Lettera Matematica

https://doi.org/10.1007/s40329-017-0205-1

The twin prime conjecture and other curiosities regarding prime

numbers
Renato Betti1

© Centro P.RI.ST.EM, Università Commerciale Luigi Bocconi 2017

Abstract
The paper begins with a reference to Riemann’s hypothesis on the sequence of prime numbers, still unproven today, and goes
on to illustrate the twin prime conjecture and the more general Polignac’s conjecture; we then recount the recent result by
Yitang Zhang about it, and the improvements obtained thanks to the online mathematical collaboration called Polymath 8.

Keywords Prime numbers · Riemann’s hypothesis · Twin primes · Polignac’s Conjecture · Bounded gaps between prime
numbers

Since our earliest years at school we study the fundamental and then continued his argument about a precisely accurate
operations of arithmetic: addition and multiplication. Then, description of the sequence of prime numbers, which entails
when we meet the prime numbers, we discover the strange considerable implications not only for number theory, but
and magical feeling of an integer sequence so easy to define, for the whole of mathematics and physics. Since then, gen-
and yet so hard to see in its internal development from one erations of mathematicians have been struggling to find a
member to the next one, at least if we examine it some- proof for it.
what later in the sequence. Who among us has not tried to The Riemann hypothesis is now the keystone of number
find some pattern, a law, a rule? Who was not fascinated by theory and its history, and the famous list of 23 problems
a strange and unexpected property? Perhaps this happens posed by Hilbert in 1900, showing great foresight, is the best
because prime numbers are defined in terms of what they illustration of this. Today, only the few of Hilbert’s prob-
are not: factorable into smaller integers. Perhaps because lems that are considered too general or more “rhetorical”
they often have delicate additive features even though they than mathematical are open, while the Riemann hypothesis
are defined by a multiplicative condition. remains unproven, so much so that it has been promoted
Elusive and intriguing, prime numbers display many fac- to one of “Millennium Prize Problems” by the Clay Math-
ets, several of which are only perceived, but have yet to be ematics Institute, which, moreover—a sign of new times for
proved, and thus we are left with conjectures. Of course, mathematicians too—assigned a 1 million dollar prize for
some of them are important and decisive, pursued for more anyone able to prove that it is true or—as the experts judge
than a century by specialists. One of these is the famous to be highly unlikely—false.
“Riemann hypothesis”, formulated with seeming noncha- Here I certainly do not intend to address this topic, which
lance in a single “little article” about number theory— would require other tools and other skills. Let us remain
indeed, it is just 10 pages long —“On the number of primes tranquilly in elementary theory, where we glimpse flashes
less than a given magnitude” [18], which soon become one of the amazing virtues of prime numbers while remaining
of the greatest contributions to the subject. The great math- sheltered from the usual language of arithmetic; where con-
ematician remarked in 1859 that he deemed the property he jectures seem simple curiosities that allow us a glance at the
had studied to hold, but that he had temporarily put it aside, overall intricate structure but, in their apparent simplicity, all
the same have remarkable surprises in store for us.
* Renato Betti One of the most famous unsolved problems about
renato.betti@polimi.it prime numbers concerns pairs of “twins”,1 that is, prime
1 1
Dipartimento di Matematica, Politecnico di Milano, Piazza According to Heinrich Tietze [20], the term “twin primes” was
Leonardo da Vinci 32, 20133 Milan, Italy coined by Paul Stäckel in 1916.

13
Vol.:(0123456789)
 Lettera Matematica

numbers pn and pn+1 whose difference is the least possible:

pn+1 − pn = 2.
It is easy to show that all twin primes, with the exception
of the first pair (3,5), are of the form 6 k ± 1. Indeed, an odd
number p has odd remainder when it is divided by 6 and, if
it is prime, the remainder cannot be 3, otherwise the number
would be divisible by 3. Thus we necessarily have p ≡ ±1
(mod 6) and, if p and p + 2 are both prime, they must be of
the form 6k–1 and 6k + 1.
This is a first, very simple property, but it is not enough
to give an idea of the distribution of pairs of twin primes, so,
since today it is possible, we resort to heuristics: automatic
computation. The first pairs are identified right away:
Fig. 1  The function 𝜋2 (n) for 0 < n < 1000
(3, 5), (5, 7), (11, 13), (17, 19), … , (101, 103), (107, 109), …
So that some deem it unlikely that ancient Greeks, so
attentive to mystical and magical properties of numbers, did to use even a pocket calculator—to find pairs higher than 1­ 09
not notice this. Yet, nobody talks about it and—it seems—in (1 billion)? Here are some of them:
the scientific literature, an explicit early reference to twin
primes only appears, and in more general terms, in 1849 by (1, 000, 000, 007, 1, 000, 000, 009)
Alphonse de Polignac,2 a French mathematician especially (1, 000, 000, 409, 1, 000, 000, 411)
known for the conjecture that bears his name:
…
Polignac’s Conjecture For every number k > 0, there are
infinitely many pairs of consecutive primes whose difference (1, 000, 001, 801, 1, 000, 001, 803).
is 2k.
Thus, if k = 1, the conjecture claims that there are infi- Not to mention the recently attained record (September
nitely many pairs of twin primes; if it were false, Polignac’s 2016): the largest pair of twin primes, as of this moment, is
whole machinery would fall with it. The result that—in some given by
sense—arrives closest to this conjecture has arrived much
2, 996, 863, 034, 895 … 21290000 ± 1,
later, and is due to the Chinese mathematician Jing-Run
Chen [2]: two numbers with no less than 388,342 digits!3
Chen’s Theorem If k is any positive number, then there At this point it is hard not to believe that the twin prime
are infinitely many primes p such that p + 2k is either prime conjecture is true, as it seems to be confirmed not only by the
or semiprime, that is, the product of at most two primes. values found up to now, but also by the trend of the function
It is worth noting that, in the same paper and with similar 𝜋2 (n) that describes them: 𝜋2 (n) equals the number of pairs
methods, Chen also comes close to the famous “Goldbach of twin primes whose smaller element is ⩽ n. Fig. 1 shows
conjecture”, showing that every “large enough” even number its behaviour when n is small (1 ⩽ n ⩽ 1000).
is the sum of either two primes or a prime and a semiprime. We have 𝜋2 (1000) = 35 and, looking at the graph, one
The value from which this property begins to hold is only may even be tempted to guess that, at some point, the func-
estimated at an incredibly large number. tion becomes constant. But then, when we consider how
Let us go back to twin primes. It was not easy for the 𝜋2 (n) varies on a larger scale, some doubt might appear (see
ancients to have an idea of what happens later in the Fig. 2). ( )
sequence of numbers and only now, thanks to computing, are Here 𝜋2 106 = 8169, and the growth is regular, with
we able to find pairs of very large twin primes. For instance, this trend, up to the maximum computed so far for n = 1016.
how long would it take with pen and paper—and I allow you To be precise [5]:
𝜋2 1016 = 10, 304, 195, 697, 298.
( )

At this point, the empirical belief that the twin prime

conjecture is valid is strengthened, but we have no proof.
Further, you can never tell with such large numbers: just
2
In his paper “Recherches nouvelles sur les nombres premiers”
[15]; it is worth noting that at the address https://books.google.fr/
books?id=O6EKAAAAYAAJ&hl=it a same-titled pamphlet by Pol-
3
ignac is available [16]. See http://www.primegrid.com/forum_thread.php?id=7021.

13
Lettera Matematica

One might think that, basically, the pairs of twins become

increasingly rare and therefore it is natural that the series of
their inverses is convergent. Yes, but this leads us to a com-
parison with the prime numbers, which are likewise known
to be increasingly rare, and yet we have known since Euler’s
time that the series of the inverses of all prime numbers is
divergent:
∑ 1
= ∞.
p prime
p

So we are left with the problem. Recently, after a long

time of stasis with regard to the twin prime conjecture,
Fig. 2  The function 𝜋2 (n) for 0 < n < 106
there is something new and interesting, both for the result
in itself and the method by which it was obtained. We cover
remember some amazing historical cases as, for instance, the method—an online collaboration, open to everybody—in
the comparison between the 𝜋(n) function, which counts the the “Appendix” to this paper, while we make you wait a bit
n
number of primes less than n, and the function Li(n) = ∫ dt
, for the result, in order to digress on Polignac’s conjecture:
2
lnt
fruits taste better when you know the tree they come from.
conjectured by Gauss as the best estimate for it (a true fact, First, let us look at some other value of the conjecture. What
proved in 1896, independently, by de la Vallée-Poussin and happens for k = 2? And for k = 3? In other words, how many
Hadamard in the celebrated prime number theorem). pairs are there of consecutive primes whose difference is 4—
Well, numerical evidence always suggested the inequality which have the rather dull name of “cousin primes”—such
𝜋(n) < Li(n), and Gauss himself was sure of it, until in 1914, as (7, 11), (13, 17) or (19, 23)? And what if the difference
to everyone’s surprise, John Edensor Littlewood [9] showed is 6—with the even duller name “sexy primes”, with an uno-
that the inequality changes sign infinitely many times, and riginal wordplay on sex, Latin for 6 –, for instance (23, 29)
in 1933 his student Stanley Skewes [19] fixed, for the num- and (31, 37)?
ber for which the sign change occurs. This upper limit was In these cases Polignac’s conjecture still holds, and the
later reduced: through a powerful numerical computation, efforts are focussed on the results that can be obtained by
Sherman Lehman [8] proved in 1966 that the phenomenon using a computer. So researchers are computing the values
occurs for sure for numbers that have 1166 decimal digits of 𝜋4 (n), the number of pairs of cousin primes ⩽ n, and 𝜋6 (n),
[3]. Today, perhaps, we have obtained a more precise limit: the number of sexy pairs ⩽ n, to discover that twins and
in any case not for numbers lesser than 1­ 019, as shown by cousins seem to have the same “asymptotic density” with
Jan Büthe [1] in 2015. The “high-flying” behaviour often respect to all prime numbers, in the sense that, within the
escapes earthly considerations. limits of the possible calculations, the two functions have
However, it is not only work with numbers large enough the same behaviour as n increases. Instead, there are many
to be beyond computation that leaves doubts about the con- more sexy primes and their density is estimated to be double
jecture. There is also a specific fact that encourages them, that of twins and cousins, as can be directly seen, at least for
since it clearly expresses that the pairs of twin primes thin small enough values of n. Fig. 3 shows the comparison up
out rapidly: in 1919, as a consequence of the prime number to n = 106: the upper curve indicates the sexy pairs, while
theorem, the Norwegian mathematician Viggo Brun proved in the lower ones, at this scale, twin and cousin pairs are
that the sum, possibly infinite, of the inverses of the twin indistinguishable.4
primes is finite: Some researchers consider two prime numbers to be cous-
ins, or sexy, even if they are not consecutive. For instance,
1 1 1 1 1 1
( ) ( ) ( )
+ + + + + + … = B. they also take as cousins 3 and 7, even though they have
3 5 5 7 11 13 the common twin 5, putting at serious risk the structures of
To be precise, the value B, which is now appropriately kinship; some even consider the pair (5,11) to be sexy, to the
called “Brun’s constant”, has been estimated heuristically: embarrassment of 7, twin of 5 and cousin of 11. The calcu-
B = 1.9021605778.. [13, 14]. Wonderful! We do not know lations abound: it turns out that, as of of 2005, the largest
whether the sum has a finite or infinite number of sum- 4
For more on these topics, see the papers by Bruno Martin [10–12].
mands, but we can say it is a finite number, and—within 5
Calculation attributed to T. Alm, M. Fleuren, and J. K. Andersen by
the precision allowed by computers—we are even able to http://mathworld.wolfram.com/CousinPrimes.html and http://math-
approximate ten of its decimal digits! world.wolfram.com/SexyPrimes.html.

13
 Lettera Matematica

Fig. 3  The functions 𝜋2 (n) and

𝜋4 (n) (lower graph) and 𝜋6 (n)
(upper graph) for 0 < n < 1016

cousin primes and sexy primes are numbers with 10,154 (11, 13, 17), (41, 43, 47) or (101, 103, 107), all of which
digits.5 An analogue of Brun’s constant is defined for the have the same final digits?7
cousin primes: With the triples of sexy primes, we begin to enter a large
landscape; starting from (31, 37, 43), we discover that two
1 1 1 1 1 1 1 1
( ) ( ) ( ) ( )
+ + + + + + + + … = B4 . generations at the beginning and the end of last century
7 11 13 17 19 23 37 41
have lived through into sexy triples: (1901, 1907, 1913) and
(omitting the initial term of 1/3 + 1/7). Its value is esti- (1987, 1993, 1999); there were no others in that century.
mated, using the terms up to 242 , to B4 = 1.1970449 … This is nothing, however, compared to what happened in
Other cases have been examined and some surprises have the eighteenth century, with a sexy set of 4 years: (1741,
been discovered. The pairs given by k = 5 are not as rare as 1747, 1753, 1759).
one might think: But let us proceed in a more orderly way. For twin primes,
one immediately finds the triple (3, 5, 7) and understands
(3, 13), (7, 17), (13, 23), (19, 29), (31, 41), (37, 47), …
that there may be no others, since one of the three numbers
Are there infinitely many? Everybody can go search would be divisible by 3. The same applies to the set of three
whether there are any greater than 1000.6 But Polignac’s cousins (3, 7,11), and an analogue for the quintuple of sexy
conjecture can be generalised: it is known that mathemati- numbers (5, 11, 17, 23, 29): given five such numbers, one of
cians are very inclined towards this type of activity. This them is always divisible by 5, as we can see from the remain-
way we reach a whole series of conjectures due to Hardy and ders modulo 5 (shown on the rows of Table 1, for p > 5).
Littlewood and dating back to the 1920s, which we will not This leads to a necessary condition for building k-ples
delve into here. But it is worth following for a while a simple of prime numbers: the whole family of even integers
extension of Polignac’s conjecture: why limit ourselves to a1 < a2 < … < ak−1 (such as 6, 12, 18 and 24 in the last
two primes at a time? What about the triples of consecutive example) is said to be “admissible” if it does not contain all
primes at a distance 2 k from each other? And the 4-tuples? the possible remainders of the division by p, whatever the
And so on. And why not also admit k-tuples of consecutive prime number p. In particular, it is sufficient to verify that
primes at different distances between them, as in the case this happens for all primes p ⩽ k , since there are at most
k–1 different remainders and we must take into account the
initial value.
6
A partial answer: (1009, 1019), (1021, 1031), (1039, 1049), (1051,
1061), …
7
We also have (1301, 1307, 1309). Are there other ones of the form
(10n + 1, 10n + 3, 10n + 7)?

13
Lettera Matematica

Table 1  Remainders modulo 5 for 5-tuples of sexy primes The venture of this method of collaboration is relatively
p p+6 p + 12 p + 18 p + 24
new and opens up a whole host of problems and questions:
we discuss it briefly in the “Appendix”. Introduced by the
1 2 3 4 0 mathematician Timothy Gowers in 2009 under the collective
2 3 4 0 1 label of Polymath to study other topics, in the case of the
3 4 0 1 2 upper limit between consecutive primes it soon produced
4 0 1 2 3 excellent results: thanks to a series of improvements in the
original methods and new ideas that appeared later, this pro-
ject, called Polymath 8, was able in a few months to lower
Experts in number theory suggest optimistically that the limit from 70,000,000 to 246.
when these trivial obstructions do not hold, the condition What a strange number! It is hard to believe that this is
is also sufficient and therefore there are endless k-ples of the final answer: 246 is certainly not a kind of impassable
primes: integer constant, such as 5 for the degree of algebraic equa-
tions that are solvable algebraically, 17 for the number of
Conjecture If a1 < a2 < ... < ak−1 is an admissible set of wallpaper groups or 26 for the sporadic simple groups. Per-
even integers, there are infinitely many primes p such that haps it is the limit that can be achieved with current tech-
p + a1, p + a2,..., p + ak−1are all prime numbers. niques... Let us wait and see. For now, it has not been pos-
sible to improve it in any way; then again, just a short time
We have no precise results. It has been proved only has elapsed.
recently that there is an upper limit N for which there are So, here is the (currently) final result:
infinitely many pairs pn and pn+1 of consecutive prime num- There are infinitely many consecutive primes whose dif-
bers such that pn+1 − pn ⩽ N . A possible value for N has ference is no greater than 246.
also been calculated, namely N = 70,000,000. This is a—so Translated from the Italian by Daniele A. Gewurz.
to speak—“weak” version of the twin primes conjecture:
there are infinitely many pairs of consecutive primes with
the property that their difference is 2 or 4 or 6 or ... up, up Appendix
to 70 million.
The author of this result is the US-based Chinese math- New forms of mathematical research?
ematician Yitang Zhang [21], who was previously little
known for his work on these topics and who worked in com- In recent times, largely due to computing tools, email and
plete isolation, grafting original techniques on known works. so on, mathematical research is increasingly taking place
Of course, when the result was announced, some raised in collaboration. It is easy, and sometimes even pleasant, to
doubts that he could have succeeded precisely where many communicate our ideas to colleagues, ask for a suggestion
had failed, but the referee of Annals of Mathematics, the and perhaps discover unforeseen and unexpected contribu-
journal to which the paper had been submitted, was clear: tions. How far is it possible to extend this form of collabora-
“The main results are of the first rank ... The author has suc- tion and to what extent can it be planned?
ceeded to prove a landmark theorem in the distribution of Of course, there are precedents: we all know the story of
prime numbers”, Nature reported in 2013, even before the Bourbaki, a group of people who decided to work together
appearance of the paper.8 to refound the principles and tools of mathematics; we are
This result caused great excitement among specialists. Of familiar with the quest that in the last century led to the clas-
course, the limit value for N is very high, but reaching a gap sification of finite simple groups through the collaboration of
between 2 and 70,000,000 is a huge conceptual leap with many mathematicians; not to mention the research involving
respect to that between 2 and infinity that was known previ- the massive, well-coordinated use of a large computer net-
ously. And then, knowing the existence of an upper limit, it work, for instance in number theory—prime numbers, twin
can certainly be improved—as Zhang himself admitted since numbers etc., such as the efforts discussed in the article—or
the beginning—perhaps through the use of computing. And aimed at decrypting complex cryptographic systems.
so it has been, with the addition of an original method of Even in the most complex cases it had been possible to
work: intensive collaboration by means of a public discus- recognise that the problem could be naturally divided into a
sion forum on the Internet, with the participation of several number of sub-problems, perhaps several ones, but in a way
researchers. that implies a “hierarchical”, fairly clear organisation. In
2009, the mathematician Timothy Gowers, a Fields Medal
8
https://www.nature.com/news/first-proof-that-infinitely-many- winner in 1998, proposed a new model of a forum open to
prime-numbers-come-in-pairs-1.12989. all, not in order to make useful and fine debates—ok, for this

13
 Lettera Matematica

too—but especially to address well-defined mathematical References

research problems without such a structural character [7].
And the proposal obtained an unexpected success. 1. Büthe, J.: An analytic method for bounding ψ(x).
The first project, consisting in the search for an elemen- arXiv:1511.02032 [math.NT] (2015)
2. Chen, J.R.: On the representation of a large even integer as
tary proof of the Hales–Jewett theorem, a combinatorial the sum of a prime and the product of at most two primes. Sci.
result that was well known but hard to grasp, was success- Sinica. 16, 157–176 (1973)
ful after a few weeks, thanks to the collaboration of dozens 3. Conway, J.H., Guy, R.K.: The book of numbers. Springer, New
of mathematicians, including graduate students. Subsequent York (1996)
4. Cranshaw, J., Kittur, A.: The Polymath Project: Lessons from
projects have had varying degrees of success, on occasions a Successful Online Collaboration in Mathematics. Proc.
with partial results, but sometimes excellent ones, as in the SIGCHI Conf. on Human Factors in Computing Systems, ACM,
case of project no. 8, attempting to improve Zhang’s result 1865–74: (2011), available at: http://www.cs.cmu.edu/~jcransh/
on the difference between consecutive primes.9 The idea is papers/cranshaw_kittur.pdf. Accessed 31 Oct 2017
5. Gourdon, X., Sebah, P.: Introduction to twin primes and Brun’s
that anyone who has something to say about the problem constant (2002). Available at: http://numbers.computation.free.
should have no qualms about participating. Of course, eve- fr/Constants/Primes/twin.html. Accessed 31 Oct 2017
rybody should make sure to express their comments—what- 6. Gowers, T.: Is massively collaborative mathematics possible?
ever form they take—in a reasonable time and in concise Gowers’s Weblog: (2009), available at: https://gowers.word-
press.com/2009/01/27/is-massively-collaborative-mathematics-
terms, because the aim is not to write a long and detailed possible/. Accessed 31 Oct 2017
paper but to encourage the debate. The project itself raised 7. Gowers, T., Nielsen, M.: Massively collaborative mathematics.
a number of issues, such as who gets credit for the results Nature. 461, 879–881 (2009)
and the papers to which it gives rise, how to acknowledge 8. Lehman, R.S.: On the difference π(x)-Li(x). Acta Arith. 11,
397–410 (1966)
those who participated only partially and so on. For now, 9. Littlewood, J.E.: Sur la distribution des nombres premiers.
the temporary solution is to sign the articles under the pseu- Comptes Rendus. 158, 1869–1872 (1914)
donym D.H.J. Polymath and preserve the whole history so 10. Martin, B.: Des jumeaux dans la famille des nombres premiers
that the contribution of everyone is clear (the “name” D.H.J. I. Images des mathématiques, 20 March: (2015), available at:
http://images.math.cnrs.fr/Des-jumeaux-dans-la-famille-des-
comes from the initial project, Polymath 1, about the already nombres-premiers-I. Accessed 31 Oct 2017
mentioned “density” Hales-Jewett theorem). 11. Martin, B.: Des jumeaux dans la famille des nombres premiers
But the real challenge—it seems to me—is that of organ- II. Images des mathématiques, 21 June: (2015), available at:
ising work, starting with a problem that lends itself to be http://images.math.cnrs.fr/Des-jumeaux-dans-la-famille-des-
nombres-premiers-II. Accessed 31 Oct 2017
worked on collectively, to provide for the action of a mod- 12. Martin, B.: Des jumeaux dans la famille des nombres premiers
erator and methods to distinguish, since the beginning, III. Images des mathématiques, 20 September: (2015). http://
valid contributions from useless ones or even deliberate images.math.cnrs.fr/Des-jumeaux-dans-la-famille-des-nombres-
obstruction. In short, how to organise a genuine collective premiers-III. Accessed 31 Oct 2017
13. Nicely, T.: Enumeration to ­1014 of the twin primes and Brun’s
research.10 constant. Virginia J. of Science 46, 195–204: (1996). http://
For now it seems that there have not been too many prob- www.trnicely.net/twins/twins.html; Accessed 31 Oct 2017
lems, thanks partly to the fact that the filtering of the vari- 14. Nicely, T.: A new error analysis of Brun’s constant. Virginia
ous contributions was done by great mathematicians. The J. of Science 52, 45–55: (2001). http://www.trnicely.net/twins/
twins4.html. Accessed 31 Oct 2017
advantages of this approach to research are obvious. In the 15. Polignac, A.: Recherches nouvelles sur les nombres premiers.
words of the man who proposed it: “if a large group of math- Comptes Rendus Paris 29, 400 (1849) (738–39 [Rectification])
ematicians could connect their brains efficiently, they could 16. Polignac, A.: Recherches nouvelles sur les nombres premiers.
perhaps solve problems very efficiently as well” [6]. Bachelier, Paris (1851)
17. Polymath, D.H.J.: The “bounded gaps between primes” Poly-
At present, there are few studies about this original math project—a retrospective. arXiv:1409.8361 [math.HO]
attempt. I merely call the reader’s attention to the paper [4]. (2014)
18. Riemann, B.: Über die Anzahl der Primzahlen unter einer gege-
benen Grosse. Monatsberichte der Berliner Akademie, 671–680
(1859)
19. Skewes, S.: On the difference π(x)–li(x). J. Lond. Math. Soc. 8,
277–283 (1933)
20. Tietze, H.: Famous problems of mathematics: solved and
unsolved mathematics problems from antiquity to modern times.
Graylock Press, Baltimore (1965)
9 21. Zhang, Y.: Bounded gaps between primes. Ann. Math. 179,
In [17] the opinions of numerous researchers who participated in
1121–1174 (2014)
the project are collected.
10
The post that originated this adventure [6] contains a series of rec-
ommendations in this regard, in order to encourage collaboration and
facilitate the work of those who coordinate it.

13
Lettera Matematica

Renato Betti is a former professor Angeli 2015). He is an ordinary member of the Accademia Nazionale
of geometry at the Politecnico di Virgiliana in Mantua.
Milano. His research concerns
category theory and its applica-
tions to algebra and geometry.
His recent publications include
La matematica come abitudine
del pensiero. Le idee scientifiche
di Pavel Florenskij (Libri del
Pristem, 2009) and Geometria
leggera. Introduzione all’idea di
spazio matematico (Franco

13