0387986995Basic TheoryC
.pdf128 |
V. SINGULARITIES OF THE FIRST KIND |
|||
Set |
|
|
3 |
3 |
|
1 |
1 |
||
|
4 |
2 |
32 |
32 |
|
0 |
1 |
1 |
3 |
|
|
4 |
16 |
32- |
M2 (x) = 1 o + xvl + x2v2
and
|
192 |
-320 |
P1 |
-16 16 ] |
|
[0 1 |
|
(V.5.19) |
PO = [ 64 |
-18] ' |
= [ -32 48 , |
P2 = |
01 ' |
||
|
Q2(r) = P° + xPl + x2P2. |
|
|
|
|||
Then, Qz-' 11f2(x)Q2(x) |
] |
+ O(x3). Note that |
|
|
|||
|
|
-1 2 |
|
|
|
|
|
|
1 |
5 |
|
|
|
|
|
1,° |
12 |
634 |
|
[192 |
-128] = |
[--21 |
4 |
|
|
[0 |
|
64 |
|
2]. |
|
|
|
|
01] |
|
|
|
|
|
64 |
64 |
|
|
|
|
|
Thus, we arrived at the following conclusion.
Conclusion V-5-10. There exists a unique 2 x 2 matrix P(x) |
00 |
|||
Zx'Pr such |
||||
that |
|
|
|
e=o |
|
|
|
|
|
(1) the matrices Pt for i = 0, 1, 2 are given by (V.5.19). |
|
|||
(2) the power series P(x) conueryes for every x, |
|
|||
(3) the transformation y" = P(x)u" changes the system |
|
|||
|
dy |
0 |
-1 |
|
|
+ X |
0 Y = 0' |
|
|
(V.5.20) |
xdx |
|
||
|
|
4 |
|
|
to |
|
|
|
|
|
d6 |
+ [_-1 2, u" = 0. |
|
|
(V.5.21) |
x3i |
|
Furthermore, the transformation y" = P(x)Po iv changes (V.5.20) to
(V.5.22) |
x |
+ 1 0 |
" = 0. |
|
|
0 |
5. A NORMAL FORM OF A DIFFERENTIAL OPERATOR |
129 |
In the case when a=1, wefixm=2. Then
1 0 -1 0 |
0 |
0 |
|
0 |
|
|
0 |
|
-1 |
0 |
0 |
0 |
0 |
|||
1 |
0 |
0 |
0 |
|
|
0 |
|
|
|
|
0 |
0 |
0 |
0 |
0 |
|
4 |
0 |
|
|
|
|
|
||||||||||
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
||
0 |
0 |
1 |
-1 0 |
|
0 |
|
|
|
|
1 |
|
-1 |
0 |
0 |
||
|
|
|
8 |
|
4 |
1 |
||||||||||
A2 = |
|
|
|
|
|
|
132= |
|
|
|
|
|
||||
0 |
-4 |
|
0 |
|
|
3 |
|
1 |
1 |
1 |
|
|
||||
|
1 |
|
0 |
|
|
|
0 |
0 |
||||||||
|
|
|
|
|
|
|
|
|||||||||
|
|
|
|
16 |
|
8 |
4 |
|
||||||||
|
|
|
|
|
|
|
|
|
|
|
|
|
||||
0 |
0 |
0 |
0 |
2 |
|
-1 |
|
|
1 |
|
1 |
1 |
1 |
2 |
-1 |
|
|
|
|
|
|
|
|
|
|||||||||
|
|
|
|
|
|
|
|
|
16 |
|
16 |
|
4 |
|||
|
|
|
|
|
|
|
|
|
|
|
|
|
||||
0 |
0 |
|
0 |
-4 2 J |
|
|
3 |
|
1 |
3 |
1 |
1 |
2 |
|||
|
|
|
|
|
|
|
|
|
64 |
|
16 |
16 |
8 |
4 |
|
|
and |
|
4 |
|
|
0 |
0 |
0 |
|
0 |
|
01 |
|
|
|
|
|
|
|
|
|
|
|
0 |
|
|
|
|
||||||
|
|
|
|
|
0 |
0 |
0 |
|
0 |
0 |
0 |
|
|
|
|
|
|
|
|
|
|
1 |
1 |
0 |
|
0 |
0 |
0 |
|
|
|
|
|
|
|
|
|
|
8 |
4 |
|
|
|
|
|
|||||
|
|
Ar2 |
= |
|
|
|
|
|
|
|
|
|
|
|||
|
|
|
1 |
1 |
0 |
|
0 |
0 |
0 |
|
|
|
|
|||
|
|
|
|
16 |
8 |
|
|
|
|
|
||||||
|
|
|
|
|
|
|
|
|
|
|
|
|
||||
|
|
|
|
- |
1 |
1 |
1 |
|
1 |
0 0 |
|
|
|
|
||
|
|
|
|
|
16 |
16 |
8 |
|
4 |
|
|
|
|
|
|
|
|
|
|
|
-3 |
1 |
1 |
|
1 |
0 |
0 |
|
|
|
|
||
|
|
|
|
|
64 |
16 |
16 |
|
8 |
|
|
|
|
|||
|
|
|
|
|
|
|
|
|
|
|
|
|||||
|
|
|
|
|
|
|
|
|
|
|
|
384 |
|
32 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
192 |
|
-16 |
|
|
Calculating eigenvectors of S2, we find a 6 x 2 matrix P2 = _72 |
-124 |
such |
||||||||||||||
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
||
|
|
|
|
|
|
|
|
|
|
|
|
2 |
|
0 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
5 |
|
1 |
|
|
|
|
|
|
|
|
|
|
|
|
1 |
0 |
|
|
|
|
|
that S2P2 = P2Ao. Note that, in this case, A0 = 1 2 |
. Set |
|
|
|
||||||||||||
1 |
|
|
|
|||||||||||||
|
|
|
|
|
|
|
|
|
L0 |
|
1 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2 |
|
|
|
|
||
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
||
|
|
|
|
1 |
|
1 |
|
|
|
|
1 |
1 |
|
|
|
|
|
|
vl = |
8 |
|
4 |
v2 = |
|
16 |
16 |
|
|
|
||||
|
|
1 |
|
1 |
|
|
1 |
|
|
|
||||||
|
|
|
|
|
|
|
|
|
|
|
3 |
|
|
|
|
|
|
|
|
|
16 |
|
8 |
|
|
|
|
64 |
16 |
|
|
|
|
|
|
M2(x) = xl/1 + x2v2, |
|
|
|
|
|
|
|
|
||||||
and |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Po |
|
84 |
62 |
|
|
|
-48 -12 |
1'2 = [2 1j , |
|||||||
(V.5.23) |
[19 |
3 |
J' Pl - |
[-72 |
-14] , |
|||||||||||
|
Q2(x) = Po + xP1 + T2 p2.
130 |
V. SINGULARITIES OF THE FIRST KIND |
|
|
0 |
x |
|
48 + Q(x3). Thus, we arrived at the following |
|
Then, Q2-'M2(X)Q2(X) = |
||
conclusion. |
0 |
0 |
|
|
+oo
Conclusion V-5-11. There exists a unique 2 x 2 matrix P(x) = >xtPt such
e=o
that
(1)the matrices Pt for P = 0, 1, 2 are given by (V.5.23),
(2)the power series P(x) convet es for every x,
(3)the transformation y = P(x)u changes the system
|
|
|
-1 |
(V.5.24) |
xdx + |
|
V = 0 |
|
|
|
0 |
to |
|
|
|
(V.5.25) |
x- + |
2 |
48I u = U. |
|
|
0 |
1 |
|
|
|
2 |
For a further discussion, see IHKS[. A computer might help the reader to calculate
S2i N2, and P2 in the cases when a = 0 and a = 1. Such a calculation is not difficult in these cases since eigenvalues of A2 are found easily (cf. §IV-1).
V-6. Calculation of the normal form of a differential operator
In this section, we present another proof of Theorem V-5-1. The main idea is to construct a power series P(x) as a formal solution of the system
dP(x) = P(x)(Ao + vo(x)) - Q(x)P(x)
(cf. Remark V-5-2).
Another proof of Theorem V-5-1.
To simplify the presentation, we assume that So = A0 = diag[,ulI,, µ2I2, ... ,
Aklk[, where pi, p2, ... , µk are distinct eigenvalues of no with multiplicities m1, respectively, and the matrix I, is m, x m j identity matrix. Since AOA(o = NoAo,
the matrix No must have the form No = diag(N01,No2, ... ,Nok[, Where Nor is an
00
mi x m1 nilpotent matrix. Let us determine two matrices P(x) = I,a + E xmP,,,
M=1
|
6. CALCULATION OF THE NORMAL FORM |
131 |
|
|
00 |
|
|
and B(x) = Sto + E xmBm by the equation |
|
|
|
|
m=1 |
|
|
d |
(in+mm) (Oo |
|
|
xa |
+ *n=1 xmBm) |
|
|
|
|
|
|
|
Cep + m=1 |
(In + m 'nPm) |
|
This equation is equivalent to |
|
|
|
|
m-1 |
|
(m > 1). |
mPm = PmS2o - S20Pm + Bm - 1m + ( PhBm-h - fl.-hPh) |
[=1
Therefore, it suffices to solve the equation
(V.6.1) mX + (A0 +N4)X - X(Ao+No) - Y = H,
where X and Y are n x n unknown matrices, whereas the matrix H is given. If we write X, Y, and H in the block-form
X11 ... |
Xlk |
[Y11 |
|
||
Xkl |
Xkk |
Ykl |
... |
Xlk |
H11 |
|
||
... Ykk |
Hkl |
...
...
Xlk
Hkk
where XXh, YYh, and H,h are m, x mh matrices, equation (V.6.1) becomes
(m + P) - µh)X)h + NO,Xjh - X,hNOh - Yjh = Hjh,
where j, h = 1,... , k. We can determine XJh and Y,h by setting Y,;h = 0 if m + u, - Ah 4 0, and X,,h = 0 if m + p, - µh = 0. More precisely speaking, if m + p - Ph 96 0, we determine X3h uniquely by solving
(m + p, - µh)Xlh + NojXjh - X3hNoh = H)h.
If M + Aj - Ah = 0, we set Y,h = H,h. In this way, we can determine P(z) and
B(x). In particular, go + B(x) has the form x ^°I'xA0, where r is a constant
n x n nilpotent matrix. Furthermore, the operator x d +A0 and the multiplication
ds
operator by No + B(x) commute. D
The idea of this proof is due to M. Hukuhara (cf. [Si17, §3.9, pp. 85-891).
Remark V-6-1. In the case of a second-order linear homogeneous differential equation at a regular singular point x = a, there exists a solution of the form
+00
01(z) = (x - a)aE c, (x - a)n, where the coefficients c,, are constants, co 0 0,
n=O
132 |
V. SINGULARITIES OF THE FIRST KIND |
and the power series is convergent. If there is no other linearly independent solution of this form, a second solution can be constructed by using the idea explained in Remark IV-7-3. This second solution contains a logarithmic term. Similarly, a third-order linear homogeneous differential equation has a solution of the form
+00
01(x) _ (x - a)° c, (x - a)" at a regular singular point x = a. Using this solu-
n=0
tion, the given equation can be reduced to a second-order equation. In particular, if there exists another solution ¢(x) of this kind such that ¢1 and ¢2 are linearly independent, then the idea given in Remark IV-7-4 can be used to find a fundamental set of solutions.
In general, a fundamental matrix solution of system (V.5.14) can be constructed if the transformation y" = P(x)xLV of Theorem V-5-4 is found. In fact, if the
definition of fo(x) of Observation V-5-5 is used, |
P(x)xL2-no-L-r is a fundamental |
matrix solution of (V.5.14). The matrix P(x) can be calculated by using the method of Hukuhara, which was explained earlier.
V-7. Classification of singularities of homogeneous linear systems
In this chapter, so far we have studied a system
(V.7.1) |
x dt = A(x)y, |
y E C", |
where the entries of the n x n matrix A(r) are convergent power series in x with complex coefficients. In this case, the singularity at x = 0 is said to be of the first kind. If a system has the form
(V.7.2) |
xk+1 |
= A(x)y", |
y" E C", |
where k is a positive integer and the entries of the n x n matrix A(x) are convergent power series in x with complex coefficients, then the singularity at x = 0 is said to be of the second kind
In §V-5, we proved the following theorem (cf. Theorem V-5-4).
Theorem V-7-1. For system (V.7.1), there exist a constant n x n matrix A0 and an n x n invertible matrix P(x) such that
(i)the entries of P(x) and P(x)-1 are analytic and single-valued in a domain
0 < jxi < r and have, at worst, a pole at x = 0, where r is a positive number,
(ii) the transformation
(V.7.3) |
y' = P(x)i7 |
changes (V.7.1) to a system |
|
(V.7.4) |
du = Aou". |
Theorem V-7-1 can be generalized to system (V.7.2) as follows.
7. CLASSIFICATION OF SINGULARITIES OF LINEAR SYSTEMS 133
Theorem V-7-2. For system (V.7.2), there exist a constant n x n matrix A0 and an n x n invertible matrix P(x) such that
(i)the entries of P(x) and P(x)-1 are analytic and single-valued in a domain
V = {x : 0 < lxi < r}, where r is a positive number,
(ii)the transformation
(V.7.5) |
y = P(x)i |
changes (V.7.2) to (V.7.4).
Proof.
Let 4i(x) be a fundamental matrix of (V.7.2) in D. Since A(x) is analytic and single-valued in D, fi(x) _ I (xe2"t) is also a fundamental matrix of (V.7.2). Therefore, there exists an invertible constant matrix C such that 46(x) = 4i(x)C (cf. (1) of Remark IV-2-7). Choose a constant matrix Ao so that C. = exp[2rriAo] (cf. Ex- ample IV-3-6) and let P(x) = 4(x)exp[-(logx)Ao]. Then, P(x) and P(x)-1 are analytic and single-valued in D. Furthermore,
dP(x) = dam) exp(_(logx)Ao] - P(x)(x-'Ao)
= x-(k+1)A(x)P(x) - P(x)(x-'Ao).
This completes the proof of the theorem. 0
An important difference between Theorems V-7-1 and V-7-2 is the fact that the matrix P(x) in Theorem V-7-2 possibly has an essential singularity at x = 0.
The proof of Theorem V-7-2 immediately suggests that Theorem V-7-2 can be extended to a system
(V.7.6) |
dg |
= F(x)y, |
dx
where every entry of the n x n matrix F(x) is analytic and single-valued on the domain V even if such an entry of F(x) possibly has an essential singularity at x = 0. More precisely speaking, for system (V.7.6), there exist a constant n x n matrix Ao and an n x n invertible matrix P(x) satisfying conditions (i) and (ii) of
Theorem V-7-2 such that transformation (V.7.5) changes (V.7.6) to (V.7.4). Now, we state a definition of regular singularity of (V.7.6) at x = 0 as follows.
Definition V-7-3. Let P(x) be a matrix satisfying conditions (i) and (ii) such that transformation (V.7.5) changes (V.7.6) to (V.7.4). Then, the singularity of
(V.7.6) at x = 0 is said to be regular if every entry of P(x) has, at worst, a pole
atx=0.
Remark V-7-4. Theorem V-7-1 implies that a singularity of the first kind is a regular singularity. The converse is not true. However, it can be proved easily that a regular singularity is, at worst, a singularity of the second kind. Furthermore, if (V.7.2) has a regular singularity at x = 0, then the matrix A(0) is nilpotent. This
is a consequence of the following theorem.
134 |
V. SINGULARITIES OF THE FIRST KIND |
Theorem V-7-5. Let A(x) and B(x) be two n x n matrices whose entries are formal power series in x with constant coefficients. Also, let r and s be two positive integers. Suppose that there exists an n x n matrix P(x) such that
(a)the entries of P(x) are formal power series in x with constant coefficients,
(b)det(P(x)] ,E 0 as a formal power series in x,
16-1 |
= A(x)y to x' 2i _ |
(c) the transformation y = P(x)ii changes the system x' |
|
X |
|
B(x)il.
Suppose also that s > r. Then, the matrix B(O) must be nilpotent.
Proof
Step 1. Applying to the matrix P(x) suitable elementary row and column opera- tions successively, we can prove the following lemma.
Lemma V-7-6. There exist two n x n matrices
+oo |
|
|
+oo |
T(x) _ |
xmTm |
and S(x) _ |
x"'Sm, |
m=0 |
|
|
m=o |
and n integers 1\1,,\2, ... , an such that
(i)the entries of n x n matrices T,,, and Sm are constants.
(ii)det To 0 0 and det So 0 0,
(iii)
T(x)P(x)S(x) = A(z) = diag(x''',zA2,...
(iv)A1<A2<...<An .
The proof of this lemma is left to the reader as an exercise.
Step 2. Change the two systems x'd = A(x)y and x'd6 = B(x)u, respectively,
to xr d = C(x)a and x' d = D(x)ii by the transformations z' = T(x)y and
ii = S(x)v. Then, F = A(x)v. Furthermore, if the matrix D(0) is nilpotent, the matrix B(0) is also nilpotent.
Step 3. Look at D(x) = x'-'A(x)-1C(x)A(x) - x'A(x)-ld ). This shows clearly that if s > r, the matrix D(0) is nilpotent.
The following corollary of Theorem V-7-5 is important.
Corollary V-7-7. Assume that conditions (a), (b), and (c) of Theorem V-7-5 are satisfied. Also, assume that A(O) and B(O) are not nilpotent. Then, r = s.
Definition V-7-8. If a singularity of the second kind is not a regular singularity, this singularity is said to be irregular. In particular, if the matrix A(O) of (V.7.2) is not nilpotent, the singularity of (V.7.2) at x = 0 is said to be irregular of order k. Also, a regular singularity is said to be of order zero.
In order to define the order of singularity at r = 0 for all systems (V.7.2), the
independent variable x must be replaced by x'1P with a suitable positive integer p.
|
7. CLASSIFICATION OF SINGULARITIES OF LINEAR SYSTEMS 135 |
For example, for a differential equation |
|
(V.7.7) |
ah(x)bhl/ = 0, |
|
h=0 |
assume that the coefficients ah are convergent power series in x and that an (x) # 0. Set y; = 6'-1y (j = 0,... , n - 1). Then, (V.7.7) becomes the system
|
f |
0 |
1 |
0 |
0 |
|
|
|
0 |
0 |
0 ... |
0 |
yl |
(V.7.8) |
6y = |
|
|
|
|
/ |
|
|
0 . |
91 |
i! = |
||
|
|
0 |
0 |
1 |
yn |
|
|
|
|
|
|
|
|
|
|
a0 |
|
|
an-1 |
|
|
|
an |
|
|
as |
|
If we change (V.7.8) by the transformation y" = diag[l, x-O, |
'.T-(n-1)0 |
|||||
Iii, |
||||||
then |
|
|
|
|
|
|
(V.7.9) |
|
|
|
|
|
|
x°6u |
|
|
|
|
|
|
0 |
1 |
0 ... |
0 |
|
|
|
0 |
0 |
0 |
|
0 |
|
|
|
|
|
|
|
+ x°diag[l, a, 2a,... , (n - 1)a) u'. |
|
0 |
0 |
0 ... |
1 |
|
|
|
-xna |
ao |
|
|
an-1 |
|
|
a. |
|
|
-x' |
|
|
|
|
|
|
an |
|
|
Set bh(x) = ah(x)x(n-h)O . Choose a non-negative rational number a so that bh(X) an(x)
(h = 0,... , n - 1) are bounded in a neighborhood of x = 0. Also, if a must be positive, choose a so that bh(0) 0 0 for some h. Then, the matrix on the right-hand side of (V.7.9) is not nilpotent at x = 0 if or > 0. Hence, we define a as the order of the singularity of (V.7.7) at x = 0. System (V.7.2) can be reduced to an equation
(V.7.7) (cf. §XIII-5).
The following theorem due to J. Moser [Mo) concerns the order of a given sin- gularity-
Theorem V-7-9. Let A(x) = x-a E x'A be an n x n matriz, where p is an
=o
integer, A are n x n constant matrices, A0 y6 0, and the power series is convergent.
Set
Im(A) = max (P_ 1 + n,0) ,
µ(A) = minlm(T-1AT-T-1dx)J
136 |
V. SINGULARITIES OF THE FIRST KIND |
Here, r = rank(Ao) and min is taken over all n x n invertible matrices T of the
form T(x) = x-9 E x°T,,, where q is an integer, T are constant n x n matrices,
and the power series is convergent. Note that det T(x) 0 0, but det To may be 0. Assume that m(A) > 1. Then, we have m(A) > p(A) if and only if the polynomial
P(A) =
vanishes identically in A, where In is the n x n identity matrix.
The main idea is to find out p(A) in a finite number of steps. Using the quantity p(A), we can calculate the order of the singularity at x = 0. The criterion for m(A) > p(A) is given in terms of a finite number of conditions on the coefficients of A(x). There is a computer program to work with those conditions. In particular, in this way, we can decide, in a finite number of steps, whether a given singularity at x = 0 is regular.
Notice that IP(x)I can be estimated at z = 0 for the matrix P(x) of Theorem V-7- 2, using similar estimates for fundamental matrix solutions of (V.7.2) and (V.7.4).
Therefore, an analytic criterion that a singularity of second kind at x = 0 be a regular singularity is that in any sectorial domain V with the vertex at x = 0, every solution fi(x) satisfies an estimate
c K[xj' |
(x E V) |
for some positive number K and a real number m. These two numbers may depend on 0 and V. In fact, Theorem V-7-2 and its remark imply that a fundamental matrix solution of (V.7.6) is P(x)xA0. The matrix P(x) has, at worst, a pole at x = 0 if and only if !P(x)I < Kjxjr in a neighborhood of x = 0 for some positive number
K and a real number p (cf. [CL, §2 of Chapter 4, pp. 111-1411). Another criterion which depends only on a finite number of coefficients of power series expansion of the matrix A(x) of (V.7.2) was given in a very concrete form by W. Jurkat and D.
A. Lutz [JL] (see also 15117, Chapter V, pp. 115-1411).
Now, let us look into the problem of convergence of the formal solution which was constructed in Theorem V-1-3. Let
n
P = Eah(x)Jh
h=0
be a differential operator with coefficients ah(x) in C{x). Assume that n > I and
a,(x) 0 0. Defining the indicial polynomial |
of the operator P as in §V-1, |
we prove the following theorem.
Theorem V-7-10.
(i)In the case when the degree of f b in s is equal to n, if we change the equation
P[y1 = 0 to a system by setting yt = y and yj = 61-1 [y) (j = 2, ... , n), the system has a singularity of the first kind at x = 0.
EXERCISES V |
137 |
(ii)In the case when the degree of fn ins is less than n, if we change the equation
P[y] = 0 by setting y1 = y and y, = b'-1 [y] (j = 2,... , n), the system has an irregular singularity at x = 0.
Proof
n |
Ltahi |
00 |
|
Tn=r+o |
|
Look at P(xd] _ >ah(x)bh[x°] = xe= x" |
fm(S)xm and set |
|
|
|
ah(x) = xnobh(x). Then, the functions bh(x) are analytic at x = 0. Furthermore,
if the degree of |
is n, we must have bn(0) # 0. Therefore, in this case we can |
|
n-1 |
write the equation P(y) = 0 in the form bo[y] = -bn(Ebh(x)bh(yJ. Claim (i)
h=0
follows immediately from this form of the equation.
If the degree of fn0 is less than n, we must have bn(0) = 0 and b,(0) - 0 for
some j such that 0 < j < n. To show (ii), change Yh further by zh = |
x(h-1)ayh |
|
with a suitable positive rational number a so that the system for (z1, ... , zn) has a singularity of the second kind of a positive order (cf. the arguments given right after Definition V-7-8, and also §XIII-7).
From Theorem V-7-10, we conclude that the formal solution x''(1 + O(x)) of
Theorem V-1-3 is convergent if the degree of f,, (s) in s is n. The following corollary of Theorem V-7-10 is a basic result due to L. Fuchs [Fu].
Corollary V-7-11. The differential equation Ply] = 0 has, at worst, a regular
singularity at x = 0 if and only if the functions ah(x) (h = 0, ... , n - 1) are x
analytic at x = 0.
Some of the results of this section are also found in [CL, Chapter 4, pp. 108-137].
EXERCISES V
V-1. Show that if A is a nonzero constant, H is a constant n x m matrix, and N1 and N2 are n x n and m x in nilpotent matrices, respectively, then there exists one and only one n x m matrix X satisfying the equation AX + NIX - XN2 = H.
V-2. Show that the convergent power series
y = F(a,x) = |
1+a (a+1)...(a+m-1Q(Q+1) Q+m-1) x,, |
|
m=1 |
|
|
|
|
|
satisfies the differential equation |
|
|
x(1-x)!f2 + |
+1)J Y - c3y = 0, |
where a, /3, and -y are complex constants.