Abstract
The exponential integral function Ei(x) is given as an indefinite integral of an elementary expression. This allows a second-order linear differential equation for the function to be constructed, which is of conventional form. A limitless number of differential equations can be derived from the original by elementary transformations, and many integrals are given by applying the method of fragments to some of these transformed equations. Results are presented here both for simple transformations and other transformations obtained by solving simple Riccati equations. Some of the Integrals are presented combine Ei(x) with Bessel functions, modified Bessel functions and Whittaker functions. All results have been checked by differentiation using Mathematica.
1. Introduction
The exponential integral function is defined by the integral [Citation1,Citation2] (1.1) (1.1) for which (1.2) (1.2) Any arbitrary function trivially obeys a differential equation of the form (1.3) (1.3) which has the general solution From Equations (Equation1.1(1.1) (1.1) )–(Equation1.3(1.3) (1.3) ) the function obeys the differential equation (1.4) (1.4) which is of conventional form. This paper applies the method of fragments introduced in [Citation3,Citation4] to derive indefinite integrals involving . As Equation (Equation1.4(1.4) (1.4) ) by itself provides only a limited number of fragments, various transformations will be applied to this equation, which provide in principle an unlimited number of cases, but only a limited number of these are interesting.
1.1. The method of fragments
The general second-order linear homogeneous differential equation is (1.5) (1.5) and in [Citation3,Citation4] the integration formula (1.6) (1.6) was derived, where is any solution of Equation (Equation1.1(1.1) (1.1) ) and is an arbitrary twice differentiable complex- valued function of x. The function in Equation (Equation1.2(1.2) (1.2) ) is the reciprocal of the Wronskian for Equation (Equation1.1(1.1) (1.1) ) and is also the integrating factor for the two leftmost terms of this equation. It also appears in the Lagrangian form [Citation3] of Equation (Equation1.1(1.1) (1.1) ) and is given by (1.7) (1.7) Suitable choices of the arbitrary function in Equation (Equation1.6(1.6) (1.6) ) give integrals involving . A useful technique for obtaining interesting integrals is to take to be a solution of a fragment of Equation (Equation1.5(1.5) (1.5) ), where a fragment is defined [Citation3] as the differential equation with one or more terms deleted or modified, for example For an equation of the form (Equation1.3(1.3) (1.3) ), the factor in Equation (Equation1.7(1.7) (1.7) ) is given as and as any constant multiplicative factor in the definition of would cancel in Equation (Equation1.2(1.2) (1.2) ), we can always take and for Equation (Equation1.4(1.4) (1.4) ) this gives Equation (Equation1.4(1.4) (1.4) ) is simple with few terms, and the small number of obvious fragments are Substituting these results into Equation (Equation1.6(1.6) (1.6) ) gives the three integrals (1.8) (1.8) (1.9) (1.9) and subtracting Equation (Equation1.9(1.9) (1.9) ) from Equation (Equation1.8(1.8) (1.8) ) gives the additional integral All of these cases are known integrals [Citation2].
1.2. Transformations of the differential equation
Equation (Equation1.4(1.4) (1.4) ) is of the baseline form with (1.10) (1.10) Equation (Equation1.10(1.10) (1.10) ) can be transformed [Citation3] to an equation in by the substitution , which gives (1.11) (1.11) In [Citation3] there was a typographical error equivalent to stating , but all the related formulas were given correctly. Defining then and Equation (Equation1.11(1.11) (1.11) ) can be expressed as [Citation3] (1.12) (1.12) The general solution of Equation (Equation1.12(1.12) (1.12) ) is given in terms of the general solution of Equation (Equation1.4(1.4) (1.4) ) by [Citation3] For Equation (Equation1.4(1.4) (1.4) ) for , the transformed Equation (Equation1.12(1.12) (1.12) ) becomes (1.13) (1.13) Employing Equation (Equation1.13(1.13) (1.13) ) in Equation (Equation1.6(1.6) (1.6) ) gives a large number of integrals involving , as is arbitrary, but only sample results can be given here. As in [Citation3,Citation4], interesting integrals can be obtained by applying the method of fragments to various forms of Equation (Equation1.13(1.13) (1.13) ). Equation (Equation1.13(1.13) (1.13) ) can be given concrete form either by directly specifying the function or by solving a Riccati equation for such that the term in square brackets takes some desired simpler form. Sample cases obtained by specifying are examined in Section 2 below, and some cases where is the solution of a Riccati equation are examined in Section 3. All results presented have been checked using Mathematica [Citation5].
2. Integrals from specifying directly
Choosing in Equation (Equation1.13(1.13) (1.13) ) gives the differential equation (2.1) (2.1) with the general solution (2.2) (2.2) and with the first derivative Equation (Equation2.1(2.1) (2.1) ) is a special case of the general Whittaker equation (2.3) (2.3) which has the general solution It might be expected that the solution given by Equation (Equation2.2(2.2) (2.2) ) could also be expressed in terms of the two Whittaker functions and , but this is not the case. An equation for in terms of and is given in [Citation6] as and for and this equation reduces to , so that Equation (Equation2.1(2.1) (2.1) ) is a degenerate case of the Whittaker equation. The relation [Citation6] reduces for n = 0 and to Hence for this degenerate case and both reduce to the elementary function in Equation (Equation2.2(2.2) (2.2) ) and the extra solution of the degenerate equation is .
Taking the differential equation in Equation (Equation1.6(1.6) (1.6) ) to be Equation (Equation2.1(2.1) (2.1) ) and specifying to be any solution of the general Whittaker Equation (Equation2.3(2.3) (2.3) ) gives the integral For this integral reduces to and for it reduces to Further simplifications of these integrals are obtained for or and or .
Choosing to be a solution of the equation (2.4) (2.4) gives in terms of a general modified Bessel function of order zero as where The modified Bessel function and the MacDonald function both obey Equation (Equation2.4(2.4) (2.4) ), but they have different formulas for their derivatives [Citation7], with and when deriving explicit integration formulas it is simpler to treat the two cases separately. We have the alternative formulas Substituting substituting these choices into Equation (Equation1.6(1.6) (1.6) ) and simplifying gives the integrals (2.5) (2.5) (2.6) (2.6) For Equations (Equation2.5(2.5) (2.5) )–(Equation2.6(2.6) (2.6) ) reduce to which are known integrals [Citation2]. For , Equations (Equation2.5(2.5) (2.5) )–(Equation2.6(2.6) (2.6) ) reduce to (2.7) (2.7) (2.8) (2.8) and Equations (Equation2.7(2.7) (2.7) )–(Equation2.8(2.8) (2.8) ) appear to be new.
Choosing to be a solution of the fragment gives (2.9) (2.9) where is the general cylinder function of order zero. Substituting Equations (Equation2.9(2.9) (2.9) ) into Equation (Equation1.6(1.6) (1.6) ) gives the integral (2.10) (2.10) For Equation (Equation2.10(2.10) (2.10) ) reduces to and for this equation reduces to Choosing gives the differential equation (2.11) (2.11) with and the general solution is Only the special case will be considered here, for which Some fragments of Equation (Equation2.11(2.11) (2.11) ) are and these give the respective integrals below for the special case . for . Choosing gives the differential equation for which For the fragments give the integrals
3. Equations from solutions of Riccati equations
The expression in square brackets in Equation (Equation1.13(1.13) (1.13) ) is and this can be simplified by setting various combinations of the terms equal to zero and solving the resulting Riccati equations for . One case is which is separable in , such that This equation gives (3.1) (3.1) where c is an arbitrary constant. Setting c = 0 and taking the tanh of both sides of Equation (Equation3.1(3.1) (3.1) ) gives the solution (3.2) (3.2) Equation (Equation3.2(3.2) (3.2) ) gives the differential equation (3.3) (3.3) with (3.4) (3.4) and solution (3.5) (3.5) The most interesting case of this solution is (3.6) (3.6) with Employing Equations (Equation3.3(3.3) (3.3) )–(Equation3.6(3.6) (3.6) ) with in Equation (Equation1.6(1.6) (1.6) ) gives, after some simplification, the integral A simpler integral can be obtained by taking to be a solution of the fragment for which Employing these results in Equation (Equation1.6(1.6) (1.6) ) gives the integral Another separable case is which has a solution and this gives the differential equation (3.7) (3.7) with the solution and For and then and The fragment of Equation (Equation3.7(3.7) (3.7) ) is identical to Equation (Equation2.1(2.1) (2.1) ) and has the same solution given by Equation (Equation2.2(2.2) (2.2) ). For and then and substituting these results in the integral formula (Equation1.6(1.6) (1.6) ) allows an integral in to be derived. Employing these formulas in Equation (Equation1.6(1.6) (1.6) ) gives initially which can be simplified using Equation (Equation1.8(1.8) (1.8) ) to give (3.1) (3.1)
3.1. Non separable cases of the Riccati equation
A Riccati equation of the form can be solved by the substitution which gives Euler [Citation8] showed that solutions of the Riccati equation are given by (3.8) (3.8) where obeys the linear equation The form of Equation (3.10) means that is trivially given by Hence solutions of the equation (3.9) (3.9) are given by solutions of the equation which has the general solution where is the general cylinder function of order one, and Hence a solution of Equation (Equation3.10(3.9) (3.9) ) is given by and a transformed form of Equation (Equation1.13(1.13) (1.13) ) is for which Equation (3.10) has the general solution with For the differential equation (Equation3.8(3.1) (3.1) ), taking in the integration formula (Equation1.6(1.6) (1.6) ) gives the integral: For this equation reduces to and for it reduces to Both of these integrals appear to be new.
The equation has the solutions These equations give the respective differential equations, solutions and derivatives: (3.11) (3.11) and (3.12) (3.12) The fragment of both Equation (Equation3.11(3.11) (3.11) ) and (Equation3.12(3.12) (3.12) ) has the exact solution and derivative Employing these results in the integration formula (Equation1.6(1.6) (1.6) ) gives the integrals The four main special cases of these integrals are All of these integrals appear to be new.
For the simpler where case we obtain the integrals and these have the special cases (3.13) (3.13) (3.14) (3.14) (3.15) (3.15) (3.16) (3.16) The integrals (Equation3.13(3.13) (3.13) ) and (Equation3.15(3.15) (3.15) ) appear to be new, but the integrals (Equation3.14(3.14) (3.14) ) and (Equation3.16(3.16) (3.16) ) are given in [Citation2].
Disclosure statement
No potential conflict of interest was reported by the author(s).
References
- Spanier K, Oldham KB. An atlas of functions. New York (NY): Hemisphere; 1987.
- Prudnikov AP, Brychkov YuA, Marichev OI. Integrals and series, Vol. 2, special functions. New York (NY): Gordon and Breach; 1986.
- Conway JT. A Lagrangian method for deriving new indefinite integrals of special functions. Integral Transforms Spec Funct. 2015;26(10):812–824.
- Conway JT. Indefinite integrals of some special functions from a new method. Integral Transforms Spec Funct. 2015;26(11):845–858.
- Wolfram S. The mathematica book. 5th ed. Champaign (IL): Wolfram Media; 2003.
- Gradshteyn IS, Ryzhik IM. Table of integrals, series and products. New York (NY): Academic; 2007.
- Brychkov YuA. Handbook of special functions: derivatives, integrals, series and other formulas. Boca Raton (FL): Chapman & Hall/CRC; 2008.
- Euler L. Institutionum calculi integralis. Vol. 2. St. Petersburg: Imperial Academy of Science; 1769.