Almost Integer

DOWNLOAD Mathematica Notebook

An almost integer is a number that is very close to an integer.

Surprising examples are given by

 sin(11)=-0.999990206...,
(1)

which equals -1 to within 5 digits and

 sin(2017RadicalBox[2, 5])=-0.9999999999999999785...,
(2)

which equals -1 to within 16 digits (M. Trott, pers. comm., Dec. 7, 2004). The first of these comes from the half-angle formula identity

 sin^211=1/2(1-cos22),
(3)

where 22 is the numerator of the convergent 22/7 to pi, so cos22 approx cos(7pi)=cospi= approx -1. It therefore follows that any pi approximation x gives a near-identity of the form cosx approx -1.

Another surprising example involving both e and pi is

 e^pi-pi=19.999099979...,
(4)

which can also be written as

(pi+20)^i=-0.9999999992-0.0000388927i
(5)
cos(ln(pi+20)) approx -0.9999999992.
(6)

Here, e^pi is Gelfond's constant. Applying cosine a few more times gives

 cos(picos(picos(ln(pi+20)))) approx -1+3.9321609261×10^(-35).
(7)

This curious near-identity was apparently noticed almost simultaneously around 1988 by N. J. A. Sloane, J. H. Conway, and S. Plouffe, but no satisfying explanation as to "why" e^pi-pi approx 20 is true has yet been discovered.

Another nested cosine almost integer is given by

 2cos(cos(cos(cos(cos(cos(cos5))))))^2 
 =0.999995254797000...
(8)

(P. Rolli, pers. comm., Feb. 19, 2004).

An example attributed to Ramanujan is

 22pi^4=2143+2.748...×10^(-6).
(9)

Some near-identities involving integers and the logarithm are

510log_(10)7=431.00000040...
(10)
88ln89=395.00000053...
(11)
272log_pi97=1087.000000204...,
(12)

which are good to 6, 6, and 6 decimal digits, respectively (K. Hammond, pers. comm., Jan. 4 and Mar. 23-24, 2006).

An interesting near-identity is given by

 1/4[cos(1/(10))+cosh(1/(10))+2cos(1/(20)sqrt(2))cosh(1/(20)sqrt(2))]=1+2.480...×10^(-13)
(13)

(W. Dubuque, pers. comm.).

Near-identities involving e and pi are given by

 e^6-pi^4-pi^5=0.000017673...
(14)

(D. Wilson, pers. comm.),

 (pi^9)/(e^8)=9.9998387...
(15)

(D. Ehlke, pers. comm., Apr. 7, 2005),

 10tanh((28)/(15)pi)-(pi^9)/(e^8) approx 6.005×10^(-9)
(16)

(Povolotsky, pers. comm., May 11, 2008), and

 (e^pi-ln3)/(ln2)-4/5=31.0000000033...
(17)

(good to 8 digits; M. Stay, pers. comm., Mar. 17, 2009), or equivalently

 (10(e^pi-ln3))/(ln2)=318.000000033...,
(18)

Other remarkable near-identities are given by

 (5(1+sqrt(5))[Gamma(3/4)]^2)/(e^(5pi/6)sqrt(pi))=1+4.5422...×10^(-14),
(19)

where Gamma(z) is the gamma function (S. Plouffe, pers. comm.),

 ln2+log_(10)2=0.994177...
(20)

(D. Davis, pers. comm.),

 (163)/(ln163)=31.9999987384...
(21)

(posted to sci.math; origin unknown),

 eC^(5/7-gamma)pi^(-(2/7+gamma)) approx 1.00014678
(22)
 (C^(gamma-19/7)pi^(2/7+gamma))/(2phi) approx 1.00105
(23)
 egammaphi(Cpi)^(-(2/7+gamma)) approx 1.01979,
(24)

where C is Catalan's constant, gamma is the Euler-Mascheroni constant, and phi is the golden ratio (D. Barron, pers. comm.), and

 163(pi-e)=68.999664...
(25)
 (53453)/(ln53453)=4910.00000122...
(26)
[(2-1)^2+((5^2-1)^2)/(6^2+1)]e-[(2+1)^2+((5^2+1)^2)/(6^2-1)]^(-1)=(613)/(37)e-(35)/(991)
(27)
=44.99999999993962...
(28)

(E. Stoschek, pers. comm.). Stoschek also gives an interesting near-identity involving the fine structure constant alpha and Feigenbaum constant delta,

 (28-delta^(-1))(alpha^(-1)-137) approx 0.999998.
(29)

E. Pegg Jr. (pers. comm., Mar. 4, 2002) discovered the interesting near-identities

 ((91)/(10))^(1/4)-(33)/(19)=3.661378...×10^(-8)
(30)

and

 ((23)/9)^5=(6436343)/(59049) approx 109.00003387.
(31)

The near-identity

 3sqrt(2)(sqrt(5)-2)=1.0015516...
(32)

arises by noting that the stellation ratio 3(sqrt(5)-2) in the cumulation of the dodecahedron to form the great dodecahedron is approximately equal to 1/sqrt(2). Another near identity is given by

 zeta(3) approx gamma^(-1/3)+pi^(-1/4)(1+2gamma-2/(130+pi^2))^(-3),
(33)

where zeta(3) is Apéry's constant and gamma is the Euler-Mascheroni constant, which is accurate to four digits (P. Galliani, pers. comm., April 19, 2002).

J. DePompeo (pers. comm., Mar. 29, 2004) found

 (5phie)/(7pi)=1.0000097...,
(34)

which is equal to 1 to five digits.

M. Hudson (pers. comm., Oct. 18, 2004) noted the almost integer

 lnK-lnlnK=1.0000744...,
(35)

where K is Khinchin's constant, as well as

 (sqrt(45))^gamma=3.000060964...,
(36)

(pers. comm., Feb. 4, 2005), where gamma is the Euler-Mascheroni constant.

M. Joseph found

 erfi(erfi(1/3sqrt(3)))=1.0000208...,
(37)

which is equal to 1 to four digits (pers. comm., May 18, 2006). M. Kobayashi (pers. comm., Sept. 17, 2004) found

 10(gamma^(-1/2)-1)^2=0.9999980...,
(38)

which is equal to 1 to five digits. The related expression

 (10)/(81)(11-2sqrt(10))-gamma=-2.72×10^(-7),
(39)

which is equal to 0 to six digits (E. Pegg Jr., pers. comm., Sept. 28, 2004). S. M. Edde (pers. comm., Sep. 7, 2007) noted that

 exp[-psi_0(1/4(2+sqrt(3)))]=1.99999969...,
(40)

where psi_0(x) is the digamma function.

E. W. Weisstein (Mar. 17, 2003) found the almost integers

2.78768×10^(-6) approx 7/(64)ln2-(131)/(1728)
(41)
2.84186×10^(-6) approx (80497)/(40320)-(43)/(144)pi^2+(3293)/(1260)ln2-(43)/(24)(ln2)^2
(42)
9.80710×10^(-6) approx (2411287)/(30240)-(100)/9pi^2+(1877)/(21)ln2-(200)/3(ln2)^2
(43)

as individual integrals in the decomposition of the integration region to compute the average area of a triangle in triangle triangle picking.

ln2 and 3^(1/3) give the almost integer

 1/(3^(1/3)ln2)=1.00030887...
(44)

(E. W. Weisstein, Feb. 5, 2005).

Prudnikov et al. (1986, p. 757) inadvertently give an almost integer result by incorrectly identifying the infinite product

 product_(k=1)^infty(1-e^(-2pik/sqrt(3)))=(e^(-2pi/sqrt(3)))_infty,
(45)

where (q)_infty is a q-Pochhammer symbol, as being equal 3^(1/4)e^(-pi/(6sqrt(3))), which differs from the correct result by

 3^(1/4)e^(-pi/(6sqrt(3)))-(e^(-2pi/sqrt(3)))_infty approx 1.82668×10^(-5).
(46)

A much more obscure almost identity related to the eight curve is the location of the jump in

 Pi(1/2i(i+sqrt(7));isinh^(-1)(sqrt(-(2i)/(i+sqrt(7)))tant),k),
(47)

where

 k=sqrt((i+sqrt(7))/(i-sqrt(7)))
(48)

and Pi(n;phi,k) is an elliptic integral of the third kind, which is 1.3333292798..., or within 4.1×10^(-6) of 4/3 (E. W. Weisstein, Apr. 2006). Another slightly obscure one is the value of x needed to give a 99.5% confidence interval for a Student's t-distribution with sample size 30, which is 2.7499956..., or within 4.4×10^(-6) of 11/4 (E. W. Weisstein, May 2, 2006).

Let l^_ be the average length of a line in triangle line picking for an isosceles right triangle, then

 l^_=1/(30)[2+4sqrt(2)+(4+sqrt(2))sinh^(-1)1] approx 0.4142933026,
(49)

which is within 8×10^(-5) of sqrt(2)-1=0.4142135624....

D. Terr (pers. comm., July 29, 2004) found the almost integer

 phi/(2^(ln2))=1.0007590...,
(50)

where phi is the golden ratio and ln2 is the natural logarithm of 2.

A set of almost integers due to D. Hickerson are those of the form

 h_n=(n!)/(2(ln2)^(n+1))
(51)

for 1<=n<=17, as summarized in the following table.

nh_n
00.72135
11.04068
23.00278
312.99629
474.99874
5541.00152
64683.00125
747292.99873
8545834.99791
97087261.00162
10102247563.00527
111622632572.99755
1228091567594.98157
13526858348381.00125
1410641342970443.08453
15230283190977853.03744
165315654681981354.51308
17130370767029135900.45799

These numbers are close to integers due to the fact that the quotient is the dominant term in an infinite series for the number of possible outcomes of a race between n people (where ties are allowed). Calling this number f(n), it follows that

 f(n)=sum_(k=1)^n(n; k)f(n-k)
(52)

with f(0)=1, where (n; k) is a binomial coefficient. From this, we obtain the exponential generating function for f

 sum_(n=0)^infty(f(n))/(n!)z^n=1/(2-e^z),
(53)

and then by contour integration it can be shown that

 f(n)=1/2(-1)^(n+1)n!sum_(k=-infty)^infty1/((ln2+2piik)^(n+1))
(54)

for n>=1, where i is the square root of -1 and the sum is over all integers k (here, the imaginary parts of the terms for k and -k cancel each other, so this sum is real). The k=0 term dominates, so f(n) is asymptotic to n!/(2(ln2)^(n+1)). The sum can be done explicitly as

 f(n)=((-1)^(n+1)in!)/(pi^(n+1)2^(n+2))[i^nzeta(n+1,1+(iln2)/(2pi))-(-i)^nzeta(n+1,-(iln2)/(2pi))],
(55)

where zeta(s,a) is the Hurwitz zeta function. In fact, the other terms are quite small for n from 1 to 15, so f(n) is the nearest integer to n!/(2(ln2)^(n+1)) for these values, given by the sequence 1, 3, 13 75, 541, 4683, ... (OEIS A034172).

A large class of irrational "almost integers" can be found using the theory of modular functions, and a few rather spectacular examples are given by Ramanujan (1913-14). Such approximations were also studied by Hermite (1859), Kronecker (1863), and Smith (1965). They can be generated using some amazing (and very deep) properties of the j-function. Some of the numbers which are closest approximations to integers are e^(pisqrt(163)) (sometimes known as the Ramanujan constant and which corresponds to the field Q(sqrt(-163)) which has class number 1 and is the imaginary quadratic field of maximal discriminant), e^(pisqrt(22)), e^(pisqrt(37)), and e^(pisqrt(58)), the last three of which have class number 2 and are due to Ramanujan (Berndt 1994, Waldschmidt 1988ab).

The properties of the j-function also give rise to the spectacular identity

 [(ln(640320^3+744))/pi]^2=163+2.32167...×10^(-29)
(56)

(Le Lionnais 1983, p. 152; Trott 2004, p. 8).

The list below gives numbers of the form x=e^(pisqrt(n)) for n<=1000 for which |nint(x)-x|<=10^(-3).

n|nint(x)-x|
25-0.00066
37-0.000022
43-0.00022
58-1.8×10^(-7)
67-1.3×10^(-6)
74-0.00083
1480.00097
163-7.5×10^(-13)
232-7.8×10^(-6)
2680.00029
522-0.00015
6521.6×10^(-10)
719-0.000013

Gosper (pers. comm.) noted that the expression

 1-262537412640768744e^(-pisqrt(163))-196884e^(-2pisqrt(163))+103378831900730205293632e^(-3pisqrt(163)).
(57)

differs from an integer by a mere 1.6×10^(-59).

AlmostIntegerTriangleDissection

E. Pegg Jr. noted that the triangle dissection illustrated above has length

d=1/2sqrt(1/(30)(61421-23sqrt(5831385)))
(58)
=7+8.574×10^(-8),
(59)

which is almost an integer.

Borwein and Borwein (1992) and Borwein et al. (2004, pp. 11-15) give examples of series identities that are nearly true. For example,

 sum_(n=1)^infty(|_ntanhpi_|)/(10^n)=1/(81)-1.11...×10^(-269)
(60)

which is true since tanhpi=0.9962... and |_ntanhpi_|=n-1 for positive integer n<268. In fact, the first few doubled values of n at which |_ntanhpi_|=|_(n+1)tanhpi_| are 268, 536, 804, 1072, 1341, 1609, ...(OEIS A096613).

An example of a (very) near-integer is

sum_(k=-infty)^(infty)1/(10^((k/100)^2))=theta_3(0,10^(-1/10000))
(61)
 approx 100sqrt(pi/(ln10))+1.3809×10^(-18613)
(62)

(Borwein and Borwein 1992; Maze and Minder 2005).

Maze and Minder (2005) found the class of near-identities obtained from

 u_k=ln2sum_(n=-infty)^infty1/((2^(k/2)+2^(-k/2))^n)
(63)

as

u_1=3.14159265359518238328842...
(64)
=pi+5.3...×10^(-12)
(65)
u_2=1.00000000004885109041382...
(66)
=1+4.8...×10^(-11)
(67)
u_3=pi/(2^3)+2.2...×10^(-10)
(68)
u_4=1/6+6.7...×10^(-10)
(69)
u_5=(3pi)/(2^7)+1.5...×10^(-9)
(70)
u_6=1/(30)+2.9...×10^(-9)
(71)

(OEIS A114609 and A114610). Here, the excesses can be computed as exact sums connected by a recurrence relation, with the first few being

r_1=2pisum_(k=1)^(infty)sech((2kpi^2)/(ln2))
(72)
r_2=(2pi)/(ln2)sum_(k=1)^(infty)2kpicsch((2kpi^2)/(ln2))
(73)

(Maze and Minder 2005). These sums can also be done in closed form using q-polygamma functions psi_q^((k))(z), giving for example

r_1=-2ipsi_(sqrt(2))(-ipil_2)-2ipsi_(sqrt(2))(ipil_2)-1/2l_2^(-1)-3pi
(74)
r_2=-2l_2psi_(sqrt(2))^((1))(-ipil_2)-2l_2psi_(sqrt(2))^((1))(ipil_2)-1/4l_2^(-1)+1,
(75)

with l_2=ln2.

An amusing almost integer involving units of length is given by

 (inches/mile)/(astronomical units/light year)=0.99812....
(76)

If combinations of physical and mathematical constants are allowed and taken in SI units, the following quantities have a near-integer numeric prefactor

(cek)/h=1.0008m A/(s K)
(77)
(P_b+7/9)(epsilon_0R_infty^2)=1000F/m^3
(78)

(M. Trott, pers. comm. Apr. 28, 2011), the first of which was apparently noticed by Weisskopf. Here, c is the speed of light, e is the elementary charge, k is Boltzmann's constant, h is Planck's constant, P_b is the bond percolation threshold for a 4-dimensional hypercube lattice, epsilon_0 is the vacuum permittivity, and R_infty is the Rydberg constant. Another famous example of this sort is Wyler's constant, which approximates the (dimensionless) fine structure constant in terms of fundamental mathematical constants.

Wolfram Web Resources

Mathematica »

The #1 tool for creating Demonstrations and anything technical.

Wolfram|Alpha »

Explore anything with the first computational knowledge engine.

Wolfram Demonstrations Project »

Explore thousands of free applications across science, mathematics, engineering, technology, business, art, finance, social sciences, and more.

Computerbasedmath.org »

Join the initiative for modernizing math education.

Online Integral Calculator »

Solve integrals with Wolfram|Alpha.

Step-by-step Solutions »

Walk through homework problems step-by-step from beginning to end. Hints help you try the next step on your own.

Wolfram Problem Generator »

Unlimited random practice problems and answers with built-in Step-by-step solutions. Practice online or make a printable study sheet.

Wolfram Education Portal »

Collection of teaching and learning tools built by Wolfram education experts: dynamic textbook, lesson plans, widgets, interactive Demonstrations, and more.

Wolfram Language »

Knowledge-based programming for everyone.