
An unusual alternating reflection method on conics is presented to evaluate inverse trigonometric and hyperbolic functions.
Citation: François Dubeau. Alternating reflection method on conics leading to inverse trigonometric and hyperbolic functions[J]. AIMS Mathematics, 2022, 7(7): 11708-11717. doi: 10.3934/math.2022652
[1] | Tianyong Han, Zhao Li, Jun Yuan . Optical solitons and single traveling wave solutions of Biswas-Arshed equation in birefringent fibers with the beta-time derivative. AIMS Mathematics, 2022, 7(8): 15282-15297. doi: 10.3934/math.2022837 |
[2] | Huo Tang, Kadhavoor Ragavan Karthikeyan, Gangadharan Murugusundaramoorthy . Certain subclass of analytic functions with respect to symmetric points associated with conic region. AIMS Mathematics, 2021, 6(11): 12863-12877. doi: 10.3934/math.2021742 |
[3] | Chunyan Liu . The traveling wave solution and dynamics analysis of the fractional order generalized Pochhammer–Chree equation. AIMS Mathematics, 2024, 9(12): 33956-33972. doi: 10.3934/math.20241619 |
[4] | Naseem Abbas, Amjad Hussain, Mohsen Bakouri, Thoraya N. Alharthi, Ilyas Khan . A study of dynamical features and novel soliton structures of complex-coupled Maccari's system. AIMS Mathematics, 2025, 10(2): 3025-3040. doi: 10.3934/math.2025141 |
[5] | Mustafa Inc, Mamun Miah, Akher Chowdhury, Shahadat Ali, Hadi Rezazadeh, Mehmet Ali Akinlar, Yu-Ming Chu . New exact solutions for the Kaup-Kupershmidt equation. AIMS Mathematics, 2020, 5(6): 6726-6738. doi: 10.3934/math.2020432 |
[6] | Hajar F. Ismael, Hasan Bulut, Haci Mehmet Baskonus, Wei Gao . Dynamical behaviors to the coupled Schrödinger-Boussinesq system with the beta derivative. AIMS Mathematics, 2021, 6(7): 7909-7928. doi: 10.3934/math.2021459 |
[7] | Zeliha Korpinar, Mustafa Inc, Ali S. Alshomrani, Dumitru Baleanu . The deterministic and stochastic solutions of the Schrodinger equation with time conformable derivative in birefrigent fibers. AIMS Mathematics, 2020, 5(3): 2326-2345. doi: 10.3934/math.2020154 |
[8] | Junjie Li, Gurpreet Singh, Onur Alp İlhan, Jalil Manafian, Yusif S. Gasimov . Modulational instability, multiple Exp-function method, SIVP, solitary and cross-kink solutions for the generalized KP equation. AIMS Mathematics, 2021, 6(7): 7555-7584. doi: 10.3934/math.2021441 |
[9] | Chun Huang, Zhao Li . Soliton solutions of conformable time-fractional perturbed Radhakrishnan-Kundu-Lakshmanan equation. AIMS Mathematics, 2022, 7(8): 14460-14473. doi: 10.3934/math.2022797 |
[10] | Rajiniganth Pandurangan, Sabri T. M. Thabet, Imed Kedim, Miguel Vivas-Cortez . On the Generalized $ \overline{\theta({\tt{t}})} $-Fibonacci sequences and its bifurcation analysis. AIMS Mathematics, 2025, 10(1): 972-987. doi: 10.3934/math.2025046 |
An unusual alternating reflection method on conics is presented to evaluate inverse trigonometric and hyperbolic functions.
It is known that circular reflections on the unit circle leads to the computation of the digits of the trigonometric constant π [1,2]. A similar analysis shows that hyperbolic reflections on the unit hyperbola leads to the computation of the digits of the logarithmic constant ln(2) [3]. Looking at those two situations, we obtain a unified way to present an unusual alternating reflection method on conics to evaluate inverse trigonometric and hyperbolic functions. A lot of research has been done on the efficient evaluation of elementary functions, see for example [4,5,6]. The method we obtain is much more a curiosity than an efficient practical method, but since it uses only an elementary geometric operation (reflection), its presentation can be interesting.
In Section 2, we introduce parametrization of the conics and express reflections and rotations in terms of the parameter. The numerical method is presented in Section 3. The way to compute the digits of the solution of the problem is explained in Section 4. In Section 5 we present some examples to illustrate the unusual methods we obtained to evaluate inverse trigonometric and inverse hyperbolic functions.
Let G=(100δ), where δ=±1, so G2=I, and
S={(x,y)∈R2 | (x,y)G(xy)=x2+δy2=1, (1−δ)x≥0 }. |
So S is the unit circle for δ=+1, and S is the right branch (x≥0) of the unit hyperbola for δ=−1.
Suppose we have two regular functions c(⋅) and s(⋅) defined in such a way that to every point (x,y)∈S there exist θ such that (x,y)=(c(θ),s(θ)), so
c2(θ)+δs2(θ)=1. |
It is clear here that for δ=+1 we can take
(c(θ),s(θ))=(cos(θ),sin(θ)), |
and for δ=−1 we can take
(c(θ),s(θ))=(cosh(θ),sinh(θ)). |
For a given point (c(⋅),s(⋅))∈S we observe that
(c(θ),s(θ))G(−δs(θ)c(θ))=0, |
we say that (−δs(θ),c(θ)) is G-orthogonal to (c(⋅),s(⋅)). From the regularity of c(⋅) and s(⋅) we get that
0=c(θ)c′(θ)+δs(θ)s′(θ)=(c(θ),s(θ))G(c′(θ)s′(θ)), |
and since (c′(θ),s′(θ)) and (−δs(θ),c(θ)) are both G-orthogonal to (c(θ),s(θ)), we can suppose that
(c′(θ),s′(θ))=(−δs(θ),c(θ)) |
then (−δs(θ),c(θ)) is the direction of the tangent to S at (c(θ),s(θ)).
Let us also suppose that
{c(α+β)=c(α)c(β)−δs(α)s(β),s(α+β)=s(α)c(β)+c(α)s(β),and{c(−θ)=c(θ),s(−θ)=−s(θ). |
In particular (c(0),s(0))=(1,0).
We introduce the following definitions and notations
rot(α)=(c(α)−δs(α)s(α)c(α)), |
and
ref(α)=(c(2α)δs(2α)s(2α)−c(2α)). |
As expected we have
rot(α)(c(θ)s(θ))=(c(θ+α)s(θ+α)), |
and
ref(α)(c(α+θ)s(α+θ))=(c(α−θ)s(α−θ)). |
We can verify that rot(α)rot(β)=rot(α+β), and
{ref(α)=rot(2α)ref(0),ref(0)ref(α)=rot(−2α),ref(α)rot(β)=rot(2α+β)ref(0). |
Let (c(α),s(α))∈S be a fixed point in the first quadrant, so c(α)>0 and s(α)>0. Let us start with an admissible x value such that there exists a point P=(x,y)∈S with y≥0. Then we look for θ such that c(θ)=x and y=s(θ)=√δ(1−x2)≥0, for an unknown θ. We apply to P=P0 a sequence of reflections. We start with ref(α) followed by ref(0), and we repeat the process. Starting with P0=P∈S, after k reflections we get the point Pk=(c(θk),s(θk))∈S. It follows that after 2n reflections P2n=(c(θ−2nα),s(θ−2nα)), so θ2n=θ−2nα, and after 2n+1 reflections P2n+1=(c(2(n+1)α−θ),s(2(n+1)α−θ)), so θ2n+1=2(n+1)α−θ.
Starting with θ=θ0>α, we would like to stop when θk∈[0,α] for the first time. The trajectory of the points Pk (k=0,1,2,…) is illustrated for δ=+1 on Figure 1 and for δ=−1 on Figure 2, where Pf is the final point.
If K is the total number of reflections (2n or 2n+1), it means that
0<θ−Kα≤α. |
Indeed, for an even number of reflections K=2n, the last reflection is with respect to the axis y=0, so θ2n cannot be 0 and we have θ2n∈(0,α], or 0<θ−2nα≤α. For an odd number of reflections K=2n+1, the last reflection is with respect to the line of direction α, so θ2n+1 cannot be α, and we have θ2n+1∈[0,α), so 0≤2(n+1)α−θ<α or 0<θ−(2n+1)α≤α.
So the total number K of reflections is
K={θα−1ifθαisaninteger,⌊θα⌋ifθαisnotaninteger. |
In the applications the coordinates of P are known but not θ. Moreover, α is not known and only some information about it is given that allow us to determine c(α) and s(α). The problem here is to find an approximation of θ from K and an approximation of α. So as a result we get a method to find θ=c−1(x), i.e. an evaluation of the inverse function c−1(⋅).
The preceding result suggests a way to compute the digits of θ. Since the integer part of θ⋅10N, noted ⌊θ⋅10N⌋, add the first N digits of the fractional part of θ, for an angle α≈10−N we will be near the goal.
To get the digits of θ we could take α=10−N, then
K={θ⋅10N−1ifθ⋅10Nisaninteger,⌊θ⋅10N⌋ifθ⋅10Nisnotaninteger. |
Unfortunately it is not the interesting situation for the applications.
In the applications, α is not exactly given, but there is a function T(⋅) for which the value T(α)=σ can be fixed, so α=T−1(σ)≈σ. It is possible to take σ=10−N. We will consider the next two situations using the Taylor's expansion of T−1(⋅).
(a) Let us consider T(⋅)=sin(⋅) for δ=+1, and T(⋅)=tanh(⋅) for δ=−1. Taylor's expansions of their corresponding T−1(⋅) are
arcsin(σ)=+∞∑ℓ=0(2ℓ)!4ℓ(ℓ!)2σ2ℓ+1(2ℓ+1), |
and
arctanh(σ)=+∞∑ℓ=0σ2ℓ+1(2ℓ+1). |
Thanks to the geometric series, we can consider the following lower and upper bounds for both series
σ<T−1(σ)<σ[1+σ23(11−σ2)], |
for 0<σ<1. Moreover
1+σ23(11−σ2)≤11−σ22, |
for 0<σ<1/√2. So we can write
1σ−σ2<1T−1(σ)<1σ |
for 0<σ<ρ where 1/2<ρ=1/√2<1. We set σ=10−N, α=T−1(10−N), and multiply by θ to get
θ⋅10N−θ2⋅10−N<θα<θ⋅10N |
As long as θα is not an integer and that θ⋅10N−θ2⋅10−N>⌊θ⋅10N⌋ we have K=⌊θ⋅10N⌋. In general in this case we get K=⌊θ⋅10N⌋−1 or K=⌊θ⋅10N⌋.
(b) Let us consider T(⋅)=tan(⋅) for δ=+1, and T(⋅)=sinh(⋅) for δ=−1. Taylor's expansions of their corresponding T−1(⋅) are
arctan(σ)=+∞∑ℓ=0(−1)ℓσ2ℓ+1(2ℓ+1), |
and
arcsinh(σ)=+∞∑ℓ=0(−1)ℓ(2ℓ)!4ℓ(ℓ!)2σ2ℓ+1(2ℓ+1). |
We can consider the following lower and upper bounds for both series
σ−σ33<T−1(σ)<σ, |
for 0<σ<1. Moreover
1−σ23>11+σ22, |
for 0<σ<1. So we can write
1σ<1T−1(σ)<1σ+σ2 |
for 0<σ<ρ where 1/2<ρ<1. We set σ=10−N, α=T−1(10−N), and multiply by θ to get
θ⋅10N<θα<θ⋅10N+θ2⋅10−N |
As long as θα is not an integer and that θ⋅10N+θ2⋅10−N<⌊θ⋅10N⌋+1 we have K=⌊θ⋅10N⌋. In general in this case we get K=⌊θ⋅10N⌋ or K=⌊θ⋅10N⌋+1.
In both cases, the result K=⌊θ⋅10N⌋ depends of the expansion of θ. let
{θ=a0.a1a2⋯aNaN+1⋯a2Na2N+1⋯θ2=˜a0.˜a1˜a2⋯˜aN˜aN+1⋯˜a2N˜a2N+1⋯ |
and
{θ⋅10N=a0a1a2⋯aN.aN+1⋯a2N−1a2Na2N+1⋯θ2⋅10−N=0.0 ⋯ 0⏟(N−1)times˜a0 ˜a1⋯. |
A sufficient condition to get K=⌊θ⋅10N⌋ can be given for the two cases. The conditions are:
(a) there exists an index n∈[N+1,2N−1] such that an>0,
(b) there exists an index n∈[N+1,2N−1] such that an<9.
For example, up to now and with modern computational facilities and methods [7], for small values of N and up to very large values of N it has not been observed sequences such that:
(a') an=0 for n∈[N+1,2N−1],
(b') an=9 for n∈[N+1,2N−1],
in the expansion of π and ln(2). So we could claim that K=⌊θ⋅10N⌋ holds up to very large values of N for θ=π and θ=ln(2).
The algorithm is as follows :
Alternating reflection method
Step 0. Enter P=(x,y)∈S with y≥0, and σ=T(α).
Step 1. Determine c(α) and s(α) from the data σ.
Step 2. The target zone is { P∈S | Pisbetween(1,0)and(c(α),s(α)) }.
Step 3. Compute c(2α)=c2(α)−δs2(α) and s(2α)=2s(α)c(α).
Step 4. Set
ref(0)=(100−1)andref(α)=(c(2α)δs(2α)s(2α)−c(2α)) |
Step 5. With P0=P, compute Pk+1 as long as Pk is not in the target zone
(i) for k=2n, compute Pk+1=P2n+1=ref(α)P2n=ref(α)Pk,
(ii) for k=2n+1, compute Pk+1=P2n+2=ref(0)P2n+1=ref(0)Pk.
We have considered the two situations δ=±1 and we got the results given in Table 1 for π and ln(2).
θ=π=3.14159265358979... | θ=ln2=0.69314718055995... | |||
N | σ=10−N | K=⌊π⋅10−N⌋ | K=⌊ln(2)⋅10−N⌋ | |
1 | 10−1 | 31 | 6 | |
2 | 10−2 | 314 | 69 | |
3 | 10−3 | 3141 | 693 | |
4 | 10−4 | 31415 | 6931 | |
5 | 10−5 | 314159 | 69314 |
S is the unit circle, and (c(⋅),s(⋅))=(cos(⋅),sin(⋅)). To get θ=π, we start with x=cos(π)=−1, so we consider P=(−1,0). K is computed using σ=10−N for N=1,2,3,4,5, for T(α)=sin(α)=σ and T(α)=tan(α)=σ. As expected, both situations generate the same value of K.
S is the right branch (x≥0) of the unit hyperbola, and (c(⋅),s(⋅))=(cosh(⋅),sinh(⋅)). To get θ=ln(2), we start with x=cosh(ln(2))=5/4, so we consider P=(5/4,3/4). K is computed using σ=10−N for N=1,2,3,4,5, for T(α)=tanh(α)=σ and T(α)=sinh(α)=σ. As expected, both situations generate the same value of K.
We not only have an unusual method to find the digits of the trigonometric constant π and the logarithmic constant ln(2) but also to evaluate the inverse trigonometric functions (for δ=+1) and inverse hyperbolic functions (for δ=−1). Indeed the method can be extended to solve for θ the equation c(θ)=x. If the given data for α allows us to say that α=10−N+O(10−2N), we get that
|θ−(K+12)⋅10−N|=1210−N+O(10−2N). |
The next examples illustrate that both values of K can be obtained. Numerical results are reported in Table 2 where examples were chosen to get the same figures.
θ=1.00000000 | θ=0.99999999 | |||||||
K | K | |||||||
N | ⌊θ⋅10N⌋ | (a) | (b) | ⌊θ⋅10N⌋ | (a) | (b) | ||
1 | 10 | 9 | 10 | 9 | 9 | 10 | ||
2 | 100 | 99 | 100 | 99 | 99 | 100 | ||
3 | 1000 | 999 | 1000 | 999 | 999 | 1000 | ||
4 | 10000 | 9999 | 10000 | 9999 | 9999 | 10000 | ||
5 | 100000 | 99999 | 100000 | 99999 | 99999 | 99999 |
We fix x and compute y=√1−x2≥0 for the point P=P0. We use for (a) T(α)=sin(α), and for (b) T(α)=tan(α). We consider two situations to illustrate that we can get the two values of K. Firstly, we consider x=0.540302305...=cos(1) (and y=√1−x2≥0), so θ=1=arccos(x). We get for (a) K=⌊1⋅10N⌋−1 and for (b) K=⌊1⋅10N⌋. Secondly we consider x=0.540302314...=cos(0.99999999), so θ=0.99999999=arccos(x). We get for (a) K=⌊0.99999999⋅10N⌋ for N=1,2,3,4, and for (b) K=⌊0.99999999⋅10N⌋+1 for N=1,2,3,4, but not for N≥5.
We fix x and compute y=√x2−1≥0 for the point P=P0. We use for (a) T(α)=tanh(α), and for (b) T(α)=sinh(α). We consider two situations to illustrate the different values of K. Firstly, we consider x=1.543080635...=cosh(1), so θ=1=arccosh(x). We get for (a) K=⌊1⋅10N⌋−1 and for (b) K=⌊1⋅10N⌋. Secondly, we consider x=1.543080623...=cosh(0.99999999), so θ=0.99999999=arccosh(x). We get for (a) K=⌊0.99999999⋅10N⌋ and for (b) K=⌊0.99999999⋅10N⌋+1 for N=1,2,3,4, but not for N≥5.
The computational cost of this method is quite low at each iteration. It requires 4 multiplications and 2 additions to compute P2n+1 from P2n, and only a sign change to get P2n+2 from P2n+1. Also, at each iteration a test is required to eventually terminate the process.
We used MATLAB with single-precision computation for the numerical examples. For σ=10−N, we had c(α)=1+O(10−2N) and s(α)=O(10−N), and the matrix ref(α) looked like
(1+O(10−2N)δO(10−N)O(10−N)1+O(10−2N)), |
so it explains why we stopped at N=5 in the computation. We could increase N with multi-precision computation.
In this paper we have presented an unusual method to find θ=c−1(x). We have considered the following problem:
{Supposegivenanadmissiblevaluex,andsety=√δ(1−x2)≥0,thenfindθsuchthatθ=c−1(x),so(x,y)=(c(θ),s(θ)). |
We can extend this problem to find also θ=s−1(y) and θ=t−1(z), where t(⋅)=s(⋅)/c(⋅). Indeed we have θ=s−1(y)=sign(y)c−1(√δ(1−δy2)) for an appropriate y value, and θ=t−1(z)=sign(z)c−1(1/√1+δz2) for an appropriate value of z.
This work has been financially supported by an individual discovery grant from the Natural Sciences and Engineering Research Council of Canada.
The author declares no conflict of interests.
[1] |
G. Galperin, Playing pool with π (the number π from the billiard point of view), Regul. Chaotic Dyn., 8 (2003), 375–394. http://dx.doi.org/10.1070/RD2003v008n04ABEH000252 doi: 10.1070/RD2003v008n04ABEH000252
![]() |
[2] | F. Dubeau, Collisions, or reflections and rotations, leading to the digits of π, Open Journal of Mathematical Sciences, to appear, 2022. |
[3] |
F. Dubeau, Hyperbolic reflections leading to the digits of ln(2), J. Appl. Math. Phys., 10 (2022), 112–131. http://dx.doi.org/10.4236/jamp.2022.101009 doi: 10.4236/jamp.2022.101009
![]() |
[4] | C. T. Fike, Computer Evaluation of Mathematical Functions, Prentice-Hall Inc., Englewood Cliffs, New Jersey, 1968. |
[5] | R. J. Pulskamp, J. A. Delaney, Computer and Calculator Computation of Elementary Functions, UMAP J., 12 (1991), 317–348. |
[6] | J. M. Muller, Elementary Functions: Algorithms and Implementation, Third Edition, Birkhäuser Boston Inc., Boston, MA, 2016. http://dx.doi.org/10.1007/978-1-4899-7983-4 |
[7] |
D. H. Bailey, P. B. Borwein, S. Plouffe, On the rapid computation of various polylogarithmic constants, Math. Comp., 66 (1997), 903–913. http://dx.doi.org/10.1090/S0025-5718-97-00856-9 doi: 10.1090/S0025-5718-97-00856-9
![]() |
1. | François Dubeau, Elastic interactions in physics are reflections in geometry, 2022, 12, 2158-3226, 085307, 10.1063/5.0087903 |
θ=π=3.14159265358979... | θ=ln2=0.69314718055995... | |||
N | σ=10−N | K=⌊π⋅10−N⌋ | K=⌊ln(2)⋅10−N⌋ | |
1 | 10−1 | 31 | 6 | |
2 | 10−2 | 314 | 69 | |
3 | 10−3 | 3141 | 693 | |
4 | 10−4 | 31415 | 6931 | |
5 | 10−5 | 314159 | 69314 |
θ=1.00000000 | θ=0.99999999 | |||||||
K | K | |||||||
N | ⌊θ⋅10N⌋ | (a) | (b) | ⌊θ⋅10N⌋ | (a) | (b) | ||
1 | 10 | 9 | 10 | 9 | 9 | 10 | ||
2 | 100 | 99 | 100 | 99 | 99 | 100 | ||
3 | 1000 | 999 | 1000 | 999 | 999 | 1000 | ||
4 | 10000 | 9999 | 10000 | 9999 | 9999 | 10000 | ||
5 | 100000 | 99999 | 100000 | 99999 | 99999 | 99999 |
θ=π=3.14159265358979... | θ=ln2=0.69314718055995... | |||
N | σ=10−N | K=⌊π⋅10−N⌋ | K=⌊ln(2)⋅10−N⌋ | |
1 | 10−1 | 31 | 6 | |
2 | 10−2 | 314 | 69 | |
3 | 10−3 | 3141 | 693 | |
4 | 10−4 | 31415 | 6931 | |
5 | 10−5 | 314159 | 69314 |
θ=1.00000000 | θ=0.99999999 | |||||||
K | K | |||||||
N | ⌊θ⋅10N⌋ | (a) | (b) | ⌊θ⋅10N⌋ | (a) | (b) | ||
1 | 10 | 9 | 10 | 9 | 9 | 10 | ||
2 | 100 | 99 | 100 | 99 | 99 | 100 | ||
3 | 1000 | 999 | 1000 | 999 | 999 | 1000 | ||
4 | 10000 | 9999 | 10000 | 9999 | 9999 | 10000 | ||
5 | 100000 | 99999 | 100000 | 99999 | 99999 | 99999 |