Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

Selfrefering sentencies embedding selrefering sentences, etc.

40 views
Skip to first unread message

ericang...@gmail.com

unread,
May 22, 2012, 12:52:49 PM5/22/12
to
Hello Rec.puzzlers (reposted from SeqFans mailing list
as I've received not one single answer);,
a friend of mine, Jean-Luc Piedanna, has discovered
an amazing sequence for the French language.
Here is the link (in French) -- and below, a short
explanation (I hope to be clear):
http://www.cetteadressecomportecinquantesignes.com/Piedanna.htm

Jean-Luc starts with a selfrefering sentence like
this:

"This sentence has thirty-one letters"

He embeds now this sentence in another selfrefering
sentence:

"This sentence, together with "This sentence has
thirty-one letters", has seventy-seven letters"

And again:

"This sentence, together with "This sentence,
together with "This sentence has thirty-one
letters", has seventy-seven letters" has ...
[X] ... letters".
Etc.

The interesting part is that Jean-Luc has found _the longest
such sequence in French_ (no matter if you start with the
31-letter long sentence like here, or with the 33-letter long
"This sentence has thirty-three letters", or whatever -- and
no matter of the words you select to write the "envelope" --
which is here "This sentence, together with... has... letters").
How did Jean)Luc find this? He has brute-force tested all
possible couples (A;B) where A is the lenght of the "seed"
and B the letter-length of the "envelope" - and when I say
"tested" I mean "exploring all possible trees" also! The
longest sequence in French is produced by the couple where
A=30 and B=43, it has 79 terms. Could someone compute the
same sequence for the English language?

Best,
É.

Mark Brader

unread,
May 22, 2012, 2:58:24 PM5/22/12
to
Éric Angelini:
> ...
> http://www.cetteadressecomportecinquantesignes.com/Piedanna.htm
>
> Jean-Luc starts with a selfrefering sentence like
> this:
>
> "This sentence has thirty-one letters"
>
> He embeds now this sentence in another selfrefering
> sentence:
>
> "This sentence, together with "This sentence has
> thirty-one letters", has seventy-seven letters"
>
> And again:
>
> "This sentence, together with "This sentence,
> together with "This sentence has thirty-one
> letters", has seventy-seven letters" has ...
> [X] ... letters".
> Etc.
>
> The interesting part is that Jean-Luc has found _the longest
> such sequence in French_ ...

Interesting that it exists, yes.

I wonder how it would change the results if "de" was changed to "par".
--
Mark Brader, Toronto "Verbose better."
m...@vex.net -- David M. Sherman

My text in this article is in the public domain.

Eric Angelini

unread,
May 23, 2012, 3:36:09 AM5/23/12
to
Hello Mark,
you will not improve the result if you change "de" into "par" -- as doing so you only test a different value of B (the size of the "envelope"), and Jean-Luc has (brute force) proved that for the
French language, the longest sentence (79 terms) is produced by the seed/envelope couple A/B with A=30 and B=43. Turning "de" into "par" would make B=44 -- and this value, as _all_ other ones, will produce a shorter sentence. This is exactly the beauty of Jean-Luc's result: he has found a kind of "marker" for the way the French language writes it's own integers. Thus the interest of such "markers" for other langages in order to compare.
Best,
É.

ken

unread,
Jun 3, 2012, 1:19:57 AM6/3/12
to
Using the seed "This sentence has - letters",
and embedding with "This sentence, together with "-", has - letters",
the process fails after 82 embeddings, with a longest sentence of 5335
letters.



This sentence, together with "This sentence, together with "This
sentence, together with "This sentence,

together with "This sentence, together with "This sentence, together
with "This sentence, together with

"This sentence, together with "This sentence, together with "This
sentence, together with "This

sentence, together with "This sentence, together with "This sentence,
together with "This sentence,

together with "This sentence, together with "This sentence, together
with "This sentence, together with

"This sentence, together with "This sentence, together with "This
sentence, together with "This

sentence, together with "This sentence, together with "This sentence,
together with "This sentence,

together with "This sentence, together with "This sentence, together
with "This sentence, together with

"This sentence, together with "This sentence, together with "This
sentence, together with "This

sentence, together with "This sentence, together with "This sentence,
together with "This sentence,

together with "This sentence, together with "This sentence, together
with "This sentence, together with

"This sentence, together with "This sentence, together with "This
sentence, together with "This

sentence, together with "This sentence, together with "This sentence,
together with "This sentence,

together with "This sentence, together with "This sentence, together
with "This sentence, together with

"This sentence, together with "This sentence, together with "This
sentence, together with "This

sentence, together with "This sentence, together with "This sentence,
together with "This sentence,

together with "This sentence, together with "This sentence, together
with "This sentence, together with

"This sentence, together with "This sentence, together with "This
sentence, together with "This

sentence, together with "This sentence, together with "This sentence,
together with "This sentence,

together with "This sentence, together with "This sentence, together
with "This sentence, together with

"This sentence, together with "This sentence, together with "This
sentence, together with "This

sentence, together with "This sentence, together with "This sentence,
together with "This sentence,

together with "This sentence, together with "This sentence, together
with "This sentence, together with

"This sentence, together with "This sentence, together with "This
sentence, together with "This

sentence, together with "This sentence, together with "This sentence
has thirty-one letters" has

seventy-seven letters" has one hundred and thirty letters" has one
hundred and eighty-eight letters" has

two hundred and forty letters" has two hundred and ninety-six letters"
has three hundred and fifty-four

letters" has four hundred and seven letters" has four hundred and
sixty letters" has five hundred and

seventeen letters" has five hundred and seventy-seven letters" has six
hundred and thirty letters" has

six hundred and eighty-six letters" has seven hundred and forty
letters" has seven hundred and ninety-

nine letters" has eight hundred and fifty-eight letters" has nine
hundred and twelve letters" has nine

hundred and sixty-nine letters" has one thousand and twenty-six
letters" has one thousand and eighty

letters" has one thousand, one hundred and forty-eight letters" has
one thousand, two hundred and

fourteen letters" has one thousand, two hundred and eighty-three
letters" has one thousand, three

hundred and fifty-three letters" has one thousand, four hundred and
twenty-three letters" has one

thousand, four hundred and ninety-one letters" has one thousand, five
hundred and fifty-nine letters"

has one thousand, six hundred and twenty-six letters" has one
thousand, six hundred and ninety letters"

has one thousand, seven hundred and fifty-nine letters" has one
thousand, eight hundred and twenty-nine

letters" has one thousand, eight hundred and ninety-nine letters" has
one thousand, nine hundred and

sixty-eight letters" has two thousand and twenty-seven letters" has
two thousand and eighty-five

letters" has two thousand, one hundred and fifty-one letters" has two
thousand, two hundred and sixteen

letters" has two thousand, two hundred and eighty letters" has two
thousand, three hundred and forty-

nine letters" has two thousand, four hundred and fifteen letters" has
two thousand, four hundred and

eighty-four letters" has two thousand, five hundred and fifty-one
letters" has two thousand, six hundred

and sixteen letters" has two thousand, six hundred and eighty letters"
has two thousand, seven hundred

and forty-nine letters" has two thousand, eight hundred and sixteen
letters" has two thousand, eight

hundred and eighty-seven letters" has two thousand, nine hundred and
fifty-five letters" has three

thousand and eleven letters" has three thousand and seventy-three
letters" has three thousand, one

hundred and forty-three letters" has three thousand, two hundred and
six letters" has three thousand,

two hundred and seventy-eight letters" has three thousand, three
hundred and forty-nine letters" has

three thousand, four hundred and eighteen letters" has three thousand,
four hundred and eighty-nine

letters" has three thousand, five hundred and fifty-nine letters" has
three thousand, six hundred and

twenty-nine letters" has three thousand, six hundred and ninety-nine
letters" has three thousand, seven

hundred and seventy-three letters" has three thousand, eight hundred
and forty-four letters" has three

thousand, nine hundred and thirteen letters" has three thousand, nine
hundred and eighty-four letters"

has four thousand and forty-three letters" has four thousand, one
hundred and seven letters" has four

thousand, one hundred and seventy-six letters" has four thousand, two
hundred and forty letters" has

four thousand, three hundred and five letters" has four thousand,
three hundred and seventy-eight

letters" has four thousand, four hundred and forty-six letters" has
four thousand, five hundred and

twelve letters" has four thousand, five hundred and eighty-one
letters" has four thousand, six hundred

and forty-nine letters" has four thousand, seven hundred and eighteen
letters" has four thousand, seven

hundred and eighty-nine letters" has four thousand, eight hundred and
fifty-nine letters" has four

thousand, nine hundred and twenty-nine letters" has four thousand,
nine hundred and ninety-nine letters"

has five thousand and fifty-eight letters" has five thousand, one
hundred and twenty-eight letters" has

five thousand, one hundred and ninety-six letters" has five thousand,
two hundred and sixty-four

letters" has five thousand, three hundred and thirty-five letters.

-ken

Eric Angelini

unread,
Jun 12, 2012, 10:15:20 AM6/12/12
to

Hello Ken,
nice work! I'll publish yr answer on my page in
a couple of hours, if you don't mind.
Did yr program test by brute-force _any_ seed
and _any_ "embedding sentence"? If yes this would
mean that your sequence of sentences is the longest
possible one in the English language, yes?
Many thanks again,
Best,
É.

ken

unread,
Jun 13, 2012, 3:06:48 AM6/13/12
to
Eric, I tried it with different embedding lengths(E),
although I'm not sure what the reasonable boundaries are.
The results seemed to be pretty random.
eg E=34->82 levels, E=35->227 levels, E=36->0 levels
E=45 was last seen with 88 branches at level 1000, no sign of stopping

I looked at the French page and my solution would be easier to see in
their format
P82= 77, 134, 188, 244, 298, 357, 413, 473, 529, 588,
644, 698, 757, 814, 874, 933, 989, 1047, 1113,
1178, 1245, 1313, 1382, 1448, 1510, 1583, 1647,
1710, 1787, 1857, 1927, 1997, 2054, 2123, 2189,
2257, 2328, 2397, 2465, 2534, 2599, 2667, 2738,
2803, 2874, 2943, 2999, 3028, 3087, 3157, 3228,
3297, 3362, 3432, 3489, 3559, 3629, 3699, 3773,
3844, 3913, 3984, 4043, 4107, 4178, 4247, 4317,
4387, 4457, 4528, 4597, 4665, 4737, 4803, 4875,
4944, 4999, 5058, 5128, 5198, 5267, 5335
This happens to end on 5335, as does the French P79.
Probably just a mysterious fluke(?)

-ken

Eric Angelini

unread,
Jun 13, 2012, 3:55:47 AM6/13/12
to
Ok, many thanks again, Ken. I'm amazed by the fact that there is a branch in some tree that still grows after 1000 iterations, showing no halting sign! Just for my records: what are the lengthes of the "seed" and of the "enveloppe" of the said sentence? (this is: how many letters there are in
the first selfrefering sentence and how many letters there are in the embedding phrasing like "this sentence, together with _ has _ letters") Please forgive my bad English, Ken -- I hope I am clear and that I've understood you're last message.
Best,
É.

ken

unread,
Jun 13, 2012, 5:15:28 AM6/13/12
to
I just used the suggested patterns:-
SEED = "This sentence has - letters" -22 letters
EMBED = "This sentence, together with -, has - letters" -34 letters

The 1000+ iterations are for EMBED=45
eg "This long English sentence, together with -, has - letters

-ken

Eric Angelini

unread,
Jun 13, 2012, 6:07:09 PM6/13/12
to
Ok, many thanks -- I'll update my page
in a couple of days, Ken.
Best,
É.

Eric Lafontaine

unread,
Jun 15, 2012, 8:01:44 AM6/15/12
to
This is of course based on the standard (for France) naming of numbers.
I wonder if the results would differ significantly for the variants from
Belgium (septante and nonante instead of soixante-dix and
quatre-vingt-dix) and Switzerland (same, plus huitante instead of
quatre-vingt(s)).

--
Eric Lafontaine

Eric Angelini

unread,
Jun 15, 2012, 5:49:11 PM6/15/12
to
Good point, Éric (I live in Brussels, Belgium). My question about foreign languages aims at understanding if there is a kind of "number naming constant" for every language. Ken's discovery of an English such sentence that doesn't stop after _thousand_ iterations (or more) shakes me! Is the English number naming so different (in term of number names average length) from the French one that the longest possible sentence in French is only about 80 terms long?!
Best ans thanks for yr post,
É.

Eric Lafontaine

unread,
Jun 16, 2012, 10:25:46 AM6/16/12
to
From what I can figure, one major difference between English and French
is the -2 difference between the lengths of the consecutive 'quatre' and
'cinq'. In English, this does not happen until 'nineteen' and 'twenty'.
I would expect Dutch to behave as English, German as French ('sieben'
and 'acht').
Italian might be quite interesting (-3 between 'cinque' and 'sei').


--
-- <br>
Eric Lafontaine<br>
http://ventre-a-pattes.com

Eric Angelini

unread,
Jun 16, 2012, 1:10:41 PM6/16/12
to

> Dutch, German, Italian

... yes, I should post this problem there, on
Dutch, Getman or Italian speaking maths
forums. Thanks for your 4-5 remark!
Best,
E.

Eric Lafontaine

unread,
Jun 23, 2012, 5:19:34 PM6/23/12
to
Hello Ken

I guess the results might somewhat differ depending on the conventions
used for writing number words.
Using SEED = 22 and relying on English usage from
http://en.wikipedia.org/wiki/English_numerals (e.g. nine hundred and
ninety-nine thousand and nine hundred and ninety-nine), I can't reach
anything for EMBED in {16, 24, 35, 44, 56, 61, 87}.
By this, I mean the iterations are well beyond several thousand, so that
I am not even sure that my memory is fit to handle the integers.

For EMBED = 57 (same conventions), I get 999976 letters with 17478
steps. Not quite sure if the last few hundred are still correct. And
pretty sure it's useless, but there it is:
http://ventre-a-pattes.net/CARATEN.html

When I still believed there might be a limit, I planned to try Dutch and
German too, but seeing how (what I hope to be) English and at least two
variants of French can produce _very_ long (actually: unmanageable)
sequences, I doubt there is any use.

--
Éric Lafontaine

ken

unread,
Jun 23, 2012, 7:28:14 PM6/23/12
to
On Jun 24, 7:19 am, Eric Lafontaine <y...@ventre-apathique.com> wrote:
> On 2012-06-13 11:15, ken wrote:
>
> > On Jun 13, 5:55 pm, Eric Angelini <ericangelini...@gmail.com> wrote:
> >> Ok, many thanks again, Ken. I'm amazed by the fact that there is a branch in some tree that still grows after 1000 iterations, showing no halting sign! Just for my records: what are the lengthes of the "seed" and of the "enveloppe" of the said sentence? (this is: how many letters there are in
> >> the first selfrefering sentence and how many letters there are in the embedding phrasing like "this sentence, together with _ has _ letters") Please forgive my bad English, Ken -- I hope I am clear and that I've understood you're last message.
> >> Best,
> >> É.
>
> > I just used the suggested patterns:-
> > SEED  = "This sentence has - letters"  -22 letters
> > EMBED = "This sentence, together with -, has - letters"  -34 letters
>
> > The 1000+ iterations are for EMBED=45
> > eg "This long English sentence, together with -, has - letters
>
> > -ken
>
> Hello Ken
>
> I guess the results might somewhat differ depending on the conventions
> used for writing number words.
> Using SEED = 22 and relying on English usage fromhttp://en.wikipedia.org/wiki/English_numerals(e.g. nine hundred and
> ninety-nine thousand and nine hundred and ninety-nine), I can't reach
> anything for EMBED in {16, 24, 35, 44, 56, 61, 87}.
> By this, I mean the iterations are well beyond several thousand, so that
> I am not even sure that my memory is fit to handle the integers.
>
> For EMBED = 57 (same conventions), I get 999976 letters with 17478
> steps. Not quite sure if the last few hundred are still correct. And
> pretty sure it's useless, but there it is:http://ventre-a-pattes.net/CARATEN.html
>
> When I still believed there might be a limit, I planned to try Dutch and
> German too, but seeing how (what I hope to be) English and at least two
> variants of French can produce _very_ long (actually: unmanageable)
> sequences, I doubt there is any use.
>
> --
> Éric Lafontaine

Hello EricThe conventions of writing large numbers in English do vary
from UK to US.
my Australian English (mostly UK English) uses commas between 'the
triplets'
eg 123,456,789 is one hundred and twenty-three million, four hundred
and fifty-six thousand, seven hundred and eighty-nine.
and (just to make it complicated) , there is an exception: when there
are no hundreds in the last triplet use 'and' anyway
eg1,011,011 is one million, eleven thousand, and eleven

I will redo my very slow program, I spent most of my time fiddling
with this stuff and not enough on speeding up the searches.
-ken

Eric Lafontaine

unread,
Jun 24, 2012, 2:08:58 AM6/24/12
to
On 2012-06-24 01:28, ken wrote:
[...]
>
> Hello EricThe conventions of writing large numbers in English do vary
> from UK to US.
> my Australian English (mostly UK English) uses commas between 'the
> triplets'
> eg 123,456,789 is one hundred and twenty-three million, four hundred
> and fifty-six thousand, seven hundred and eighty-nine.
> and (just to make it complicated) , there is an exception: when there
> are no hundreds in the last triplet use 'and' anyway
> eg1,011,011 is one million, eleven thousand, and eleven
>
> I will redo my very slow program, I spent most of my time fiddling
> with this stuff and not enough on speeding up the searches.
> -ken
>

Hello Ken

I have relied on the UK variant, without caring for the exception you
mention. Didn't use commas either, but that would be irrelevant since I
do not regard them as letters.
I'll look at it again on a smaller set: I am currently generating names
and letter counts up to 250 million (practical limit set by the
compiler), but the result files grow so large that it doesn't make sense
to go beyond one million.
Still, I expect quite a lot of combinations to go through the roof.

--
Éric Lafontaine

Eric Lafontaine

unread,
Jun 24, 2012, 6:17:37 AM6/24/12
to
On 2012-06-24 01:28, ken wrote:
> Hello EricThe conventions of writing large numbers in English do vary
> from UK to US.
> my Australian English (mostly UK English) uses commas between 'the
> triplets'
> eg 123,456,789 is one hundred and twenty-three million, four hundred
> and fifty-six thousand, seven hundred and eighty-nine.
> and (just to make it complicated) , there is an exception: when there
> are no hundreds in the last triplet use 'and' anyway
> eg1,011,011 is one million, eleven thousand, and eleven
>
> I will redo my very slow program, I spent most of my time fiddling
> with this stuff and not enough on speeding up the searches.
> -ken
>

I redid it differently, no longer looking at the full set of solutions
but only at improvements (within one pair seed-embedded).

For seeds between 11 (I got ... letters) and 255, embedded between 12
and 255, and not checking beyond one million letters or so, I end up
with 1311 pairs that do not complete, the number of iterations at that
moment ranging from 3917 (13 times) to 82717 (seed 12 "I have ...
letters", embedded 14 "I and ... got ... letters").

No result at all for seed > 104, but there does not seem to be any such
limitation for embedded.

--
Éric Lafontaine

Eric Lafontaine

unread,
Jun 24, 2012, 10:35:42 AM6/24/12
to
On 2012-06-24 01:28, ken wrote:
>
> Hello EricThe conventions of writing large numbers in English do vary
> from UK to US.
> my Australian English (mostly UK English) uses commas between 'the
> triplets'
> eg 123,456,789 is one hundred and twenty-three million, four hundred
> and fifty-six thousand, seven hundred and eighty-nine.
> and (just to make it complicated) , there is an exception: when there
> are no hundreds in the last triplet use 'and' anyway
> eg1,011,011 is one million, eleven thousand, and eleven
>
> I will redo my very slow program, I spent most of my time fiddling
> with this stuff and not enough on speeding up the searches.
> -ken
>

Oops

Initialisation error.
Seed 12 and embed 14 provide only 15000 or so iterations, and the same
must apply to the rest.
Comes a bit closer to French, where I get just short of 20000 iterations
for one million letters (over the whole range).
--
Éric Lafontaine

Eric Angelini

unread,
Jun 25, 2012, 3:53:55 PM6/25/12
to
Many thanks Eric and Ken for
your impressive results! And,
Eric, for the "ventre-a-pattes"
link -- which contradicts the
claim of Jean-Luc Piedanna, BTW.
I'll contact JLP next week and
will show him yr French (Belgium)
sequence...
Best
'É.

Eric Lafontaine

unread,
Jun 25, 2012, 4:19:13 PM6/25/12
to
> '�.
>

I have contacted Jean-Luc as well.
No longer any plans to handle the Belgian variant so far: Franch, Swiss
and English are so similar that I don't expect Belgian to be any
different (nor any Germanic, Latin, Scandinavian, and so on). Chinese is
another cup of tea, but only because the size of the literals grows much
slower than in Western languages, so that putting a limit at one million
"characters" allows many more steps (best I could reach is 55746) before
hitting the wall. Got my disk full that way (starting with 300 GB free
space ...)

--
�ric Lafontaine
0 new messages