go onto the page there (citing pairs of particles) or should a new page be prepared citing total particle numbers? What's the rule for this sort of thing?

Example: Doubled triangular numbers are 2,6,12,20,30,42,56,72,90,110... and are intervals between idealized magic numbers for the harmonic oscillator model of the nucleus, for either neutrons or protons, those magic numbers being doubled tetrahedral numbers 2,8,20,40,70,112,168,240 and so forth. In more realistic nuclear models including extra terms in the Hamiltonian we do find certain doubled sequences already in OEIS: as ...26,42,62,86,114,146,182... These numbers appear as count differences between certain suborbital components whose only quantum number difference is 1, for quantum number n. These are summed pairs of every other doubled triangular number. Other such sequences (as id. of every third, fourth, fifth, etc.) are not found, but their singular equivalents ARE.

Thanks for suggestions!

Answer from N. J. A. Sloane, May 10 2013.

Generally we discourage adding a sequence whose terms are twice an existing sequence. But there are two situations where it would be a good idea:

1. if the doubled sequence occurs in an important problem,


2. if there are sequences that can more easily be derived from the doubled sequence than from the original sequence. You see, the program Superseeker, when trying to identify a query sequence, looks at all the sequences in the OEIS, and if the query sequence is closer to a doubled sequence, then it would be a good idea to have the doubled sequence in the OEIS.

Certainly there are a large number of examples where doubled sequences exist in the OEIS in their own right — the first two examples you gave being A002378 (the oblong numbers) and A007290 (which has a number of applications).

In short, use your judgement!

Q: What are the rules for approving a newly submitted sequence, which might be potentially duplicate of already existing one, when both have the same terms?

(Question is posted by Alexander R. Povolotsky 17:07, 17 May 2013 (UTC), May 17 2013.)

Please clarify under what circumstances the newly submitted sequence, which might be potentially duplicate of the already existing sequence, should be admitted into OEIS and what process will prompt for the resolution of this potential duplication ? Consider for example the case, which involves already existing sequence A200782 and newly just entered sequence A225381.

Answer from N. J. A. Sloane, May 17 2013: It depends on several factors.

If it is only a conjecture that the two sequences are the same (as in the example you mention) then in general they should not be merged.

If they have been proved to be the same, and both sequences are recent additions, they they should be merged and the youngest should be deleted.

If they have been proved to be the same, and both sequences have been in the OEIS for years, then they should be merged into the A-number of the more senior entry, and the other marked as a duplicate, but should not be deleted.

(I will fill in more details later.)

Answer from Charles R Greathouse IV 03:31, 25 May 2013 (UTC):

The general rule: if the sequences have been proved to be the same and are of different ages, the sequences should be merged into the most senior one (oldest or lowest A-number). Any recent duplicates should be recycled while established sequences should be marked dead. Charles R Greathouse IV 03:31, 25 May 2013 (UTC)

Q: What should the offset be for sequences with a prime power parameter?

(Posted by Eric M. Schmidt 09:21, 21 May 2013 (UTC))

For sequences whose parameter is naturally a prime power, such as a finite field size, what should the offset be? Existing sequences do not have a consistent convention. For example, A177744 has offset 1 while A008914 has offset 2.

Answer from N. J. A. Sloane, May 21 2013:

Probably it should be 1, thinking of the index as running through the list of prime powers (and lists have index 1 by convention). I have corrected the offset of A008914.

Q: What requirements are imposed on programs ?

(Posted by Alexander R. Povolotsky 12:34, 24 May 2013 (UTC))

What is major purpose of the program ? Are programs mostly intended for demonstration purpose, that is to generate 3 lines of terms in the data section ... or they are intended for efficient calculation of b files ... or unlimited (large) number of terms ? Could two variants of the same language program (one for demonstration purpose and another for very efficient calculation of unlimited (large) number of terms be presented in the same sequence ?

Answer from Charles R Greathouse IV 03:40, 25 May 2013 (UTC):

Many purposes are possible, including all of the purposes you mention and more beside: efficient testing or generation of isolated terms, for example. Use your best judgment for the sequence in question. An easy sequence generally does not need many programs, unless it is of special importance, but a sequence which is difficult might merit many implementations in a single language to cover different purposes that its readers may have.

Addendum: I have written a page giving my views on the subject: User:Charles R Greathouse IV/Programs.

Q: Large terms in A007651 or A005150 ?

(Posted by Zak Seidov, May 26 2013)

What is the correct procedure in "Describe the previous term! (method B)" if some frequency in a[n-1] is >9 ?

Answer from N. J. A. Sloane, May 27 2013. In that case the standard way of representing these strings in the OEIS by concatenating the digits will not work, and you would have to use some other method, such as displaying the sequence as an array.

Suppose the first term (really a string) is

11111111111111111111 (that is, twenty ones)

Then the sequence of strings would be

1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1,
1 20,
1 1 20 1,
1 2 20 1 1 1,
1 1 2 1 20 1 1 3,
1 2 2 1 20 1 1 2 3 1,

A similar remark applies to the first sequence of this type, A005150.

Q: Some questions

(Posted by Wesley Ivan Hurt, May 28 2013)

Question 1: What is the difference between a sequence that has been marked "reviewed" and one that is marked "proposed for review"?

Answer 1: An associate editor, on reviewing a sequence for correctness, moves it from "proposed for review" to "reviewed". It still doesn't show up in the main OEIS until approved by an Editor-in-Chief. Charles R Greathouse IV 18:33, 29 May 2013 (UTC)

Question 2: What is the significance of the "+" and "-" numbers that appear next to a sequence and under the Status heading? For example, if a proposed sequence has a +5 -3 beside it, what does this signify?

Answer 2: Lines added/lines removed. If you correct a comment and add a formula the sequence should have "+2 -1". Charles R Greathouse IV 18:33, 29 May 2013 (UTC)

Question 3: Suppose that an editor can verify everything in a particular submission except for one or two particular fields (in particular, software that you may not own and/or programming languages like Java that you may not be able to verify). Should this simply be stated in the comment line before it is proposed for further review? Are there online tools/converters that can circumvent this issue? In particular, I currently do not have a copy of Mathematica and would like to verify Mathematica code for submitted work in the future. Is there a website where one can run short Mathematica programs, similar to the MAGMA calculator at ?

Answer 3: You can leave this information in the pink box for the editors (and the author). You may be able to use Mathics online but I haven't used it much. For other possibilities see [1][2] about which I know nothing. Charles R Greathouse IV 18:33, 29 May 2013 (UTC)

Question 4: Is there a list of all the programming languages that appear in the OEIS?

Q: What is a b-file?

Answer from N. J. A. Sloane, Jun 02 2013:

The Data section of an entry in the OEIS usually gives about three lines of terms (200 or 300 characters).

Often one wants more terms than this, so many sequences now have "b-files", giving 10000 or more terms. If there is a b-file, it will be the first item in the Links section.

For an example, see the entry for the Fibonacci numbers, A000045, and click the top link.

See for the format. You can always get to this page by clicking the "Contribute new seq. or comment" link at the foot of any OEIS page.

Difference between A006056 and A006057

What is the difference between A006056 and A006057? The title of both is "Number of labeled topologies with n points." The sequences are certainly different, but the descriptions do not explain the way in which they are counting different things. —Brian Kell 06:40, 9 January 2014 (UTC)

  • (From Neil Sloane, Jan 09 2014) I have the Erne' paper and I will clarify the definitions later today.

Two suggestions

(Posted by Jonathan Sondow 15:39, 3 February 2014 (UTC))

1. On the Hints page after "You can separate search terms with | (no spaces around the |) and it means sequences that match either term." add something like "In other words, | is the logical OR." That way someone who searches for " or " on the Hints page will find the answer.

2. Invent and build a way to search for a subsequence. Of course, one can omit commas, but that loses the ordering and results in too many hits. One solution to the searching-for-subsequence problem might be to replace commas with some other symbol, e.g., asterisks "*".

Answer: Use the subseq operator, like subseq:17,8749,246401. - Charles R Greathouse IV 00:31, 4 February 2014 (UTC)
Thanks, Charles! - Jonathan Sondow 14:45, 7 February 2014 (UTC)

Q: Can I add an OEIS sequence to my watchlist directly from the sequence page?

(Posted by Jonathan Sondow 15:30, 7 February 2014 (UTC).)

If yes, how do I do it? (I don't see an "Add to watchlist" button on the sequence page.) If no, (how) can I add the sequence to my watchlist without using "Edit raw watchlist"?

Answer: It's worse than that -- you can't watch sequences pages at all. The wiki half of the OEIS doesn't know about the sequence half. - Charles R Greathouse IV 20:49, 11 July 2014 (UTC)

Removing accidental sequences and getting past active edit limit

What is the proper way to remove/release your sequence that has not been submitted yet?

Answer: Add a pink box comment requesting deletion. If it's not handled in a day or two, feel free to leave a note on my wiki Talk page and I'll try to get to it. - Charles R Greathouse IV 22:03, 11 July 2014 (UTC)

Is the active edit limit restricted to only 7 edits? As a submitter, how can we work on other sequences while we are waiting for the editors to approve the outstanding ones?

Answer: 7 simultaneous sequences, but an unlimited number of edits per sequence. Usually two or three suffice, but the limit is more generous just in case. Limits are useful, not only to encourage people to submit high-quality sequences (and edits) but to divide scarce editorial resources over a large number of contributors.
In special cases we consider requests to increase editing limits. For example, you may be working on a collection of 20 closely-related sequences which reference each other in a complex pattern and it's easier to set them up in advance rather than sequentially. - Charles R Greathouse IV 22:03, 11 July 2014 (UTC)

Large data terms

Are there any guidelines about when large terms are permitted in the sequence page and not just a b-file? There are currently 138 data field terms above 120 digits: User:Jens Kruse Andersen/Large terms. Decimal expansions on sequence pages are wrapped at 120 digits, but the line length is increased long before that. See for example Many of the large terms seem unimportant, for example primes of form f(n) for some function where the n values can just be given in a comment or cf, and often are. The largest current term is 285 digit in A119555 (there are currently draft edits by me). Jens Kruse Andersen 13:58, 4 July 2014 (UTC)

I think this is a question which should be addressed/answered. My personal opinion would be that terms should (in general) be kept short enough not to enlarge the default width used to display the record(s). Of course it may be possible to make an exception in special cases where it appears justified for good reasons. MFH 22:29, 11 August 2015 (UTC)

Number of nondegenerate Boolean functions of n variables

A search for the numbers in the title returns the three distinct sequences A000371, A000618, and A003181. Sequence A000371 (and the underlying definition) seems to be the most widely used one. It is also the "complement" of A005530, the "number of degenerate Boolean functions of n variables" (among others). What definition of "nondegenerate" is the basis for A000618 and A003181? --Konstantin Ziegler 15:05, 11 July 2014 (UTC) Answer: Please see the Muroga or Sklansky references. - Neil Sloane.

Offset in A000533

My opinion is this sequence is not correct the way it is currently entered. The definition says '10^n + 1, n >= 1', which to me means the first value for n is 1 (i.e. offset is 1). By that definition, the first term is 11, not 1. With offset 0, the given definition produces the sequence 2, 11, 101, 1001, .... In any case, I don't see how '1' could be the first term. I was told the 'name' should not be changed (see history), but given name, offset and terms just don't suit together. Should something be done about this? If so, what should be changed? Felix Fröhlich 14:45, 16 July 2014 (UTC). Answer: I stated explicitly that a(0)=1. Neil Sloane

User contributions links

Is there a better page to discuss the wiki and make suggestions like the below?

I have created Template:Sp-contributions-footer with various user links. The template page uses N. J. A. Sloane as default example. Here is the result of {{Sp-contributions-footer|Jens Kruse Andersen}}:

I suggest the template is displayed at the bottom of user contributions pages like Special:Contributions/Jens Kruse Andersen, with links for the username. As far as I can tell, this will happen if an administrator creates MediaWiki:Sp-contributions-footer with this content (as displayed here and not in the edit window):


If the OEIS search links seem excessive then they could stop after "Author". Jens Kruse Andersen 11:21, 31 July 2014 (UTC)

Searching the OEIS by keyword

Is there a way to search the OEIS by keyword? Say, for example, I want to find all sequences with keyword 'more' and only those where the term 'more' appears as a keyword, not anywhere else on the sequence page. If possible, how can this be done? Felix Fröhlich 08:14, 12 August 2014 (UTC)

keyword:more. Click the "Hints" link next to the OEIS Search button to see search options. Jens Kruse Andersen 09:02, 12 August 2014 (UTC)
Awesome, thank you. Felix Fröhlich 06:13, 13 August 2014 (UTC)

Difference between A240555 and A005837?

I'm unsure as to what exactly "earliest possible" means (which sequence A240555 claims to be); a natural interpretation is to take the earliest terms, but my code agrees with A005837 and not A240555. Both sequences deal with 4-AP avoiding sequences. --Lewis Chen 04:42, 3 January 2016 (UTC)

Answer from N. J. A. Sloane 08:37, 3 January 2016 (UTC). The definition of A240555 should have said "lexicographically earliest". I have made the change. Thank you for noticing this.

Forgive me for being unnecessarily obtuse, but I don't see how A240555 is lexicographically earlier than A005837. They first differ on term 6 for which A005837(6)=8 whereas A240555(6)=9. --Lewis Chen 21:11, 3 January 2016 (UTC)

You are right, and there is something wrong. I have restored the original definition of A240555, which I admit I do not understand. This needs further investigation. I will post a message here when the matter is resolved. - N. J. A. Sloane 22:40, 3 January 2016 (UTC)

It turned out that A240555 and three related sequences were wrong. This has now been fixed. See the cross-references in A005837 for a summary of all these sequences that avoid arithmetic progressions of specified length. Thank you again for noticing that there was a problem. As a result of your message, 20 sequences have been revised. - N. J. A. Sloane 21:03, 4 January 2016 (UTC)

Symmetric tables

When seeking to add a new two-parameter sequence A(m,n) with the symmetry property A(m,n) = A(n,m), which of the following is preferred?

  1. Add the sequence as a lower-triangular matrix read by rows: A(1,1), A(2,1), A(2,2), A(3,1), A(3,2), ...
  2. Add the sequence as a full table read by anti-diagonals: A(1,1), A(2,1), A(1,2), A(3,1), A(2,2), ...
  3. Add both sequences and cross-reference them.

Peter J. Taylor 23:09, 11 January 2016 (UTC)

Answer from N. J. A. Sloane 23:23, 11 January 2016 (UTC): If it a published table, then the author has often faced that problem, and we recommend doing what he or she did. Otherwise, to be safe, we recommend your third solution: Add both sequences and cross-reference them

3X+1 problem

I am unable to load this page of the wiki (shows error 500)...I tried several times--Bill McEachen 17:23, 31 January 2016 (UTC)

Answer from Felix Fröhlich 13:37, 15 May 2016 (UTC): Does it work now? I can load the page in my browser, although it is quite slow. The edit window contains a warning, suggesting to split the page into several smaller pages, so maybe it would be a good idea to do that. There are many tables, graphics and Latex formulas on the page. {{Vertical bar graph from lists}}, which is transcluded onto that page, seems to be especially problematic.

I have trimmed down the section about the reduced Collatz function. It's less than 7K of the source code, but since it loads a lot of images of typeset math, the actual bandwidth savings should be greater. - Alonso del Arte 16:29, 15 May 2016 (UTC)

Erdős Number

Having entered a couple of minor sequences (and had at least one accepted), I would like to ask the question: does this count in any way as "collaboration"? I ask because (as an amateur mathematician who works solo, mainly on a completely different wiki), I want to see whether I can get an Erdős number as a result. This is purely for my own gratification -- as I already have a surprisingly small Pegg number and an even more ridiculously tiny Bacon number, I wondered whether I can get the hat trick.

If OEIS does not qualify in this way for "collaboration", can anyone offer advice as to how an amateur mathematician, flying solo through a complete lack of any face-to-face interaction with "real" mathematicians, get such an Erdős number? Like: anyone out there with a finite Erdős number who is prepared to offer me a collaboration opportunity?

Apologies if this is too frivolous for this site. --Prime Mover 06:38, 23 October 2016 (UTC)

Maybe it is frivolous but you're neither the first nor the last to wonder about it. I started writing a paper about hypersigma and asked Charles to be a co-author, but that didn't go anywhere because the paper was of a purely expository nature, with no results worthy of publication in a reviewed journal. If you can find such results, ask an OEIS Editor who does have an Erdős number to review your results. If not, just be happy about your Pegg and Bacon numbers. - Alonso del Arte 15:58, 23 October 2016 (UTC)
Publishing OEIS sequences doesn't meet the requirements of the Erdős number game as it's usually understood. But you could play by whatever rules make you happy, of course. My suggestion: use OEIS collaboration as a springboard to write a short paper with some of the people you meet here... regardless of games and rules, it should be fun and maybe useful to someone. - Charles R Greathouse IV (talk) 01:19, 16 January 2019 (EST)

Sequence defined from a sequence that is unsuitable to be added itself

I'm preparing a sequence where the terms are defined as being rounded from a real-valued sequence, the terms of which are defined from earlier terms of the real sequence, ie recursively. What would be a good way to word the definition? Could you point me to any good examples of similarly defined sequences? - Peter Munn 13:53, 25 April 2017 (UTC)

A000319 is an example (in this case, defining an auxiliary sequence). - Charles R Greathouse IV (talk) 01:19, 16 January 2019 (EST)

What is A000157 really?

A000157 has the title "Number of Boolean functions of n variables" and no comments elaborating on that. It clearly doesn't count Boolean functions: for a start, the values are less than A000616 (number of irreducible Boolean functions of n variables). What are the equivalence classes which A000157 is really counting? Peter J. Taylor (talk) 04:22, 17 February 2018 (EST)

Is there an accepted way to download a machine-readable version of an OEIS entry?

I don't want to download and store the multi-gigabyte full database archive, I only want to retrieve a small number of entries that I'm interested in. I can download entries by retrieving the HTML page and parsing it. However I would like to know - is there any way to get the raw data, for example as XML, so that I don't have to parse the page? Secondly, is it acceptable to download these pages with a script? It would only be a few dozen or so, so not a huge load on the server, but recognizably done by a automated script rather than a human in a browser. Jack W Grahl (talk) 16:08, 29 April 2018 (EDT)

This question is answered here:,_Compressed_Files

Offset of A000642

Q. A000642 has offset 0,4 and initial terms 0,1,1,2. Its name is "Number of alkyl derivatives of acetylene X^{II} C_n H_{2n+2} with n carbon atoms." and it has the comment "Number of structural isomers of alkynes C_n H_{2n-2} with n carbon atoms." I don't know enough chemistry to know whether the offset is correct for the name. What I do know is that (a) it's a logical impossibility to have an alkyne with 1 carbon atom, because it can't triple-bond to itself; (b) the linked reference has a single table with title "Number of isomeric hydrocarbons of the acetylene series" which starts 1,1,2 with offset 2,3. Which is incorrect: the comment or the offset? The answer seems to hinge on the meaning of the name. Note that if the offset is incorrect, so is the g.f. Peter J. Taylor (talk) 18:34, 12 January 2019 (EST)

A. Polya's theory enable one to solve many problems like this by using generating functions. However, you must remember that the initial term or two of the generating function is often defined by convention, and need not correspond to any physical structure. Second, the entry has 5 or 6 references, so you could start by studying them - especially the Polya article. Third, since this is one of the original entries in the database, going back to the 1960's, I am certain the terms are correct. - N. J. A. Sloane 21:27, 12 January 2019 (EST)

As noted in the original query, I noticed the discrepancy as a result of studying the Coffman, Blair and Henze reference, whose table starts 1,1,2 with offset 2,3. The same discrepancy can be observed when reading the second Read reference, which gives (bottom of p31) the g.f. for alkynes as x^2 Z(S_2; A598(x)), which disagrees by a factor of x with the g.f. listed in OEIS. There is most definitely an offset error. (NB I don't dispute the terms: just the offset). The question is whether the error is only in the alkynes, and should be corrected by amending the comment to "Number of structural isomers of alkynes C_{n+1} H_{2n+2} with n+1 carbon atoms." or whether it is also in the alkyl derivations of acetylene, and should be corrected by changing the offset to 1,4 and adding the factor of x to the g.f. Peter J. Taylor (talk) 12:37, 13 January 2019 (EST)
I changed the definition and offset so as to agree with Coffman et al. (1933). - N. J. A. Sloane 13:31, 13 January 2019 (EST)

Notation for rectangular arrays (question about A083064)

For a sequence given as a table read by antidiagonals, defined with the notation T(n, k), is it required that n refer to the row and k to the column (as in standard matrix notation)? If so, then the Example in A083064 appears to be transposed. The sequence is defined as "Square number array T(n,k)=(k(k+2)^n+1)/(k+1) read by antidiagonals." By that definition, we get:

T(2,7)=(7*81 + 1)/8 = 71

T(7,2)=(2*16384 + 1)/3 = 10923

So 71 should appear in row 2 and 10923 should appear in row 7. But in the Example, it's the other way around. Perhaps the definition of the sequence should be changed to T(n,k)=(n(n+2)^k+1)/(n+1). Mathew Englander (talk) 03:42, 21 October 2020 (EDT)

Enter further questions

This is a place where users may post questions. (Click "Edit" next to this subsection title [after logging in], and add your question at the top of the editing area, before the current title).

See also