]\gt ; ' [ 759 ] XVIII .
On RICCATI'S and its Transforrnations , some Integrals ) , satisfy them .
J. W. L. GLAISHER , M.A. , F.R.S. , Fellow of College , Ccnnbridge .
Received and read June 16 , 1881 .
Introduction .
THE present memoir relates chiefly to the different forms of the iutegrals of the differential equation . . . . . . .
( 1 ) , and to the evaluation of certain defnite integrals which are connected with this equation .
Transforming ( 1 ) by assuming , it becomes ( 2 ) , that is , writing for ( ) , and this equation may be transformed into ( 4 ) , by the substitution where equation ( 4 ) may be regarded the standard form of RICCATI'S equation ( see S III .
, art .
17 ) .
It is well-known that these equations admit of integration in finite form if ) an integer , an uneven integer , the reciprocal of an uneven integer , respectively .
The contents of the memoir are as follows : In the first section ( S I. ) six particular integrals of the are obtained , and the relations between them are examined .
When is not an integer , the six integrals extend to infinity , and in this case the relations between them present no special peculiarity .
When is all integel , two of the series terminate , nd we obtain two particular integrals of ( 1 ) which contain finite number of terms .
Tho series terminate in consequence of the occurrence of zero factors in the coefficients of the terms , but if they are continued , zero factors occur also in the denominators , so that , after a finite number of zero terms , the series may be regarded as recommencing and extending to infinity .
If the terminating series are supposed to recommence in this manner , so that all the series extelld to infinity , then the relations between the particular integrals are the same as when is not an integer ; but if the series are supposed to terminate absolutely when the zero terms occur , the relations are quite different .
As the finite portions of the particular integrals satisfy the differential equation , it is more natural to regard the series as terminating absolutely , and on this supposition the relations between the particular integrals exhibit a remarkable diversity of form according as is or is not an integer .
The second section contains what is believed to be a new form of the solution of ( 1 ) in the case of an integer .
It is shown that if , a positive integer , this equation is satisfied by the coefficient of in the expansion of in ascending powers of .
The six particular integrals given in S I. of the equation ( 1 ) and the relations connecting them are obtained by different expansions of this expression .
The third section contains the six particular integrals of ( 3 ) and ( 4 ) corresponding to those of ( 1 ) , from which they are deduced by means of the transformations stated above .
The fourth section relates to the particular cases in which the differential equations admit of integration in a finite form .
If a differential equation is satisfied by an infinite series , and if for certain values of a quantity involved in it the series termi .
nates , then in this case we may present the integral in a different form by commencin the finite series at the other end , and writing the terms in the reverse order .
Thus , for example , a particular integral of ( 1 ) is , where but , if positive integer , then commencing the series at the other end , . .
These reverse forms in the case of the equations ( 1 ) .
( 3 ) , ( 4 ) are given in this section .
It is worthy of remark that if we are given a particular integral of a differential equation in the form of a terminating series , such as , for example , EQUATION AND ITS TRANSFORMATIONS .
\mdash ; \amp ; c. , being a positive integer , then we might suppose that the corresponding particular integral , when was not an integer , would be obtained by the series , which does not then terminate , to infinity .
This infinite series , when is not an integer , still satisfies the differential equation , but is divergent ; and the true integral is obtained by commencing the series at the other end and continuing it to infinity backwards .
In general , when we have a series which terminates of itself for a particular form of , we may derive from it two infinite series , when has not this form , by commencing it at either end .
One of these will be an ascending series and the other a descending series ; and we can thus , as it were , pass from the one to the other through the intervention of the finite series .
The fifth section the evaluations of the definite integrals where denotes any real quantity and any positive quantity .
These intg.als have been evaluated when is of the form and when is a positive integer ; but , so far as I know , the general formulae given are new .
It is known that these integrals satisfy differential equations of the forms ( 4 ) and ( 1 ) respectively , so thab their values are necessarily connected with the solutions of these equations considered in I. and III .
The results are curious , as they exhibit changes of form to those refel.red to in describing the contents of S I. , and which are due to the same cause\mdash ; viz. the recommencement of the terminating series after the zero terms .
When is unrestricted it is shown that we have ; but when is a positive integer the first series is to be continued till it terminates , and the second is to be ignored ; and is a negative integer the second is to be continued till it terminates , and the first is to be ignored .
The well-known value of the integral when even integer , viz. does not suggest the general formula , the terms of the finite series being written in the reverse order .
Certain formulae of BOOLE'S and CAUCHY'S are also considered and extended in this section .
The sixth section , which is the longest in the memoir , relates to the numerous symbolic solutions of the equation ( 1 ) and its transformations ( 3 ) and ( 4 ) in the cases in which they are integrable in finite terms .
In this section these symbolic solutions are derived from the definite integrals considered in S V. ; and the various symbolic theorems to which they lead by comparing different forms of the results are examined .
A great many symbolic solutions of the differential equations have been given by R. L. LLIS , BOOLE , , HARGREAVE , WILLIAMSON , DONKIN , \amp ; c. , and these are briefly noticed and connected with one another .
It may be observed that the solution , which has been several times independently discovered , seems to have been first published by Mr. GASKIN , who iu effect gave it in a problem set in the Senate House Examination at Cambridge in 1839 .
The seventh section relates to the connexion between the results given in SS and the formulae of BESSEL'S Functions .
BESSEL'S equation may be derived from ( 1 ) by the simple substitutions ; so that all the theorems relating to the solutions of ( 1 ) have analogues in the solutions of BESSEL'S equation , which are deducible from them by these transformations .
In this section the formulae in BESSEL'S Functions which correspond to those considered in the memoir are stated in a convenient form for comparison .
The number of such formulae is not great , and the substitution of for , which converts exponentials into sines and cosines , and a single series multiplied by an exponential factor into the sum of two series multiplied respectively by a sine and a cosine , changes considerably the appearance of the results , which , from an analytied point of view , are less simple when the differential equation is of BESSEL'S form .
The principal case considered in the theory of BESSEL'S Functions is that of an integer : this corresponds to the case of an integer , which is generally excluded in this memoir , as it renders certain of the particular integrals infinite ( S I. , arts .
1 , 3 ) .
The case of EQUATION AND ITS TRANSFORMATIONS .
finite solution corresponds in ( 1 ) to an integer and in BESSEL'S equation to an integer .
The fact that metion J is expressible in a finite form when , and the finite expression itself , are well known , and the is an important one in physical investigations ; but , so far as I know , the recommencement of the series after the zero terms has not been specially noticed in connexion with the subject of BESSEL'S Functions .
The eighth ( and last ) section contains a list of writings the contents of which are closely connected with the subject of the memoir , arranged in order of date and classed under the sections in which they are noticed .
There is also in each case a short account of the portion of the paper for which it has been referred to , with the numbers of the articles in which the referen ces occur .
The section does not contain a list of all the papers referred to in the memoir ; only those papers which are closely connected with it , and portions of which are , in most cases , to some extent reproduced in it , being included .
The part of the list which relates to S VI .
is intended to be supplementary to that section : it is not in any sense a bibliography of the nbolic solutions , but it probably contains references to the more important papers on the subject .
In the ' Philosophical Magazine ' for 1868 CAYLEY gave the four particular rals ( SIII .
) of RICCATI'S equation ( 4 ) ; and in the same journal for 1872 I investigated the relations between these four partir integrals and the well-known particular integrals .
The are the same as those giyen in SIII .
, and the method is similar to that employed in SI .
I afterwards found that the process of obtaining and connecting the particular integrals assumed a much more simple form when the differential equation was taken to be ( 1 ) than when it was ( 4 ) ; it seemed desirable to re-write the whole investigation , taking ( 1 ) as the equation .
This investigation forms ST .
; it is similar in every Qpect to that contained in the 'Philosophical Magazine , ' but is much more complete .
The correS } ) onding results for the equations ( 3 ) and ( 4 ) are deduced in S III .
The fact that , in the solution io series of a differential equation , if the series terminates but when continued recommences , the latte ] .
portion as well as the finite series satisfies the differential equation , was pointed out by in the ' } of Mathematics ' for 1869 .
The formula ( 8 ) of Sy .
was published in the ' British tion Report ' 1872 , with a brief account of the process given in , 21 .
The principal portion of two short papers , ' On Equation\ldquo ; On certain Differential tions allied to RICCATI'S , \ldquo ; which were published in the trterly Journal of Mathematics ' for 1871 and 1872 , are incorporated in SVI .
The memoir thus includes the results in scattered notes papers .
In these the differential equation considered was generally in form ( 4 ) , but the advantage of adopting ( l ) as the standard form in preference to ( 4 ) considerable .
As far as the differential equation is concerned ( 4 ) , which cf only two terms , is the simplest form ; but as regards the expression of the results , both ( 1 ) and ( 3 ) are superior in every respect .
The equation ( 3 ) was adopted as the standard form by M. BACH in his paper of 1874 ( see SVIII The form for the exponent in RICCATI'S equation ( 4 ) was first employed , I believe , in paper in the ' Philosophical Magazine ' for 1868 , which has been already referred to .
The use of the quantity greatly simplifies the formulae relating to the solution of the equation .
With the exception of S , the memoir was written about three years ago , the delay in communicating it to the Society being due to the fact that it seemed desirable to connect the results more closely with BESSEL'S Functions .
As the theory of these functions forms a distinct , ) recognised branch of analysis , and the differential equations considered are transformable into BESSEL'S equation.by very simple changes in the variables , it was clearly of importance to examine with some care the connexion of the formulae with those of BESSEL'S Functions , and it even seemed possible that it might be advisable to adopt BESSEL'S equation as the standard form .
For the reasons already stated it appeared that this was not the case , and that the analytical treatment of the subject was complicated by the change to BESSEL'S equation .
It is well known that the integrals of the differential equations ( 1 ) , . .
( 4 ) can be expressed in terms of BESSEL'S Functions ; and LOMMEL has specially considered these solutions in several papers in the 'Mathematische Annalen .
In these papers , however , the points to which the memoir relates are not referred to .
It therefore seemed sufficient to give in SVII .
the connexion between the principal formulae , reserving for a separate paper , if it should appear desirable , the examination of the relations which the series considered in the memoir stand to BESSEL'S Functions with negative indexes and to the functions of the second kind introduced by LOMMEL and by NEUMANN .
During the time that the memoir has been in manuscript I have published two extracts from it , viz. the theorem in S , arts .
8 , 9 , in the 'British Association Report ' for 1880 , and the theorem ( 50 ) and its proof ( SVI .
, art .
41 ) in the ' Proceedings of the Cambridge Philosophical Society ' for 1879 .
The differential equations ( 1 ) , . .
( 4 ) present three distinct peculiarities , viz. they are finitely integrable only in special cases ; ( ii .
) they are satisfied by certain remarkable definite integrals , which attracted attention quite independently of the differential equations ; and the solutions when finite admit of being exhibited in various symbolic forms .
In reference to the third of these properties , it is remarkable how much attention has been devoted to the solutions of the equations in these finite cases during the last fifty years .
The differential equations ( 1 ) , .
( 4 ) have however been frequently discussed not as simple transformations of one another .
( 1870 ) , pp. 624-635 ; vol. ui .
( 1871 ) , pp. 475-487 ; vol. , pp. 510-536 .
EQUATION AND ITS TRANSFORMATIONS .
76.5 but as if they were essentially different , and the processes of solution have been applied to them independently .
Also many of the forms have been -discovered several times ; and it therefore seemed to be worth while to collect together , as in SVI .
, the different symbolic formulae , and exhibit the nature of the relations between them .
Although the equation ( 1 ) is connected with BESSEL'S equation by such simple relations , the methods of treatment of the two equations by mathematicians have been very different .
In the case of ( 1 ) and its transformations ( 2 ) , ( 3 ) , ( 4 ) , the purely analytical part of the theory and the forms of the solutions have chiefly attracted attention ; while in the case of BESSEL'S equation the expansion of the results in series suitable for calculation has been one of the 1nain objects .
The theories of the two equations have been developed from very different points of view : the one has been considered in reference to the methods of solution and the peculiarities already referred to , and the other has been considered almost wholly in connexion with the functions which satisfy it , and their applications in astronomy and physics .
It is cnrious that two such very distinct classes of analytical investigation should have been formed having reference to differential equations so closely connected .
It is proper to remark here that in the differential equation ( 1 ) and throughout the memoir the constant may be put equal to unity hout loss of generality .
It was found to be desirable to retain it , as there is some in having present in the solutions a letter whose sign can be changed at pleasure , and also because the transition to the differential equations ' \amp ; .
, which sign changed ) is thus rendered more convenient .
The ordinary differential equations ( 1 ) , . .
( 4 ) are COllsidered throughout , and no reference is made to the corresponding partial differential , \amp ; c. , the solutions of which may be deduced in the usual manner by cing by ' and and by and .
No point of interest arises in connexion with this transition .
Following the notation usually adopted in connexion with the differential tion ( 1 ) , is used throughout to denote a positive integer .
The expression 1 ) , which occurs only towards the end of S .
and in S , is denoted by headings of the eight sections , with the numbers of the articles which they contain and the pages , are as follows : S I. Direct integration of the differential equation in series , and connexion between the particular integrals .
Arts .
1-7 ; pp. 766-774 .
S IT .
Integration of the differential equation when an integer .
Arts .
8-15 ; pp. 774-779 .
S .
Transformations of the original differential equation .
RICCATI'S equation .
Arts .
16 , 17 ; pp. S .
Special forms of the particular integrals in cases in which the differential equations admit of integration in a finite form .
Arts .
] , 19 ; pp. 783 , 784 .
S .
Evaluation of definite integrals satisfying the differential equat ) .
Arts .
20-28 , pp. 784-797 .
S .
Symbolic forms of the particular integrals in the cases in which the equations admit of integration in a finite form .
Arts .
29-42 ; pp. 798-819 .
S .
Connexion with BESSEL'S Functions .
Arts .
48-48 ; pp. 819-822 .
S VIII .
Writings specially connected with the contents of the memoir .
Pp .
823-828 .
S I. Direct of the equation in series , and connexion between thoe particular .
Arts .
1-7 .
1 .
The most direct method of integrating the differential equation ( 1 ) , and the relations that exist between the different particular integrals , appears to be as follows .
Let the summation extending to all positive integral values of then , substituting in the differential equation , we have whence , putting or 1 , or aking the first root , the equations giving .
are EQUATION AND ITS whence so that the solution corresponding to the root ' is where , as throughout this memoir , denotes 1.2.3 .
Similarly , taking the root , the other solution is found to be , as and V are independent , the complete integral of the differential equation is , A and being arbitrary constants .
There is nothing in the form of these series to indicate that for any values of the integral of the differential equation of being expressed in a finite form .
They show , however , that if the half of an uneven integel ( the case alone excepted ) the solution assumes a different form , .
if , say , in the terms aftel certain point become infinite , the solution is of the form cx , being a new .
This case is excluded in what follows ; and throughout the melnoi ] is supposed not to be of the forms .
If , however , is of either of these forms only certain of the series considered will involve innfinite terms , and th relations connecting those series which do not involve infinite terms will still remain true .
2 .
Transforming the differential equation ( 1 ) by assuming , a ) substitution suggested by the form of the first member of the equation , we obtain the differential equation in Putting as before we have whence or Taking the first root , the equations are giving , .
; and we obtain the particular integral Similarly , the other particular integral is found to be If we had transformed ( 1 ) by assuming , we should have obtained a differential equation in differing from that given above only in having the of changed : and the two particular integrals would differ from thosejust written only in having the signs of the alternate terms negative .
3 .
Thus , of the differential equation we have obtained the six particular integrals , where EQUATION AND ITS TRANSFORMATIONS .
These integrals form three pairs and and and , either of the integrals in each pair being deducible from the other by the substitution of for : and , since the differential equation involves only in the form , it is evident priori that if in any expression satisfying the differential equation , is replaced by , the new expression must still satisfy the differential equation .
Also the pairs and and , are convertible the one into the other by changing the sign of 4 .
If is a positive the in and terminate and the general integral of the differential equation is ; and if is a negative integer , the series in and terminate and the general integral is Thus , if , the general integral is ; and , if , the general integral is 5 .
As however we have six particular integrals , of which , for any given value of only two can be independent , it renlains to investigate the relations between the particular rals in the different cases that arise .
Suppose unrestricted ( except as mentioned in art .
1 ) , but not equal to an integer .
In this case all the series extend to infinity , and for , leaving out of consideration the factor that occurs in both , the coefficient of in .
and we see that the expression in brackets is equal to the coefficient of in } expansion of \mdash ; p. that is , in If , therefore , is uneven the coefficient of in is zero , and if is even the coefficient which is the coefficient of in U. Since differs from only in having the sign of changed , and since is a function of only , it follows that .
Also , since differ from only in having in place of , it follows that ( 2 ) .
Suppose a positive integer , , say .
In this case the term of the in , including the factor , is and the next term yanishes owing to the presence of the factor or in the numerator .
For the same reason all the succeeding terms vanish until the factor appears in the denominator also , when the zero factors cancel one another and the series recommences , the first term of the new series being 1 . !
, where EQUATION AND ITS TRANSFORMATIONS .
The new series , multiplied by the factor , thus becomes Denoting then by the finite part of , the series being supposed to end at the term immediately preceding the first term which contains a zero factor in the numerator , viz. putting we have found that Similarly , if denotes the finite part of , the series ending at the term immediately preceding the first term which contains a zero in the numerator , we find that and also , as before , The proof in that does not apply as it stands when , but it be extended so as to include this case by putting , and ] indefinitely smail .
The equality of and for all values of oved without the use of limits by showing that the coefficient of in to the coefficient of in P. To prove this ; first suppose to be eyen and , then the of in is equal to The last term and the expression in brackets is equal to the coefficient of in 3 The coefficient of in the expansion of is equal to and therefore the coefficient of in is eqnal to is the coefficient of in when the factors are divided out from the numerator and denominator .
Similarly , if ] , the coefficient of in is found to be equal to , is the coefficient of in when the factors are divided out from the numerator and denominator .
Thus , if , the coefficients of the terms involving , . .
in the series in vanish , and we have ( 3 ) If a negative integer , then and involve zero terms , and , denoting by and the values of and when the series are supposed to terminate at the term preceding the first term involving a zero factor in the numerator , V , and become equal to , and when is put equal to , that is , to the and of and vice versed .
In this case , therefore , and The l.elations between the particular integrals in the three cases are therefore ( 1 ) not an integer , ( 2 ) a positive integer , , ( 3 ) negative integer , , ; ; 6 .
If we suppose the series always to terminate directly a zero factor appears in a numerator that denoted , the relations are ( 1 ) not ; ( 2 ) positive ; ( 3 ) negative integer , ; The change of form of the relations , which in this mode of statement appears so remarkable , does not , as we have seen , occur if the series be supposed to extend to infinity in all cases .
It may be observed that it is ] from the manner in which the series were obtained in arts .
1 and 2 that we are always at liberty to stop at the term immcdiately preceding the first term containing a zero factor in the numel.ator , as this finite portion of the series satisfies the differential equation , and that the , second series obtained by allowing the terms to recommence and to extend to infinity also satisfies the diffe ] Gntial equation .
The phrase ' term preceding the first term .
a zero factor in the numerator\ldquo ; has been used in p1eference to preceding the zero term\ldquo ; in order to include the cases of or , in which no zero term occurs .
7 .
It was shown in art .
5 that Putting in this identity , we have The right-hand side of this equation is unaltered by a change of sign of , and therefore , putting which is true for all values of , except a negative even integer .
Writing in place of , it follows that which is true for all values of except negative uneven integer .
Several deductions from this formula are given in a paper " " Generalised Form of Certain Series\ldquo ; Proceedings London Mathematical Society , .
S II .
Integration of the differential equation when inte.ger .
Arts .
8-15 .
8 .
A particular integral of the equation is for , from this value of we find at once by differentiation EQUATION AND ITS TRANSFORMATIONS .
775 whence 9 .
Let the above value of be expanded in powers of , so that then .
and therefore satisfies the differential equation Thus the general integral of this differential equation is A. coefficient of in expansion of .
coefficient of in expansion of The particular integrals to which the different modes of expansion of lead will now be examined , and connected with the forms already obtained in S I. 10 .
The coefficient of in the expansion of is equal to the coefficient of in and the coefficient of in Thus the coefficient of in the terms involving uneven powers of . .
where Of the terms involving even powers of the first that contains a term in is so that the coefficient of in the terms even powers of The complete coefficient of in the expansion of therefore being the same as in art 5 .
11 .
Now and we obtain other forms of the integral by finding the coefficients of in the expansion of } and of } , and multiplying them by and respectively .
It is well known that and where in the second series , if is an even positive integer ffie coefficients of the terms involving are zero , and if is an uneven positive integer the coefficients of the terms involving .
are zero .
Putting hese formulae become EQUATION AND ITS TRANSFORMATIONS .
The coefficient of in therefore and the coefficient of in 12 .
The coefficient of in , that is in is , by the last rticle , equal to for , when is greater than , there is no term involving This expression for the constant multiplier ' 1.3.5 . . .
which is the quantity denoted by in art .
10 .
The coefficient of in the of is therefore 13 .
The coefficient of in in is , by art .
10 , equal to MR. J. W. L. GLAISHER ON RICCATI'S The coefficient of in the expansion of } is therefore ; and we have so that the coefficient of .
14 .
Thus the three forms of the same integral which are obtained by the expansion of are Changing the sign of , we obtain as the coefficient of in the expansion of the values , , giving the three equal integrals FQUATION AND ITS TRANSFORMATIONS .
Therefore whence , which with the relations found for the case of a positive integer in art .
5 .
If is a tive integer , then ; we may therefore replace by throughout in the ' just obtained , and thus deduce the system of rals considered in of 15 .
It may be that , since the series for in art .
11 terminate and recommence when is respectively a negative or positive integer , it is evident that the solutions in series of the equation satisfied by them will present points of similarity to the solutions and of ( 1 ) .
The former differential equation is and its integration in series is considered in a paper " " mple Illustratiye of a Point in the Solution of Differential in Series\ldquo ; Iessenger of ' vol. yiii .
, pp. 20-23 ) .
mations of the .
RICCATT'S .
Arts .
1 16 .
If the differential equation is transformed by assuming , it becomes This equation therefore admits of integration in a finite form when an integer , and the six particular integrals , which are equal respectively to are connected with one another , in the different cases , by the same relations as those found for in art .
5 .
If we put , so that the differential equation becomes ( 3 ) , then the six integrals take the forms The differential equation admits of integration in a finite form if an uneven integer , and the relations between the particular integrals are the same as in art .
5 , viz. , accented letters denoting the terminated series before , not an integer , ; a positive uneven integer , , ; EQUATION AND ITS TRANSFORMATTONS .
781 a negative uneven rrer , ; where This is perhaps the simplest form in which the six integrals can be exhibited ; ctnd , having regard merely to the simplicity of the series and to the expression of the manner in which they are related to one another , ( 3 ) should be preferred as the standard form of the differential equation both to the original form ( 1 ) and to RICCATI'S equation ( 4 ) , which is sidered in the next article .
The form is thab adopted by BACH in his memoir ( see .
of It may be )bserved that if , a positiye , the differential equation ( 2 ) is satisfied by coefficient of in the expansion of , and if , by coefficient of in the expansion of ; these results follow from SII .
17 . .
the equation ( 3 ) by , it ) ecomes or , putting ' ( 4 ) .
RICCATI'S equation in its original form is ; it may nout loss of generality be written d- .
and , putting ) becomes Thus ( 4 ) is the equation derived from ' by assuming ' and it is to it as form ] ) ' equation .
MR. J. W. L. GLAISHhR ON RICCATI'S The six particular integrals of ( 4 ) are The differential equation admits of integration in a finite form if the reciprocal of an uneyen integer , and , the terminated series being denoted by accented letters as the relations between the particular integrals are the same as in art .
5 , viz. ( 1 ) not the reciprocal of an uneven integer , ; ( 2 ) the reciprocal of an uneven positive integer , ; ( 3 ) the reciprocal of an uneven negative integer , ; where The integrals were given by CAYLEY in the 'Philosophical Magazin ' Fourth series , vol. 36 , pp. 348-351 ( November , 1868 ) .
EQUATION AND ITS 7S3 Special forms of the in the cases in which the admit of integration in .
Arts .
, 19 .
18 .
When the differential equations admit of tion in inin o ) finite number of terms , these finite particular be in form by commencing the terminating series the other end .
Thus in the case of the differential equation if is a positive integer , the particular integral . . .
; so that , if an integer , the finite particular ( the series being continued till it terminates of itself a zero factor ) , a similal ' expression derived this by the sign of 19 .
Similarly , if an uneven , then and a similar expression derived from this by changing the sign of , are particular integrals of the differential equation In the case of RICCATI'S equation , if the reciprocal of an uneven integer , the two particular inte , orals are and a similar expression derived from this by changing the sign of These appear to be the best forms in which the integrals can be presented when the equations admit of solution in a finite form : but they do not suggest the solutions for the general cases when bhe letters are unrestricted .
The series ultimately become divergent when they do not terminate .
S V. of definite integrals satisfying the equations .
Arts .
20-28 .
20 .
It was shown by that the definite integral ( 5 ) satisfies the RICCATI'S equation ( 6 ) , so that the yalue integral must be of the form , where are the same as in art .
17 , and ; it remains to determine the constants A and B. * ' Journal \amp ; colc Polytechnique , ' Cahier , p. 237 .
EQUATION AND ITS TRANSFORMA .
TIONS .
It is however more convenient in the first place to consider the definite integral in the form which is obtained by transforming ( 5 ) by the substitution , for the integral ( 5 ) thus becomes Comparing ( 6 ) with the standard form ( 4 ) of equation in art .
17 , we have , so that , and Let ; then , and we see that the definite integral satisfies the differential equation The value of this definite integral is therefore of the form , where are the same as in art .
16 , being substituted for and pub : viz. , writing then Suppose positive , and put ; we thus find , and therefore form the integral by assuming ; this equation then becomes N. whence , changing the sign of ; and it follows therefore that .
Thus for all values of except , course , even integer ) ( 7 ) the series extending to infinity in eyery case .
The method by which the fundamental formula ( 7 ) has obtained is open to some objections .
These will be noticed , and a complete proof of ( 7 ) given , in art .
28 .
21 .
We have , since ; and , if is an uneven integer , this being as defined in art .
16 , when is put EQUATION AND ITS TRANSFORMATIONS .
Now , if is a positive uneven integer , we have , by art .
16 , ; and is equal .
to when is written for , and put ; so that , if is a positive uneven integer , the series terminating at the term preceding the first term containing a zero factor in the numerator .
Transforming the integral by assuming , we find that , if is a negative uneven integer , the series terminating when the first zero factor appears .
Thus , generally , ( 8 ) , if is not equal to an integer : but if a positive uneven integer , the first series continued up to the first term containing a zero factor in the numerator is the alue of the integral , the second series being ignored altogether ; and if a negative uneven integer , the second series continued up to the first term containing zero factor in the numerator , is the yalue of the integral , the first series ignored altogether .
The rule may therefore be stated as follows : if neither sel.ies terllinates then ( 8 ) represents the yalue of the integral , but if one of the series terminates , the finite series represents the value of the integral , the other being ignored ; a series being supposed to terminate at the term preceding the first term that contains a zero factor in the numerator .
The apparent change of form is curious , but the reason for it fully appeared in S I. , arts .
3-6 .
In the 'British Association Peport ' for lS72 ctions of the Sections , pp. 15-17 ) I gave the formula ( 8 ) with a brief indication of the method by which it had been obtained ; this method is substantially the same as that just explained .
As far as I know the general value of the integral had not been given before ; although the value in the case of an uneven integer has been long known .
It is scarcely necessary to remark that in ( 8 ) must not an even integer : this case is specially excepted throughout ( see end of art .
1 ) .
22 .
The case when an uneven integer is included in a general formula given by CAUCHY in vol. .
cf his 'Exercices des Mathematiques ' ( 1826 ) , pp. 54-56 .
He has there shown that if being a positive integer , and an e.ven function , then .
( 9 ) .
This is the case corresponding to of a formula proved by BOOLE ( Philosophical Transactions , vol. , 1857 , p. 783 ) , ( 10 ) .
BOOLE'S formula may however be deduced from CAUCHY'S ; for , replacing by , we have and this integral , transformed by assumin , becomes in which , if we put and replace by , the expression subject to the functional sign becomes * In the ' Messenger of Mathematics , ' vol. , p. 79 , I stated that CAUCHY'S proof was not applicable to the more general theorem in which -was replaced by .
The error was corrected in a paper " " On a Formula of CAUCHI'S for the Evaluation of a Class of Definite Integrals ' Proceedings of the Cambridge Philosophical Society , ' vol. , pp. 5-12 , 1876 ) ; this paper contains also the theorem , corresponding to CAUCHY'S , in which is an uneven function .
EQUATION AND ITS TRANSFORMATIONS .
To deduce the value of the integral from BOOLE'S formula , let , then . .
; and therefore the coefficient of Thus ( 11 ) , whence ( 12 ) , the series being continued till it terminates of itself .
This formula is in effect that given by Exercices , for the evaluation of the integral .
It had however , as CAUCHY himself remarks , been previously published by LEGENDRE in vol. , p. 366 , of his 'Exercices de Calcul Integral ' ( 1811 ) .
LEGENDRE , whose method is quite rent to CAUCHY'S , adds that EULER , in vol. , p. 415 of his 'Institutiones Calculi Integralis ' ( 1794 ) mentions the integrals which correspond to the particular cases and of the integral in ( 12 ) , as apparently not admitting of evaluation by known methods ; and he gives their values .
If in the series in ( 11 ) the terms be written in the order , we have which agrees with ( 8 ) when , since * ' Supplcmentum V. ' ad tom .
1 , cap. viii .
Transformin the integral in ( 12 ) by assuming , it is seen that ( 12 ) is true also when a negative integral ; so that this formula is true when any integer .
The same transformation shows that in general , if is an even function , Putting , the formula ( 12 ) assumes the form which is true when an uneven integer , the series being continued till it terminates of itself .
23 .
The investigations of the formulae ( 9 ) and ( 10 ) given by CAUCHY and BOOLE are only applicable in the case of an integer , and do not indicate what the formulae become when is unrestricted .
A method , however , which I employed the ' Messenger of Mathematics ' ( vol. ii .
, 1872 , pp. 78 , 79 ) to prove BOOLE'S formula , and which depends on direct transformations of the integrals , leads to the general theorem .
We have ; and if we transform the second integral by assuming , then , since is an even function , we find that the original integral Now transform this integral by assuming ; we thus have and , taking the upper sign , the integral becomes If is an eyen integer , the quantity in square brackets this expression containing tertns .
EQUATION AND ITS TRANSFORMATIONS .
Thus , if is an even integer , we have which agrees with BOOLE'S formula .
But if is not an even integer , the expression in square brackets when expanded contains an infinite number of terms , and putting , as before , the general formula is inf .
inf .
24 .
This formula involves infinite terms unless is such a function that the integrals , . . .
are all finite .
This condition is not fulfilled when is infinite when or , so that we do not obtain by means of the formula a demonstration of the equation ( 8 ) .
If however we replace by in all the terms , whether the integral be really infinite not , we do in fact , as we should expect , obtain ( 8 ) .
For , putting , substituting gamma-functions for the integrals , and writing in place of , the formula gives The coefficients are readily identified with in ( s ) , for evidently This process , regarded as a method of obtaining the formula ( 8 ) , is of course unsound , and could not be rendered satisfactory without careful discussion and development .
Such substitutions , however , very frequently give correct results , and it is generally interesting to examine whether , in any case that arises , the result so derived is true or not .
In this instance also we thus obtain two new forms of the expression forming the right hand member of ( 8 ) .
25 .
Transforming ( 8 ) by the assumption , putting , and writing for , it will be found that ( 13 ) ; where , as before , if either series terminates through the presence of a zero factor in a numerator , the terminating series represents the value of the integral , and if neither series terminates , both are to be included .
When one of the series terminates , that is , when twice the reciprocal of an uneven integer , the formula may , by taking the terms of the series in the reverse order , be written Putting and becomes If for example , we have the well-known result 26 .
The definite integral has been evaluated when is a positive integer , the formula in this case being ( 14 ) .
* See ScHLolIILcH , ' Analytische Studien ' Leipzig , , part ii .
, p. 97 , and CRELLE'S Journal , vol. xxxiii .
, p. 273 , or CATALAN , ' LIOUVILLE'S Journal , ' ser. 1 , vol. , p. 110 .
EQUATION AND ITS TRANSFORMATIONS .
This result may be readily obtained by differentiating both members of the equation times with regard to : see infr , art .
31 .
I now proceed to investigate the value of the integral when is unrestricted : it is to be observed , however , that must be positive and greater than unity , for otherwise the integral is infinite in value .
It is to prove that the int , egral satisfies the differential equation ; for , by actual differentiation , ; and by a double integration by parts we find that Thus the value of the integral must be of the form , where and are as defined in art .
3 , and A and are constants to be determined .
It is howeyer more convenient to avoid the determination of the constants by deducin the value of the integral from the formula ( 8 ) of In the 'Journal de l'Ecole ique , ' Cah . .
( vol. ix p. 241 , has proved a formula which , after some unimportant transformations , may be written POISSON'S demonstrat , ion holds good for all values of such that the integral ) the right-hand side of the equation is finite .
for ?
and sforming the right-hand integral by assuming , this equation becomes * 'Quarterly Journal of Mathematics , ' vol. , p. 130 .
; whence , replacing by ( 15 ) which represents the value of the integral for all values of greater than unity .
If is a positive integer the first series terminates through the presence of a zero factor in a numerator , and this finite series is the value of the integral , the second series being ignored .
If is a positive integer , then , writing the terms of the series in the reverse order , which agrees with ( 14 ) .
EQUATION AND ITS TRANSFORMATIONS .
27 .
In a similar manner we may obtain the ] of the integral which is finite for all values of greater than unity .
For , differentiating ( 15 ) with respect to , we have where , as before , if the first series terminates , the finite portion of it represents the value of the integral .
If is a positive integer , and the terms of the series are written in the reverse order , we find which is a known result SCHLoMILCH , Anal. Stud .
It follows at once by combining ( 15 ) and ( 16 ) that , for all values of such that the integrals are not infinite , , if or It would be strange if this equation were new , but I have not met with it anyn : it is readily proyed in the case of an integer , for and whence which , differentiated times with regard to .
giyes the relation in question .
28 .
The method by which the rmula ( 7 ) was obtained in art .
20 is not satisfactory for two reasons , ( i ) because the lntegral is infinite in value when is negative , ] the gamma-function , which is supposed to satisfy the equation for all values of , is finite when is nega , tive , except when is a negative integer , so that we are not entitled to assume that we may always replace the integral by the gamma-function , and ( ii ) , because it is assumed that we may change the sign of in the equation giving the value of the integral .
The following demonstration of the formula ( 7 ) is , however , I believe , quite rigorous .
The gamma-function is supposed to be defined by the equation from to , and by the equation for all other values of This is in effect the definition of the gamma-function generally adopted in analysis .
We have seen in art .
20 that satisfies the differential equation , so that ; and by a simple transformation of the integral , it follows that ( 17 ) , where Suppose to be intermediate to and 1 .
Put in ( 17 ) and we have , whence .
Now by actual differentiation of the series represented by and we find that and similarly Transforming the integral in ( 17 ) by assuming , it becomes ; EQUATION AND ITS TRANSFORMATIONS .
whence , differentiating with regard to Now lies between and 1 , and putting , we have giving .
Thus , if lies between and 1 , it has been proved that and this equation can be readily shown to be true for all values of by differentiating both members of it any number of times with regard to or For , differentiatin with regard to ' and , differentiating with regard to If , therefore , the formula ( 7 ) is true when , it is true when ; and it has been proved to be true for all values of between and 1 : it is true for all real values of It may be remarked that whatever value may have , the integral is never infinite : so that the differentiations with regard to or are always permissible .
S Symbolic forms of the particular integrals in the cases in which the differential equations admit of integration in a finite form .
Arts .
29-42 .
29 .
It has been shown in art .
26 that .
( 18 ) satisfies the differential equation ( 1 ) .
Now and therefore , if is a positive integer , complete integral of ( 1 ) is therefore ( 19 ) ; and , since , this result be written also ( 20 ) .
Since the differential equation ( l ) remains unaltered if is substituted for it follows that the complete integral of ( 1 ) may be expressed also in the forms , and .
30 .
Putting , we see that the complete integral of the differential equation EQUATION AND ITS TRANSFORMATIONS .
when is an integer .
either positive or negative , is given by any one of the formulae , , , .
Putting now , where and , the differential equation becomes ( 4 ) ; and the integrals take the forms , If is a positive integer , so that , then , from the first and third forms , or ; and , if is a negative integer , so that , then , ffom the second and urth forms , or .
31 .
These formulae may be readily connected with the series-integrals found in S I. , for , comparing ( 20 ) with the series in arts .
3 and 5 , we see that where A is a constant .
Putting in place of , this equation becomes ; and , observing that the coefficient of is , it is evident that Writing the terms in the reverse order , as in S , we find ( 21 ) ; that is , on replacing by which is a known formula ( see , for example , SCHLoMILCH'S 'Analytische Studien ' ( 1848 ) , p. 86 ) .
The formulae which result from comparing the solutions of CATI S equation ( 4 ) in arts .
17 and 30 are EQUATION AND ITS TRANSFORMATIONS .
if and if ID the case of , we have , and therefore , . .
; also \mdash ; \mdash ; , so that ) , and similarly , \amp ; c. Thus the -coefficient which multiplies the right-hamd side of ( 22 ) ; and , writing the terms on the right-hand side of ( 22 ) in the order , the formula becomes where Treating the formula ( 23 ) in the same manner , we find where The right-hand members of these two formulae differ from one another and the last expression in S .
( art .
19 ) only by the of -hich occur as in the two former expressions .
MlJCCCLXXXL 32 .
It follows from the forms of in art .
29 that and it car ] be readily verified that , so that we have ( 24 ) .
Transforming this result by putting cted ) , it becomes If now this may be written and , on putting , we obtain the same result ; so that this formula holds good whenever is of the forms It follows from this theorem and from the two formulae at the end of the last article that where The relation ( 24 ) , or , as it be written more conveniently , .
( 25 ) , admits of being established as follows .
Suppose expanded in ascending of , and consider the term in : we have .
QUATION AND ITS TION , 80.3 and ' so that . . .
; therefore , The preceding investigation shows also that , if denotes any function , then ( 26 ) ; for this theorem has been proved to be true when is of the form .
; and as it merely asserts an identical relation between the derived functions of , it must hold good universally , since the ruth of such a relation could not be dependent on the fact of whether was or was not expressible in any particular form .
33 .
The property upon which the theorem ( 26 ) depends is that the symbols of operation , \amp ; c. are convertible as regards order ; that is to say , operating with upon , the result is the same in whatever the are performed .
This is evident , for so that the result of the operations ) and therefore , is independent of the order in which they are performed .
Now the left-hand side of ( 25 ) by is *See CAYLEY , 'Proceedings of the London [ athcmatical ' vol. viii .
, ) ' Solutions of the Cambridge Scnate-House Problems and Riders ' for 1878 , pp. and , writing the operators in a different order , this expression .
This investigation of ( 25 ) is in effect the same as that given in the last article , but the form in which the process is presented is somewhat preferable .
Denoting for the moment the operator by , then . . . .
Also , the operators in the reverse order , Thus and therefore , replacin by , ' ( 27 ) , or , writing for .
( 28 ) , 34 .
Putting , and and respectively in ( 27 ) and ( 2S ) , these formulae give ; EQUATION AN D ITS IATIONS .
80 ence The complete integral of ( l ) may therefore be written also in the form ( 29 ) or in the form ( 30 ) .
The first of these solutions , viz. ( 29 ) , is that given by BOOLE in the ' Philosophical Transactions ' for and in his tise on Differential Equations , ' chap .
xyii .
, BOOLE'S process is as follows : he shows that where denotes ' and he thence deduces that .
But if the factors are written in the reverse order , we have ( D\mdash ; 3 ) .
which is ( 19 ) .
BOOLE does not seem to have anywhere alluded to the connexion between his own form ( 29 ) and the form ( 19 ) , or to have remarked that the tter w obtainable by his own method .
Putting ( unrestricted ) in ( 27 ) and ( 28 ) , we find * ' On a General Method in lysis , \ldquo ; p. : .
; so that the solution of RICCATI'S equation may be written also in the forms , if , and if 35 .
BOOLE'S foIYn ( of the solution of the equation ( 1 ) can be obtained also from the definite integral ( 18 ) in art .
29 ; for we have , if , when is a positive integer , EQUATION AND ITS TRANSFORMATIONS .
leading to the complete integral It will be observed that both integrals being finite for every positive integral value of , and that the second integral when integrated with regard to between the limits and The of these two integrals is zero , as it can be shown that diminishes as increases , and , in the limit when is infinite , vanishes .
A similar remark applies to the second integration with regard to .
The above process does not therefore involve the assumptions , 86 .
POISSON'S theorem quoted in Sy .
, art .
20 , .
that the definite integral ( 5 ) satisfies the differential equation ( 6 ) shows that equation ( 4 ) is satisfied by the definite integral ( 31 ) .
Putting , and forming the integral by taking , we find that which ] , if so that the differential equation is satisfied by ( 32 ) .
Similarly , by transforming ( 31 ) to the form , we find that the differential equation is satisfied by ( 33 ) , if that is , if The formulae ( 32 ) and ( 33 ) , on substituting for its value in terms of , lead at once to the solutions given in art .
29 .
method was applied by POISSON* to show that the equation ( 4 ) is integrable when The formula ( 25 ) of art .
32 may be easily deduced from the equation for we have and also so that whence ( 34 ) .
* Journal de cole Polytechnique , ' vol. , pp. 236 , 237 .
POISSON'S investigation is reproduced in DE MORGAN'S 'Differential and Integral Calculus , ' pp. 703 , 704 .
The formulae ( 32 ) and ( 33 ) are obtained in the same manner as in the text from the integral ( 31 ) , and the solutions in art .
30 are deduced from them , in a paper " " On RICCATI'S Equation , Quarterly Journal of Mathematics , ' vol. xi .
, 1871 , pp. 267-273 ) .
In the paper " " On RICCATI'S equation referred to in the preceding note , the following two formulae occur These are inaccurate owing to the omission of the factor in both , and a wrong sign in the latter when these corrections are made , both become identical with ( 34 ) .
The formula ( 34 ) is given , and is deduced from it , in a paper " " Sir un propriete de la fonction \ldquo ; ( ' ] Correspondence Nathematique , ' vol. , p. 240 ) .
EQUATION AND ITS TRANSFORMATTONS .
Now , identically , denoting any function , , and therefore ( 34 ) may be transformed into ' which , putting , becomes Therefore ; replacing by , this formula becomes whence , taking ( 35 ) .
As was shown in art .
32 , this is a particular case of the more general formula ; and this formula itself admits of generalisation , as it can be proved that , , and generally , being any positive integer , .
These formulae are obtained in a paper ' On Certain Identical Relations published in the ' Proceedings of the London Mathematical cQociety , ' vol. viii .
( 1876 ) , pp. 47-51 .
.37 .
As mentioned in art .
{ 34 , the integral of the differential equation was first given by BOOLE in the ' Philosophical Transactions ' for 1844 .
The integral ( 36 ) is due to Mr. GASKIN , and was in effect given by him in a problem set in the Cambridge Senate House Examination for 1839 .
The problem is as follows : ' ' If be the greatest root of the equation or are general values of in the equation according as is an integer or fraction : and in the first case where ; apply the first or third result to solve the equation Thus Mr. GASKIN'S theorem is that the solution of is ( 37 ) , where is to be put equal to after the performance of the differentiations , being a positive integer , and that in general .
( 38 ) , being any positive quantity .
* The problem forms the second part of Question 8 of the paper set on the afternoon of Tuesday , January 8 , 1839 ( ' Cambridge University Calendar , ' 1839 , p. 319 ) .
EQUATION AND 1TS TRANSFORMATIONS .
The form ( 37 ) is readily identified with ( 36 ) , for , from , , if A method of proving the theorems contained in Mr. GASKIN'S question is given in HYMERS'S 'Treatise on DiffereIltial Equations , and on the Calculus of Finite Differences ' idge , pp. 83-85 .
The result ( 38 ) is there verified by showing that satisfies the differential equation ; and it is remarked that ( 37 ) may be verified in a similar manner by showing being put equal to , satisfies The integral ( 37 ) was subsequently obtained by R. LESLIE ELLIS by a dificrent process in the 'Cambridge Mathematical JournaL ' vol. , p. ( February , ] 841 ) .
A full account of method , with its application to the equation in question , is given in DE MORGAN'S ' Differential and Integral Calculus , ' pp. 701-703 .
In a paper , ' Remarques sir l'o'quation \ldquo ; ( ' LIOUyTLLE'S , ' vol. xi .
, 1846 , pp. 338-340 ) , M. LEBESGUE proved that the integrals of the equations * In the second edition ( 1858 ) of HYMEBS'S work , only the proof that ( 38 ) satisfies the equation is given , no reference being made to Mr. GASKIN'S other result .
An account of BOOTE'S solution and method , taken from the 'Philosophical Transactions ' for 1844 , is however pp. 99-106 .
" " On the Integration of Certain DifTerential Equations ] ) MR. J. W. L. GLAISHER ON RICCATI'S and : are respectively . . .
and .
where , and the former of the two expressions involves differentiations and the latter In the 'Philosophical May , 1856 , Mr. BENJAMIN obtained by a symbolic method the integrals of the differential equations in the respective forms , and that of the equation in the form ; and in the 'Philosophical Transactions ' for 1857 the late Professor DONKIN obtained , also by a symbolic method , the integral of this last equation in the form .
* ' On the Solution of Certain Differential Equations\ldquo ; Philosophical Magazine , ' Fourth series , vol. xi .
, pp. 364-371 ) .
" " On the Equation of LAPLACE'S Functions , \amp ; c vol. 147 , p. 44 .
A proof that the integral of the partial differential equation , which is a simple transformation of ( 1 ) , may be presented in the form is given by Professor C. NIVEN in the ' Solutions of the Senate-House Problems and Riders ' for 1878 , pp. 158 , 159 .
EQUATION AND ITS TRANSFORMATIONS .
38 .
Taking the differential equation in the form ( 1 ) , which has been adopted as the standard form in this memoir , it may be observed that , although the integrals and ( 39 ) have thus been given more than once by different mathematicians , the slightly modified form seems scarcely to have been noticed . ?
' It was this form which led me to the solution in SII .
as folows ; if is written for after the performance of the differentiations , then .
coefficient of in coefficient of in !
coefficient of in ' !
coefficient of in In my paper ' On a Differential Equation allied to RICCATI'S\ldquo ; ( ' Quarterly Journal , ' vol. , 1872 , p. 136 , I deduced by this method from the form ( 39 ) , which is the same as ( 19 ) of art .
29 , that the solution of ( 1 ) was coefficient of in but I did not then remark the far more simple form coefficient of in 39 .
It is interesting to connect Mr. GASKIN'S definite-integral ( 38 ) of art .
37 with that given in art .
26 .
The latter is .
( 40 ) , * The integral is however in cffcct expressed in this form in 's , p. 92 .
being supposed to be any positive quantity ; and the process of verifying that this is a solution of the equation is as follows .
By actual differentiation , we have , as in art .
26 , ( 41 ) , and , by a double integration by parts , ( 42 ) .
The integral ( 40 ) therefore satisfies the differential equation , since the quantity in square brackets vanishes between the limits of integration .
If these limits had been any quantities independent of , instead of , we should have obtained a result corresponding to ( 41 ) , but the quantity in square brackets in ( 42 ) would not have vanished .
Replacing by , it is clear that will satisfy the rential equation if can be so chosen that is zero .
This would be the case , but these values of are inadmissible as they are not independent of Transforming now the integral in ( 40 ) by the substitution , we have and therefore ( 43 ) also satisfies the differential equation .
To verify this , we find by differentiation .
( 44 ) *As denotes a positive integer in this memoir , is used to denote .
EQUATION AND ITS TRANSFORMATIONS .
and , by integration by parts , ( 4.5 ) , whence the right-hand member of ( 44 ) If the limits were the differential ation would still be satisfied if the quantity in square brackets in ( 45 ) vanished between these limits .
This is not the case for any other values of and besides and , but if in ( 43 ) is replaced by , so that the integral is xidt , then the quantity in square brackets , which vanishes when , and therefore , he differential equation is satisfied by the integral xtdt .
Since in this case the quantity in square brackets vanishes in virtue of the factor , we may replace by , being any constant , so that the solution of the differential equation may be written ( 46 ) .
If in the differential equation be replaced by , this integral becomes ( 47 ) , which is Mr. 's formula ( 38 ) .
40 .
This is not however , as stated by Mr. GASKIN , the general integral of the differential equation , as it in fact contains only one arbicrary constant .
For , evidently , so that the introduction of the constant does not increase the generality of the solution .
Returning to the integral ( 46 ) , we find , on putting and transforming the integral by the substitution ( 4S ) ( 49 ) .
Now , as will be shown in the next article , ( 50 ) , so that the particular integral ( 49 ) is equivalent to .
The complete solution of the differential equation is , by art .
29 , , which may therefore be written This is the complete solution in the form corresponding to ( 48 ) .
41 .
To prove the relation ( 50 ) , let EQUATION AND ITS TRANSFORMATIONS .
then , denoting for the sake of brevity by whence and therefore ( 51 ) .
Again , and ; therefore whence ( 52 ) .
If instead of the integral we start with the integral we have and , integrating the second term by parts as before , we find so that .
Thus from ( 52 ) and ( 53 ) , from ( 51 ) , which is the relation ( 50 ) .
It follows from ( 50 ) in connexion with ( 21 ) of art .
31 that .
( 54 ) , a formula given by HARGREAVE in the 'Philosophical Transactions ' for 1848 , p. 34 .
42 .
In the paper just referred to HARGREAVE obtained by a symbolic process the solution of the differential equation in the form , and thence , by ( 54 ) , deduced the solution in the expanded form .
HARGREAVE also gives on p. 45 of his memoir the complete solution of the equation in the form .
( 55 ) .
One or other of the definite integrals in ( 55 ) is however always infinite , except when lies between and 1 .
In the case of the differential equation ( 1 ) , this solution becomes or , as it may be written more conveniently , * On the Solution of Linear Differential Equations pp. EQUATION AND ITS TRANSFORMATIONS .
It is easy to connect these definite integrals with the series and of art .
3 , for whence Similarly V ; and we thus obtain expressions for and V as definite integl.als , taken between the limits 1 and , for all values of for which the integrals fnite .
S Connexion tnctions .
Arts .
43-48 .
43 .
If the equation ( 1 ) is transformed by putting , it assumes the form ( 56 ) .
The equation of BESSEL'S Functions is ( 57 ) , so that ( 56 ) becomes identical with ( 57 ) if We may therefore pass from the solutions of the equation ( 1 ) to the solutions of BESSEL'S equation ( 57 ) by multiplying by and putting , .
The BESSEL'S Function , , may be defined for real values of greater than - by either of the formulae ( 58 ) , ( 59 ) , where denotes , as throughout , .
Comparing ( 58 ) with the expression V in art .
3 , we see that if , and if is replaced by , the series in the two formulae become identical , the exact relation between V and BESSEL'S Function being , where A denotes the constant and is supposed to be positive .
The formula ( 59 ) corresponds to Mr. GASKIN'S definite integral solution ( 38 ) or to one of the definite integrals in HARGRrAVE'S solution ( 55 ) .
45 .
It is known that may be exhibited as the sum of two series multiplied respectively by and , viz .
( 60 ) , where *LOMMEL'S 'Studien uber die Bessel'schen Functionen ' ( 1868 ) , , or TODHIINrER'S 'Treatise on LAPLACE'S tions , LAME'S Functions , and BESSEL'S Functions ' ( 1875 ) , p. 292 .
EQUATION AND ITS TRANSFORMATIONS .
This formula may be written the expression on the right-hand side therefore corresponds to where and are as defined in art .
3 , so that the algebraic theorem to which the two forms of BESSEL'S Functions ( 58 ) and ( 60 ) lead is .
46 .
The formula involving descending series for BESSEL'S Function , is . .
; the descending series ultimately diverge for all values of for which they do not terminate , but the converging terms may be used for the calculation of ; and this formula was fact employed by HANSEN in the calculation of his tables of and .
If being an integer , the series terminate we finite expression for .
Replacing the sine and cosine by their exponential formula may be written where 1 \mdash ; \amp ; .
and differs from only in having all the terms positive .
' Studien uber die Bcssel'schcn unctionen , ) .
Putting , we have therefore where \mdash ; \amp ; c denotes a similar series , having all the terms positive .
If is a positive integer , this expression corresponds to , when the tenns in and are written in the reverse order , as in S IV .
If is not an integer , the series , as already mentioned , are divergent , so that , strictly speaking , the formula only has a meaning when it contains a finite number of terms .
An expression can however be found for the remainder after a finite number of terms , i. e. , for the difference between and the sum of these terms , by means of which the use of the formula in calculation may be justified .
47 .
It is a known theorem in BESSEL'S Functions that if is a positive integer , JP is equal to the coefficient of in the expansion of and it follows therefore , by means of the relation between V and in art .
44 , that , if an even positive integer coefficient of in , and if an uneven positive integer coefficient of in .
48 .
It was shown in II .
that the differential equation ( 1 ) was satisfied by the coefficient of ) in the expansion of , and we thus find that if being a positive integer , the general integral of BESSEL'S equation ( 57 ) is coefficient of in the expansion of , , , , EQUATION AND ITS ATIONS .
S VIIL Writings speciality cozmected with the of ) } [ When only a portion of a paper relates to the subject of the memoir , the pagenumbers efer only to this portion .
] rlNGS REI IN SS I. , , III .
1868 .
CAYLEY .
" " tion 'Philosophical azine , series , vol. xxxvi .
, pp. 348-351 .
The equation is written in the and the expressions ) of art .
17 obtained by of the forms in question and equating coefficients .
Two of the is the reciprocal of an uneyen integer .
( ii .
) 1869 .
\mdash ; " " Note on he tion of Certain uations by Series 'Messenger of Mathematics , ' First series , vol. , pp. It is shown that if we have a solution of a differential equation , and that if one of the factors , say ishes , then we may stop at the preceding term , the finite so obtained being a integral ; but that if we continue the , notwithstanding the evanescelJt ; and if at length a factor in a denolnir]ator , say , vanishes , then the series recommences with the term inyolying , and we have culother rticular in ) , in which may be replaced arbitrary constant B. ( iii .
) 1872 .
GLAISHER .
' On the elations between the in Solution of Eqnation 'Philosophical ' Fourth vol. xliii .
, pp. 433-438 .
The relations between ) , ) ) given in art .
17 obtained .
hesG afford an example of the principle in ( ii .
) .
See the , 1 ) .
763 .
: ( iv .
) 1874 .
BACH .
" " De l'Integration par les Series de quation \ldquo ; ' Annals Scientifiques de cole Normale Superieure .
' Deuxieme serie , vol. pp. 47-68 .
Detailed account , with developments , of and .
In is written in place of and in place of , so that the series are reduced to the forms given in art .
16 .
If the differential equation is similarly transformed it becomes This is the form of the equation adopted by M. BACH , who finally deduces the series in the case of RICCATI'S equation .
The form is a very convenient one .
See art .
16 .
1878 .
GLAISHER .
" " Example Illustrative of a Point in the Solution of Differential Equations in Series 'Messenger of Mathematics , ' vol. viii .
, pp. 20-23 .
In the well-known expansions quoted in art .
11 , viz. the series are such that if is an integer , one terminates , and after a certain number of zero terms , recommences and reproduces the other .
It follows therefore that the differential equation whose general integral is must afford an example of the principle pointed out in .
The differential equation is found to be and its integration in series affords the illustration referred to in the title .
The note was suggested by art .
11 .
See art .
15 .
1878 .
\mdash ; " " Generalised Form of Certain Series 'Proceedings of the London Mathematical Society , ' vol. ix .
, pp. 197-202 .
Theorems deduced from EQUATION AND ITS TRANSFORMATIONS .
See art .
7 .
( vii .
) 1878 .
\mdash ; " " On the Solution of a Differential Equation allied to \ldquo ; ' British Association Report ' for 1878 ( Dublin ) , pp. 469 , 470 .
Proof that the coefficient of in the expansion of satisfies the differential equation See arts .
8 , 9 .
WRITINGS To Sy .
( viii .
) 1813 .
POISSON .
" " Memoir sir les grales Definies de l'Ecole Polytechnique , ' vol. .
( cab .
xvi pp. 236-239 , 241 .
It is proved that if \ldquo ; then satisfies the differential equation , and it is deduced this result that the equation is integrable in a finite form when .
See arts .
20 , 36 .
A relation between two definite integl.als is proved .
art .
26 .
( ix .
) 1872 .
GLAISHER .
On the lation i Series of Definite ls ' British Association Report ' for 1872 ( Brighton , , Transactions of the Sections , 15-17 .
Investigation of the formula ( s ) of by the process given in , 22 .
WRITINGS To 1N SVI .
1839 .
GASKIN .
Senate House Problem .
The solution of the equation is given in the forms beiug put to after the differentiations , and See arts .
37 , 39 .
( xi .
) 1839 .
YMERS .
'A Treatise on Equations and on the Calculus of Finite Differences ' ( 1839 ) , pp. 83-85 ; also , second edition ( 1858 ) , p. 125 .
Solution of Mr. GASKIN'S problem in .
See art .
37 .
( xii .
) 1841 . .
" " On the Integration of Certain Differential Equations ' Cambridge Mathematical Journal , ' vol. , pp. 193-195 .
Independent investigation of the first of Mr. GASKIN'S forms in .
Seo art .
37 .
( xiii .
) 1841 .
DE MORGAN .
The Differential and Integral Calculus , ' pp. 702-704 .
of ELLIS'S method see and of POISSON'S determination of the rable cases of RICCATI'S equation ( see viii See arts .
36 , 37 .
1844 .
BOOLE .
" " On a Method in Analysis 'Phi]osophical Transactions ' for 1844 , pp. 251 , 252 .
This paper contains BOOLE'S general symbolic method .
The solution of the equation ( 1 ) is given in the form The general method and this solution are reproduced with only slight changes in BOOLE'S ' Differential Equations , ' chapter xyii .
See arb .
34 .
( xy .
) 1846 .
LEBESGUE .
" " Remarques sir ' LIO JVtLLE'S Journal , ' vol. xi .
, pp. 338 , 339 .
Solution of this equation in a form involving repeated differentiations with regard to .
See art .
37 .
( xvi .
) 1848 .
HARGREAVE .
" " On the Solution of Linear Differential Equations ' Philosophical Transactions ' for 1848 , pp. 34 , 35 , 45 .
The paper contains the general integral of ( 1 ) in the forms , and a development of in a series .
There are also solutions of other allied equations .
See arts .
41 , EQUATION AND ITS S. 827 ( xvii .
) 1856 .
WILLIAMSON .
" " On the Solution of Certain Differential Equations ' Philosophical azineo , ' Fourth series , vol. xi .
, pp. 364-369 .
The general integral of the equation is given in the form , and the solutions of RICCATI'S and several other equations are also obtained .
The symbolic expressions are developed by means of the theorem , of which a proof is given .
See art .
37 .
( xyiii .
) 1857 .
DONKIN .
" " On the Equation of LAPLACE'S ncticns , \amp ; c 'Philosophical Transactions ' for 1857 , p. 44 .
The integral of the equation in ( xvii .
) is given in the form .
This solution occurs in a note , as an example of the application of the genel.al method of the paper to a particular equation .
See art .
( xix .
) 1871 .
GLAISHER .
" " On EquatioI 'Quarterly Journal of Mathematics , ' vol. xi .
pp. By means of the definite integral ( 31 ) of art .
36 , the solution of RICCATI'S equation is obtained in the forms , \amp ; .
and the formulae ( 22 ) and ( 23 ) of art .
.ql arc proved .
See arts .
31 , 36 .
( xx .
) 1872 .
\mdash ; " " On a Differential Equation allied to RICCATI'S.\ldquo ; 'Quarterly Journal of Mathematics , ' vol. , pp. 129-137 .
The equation is ( 1 ) , and the definite integral is applied as in art .
to obtain the eneral integral in the form and also in BOOLE'S form ( 29 ) : the results are transformed so as to give the symbolic solution of BICCATI'S equation , which is integrated also by BOOLE'S method .
See arts .
29 , 30 , 33 , 34 , 35 , 38 .
( xxi .
) 1876 .
\mdash ; " " Sir un Proprie'fe de la Fonction 'Nouvelle Correspondence Mathematique , ' vol. , pp. 240-243 , 349-350 .
Proof of the theorem by means of the integral See art .
36 .
( xxii .
) 1876 .
\mdash ; " " On Certain Identical Differential Equations 'Proceedings of the London Society , ' vol. viu pp. 47-51 .
Generalisations of the theorem in ( xxi .
) , as for example , and other similar results .
See arts .
33 , 36 .
( xxiii .
) 1879 .
' On a bolic Theorem involving Repeated Differentiations ' Proceedings of the Cambridge Philosophical Society , ' vol. , pp. 269-271 .
The theorem is ( 50 ) of art .
40 , viz. and the proof is the same as in art .
41 .

