1 |
149 |
jeremybenn |
-- CXG2016.A
|
2 |
|
|
--
|
3 |
|
|
-- Grant of Unlimited Rights
|
4 |
|
|
--
|
5 |
|
|
-- Under contracts F33600-87-D-0337, F33600-84-D-0280, MDA903-79-C-0687,
|
6 |
|
|
-- F08630-91-C-0015, and DCA100-97-D-0025, the U.S. Government obtained
|
7 |
|
|
-- unlimited rights in the software and documentation contained herein.
|
8 |
|
|
-- Unlimited rights are defined in DFAR 252.227-7013(a)(19). By making
|
9 |
|
|
-- this public release, the Government intends to confer upon all
|
10 |
|
|
-- recipients unlimited rights equal to those held by the Government.
|
11 |
|
|
-- These rights include rights to use, duplicate, release or disclose the
|
12 |
|
|
-- released technical data and computer software in whole or in part, in
|
13 |
|
|
-- any manner and for any purpose whatsoever, and to have or permit others
|
14 |
|
|
-- to do so.
|
15 |
|
|
--
|
16 |
|
|
-- DISCLAIMER
|
17 |
|
|
--
|
18 |
|
|
-- ALL MATERIALS OR INFORMATION HEREIN RELEASED, MADE AVAILABLE OR
|
19 |
|
|
-- DISCLOSED ARE AS IS. THE GOVERNMENT MAKES NO EXPRESS OR IMPLIED
|
20 |
|
|
-- WARRANTY AS TO ANY MATTER WHATSOEVER, INCLUDING THE CONDITIONS OF THE
|
21 |
|
|
-- SOFTWARE, DOCUMENTATION OR OTHER INFORMATION RELEASED, MADE AVAILABLE
|
22 |
|
|
-- OR DISCLOSED, OR THE OWNERSHIP, MERCHANTABILITY, OR FITNESS FOR A
|
23 |
|
|
-- PARTICULAR PURPOSE OF SAID MATERIAL.
|
24 |
|
|
--*
|
25 |
|
|
--
|
26 |
|
|
-- OBJECTIVE:
|
27 |
|
|
-- Check that the ARCTAN function returns a
|
28 |
|
|
-- result that is within the error bound allowed.
|
29 |
|
|
--
|
30 |
|
|
-- TEST DESCRIPTION:
|
31 |
|
|
-- This test consists of a generic package that is
|
32 |
|
|
-- instantiated to check both Float and a long float type.
|
33 |
|
|
-- The test for each floating point type is divided into
|
34 |
|
|
-- several parts:
|
35 |
|
|
-- Special value checks where the result is a known constant.
|
36 |
|
|
-- Exception checks.
|
37 |
|
|
--
|
38 |
|
|
-- SPECIAL REQUIREMENTS
|
39 |
|
|
-- The Strict Mode for the numerical accuracy must be
|
40 |
|
|
-- selected. The method by which this mode is selected
|
41 |
|
|
-- is implementation dependent.
|
42 |
|
|
--
|
43 |
|
|
-- APPLICABILITY CRITERIA:
|
44 |
|
|
-- This test applies only to implementations supporting the
|
45 |
|
|
-- Numerics Annex.
|
46 |
|
|
-- This test only applies to the Strict Mode for numerical
|
47 |
|
|
-- accuracy.
|
48 |
|
|
--
|
49 |
|
|
--
|
50 |
|
|
-- CHANGE HISTORY:
|
51 |
|
|
-- 19 Mar 96 SAIC Initial release for 2.1
|
52 |
|
|
-- 30 APR 96 SAIC Fixed optimization issue
|
53 |
|
|
-- 17 AUG 96 SAIC Incorporated Reviewer's suggestions.
|
54 |
|
|
-- 12 OCT 96 SAIC Incorporated Reviewer's suggestions.
|
55 |
|
|
-- 02 DEC 97 EDS Remove procedure Identity_1_Test and calls to
|
56 |
|
|
-- procedure.
|
57 |
|
|
-- 29 JUN 98 EDS Replace -0.0 with call to ImpDef.Annex_G.Negative_Zero
|
58 |
|
|
-- 28 APR 99 RLB Replaced comma accidentally deleted in above change.
|
59 |
|
|
-- 15 DEC 99 RLB Added model range checking to "exact" results,
|
60 |
|
|
-- in order to avoid too strictly requiring a specific
|
61 |
|
|
-- result.
|
62 |
|
|
--!
|
63 |
|
|
|
64 |
|
|
--
|
65 |
|
|
-- References:
|
66 |
|
|
--
|
67 |
|
|
-- Software Manual for the Elementary Functions
|
68 |
|
|
-- William J. Cody, Jr. and William Waite
|
69 |
|
|
-- Prentice-Hall, 1980
|
70 |
|
|
--
|
71 |
|
|
-- CRC Standard Mathematical Tables
|
72 |
|
|
-- 23rd Edition
|
73 |
|
|
--
|
74 |
|
|
-- Implementation and Testing of Function Software
|
75 |
|
|
-- W. J. Cody
|
76 |
|
|
-- Problems and Methodologies in Mathematical Software Production
|
77 |
|
|
-- editors P. C. Messina and A. Murli
|
78 |
|
|
-- Lecture Notes in Computer Science Volume 142
|
79 |
|
|
-- Springer Verlag, 1982
|
80 |
|
|
--
|
81 |
|
|
|
82 |
|
|
with System;
|
83 |
|
|
with Report;
|
84 |
|
|
with Ada.Numerics.Generic_Elementary_Functions;
|
85 |
|
|
with Impdef.Annex_G;
|
86 |
|
|
procedure CXG2016 is
|
87 |
|
|
Verbose : constant Boolean := False;
|
88 |
|
|
Max_Samples : constant := 1000;
|
89 |
|
|
|
90 |
|
|
-- CRC Standard Mathematical Tables; 23rd Edition; pg 738
|
91 |
|
|
Sqrt2 : constant :=
|
92 |
|
|
1.41421_35623_73095_04880_16887_24209_69807_85696_71875_37695;
|
93 |
|
|
Sqrt3 : constant :=
|
94 |
|
|
1.73205_08075_68877_29352_74463_41505_87236_69428_05253_81039;
|
95 |
|
|
|
96 |
|
|
Pi : constant := Ada.Numerics.Pi;
|
97 |
|
|
|
98 |
|
|
generic
|
99 |
|
|
type Real is digits <>;
|
100 |
|
|
Half_PI_Low : in Real; -- The machine number closest to, but not greater
|
101 |
|
|
-- than PI/2.0.
|
102 |
|
|
Half_PI_High : in Real;-- The machine number closest to, but not less
|
103 |
|
|
-- than PI/2.0.
|
104 |
|
|
PI_Low : in Real; -- The machine number closest to, but not greater
|
105 |
|
|
-- than PI.
|
106 |
|
|
PI_High : in Real; -- The machine number closest to, but not less
|
107 |
|
|
-- than PI.
|
108 |
|
|
package Generic_Check is
|
109 |
|
|
procedure Do_Test;
|
110 |
|
|
end Generic_Check;
|
111 |
|
|
|
112 |
|
|
package body Generic_Check is
|
113 |
|
|
package Elementary_Functions is new
|
114 |
|
|
Ada.Numerics.Generic_Elementary_Functions (Real);
|
115 |
|
|
|
116 |
|
|
function Arctan (Y : Real;
|
117 |
|
|
X : Real := 1.0) return Real renames
|
118 |
|
|
Elementary_Functions.Arctan;
|
119 |
|
|
function Arctan (Y : Real;
|
120 |
|
|
X : Real := 1.0;
|
121 |
|
|
Cycle : Real) return Real renames
|
122 |
|
|
Elementary_Functions.Arctan;
|
123 |
|
|
|
124 |
|
|
-- flag used to terminate some tests early
|
125 |
|
|
Accuracy_Error_Reported : Boolean := False;
|
126 |
|
|
|
127 |
|
|
-- The following value is a lower bound on the accuracy
|
128 |
|
|
-- required. It is normally 0.0 so that the lower bound
|
129 |
|
|
-- is computed from Model_Epsilon. However, for tests
|
130 |
|
|
-- where the expected result is only known to a certain
|
131 |
|
|
-- amount of precision this bound takes on a non-zero
|
132 |
|
|
-- value to account for that level of precision.
|
133 |
|
|
Error_Low_Bound : Real := 0.0;
|
134 |
|
|
|
135 |
|
|
procedure Check (Actual, Expected : Real;
|
136 |
|
|
Test_Name : String;
|
137 |
|
|
MRE : Real) is
|
138 |
|
|
Max_Error : Real;
|
139 |
|
|
Rel_Error : Real;
|
140 |
|
|
Abs_Error : Real;
|
141 |
|
|
begin
|
142 |
|
|
-- In the case where the expected result is very small or 0
|
143 |
|
|
-- we compute the maximum error as a multiple of Model_Epsilon
|
144 |
|
|
-- instead of Model_Epsilon and Expected.
|
145 |
|
|
Rel_Error := MRE * abs Expected * Real'Model_Epsilon;
|
146 |
|
|
Abs_Error := MRE * Real'Model_Epsilon;
|
147 |
|
|
if Rel_Error > Abs_Error then
|
148 |
|
|
Max_Error := Rel_Error;
|
149 |
|
|
else
|
150 |
|
|
Max_Error := Abs_Error;
|
151 |
|
|
end if;
|
152 |
|
|
|
153 |
|
|
-- take into account the low bound on the error
|
154 |
|
|
if Max_Error < Error_Low_Bound then
|
155 |
|
|
Max_Error := Error_Low_Bound;
|
156 |
|
|
end if;
|
157 |
|
|
|
158 |
|
|
if abs (Actual - Expected) > Max_Error then
|
159 |
|
|
Accuracy_Error_Reported := True;
|
160 |
|
|
Report.Failed (Test_Name &
|
161 |
|
|
" actual: " & Real'Image (Actual) &
|
162 |
|
|
" expected: " & Real'Image (Expected) &
|
163 |
|
|
" difference: " & Real'Image (Actual - Expected) &
|
164 |
|
|
" max err:" & Real'Image (Max_Error) );
|
165 |
|
|
elsif Verbose then
|
166 |
|
|
if Actual = Expected then
|
167 |
|
|
Report.Comment (Test_Name & " exact result");
|
168 |
|
|
else
|
169 |
|
|
Report.Comment (Test_Name & " passed");
|
170 |
|
|
end if;
|
171 |
|
|
end if;
|
172 |
|
|
end Check;
|
173 |
|
|
|
174 |
|
|
|
175 |
|
|
procedure Special_Value_Test is
|
176 |
|
|
-- If eta is very small, arctan(x + eta) ~= arctan(x) + eta/(1+x*x).
|
177 |
|
|
--
|
178 |
|
|
-- For tests 4 and 5, there is an error of 4.0ME for arctan + an
|
179 |
|
|
-- additional error of 1.0ME because pi is not exact for a total of 5.0ME.
|
180 |
|
|
--
|
181 |
|
|
-- In test 3 there is the error for pi plus an additional error
|
182 |
|
|
-- of (1.0ME)/4 since sqrt3 is not exact, for a total of 5.25ME.
|
183 |
|
|
--
|
184 |
|
|
-- In test 2 there is the error for pi plus an additional error
|
185 |
|
|
-- of (3/4)(1.0ME) since sqrt3 is not exact, for a total of 5.75ME.
|
186 |
|
|
|
187 |
|
|
|
188 |
|
|
type Data_Point is
|
189 |
|
|
record
|
190 |
|
|
Degrees,
|
191 |
|
|
Radians,
|
192 |
|
|
Tangent,
|
193 |
|
|
Allowed_Error : Real;
|
194 |
|
|
end record;
|
195 |
|
|
|
196 |
|
|
type Test_Data_Type is array (Positive range <>) of Data_Point;
|
197 |
|
|
|
198 |
|
|
-- the values in the following table only involve static
|
199 |
|
|
-- expressions so no additional loss of precision occurs.
|
200 |
|
|
Test_Data : constant Test_Data_Type := (
|
201 |
|
|
-- degrees radians tangent error test #
|
202 |
|
|
( 0.0, 0.0, 0.0, 4.0 ), -- 1
|
203 |
|
|
( 30.0, Pi/6.0, Sqrt3/3.0, 5.75), -- 2
|
204 |
|
|
( 60.0, Pi/3.0, Sqrt3, 5.25), -- 3
|
205 |
|
|
( 45.0, Pi/4.0, 1.0, 5.0 ), -- 4
|
206 |
|
|
(-45.0, -Pi/4.0, -1.0, 5.0 ) ); -- 5
|
207 |
|
|
|
208 |
|
|
begin
|
209 |
|
|
for I in Test_Data'Range loop
|
210 |
|
|
Check (Arctan (Test_Data (I).Tangent),
|
211 |
|
|
Test_Data (I).Radians,
|
212 |
|
|
"special value test" & Integer'Image (I) &
|
213 |
|
|
" arctan(" &
|
214 |
|
|
Real'Image (Test_Data (I).Tangent) &
|
215 |
|
|
")",
|
216 |
|
|
Test_Data (I).Allowed_Error);
|
217 |
|
|
Check (Arctan (Test_Data (I).Tangent, Cycle => 360.0),
|
218 |
|
|
Test_Data (I).Degrees,
|
219 |
|
|
"special value test" & Integer'Image (I) &
|
220 |
|
|
" arctan(" &
|
221 |
|
|
Real'Image (Test_Data (I).Tangent) &
|
222 |
|
|
", cycle=>360)",
|
223 |
|
|
Test_Data (I).Allowed_Error);
|
224 |
|
|
end loop;
|
225 |
|
|
|
226 |
|
|
exception
|
227 |
|
|
when Constraint_Error =>
|
228 |
|
|
Report.Failed ("Constraint_Error raised in special value test");
|
229 |
|
|
when others =>
|
230 |
|
|
Report.Failed ("exception in special value test");
|
231 |
|
|
end Special_Value_Test;
|
232 |
|
|
|
233 |
|
|
|
234 |
|
|
|
235 |
|
|
procedure Check_Exact (Actual, Expected_Low, Expected_High : Real;
|
236 |
|
|
Test_Name : String) is
|
237 |
|
|
-- If the expected result is not a model number, then Expected_Low is
|
238 |
|
|
-- the first machine number less than the (exact) expected
|
239 |
|
|
-- result, and Expected_High is the first machine number greater than
|
240 |
|
|
-- the (exact) expected result. If the expected result is a model
|
241 |
|
|
-- number, Expected_Low = Expected_High = the result.
|
242 |
|
|
Model_Expected_Low : Real := Expected_Low;
|
243 |
|
|
Model_Expected_High : Real := Expected_High;
|
244 |
|
|
begin
|
245 |
|
|
-- Calculate the first model number nearest to, but below (or equal)
|
246 |
|
|
-- to the expected result:
|
247 |
|
|
while Real'Model (Model_Expected_Low) /= Model_Expected_Low loop
|
248 |
|
|
-- Try the next machine number lower:
|
249 |
|
|
Model_Expected_Low := Real'Adjacent(Model_Expected_Low, 0.0);
|
250 |
|
|
end loop;
|
251 |
|
|
-- Calculate the first model number nearest to, but above (or equal)
|
252 |
|
|
-- to the expected result:
|
253 |
|
|
while Real'Model (Model_Expected_High) /= Model_Expected_High loop
|
254 |
|
|
-- Try the next machine number higher:
|
255 |
|
|
Model_Expected_High := Real'Adjacent(Model_Expected_High, 100.0);
|
256 |
|
|
end loop;
|
257 |
|
|
|
258 |
|
|
if Actual < Model_Expected_Low or Actual > Model_Expected_High then
|
259 |
|
|
Accuracy_Error_Reported := True;
|
260 |
|
|
if Actual < Model_Expected_Low then
|
261 |
|
|
Report.Failed (Test_Name &
|
262 |
|
|
" actual: " & Real'Image (Actual) &
|
263 |
|
|
" expected low: " & Real'Image (Model_Expected_Low) &
|
264 |
|
|
" expected high: " & Real'Image (Model_Expected_High) &
|
265 |
|
|
" difference: " & Real'Image (Actual - Expected_Low));
|
266 |
|
|
else
|
267 |
|
|
Report.Failed (Test_Name &
|
268 |
|
|
" actual: " & Real'Image (Actual) &
|
269 |
|
|
" expected low: " & Real'Image (Model_Expected_Low) &
|
270 |
|
|
" expected high: " & Real'Image (Model_Expected_High) &
|
271 |
|
|
" difference: " & Real'Image (Expected_High - Actual));
|
272 |
|
|
end if;
|
273 |
|
|
elsif Verbose then
|
274 |
|
|
Report.Comment (Test_Name & " passed");
|
275 |
|
|
end if;
|
276 |
|
|
end Check_Exact;
|
277 |
|
|
|
278 |
|
|
|
279 |
|
|
procedure Exact_Result_Test is
|
280 |
|
|
begin
|
281 |
|
|
-- A.5.1(40);6.0
|
282 |
|
|
Check_Exact (Arctan (0.0, 1.0), 0.0, 0.0, "arctan(0,1)");
|
283 |
|
|
Check_Exact (Arctan (0.0, 1.0, 27.0), 0.0, 0.0, "arctan(0,1,27)");
|
284 |
|
|
|
285 |
|
|
-- G.2.4(11-13);6.0
|
286 |
|
|
|
287 |
|
|
Check_Exact (Arctan (1.0, 0.0), Half_PI_Low, Half_PI_High,
|
288 |
|
|
"arctan(1,0)");
|
289 |
|
|
Check_Exact (Arctan (1.0, 0.0, 360.0), 90.0, 90.0, "arctan(1,0,360)");
|
290 |
|
|
|
291 |
|
|
Check_Exact (Arctan (-1.0, 0.0), -Half_PI_High, -Half_PI_Low,
|
292 |
|
|
"arctan(-1,0)");
|
293 |
|
|
Check_Exact (Arctan (-1.0, 0.0, 360.0), -90.0, -90.0,
|
294 |
|
|
"arctan(-1,0,360)");
|
295 |
|
|
|
296 |
|
|
if Real'Signed_Zeros then
|
297 |
|
|
Check_Exact (Arctan (0.0, -1.0), PI_Low, PI_High, "arctan(+0,-1)");
|
298 |
|
|
Check_Exact (Arctan (0.0, -1.0, 360.0), 180.0, 180.0,
|
299 |
|
|
"arctan(+0,-1,360)");
|
300 |
|
|
Check_Exact (Arctan ( Real ( ImpDef.Annex_G.Negative_Zero ), -1.0),
|
301 |
|
|
-PI_High, -PI_Low, "arctan(-0,-1)");
|
302 |
|
|
Check_Exact (Arctan ( Real ( ImpDef.Annex_G.Negative_Zero ), -1.0,
|
303 |
|
|
360.0), -180.0, -180.0, "arctan(-0,-1,360)");
|
304 |
|
|
else
|
305 |
|
|
Check_Exact (Arctan (0.0, -1.0), PI_Low, PI_High, "arctan(0,-1)");
|
306 |
|
|
Check_Exact (Arctan (0.0, -1.0, 360.0), 180.0, 180.0,
|
307 |
|
|
"arctan(0,-1,360)");
|
308 |
|
|
end if;
|
309 |
|
|
exception
|
310 |
|
|
when Constraint_Error =>
|
311 |
|
|
Report.Failed ("Constraint_Error raised in Exact_Result Test");
|
312 |
|
|
when others =>
|
313 |
|
|
Report.Failed ("Exception in Exact_Result Test");
|
314 |
|
|
end Exact_Result_Test;
|
315 |
|
|
|
316 |
|
|
|
317 |
|
|
procedure Taylor_Series_Test is
|
318 |
|
|
-- This test checks the Arctan by using a taylor series expansion that
|
319 |
|
|
-- will produce a result accurate to 19 decimal digits for
|
320 |
|
|
-- the range under test.
|
321 |
|
|
--
|
322 |
|
|
-- The maximum relative error bound for this test is
|
323 |
|
|
-- 4 for the arctan operation and 2 for the Taylor series
|
324 |
|
|
-- for a total of 6 * Model_Epsilon
|
325 |
|
|
|
326 |
|
|
A : constant := -1.0/16.0;
|
327 |
|
|
B : constant := 1.0/16.0;
|
328 |
|
|
X : Real;
|
329 |
|
|
Actual, Expected : Real;
|
330 |
|
|
Sum, Em, X_Squared : Real;
|
331 |
|
|
begin
|
332 |
|
|
if Real'Digits > 19 then
|
333 |
|
|
-- Taylor series calculation produces result accurate to 19
|
334 |
|
|
-- digits. If type being tested has more digits then set
|
335 |
|
|
-- the error low bound to account for this.
|
336 |
|
|
-- The error low bound is conservatively set to 6*10**-19
|
337 |
|
|
Error_Low_Bound := 0.00000_00000_00000_0006;
|
338 |
|
|
Report.Comment ("arctan accuracy checked to 19 digits");
|
339 |
|
|
end if;
|
340 |
|
|
|
341 |
|
|
Accuracy_Error_Reported := False; -- reset
|
342 |
|
|
for I in 0..Max_Samples loop
|
343 |
|
|
X := (B - A) * Real (I) / Real (Max_Samples) + A;
|
344 |
|
|
X_Squared := X * X;
|
345 |
|
|
Em := 17.0;
|
346 |
|
|
Sum := X_Squared / Em;
|
347 |
|
|
|
348 |
|
|
for II in 1 .. 7 loop
|
349 |
|
|
Em := Em - 2.0;
|
350 |
|
|
Sum := (1.0 / Em - Sum) * X_Squared;
|
351 |
|
|
end loop;
|
352 |
|
|
Sum := -X * Sum;
|
353 |
|
|
Expected := X + Sum;
|
354 |
|
|
Sum := (X - Expected) + Sum;
|
355 |
|
|
if not Real'Machine_Rounds then
|
356 |
|
|
Expected := Expected + (Sum + Sum);
|
357 |
|
|
end if;
|
358 |
|
|
|
359 |
|
|
Actual := Arctan (X);
|
360 |
|
|
|
361 |
|
|
Check (Actual, Expected,
|
362 |
|
|
"Taylor_Series_Test " & Integer'Image (I) & ": arctan(" &
|
363 |
|
|
Real'Image (X) & ") ",
|
364 |
|
|
6.0);
|
365 |
|
|
|
366 |
|
|
if Accuracy_Error_Reported then
|
367 |
|
|
-- only report the first error in this test in order to keep
|
368 |
|
|
-- lots of failures from producing a huge error log
|
369 |
|
|
return;
|
370 |
|
|
end if;
|
371 |
|
|
|
372 |
|
|
end loop;
|
373 |
|
|
Error_Low_Bound := 0.0; -- reset
|
374 |
|
|
exception
|
375 |
|
|
when Constraint_Error =>
|
376 |
|
|
Report.Failed
|
377 |
|
|
("Constraint_Error raised in Taylor_Series_Test");
|
378 |
|
|
when others =>
|
379 |
|
|
Report.Failed ("exception in Taylor_Series_Test");
|
380 |
|
|
end Taylor_Series_Test;
|
381 |
|
|
|
382 |
|
|
|
383 |
|
|
procedure Exception_Test is
|
384 |
|
|
X1, X2, X3 : Real := 0.0;
|
385 |
|
|
begin
|
386 |
|
|
|
387 |
|
|
begin -- A.5.1(20);6.0
|
388 |
|
|
X1 := Arctan(0.0, Cycle => 0.0);
|
389 |
|
|
Report.Failed ("no exception for cycle = 0.0");
|
390 |
|
|
exception
|
391 |
|
|
when Ada.Numerics.Argument_Error => null;
|
392 |
|
|
when others =>
|
393 |
|
|
Report.Failed ("wrong exception for cycle = 0.0");
|
394 |
|
|
end;
|
395 |
|
|
|
396 |
|
|
begin -- A.5.1(20);6.0
|
397 |
|
|
X2 := Arctan (0.0, Cycle => -1.0);
|
398 |
|
|
Report.Failed ("no exception for cycle < 0.0");
|
399 |
|
|
exception
|
400 |
|
|
when Ada.Numerics.Argument_Error => null;
|
401 |
|
|
when others =>
|
402 |
|
|
Report.Failed ("wrong exception for cycle < 0.0");
|
403 |
|
|
end;
|
404 |
|
|
|
405 |
|
|
begin -- A.5.1(25);6.0
|
406 |
|
|
X3 := Arctan (0.0, 0.0);
|
407 |
|
|
Report.Failed ("no exception for arctan(0,0)");
|
408 |
|
|
exception
|
409 |
|
|
when Ada.Numerics.Argument_Error => null;
|
410 |
|
|
when others =>
|
411 |
|
|
Report.Failed ("wrong exception for arctan(0,0)");
|
412 |
|
|
end;
|
413 |
|
|
|
414 |
|
|
-- optimizer thwarting
|
415 |
|
|
if Report.Ident_Bool (False) then
|
416 |
|
|
Report.Comment (Real'Image (X1 + X2 + X3));
|
417 |
|
|
end if;
|
418 |
|
|
end Exception_Test;
|
419 |
|
|
|
420 |
|
|
|
421 |
|
|
procedure Do_Test is
|
422 |
|
|
begin
|
423 |
|
|
Special_Value_Test;
|
424 |
|
|
Exact_Result_Test;
|
425 |
|
|
Taylor_Series_Test;
|
426 |
|
|
Exception_Test;
|
427 |
|
|
end Do_Test;
|
428 |
|
|
end Generic_Check;
|
429 |
|
|
|
430 |
|
|
-----------------------------------------------------------------------
|
431 |
|
|
-----------------------------------------------------------------------
|
432 |
|
|
-- These expressions must be truly static, which is why we have to do them
|
433 |
|
|
-- outside of the generic, and we use the named numbers. Note that we know
|
434 |
|
|
-- that PI is not a machine number (it is irrational), and it should be
|
435 |
|
|
-- represented to more digits than supported by the target machine.
|
436 |
|
|
Float_Half_PI_Low : constant := Float'Adjacent(PI/2.0, 0.0);
|
437 |
|
|
Float_Half_PI_High : constant := Float'Adjacent(PI/2.0, 10.0);
|
438 |
|
|
Float_PI_Low : constant := Float'Adjacent(PI, 0.0);
|
439 |
|
|
Float_PI_High : constant := Float'Adjacent(PI, 10.0);
|
440 |
|
|
package Float_Check is new Generic_Check (Float,
|
441 |
|
|
Half_PI_Low => Float_Half_PI_Low,
|
442 |
|
|
Half_PI_High => Float_Half_PI_High,
|
443 |
|
|
PI_Low => Float_PI_Low,
|
444 |
|
|
PI_High => Float_PI_High);
|
445 |
|
|
|
446 |
|
|
-- check the Floating point type with the most digits
|
447 |
|
|
type A_Long_Float is digits System.Max_Digits;
|
448 |
|
|
A_Long_Float_Half_PI_Low : constant := A_Long_Float'Adjacent(PI/2.0, 0.0);
|
449 |
|
|
A_Long_Float_Half_PI_High : constant := A_Long_Float'Adjacent(PI/2.0, 10.0);
|
450 |
|
|
A_Long_Float_PI_Low : constant := A_Long_Float'Adjacent(PI, 0.0);
|
451 |
|
|
A_Long_Float_PI_High : constant := A_Long_Float'Adjacent(PI, 10.0);
|
452 |
|
|
package A_Long_Float_Check is new Generic_Check (A_Long_Float,
|
453 |
|
|
Half_PI_Low => A_Long_Float_Half_PI_Low,
|
454 |
|
|
Half_PI_High => A_Long_Float_Half_PI_High,
|
455 |
|
|
PI_Low => A_Long_Float_PI_Low,
|
456 |
|
|
PI_High => A_Long_Float_PI_High);
|
457 |
|
|
|
458 |
|
|
-----------------------------------------------------------------------
|
459 |
|
|
-----------------------------------------------------------------------
|
460 |
|
|
|
461 |
|
|
|
462 |
|
|
begin
|
463 |
|
|
Report.Test ("CXG2016",
|
464 |
|
|
"Check the accuracy of the ARCTAN function");
|
465 |
|
|
|
466 |
|
|
if Verbose then
|
467 |
|
|
Report.Comment ("checking Standard.Float");
|
468 |
|
|
end if;
|
469 |
|
|
|
470 |
|
|
Float_Check.Do_Test;
|
471 |
|
|
|
472 |
|
|
if Verbose then
|
473 |
|
|
Report.Comment ("checking a digits" &
|
474 |
|
|
Integer'Image (System.Max_Digits) &
|
475 |
|
|
" floating point type");
|
476 |
|
|
end if;
|
477 |
|
|
|
478 |
|
|
A_Long_Float_Check.Do_Test;
|
479 |
|
|
|
480 |
|
|
|
481 |
|
|
Report.Result;
|
482 |
|
|
end CXG2016;
|