1 SUBROUTINE ZLATM4( ITYPE, N, NZ1, NZ2, RSIGN, AMAGN, RCOND,
2 $ TRIANG, IDIST, ISEED, A, LDA )
3 *
4 * -- LAPACK auxiliary test routine (version 3.1) --
5 * Univ. of Tennessee, Univ. of California Berkeley and NAG Ltd..
6 * November 2006
7 *
8 * .. Scalar Arguments ..
9 LOGICAL RSIGN
10 INTEGER IDIST, ITYPE, LDA, N, NZ1, NZ2
11 DOUBLE PRECISION AMAGN, RCOND, TRIANG
12 * ..
13 * .. Array Arguments ..
14 INTEGER ISEED( 4 )
15 COMPLEX*16 A( LDA, * )
16 * ..
17 *
18 * Purpose
19 * =======
20 *
21 * ZLATM4 generates basic square matrices, which may later be
22 * multiplied by others in order to produce test matrices. It is
23 * intended mainly to be used to test the generalized eigenvalue
24 * routines.
25 *
26 * It first generates the diagonal and (possibly) subdiagonal,
27 * according to the value of ITYPE, NZ1, NZ2, RSIGN, AMAGN, and RCOND.
28 * It then fills in the upper triangle with random numbers, if TRIANG is
29 * non-zero.
30 *
31 * Arguments
32 * =========
33 *
34 * ITYPE (input) INTEGER
35 * The "type" of matrix on the diagonal and sub-diagonal.
36 * If ITYPE < 0, then type abs(ITYPE) is generated and then
37 * swapped end for end (A(I,J) := A'(N-J,N-I).) See also
38 * the description of AMAGN and RSIGN.
39 *
40 * Special types:
41 * = 0: the zero matrix.
42 * = 1: the identity.
43 * = 2: a transposed Jordan block.
44 * = 3: If N is odd, then a k+1 x k+1 transposed Jordan block
45 * followed by a k x k identity block, where k=(N-1)/2.
46 * If N is even, then k=(N-2)/2, and a zero diagonal entry
47 * is tacked onto the end.
48 *
49 * Diagonal types. The diagonal consists of NZ1 zeros, then
50 * k=N-NZ1-NZ2 nonzeros. The subdiagonal is zero. ITYPE
51 * specifies the nonzero diagonal entries as follows:
52 * = 4: 1, ..., k
53 * = 5: 1, RCOND, ..., RCOND
54 * = 6: 1, ..., 1, RCOND
55 * = 7: 1, a, a^2, ..., a^(k-1)=RCOND
56 * = 8: 1, 1-d, 1-2*d, ..., 1-(k-1)*d=RCOND
57 * = 9: random numbers chosen from (RCOND,1)
58 * = 10: random numbers with distribution IDIST (see ZLARND.)
59 *
60 * N (input) INTEGER
61 * The order of the matrix.
62 *
63 * NZ1 (input) INTEGER
64 * If abs(ITYPE) > 3, then the first NZ1 diagonal entries will
65 * be zero.
66 *
67 * NZ2 (input) INTEGER
68 * If abs(ITYPE) > 3, then the last NZ2 diagonal entries will
69 * be zero.
70 *
71 * RSIGN (input) LOGICAL
72 * = .TRUE.: The diagonal and subdiagonal entries will be
73 * multiplied by random numbers of magnitude 1.
74 * = .FALSE.: The diagonal and subdiagonal entries will be
75 * left as they are (usually non-negative real.)
76 *
77 * AMAGN (input) DOUBLE PRECISION
78 * The diagonal and subdiagonal entries will be multiplied by
79 * AMAGN.
80 *
81 * RCOND (input) DOUBLE PRECISION
82 * If abs(ITYPE) > 4, then the smallest diagonal entry will be
83 * RCOND. RCOND must be between 0 and 1.
84 *
85 * TRIANG (input) DOUBLE PRECISION
86 * The entries above the diagonal will be random numbers with
87 * magnitude bounded by TRIANG (i.e., random numbers multiplied
88 * by TRIANG.)
89 *
90 * IDIST (input) INTEGER
91 * On entry, DIST specifies the type of distribution to be used
92 * to generate a random matrix .
93 * = 1: real and imaginary parts each UNIFORM( 0, 1 )
94 * = 2: real and imaginary parts each UNIFORM( -1, 1 )
95 * = 3: real and imaginary parts each NORMAL( 0, 1 )
96 * = 4: complex number uniform in DISK( 0, 1 )
97 *
98 * ISEED (input/output) INTEGER array, dimension (4)
99 * On entry ISEED specifies the seed of the random number
100 * generator. The values of ISEED are changed on exit, and can
101 * be used in the next call to ZLATM4 to continue the same
102 * random number sequence.
103 * Note: ISEED(4) should be odd, for the random number generator
104 * used at present.
105 *
106 * A (output) COMPLEX*16 array, dimension (LDA, N)
107 * Array to be computed.
108 *
109 * LDA (input) INTEGER
110 * Leading dimension of A. Must be at least 1 and at least N.
111 *
112 * =====================================================================
113 *
114 * .. Parameters ..
115 DOUBLE PRECISION ZERO, ONE
116 PARAMETER ( ZERO = 0.0D+0, ONE = 1.0D+0 )
117 COMPLEX*16 CZERO, CONE
118 PARAMETER ( CZERO = ( 0.0D+0, 0.0D+0 ),
119 $ CONE = ( 1.0D+0, 0.0D+0 ) )
120 * ..
121 * .. Local Scalars ..
122 INTEGER I, ISDB, ISDE, JC, JD, JR, K, KBEG, KEND, KLEN
123 DOUBLE PRECISION ALPHA
124 COMPLEX*16 CTEMP
125 * ..
126 * .. External Functions ..
127 DOUBLE PRECISION DLARAN
128 COMPLEX*16 ZLARND
129 EXTERNAL DLARAN, ZLARND
130 * ..
131 * .. External Subroutines ..
132 EXTERNAL ZLASET
133 * ..
134 * .. Intrinsic Functions ..
135 INTRINSIC ABS, DBLE, DCMPLX, EXP, LOG, MAX, MIN, MOD
136 * ..
137 * .. Executable Statements ..
138 *
139 IF( N.LE.0 )
140 $ RETURN
141 CALL ZLASET( 'Full', N, N, CZERO, CZERO, A, LDA )
142 *
143 * Insure a correct ISEED
144 *
145 IF( MOD( ISEED( 4 ), 2 ).NE.1 )
146 $ ISEED( 4 ) = ISEED( 4 ) + 1
147 *
148 * Compute diagonal and subdiagonal according to ITYPE, NZ1, NZ2,
149 * and RCOND
150 *
151 IF( ITYPE.NE.0 ) THEN
152 IF( ABS( ITYPE ).GE.4 ) THEN
153 KBEG = MAX( 1, MIN( N, NZ1+1 ) )
154 KEND = MAX( KBEG, MIN( N, N-NZ2 ) )
155 KLEN = KEND + 1 - KBEG
156 ELSE
157 KBEG = 1
158 KEND = N
159 KLEN = N
160 END IF
161 ISDB = 1
162 ISDE = 0
163 GO TO ( 10, 30, 50, 80, 100, 120, 140, 160,
164 $ 180, 200 )ABS( ITYPE )
165 *
166 * abs(ITYPE) = 1: Identity
167 *
168 10 CONTINUE
169 DO 20 JD = 1, N
170 A( JD, JD ) = CONE
171 20 CONTINUE
172 GO TO 220
173 *
174 * abs(ITYPE) = 2: Transposed Jordan block
175 *
176 30 CONTINUE
177 DO 40 JD = 1, N - 1
178 A( JD+1, JD ) = CONE
179 40 CONTINUE
180 ISDB = 1
181 ISDE = N - 1
182 GO TO 220
183 *
184 * abs(ITYPE) = 3: Transposed Jordan block, followed by the
185 * identity.
186 *
187 50 CONTINUE
188 K = ( N-1 ) / 2
189 DO 60 JD = 1, K
190 A( JD+1, JD ) = CONE
191 60 CONTINUE
192 ISDB = 1
193 ISDE = K
194 DO 70 JD = K + 2, 2*K + 1
195 A( JD, JD ) = CONE
196 70 CONTINUE
197 GO TO 220
198 *
199 * abs(ITYPE) = 4: 1,...,k
200 *
201 80 CONTINUE
202 DO 90 JD = KBEG, KEND
203 A( JD, JD ) = DCMPLX( JD-NZ1 )
204 90 CONTINUE
205 GO TO 220
206 *
207 * abs(ITYPE) = 5: One large D value:
208 *
209 100 CONTINUE
210 DO 110 JD = KBEG + 1, KEND
211 A( JD, JD ) = DCMPLX( RCOND )
212 110 CONTINUE
213 A( KBEG, KBEG ) = CONE
214 GO TO 220
215 *
216 * abs(ITYPE) = 6: One small D value:
217 *
218 120 CONTINUE
219 DO 130 JD = KBEG, KEND - 1
220 A( JD, JD ) = CONE
221 130 CONTINUE
222 A( KEND, KEND ) = DCMPLX( RCOND )
223 GO TO 220
224 *
225 * abs(ITYPE) = 7: Exponentially distributed D values:
226 *
227 140 CONTINUE
228 A( KBEG, KBEG ) = CONE
229 IF( KLEN.GT.1 ) THEN
230 ALPHA = RCOND**( ONE / DBLE( KLEN-1 ) )
231 DO 150 I = 2, KLEN
232 A( NZ1+I, NZ1+I ) = DCMPLX( ALPHA**DBLE( I-1 ) )
233 150 CONTINUE
234 END IF
235 GO TO 220
236 *
237 * abs(ITYPE) = 8: Arithmetically distributed D values:
238 *
239 160 CONTINUE
240 A( KBEG, KBEG ) = CONE
241 IF( KLEN.GT.1 ) THEN
242 ALPHA = ( ONE-RCOND ) / DBLE( KLEN-1 )
243 DO 170 I = 2, KLEN
244 A( NZ1+I, NZ1+I ) = DCMPLX( DBLE( KLEN-I )*ALPHA+RCOND )
245 170 CONTINUE
246 END IF
247 GO TO 220
248 *
249 * abs(ITYPE) = 9: Randomly distributed D values on ( RCOND, 1):
250 *
251 180 CONTINUE
252 ALPHA = LOG( RCOND )
253 DO 190 JD = KBEG, KEND
254 A( JD, JD ) = EXP( ALPHA*DLARAN( ISEED ) )
255 190 CONTINUE
256 GO TO 220
257 *
258 * abs(ITYPE) = 10: Randomly distributed D values from DIST
259 *
260 200 CONTINUE
261 DO 210 JD = KBEG, KEND
262 A( JD, JD ) = ZLARND( IDIST, ISEED )
263 210 CONTINUE
264 *
265 220 CONTINUE
266 *
267 * Scale by AMAGN
268 *
269 DO 230 JD = KBEG, KEND
270 A( JD, JD ) = AMAGN*DBLE( A( JD, JD ) )
271 230 CONTINUE
272 DO 240 JD = ISDB, ISDE
273 A( JD+1, JD ) = AMAGN*DBLE( A( JD+1, JD ) )
274 240 CONTINUE
275 *
276 * If RSIGN = .TRUE., assign random signs to diagonal and
277 * subdiagonal
278 *
279 IF( RSIGN ) THEN
280 DO 250 JD = KBEG, KEND
281 IF( DBLE( A( JD, JD ) ).NE.ZERO ) THEN
282 CTEMP = ZLARND( 3, ISEED )
283 CTEMP = CTEMP / ABS( CTEMP )
284 A( JD, JD ) = CTEMP*DBLE( A( JD, JD ) )
285 END IF
286 250 CONTINUE
287 DO 260 JD = ISDB, ISDE
288 IF( DBLE( A( JD+1, JD ) ).NE.ZERO ) THEN
289 CTEMP = ZLARND( 3, ISEED )
290 CTEMP = CTEMP / ABS( CTEMP )
291 A( JD+1, JD ) = CTEMP*DBLE( A( JD+1, JD ) )
292 END IF
293 260 CONTINUE
294 END IF
295 *
296 * Reverse if ITYPE < 0
297 *
298 IF( ITYPE.LT.0 ) THEN
299 DO 270 JD = KBEG, ( KBEG+KEND-1 ) / 2
300 CTEMP = A( JD, JD )
301 A( JD, JD ) = A( KBEG+KEND-JD, KBEG+KEND-JD )
302 A( KBEG+KEND-JD, KBEG+KEND-JD ) = CTEMP
303 270 CONTINUE
304 DO 280 JD = 1, ( N-1 ) / 2
305 CTEMP = A( JD+1, JD )
306 A( JD+1, JD ) = A( N+1-JD, N-JD )
307 A( N+1-JD, N-JD ) = CTEMP
308 280 CONTINUE
309 END IF
310 *
311 END IF
312 *
313 * Fill in upper triangle
314 *
315 IF( TRIANG.NE.ZERO ) THEN
316 DO 300 JC = 2, N
317 DO 290 JR = 1, JC - 1
318 A( JR, JC ) = TRIANG*ZLARND( IDIST, ISEED )
319 290 CONTINUE
320 300 CONTINUE
321 END IF
322 *
323 RETURN
324 *
325 * End of ZLATM4
326 *
327 END
2 $ TRIANG, IDIST, ISEED, A, LDA )
3 *
4 * -- LAPACK auxiliary test routine (version 3.1) --
5 * Univ. of Tennessee, Univ. of California Berkeley and NAG Ltd..
6 * November 2006
7 *
8 * .. Scalar Arguments ..
9 LOGICAL RSIGN
10 INTEGER IDIST, ITYPE, LDA, N, NZ1, NZ2
11 DOUBLE PRECISION AMAGN, RCOND, TRIANG
12 * ..
13 * .. Array Arguments ..
14 INTEGER ISEED( 4 )
15 COMPLEX*16 A( LDA, * )
16 * ..
17 *
18 * Purpose
19 * =======
20 *
21 * ZLATM4 generates basic square matrices, which may later be
22 * multiplied by others in order to produce test matrices. It is
23 * intended mainly to be used to test the generalized eigenvalue
24 * routines.
25 *
26 * It first generates the diagonal and (possibly) subdiagonal,
27 * according to the value of ITYPE, NZ1, NZ2, RSIGN, AMAGN, and RCOND.
28 * It then fills in the upper triangle with random numbers, if TRIANG is
29 * non-zero.
30 *
31 * Arguments
32 * =========
33 *
34 * ITYPE (input) INTEGER
35 * The "type" of matrix on the diagonal and sub-diagonal.
36 * If ITYPE < 0, then type abs(ITYPE) is generated and then
37 * swapped end for end (A(I,J) := A'(N-J,N-I).) See also
38 * the description of AMAGN and RSIGN.
39 *
40 * Special types:
41 * = 0: the zero matrix.
42 * = 1: the identity.
43 * = 2: a transposed Jordan block.
44 * = 3: If N is odd, then a k+1 x k+1 transposed Jordan block
45 * followed by a k x k identity block, where k=(N-1)/2.
46 * If N is even, then k=(N-2)/2, and a zero diagonal entry
47 * is tacked onto the end.
48 *
49 * Diagonal types. The diagonal consists of NZ1 zeros, then
50 * k=N-NZ1-NZ2 nonzeros. The subdiagonal is zero. ITYPE
51 * specifies the nonzero diagonal entries as follows:
52 * = 4: 1, ..., k
53 * = 5: 1, RCOND, ..., RCOND
54 * = 6: 1, ..., 1, RCOND
55 * = 7: 1, a, a^2, ..., a^(k-1)=RCOND
56 * = 8: 1, 1-d, 1-2*d, ..., 1-(k-1)*d=RCOND
57 * = 9: random numbers chosen from (RCOND,1)
58 * = 10: random numbers with distribution IDIST (see ZLARND.)
59 *
60 * N (input) INTEGER
61 * The order of the matrix.
62 *
63 * NZ1 (input) INTEGER
64 * If abs(ITYPE) > 3, then the first NZ1 diagonal entries will
65 * be zero.
66 *
67 * NZ2 (input) INTEGER
68 * If abs(ITYPE) > 3, then the last NZ2 diagonal entries will
69 * be zero.
70 *
71 * RSIGN (input) LOGICAL
72 * = .TRUE.: The diagonal and subdiagonal entries will be
73 * multiplied by random numbers of magnitude 1.
74 * = .FALSE.: The diagonal and subdiagonal entries will be
75 * left as they are (usually non-negative real.)
76 *
77 * AMAGN (input) DOUBLE PRECISION
78 * The diagonal and subdiagonal entries will be multiplied by
79 * AMAGN.
80 *
81 * RCOND (input) DOUBLE PRECISION
82 * If abs(ITYPE) > 4, then the smallest diagonal entry will be
83 * RCOND. RCOND must be between 0 and 1.
84 *
85 * TRIANG (input) DOUBLE PRECISION
86 * The entries above the diagonal will be random numbers with
87 * magnitude bounded by TRIANG (i.e., random numbers multiplied
88 * by TRIANG.)
89 *
90 * IDIST (input) INTEGER
91 * On entry, DIST specifies the type of distribution to be used
92 * to generate a random matrix .
93 * = 1: real and imaginary parts each UNIFORM( 0, 1 )
94 * = 2: real and imaginary parts each UNIFORM( -1, 1 )
95 * = 3: real and imaginary parts each NORMAL( 0, 1 )
96 * = 4: complex number uniform in DISK( 0, 1 )
97 *
98 * ISEED (input/output) INTEGER array, dimension (4)
99 * On entry ISEED specifies the seed of the random number
100 * generator. The values of ISEED are changed on exit, and can
101 * be used in the next call to ZLATM4 to continue the same
102 * random number sequence.
103 * Note: ISEED(4) should be odd, for the random number generator
104 * used at present.
105 *
106 * A (output) COMPLEX*16 array, dimension (LDA, N)
107 * Array to be computed.
108 *
109 * LDA (input) INTEGER
110 * Leading dimension of A. Must be at least 1 and at least N.
111 *
112 * =====================================================================
113 *
114 * .. Parameters ..
115 DOUBLE PRECISION ZERO, ONE
116 PARAMETER ( ZERO = 0.0D+0, ONE = 1.0D+0 )
117 COMPLEX*16 CZERO, CONE
118 PARAMETER ( CZERO = ( 0.0D+0, 0.0D+0 ),
119 $ CONE = ( 1.0D+0, 0.0D+0 ) )
120 * ..
121 * .. Local Scalars ..
122 INTEGER I, ISDB, ISDE, JC, JD, JR, K, KBEG, KEND, KLEN
123 DOUBLE PRECISION ALPHA
124 COMPLEX*16 CTEMP
125 * ..
126 * .. External Functions ..
127 DOUBLE PRECISION DLARAN
128 COMPLEX*16 ZLARND
129 EXTERNAL DLARAN, ZLARND
130 * ..
131 * .. External Subroutines ..
132 EXTERNAL ZLASET
133 * ..
134 * .. Intrinsic Functions ..
135 INTRINSIC ABS, DBLE, DCMPLX, EXP, LOG, MAX, MIN, MOD
136 * ..
137 * .. Executable Statements ..
138 *
139 IF( N.LE.0 )
140 $ RETURN
141 CALL ZLASET( 'Full', N, N, CZERO, CZERO, A, LDA )
142 *
143 * Insure a correct ISEED
144 *
145 IF( MOD( ISEED( 4 ), 2 ).NE.1 )
146 $ ISEED( 4 ) = ISEED( 4 ) + 1
147 *
148 * Compute diagonal and subdiagonal according to ITYPE, NZ1, NZ2,
149 * and RCOND
150 *
151 IF( ITYPE.NE.0 ) THEN
152 IF( ABS( ITYPE ).GE.4 ) THEN
153 KBEG = MAX( 1, MIN( N, NZ1+1 ) )
154 KEND = MAX( KBEG, MIN( N, N-NZ2 ) )
155 KLEN = KEND + 1 - KBEG
156 ELSE
157 KBEG = 1
158 KEND = N
159 KLEN = N
160 END IF
161 ISDB = 1
162 ISDE = 0
163 GO TO ( 10, 30, 50, 80, 100, 120, 140, 160,
164 $ 180, 200 )ABS( ITYPE )
165 *
166 * abs(ITYPE) = 1: Identity
167 *
168 10 CONTINUE
169 DO 20 JD = 1, N
170 A( JD, JD ) = CONE
171 20 CONTINUE
172 GO TO 220
173 *
174 * abs(ITYPE) = 2: Transposed Jordan block
175 *
176 30 CONTINUE
177 DO 40 JD = 1, N - 1
178 A( JD+1, JD ) = CONE
179 40 CONTINUE
180 ISDB = 1
181 ISDE = N - 1
182 GO TO 220
183 *
184 * abs(ITYPE) = 3: Transposed Jordan block, followed by the
185 * identity.
186 *
187 50 CONTINUE
188 K = ( N-1 ) / 2
189 DO 60 JD = 1, K
190 A( JD+1, JD ) = CONE
191 60 CONTINUE
192 ISDB = 1
193 ISDE = K
194 DO 70 JD = K + 2, 2*K + 1
195 A( JD, JD ) = CONE
196 70 CONTINUE
197 GO TO 220
198 *
199 * abs(ITYPE) = 4: 1,...,k
200 *
201 80 CONTINUE
202 DO 90 JD = KBEG, KEND
203 A( JD, JD ) = DCMPLX( JD-NZ1 )
204 90 CONTINUE
205 GO TO 220
206 *
207 * abs(ITYPE) = 5: One large D value:
208 *
209 100 CONTINUE
210 DO 110 JD = KBEG + 1, KEND
211 A( JD, JD ) = DCMPLX( RCOND )
212 110 CONTINUE
213 A( KBEG, KBEG ) = CONE
214 GO TO 220
215 *
216 * abs(ITYPE) = 6: One small D value:
217 *
218 120 CONTINUE
219 DO 130 JD = KBEG, KEND - 1
220 A( JD, JD ) = CONE
221 130 CONTINUE
222 A( KEND, KEND ) = DCMPLX( RCOND )
223 GO TO 220
224 *
225 * abs(ITYPE) = 7: Exponentially distributed D values:
226 *
227 140 CONTINUE
228 A( KBEG, KBEG ) = CONE
229 IF( KLEN.GT.1 ) THEN
230 ALPHA = RCOND**( ONE / DBLE( KLEN-1 ) )
231 DO 150 I = 2, KLEN
232 A( NZ1+I, NZ1+I ) = DCMPLX( ALPHA**DBLE( I-1 ) )
233 150 CONTINUE
234 END IF
235 GO TO 220
236 *
237 * abs(ITYPE) = 8: Arithmetically distributed D values:
238 *
239 160 CONTINUE
240 A( KBEG, KBEG ) = CONE
241 IF( KLEN.GT.1 ) THEN
242 ALPHA = ( ONE-RCOND ) / DBLE( KLEN-1 )
243 DO 170 I = 2, KLEN
244 A( NZ1+I, NZ1+I ) = DCMPLX( DBLE( KLEN-I )*ALPHA+RCOND )
245 170 CONTINUE
246 END IF
247 GO TO 220
248 *
249 * abs(ITYPE) = 9: Randomly distributed D values on ( RCOND, 1):
250 *
251 180 CONTINUE
252 ALPHA = LOG( RCOND )
253 DO 190 JD = KBEG, KEND
254 A( JD, JD ) = EXP( ALPHA*DLARAN( ISEED ) )
255 190 CONTINUE
256 GO TO 220
257 *
258 * abs(ITYPE) = 10: Randomly distributed D values from DIST
259 *
260 200 CONTINUE
261 DO 210 JD = KBEG, KEND
262 A( JD, JD ) = ZLARND( IDIST, ISEED )
263 210 CONTINUE
264 *
265 220 CONTINUE
266 *
267 * Scale by AMAGN
268 *
269 DO 230 JD = KBEG, KEND
270 A( JD, JD ) = AMAGN*DBLE( A( JD, JD ) )
271 230 CONTINUE
272 DO 240 JD = ISDB, ISDE
273 A( JD+1, JD ) = AMAGN*DBLE( A( JD+1, JD ) )
274 240 CONTINUE
275 *
276 * If RSIGN = .TRUE., assign random signs to diagonal and
277 * subdiagonal
278 *
279 IF( RSIGN ) THEN
280 DO 250 JD = KBEG, KEND
281 IF( DBLE( A( JD, JD ) ).NE.ZERO ) THEN
282 CTEMP = ZLARND( 3, ISEED )
283 CTEMP = CTEMP / ABS( CTEMP )
284 A( JD, JD ) = CTEMP*DBLE( A( JD, JD ) )
285 END IF
286 250 CONTINUE
287 DO 260 JD = ISDB, ISDE
288 IF( DBLE( A( JD+1, JD ) ).NE.ZERO ) THEN
289 CTEMP = ZLARND( 3, ISEED )
290 CTEMP = CTEMP / ABS( CTEMP )
291 A( JD+1, JD ) = CTEMP*DBLE( A( JD+1, JD ) )
292 END IF
293 260 CONTINUE
294 END IF
295 *
296 * Reverse if ITYPE < 0
297 *
298 IF( ITYPE.LT.0 ) THEN
299 DO 270 JD = KBEG, ( KBEG+KEND-1 ) / 2
300 CTEMP = A( JD, JD )
301 A( JD, JD ) = A( KBEG+KEND-JD, KBEG+KEND-JD )
302 A( KBEG+KEND-JD, KBEG+KEND-JD ) = CTEMP
303 270 CONTINUE
304 DO 280 JD = 1, ( N-1 ) / 2
305 CTEMP = A( JD+1, JD )
306 A( JD+1, JD ) = A( N+1-JD, N-JD )
307 A( N+1-JD, N-JD ) = CTEMP
308 280 CONTINUE
309 END IF
310 *
311 END IF
312 *
313 * Fill in upper triangle
314 *
315 IF( TRIANG.NE.ZERO ) THEN
316 DO 300 JC = 2, N
317 DO 290 JR = 1, JC - 1
318 A( JR, JC ) = TRIANG*ZLARND( IDIST, ISEED )
319 290 CONTINUE
320 300 CONTINUE
321 END IF
322 *
323 RETURN
324 *
325 * End of ZLATM4
326 *
327 END