summaryrefslogtreecommitdiff
path: root/src/util/floatingpoint.h.in
blob: 7e430ad877625862d9854427e290e77781dd8616 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
/*********************                                                        */
/*! \file floatingpoint.h.in
 ** \verbatim
 ** Top contributors (to current version):
 **   Martin Brain, Aina Niemetz, Haniel Barbosa
 ** Copyright (c) 2013  University of Oxford
 ** This file is part of the CVC4 project.
 ** Copyright (c) 2009-2020 by the authors listed in the file AUTHORS
 ** in the top-level source directory and their institutional affiliations.
 ** All rights reserved.  See the file COPYING in the top-level source
 ** directory for licensing information.\endverbatim
 **
 ** \brief A floating-point value.
 **
 ** This file contains the data structures used by the constant and parametric
 ** types of the floating point theory.
 **/
#include "cvc4_public.h"

#ifndef CVC4__FLOATINGPOINT_H
#define CVC4__FLOATINGPOINT_H

#include <fenv.h>

#include "util/bitvector.h"
#include "util/rational.h"

// clang-format off
#if @CVC4_USE_SYMFPU@
// clang-format on
#include <symfpu/core/unpackedFloat.h>
#endif /* @CVC4_USE_SYMFPU@ */

namespace CVC4 {
// Inline these!
inline bool CVC4_PUBLIC validExponentSize(uint32_t e) { return e >= 2; }
inline bool CVC4_PUBLIC validSignificandSize(uint32_t s) { return s >= 2; }

/* -------------------------------------------------------------------------- */

/**
 * Floating point sorts are parameterised by two constants > 1 giving the
 * width (in bits) of the exponent and significand (including the hidden bit).
 * So, IEEE-754 single precision, a.k.a. float, is described as 8 24.
 */
class CVC4_PUBLIC FloatingPointSize
{
 public:
  /** Constructors. */
  FloatingPointSize(uint32_t exp_size, uint32_t sig_size);
  FloatingPointSize(const FloatingPointSize& old);

  /** Operator overload for equality comparison. */
  bool operator==(const FloatingPointSize& fps) const
  {
    return (d_exp_size == fps.d_exp_size) && (d_sig_size == fps.d_sig_size);
  }

  /** Implement the interface that symfpu uses for floating-point formats. */

  /** Get the exponent bit-width of this floating-point format. */
  inline uint32_t exponentWidth(void) const { return this->d_exp_size; }
  /** Get the significand bit-width of this floating-point format. */
  inline uint32_t significandWidth(void) const { return this->d_sig_size; }
  /**
   * Get the bit-width of the packed representation of this floating-point
   * format (= exponent + significand bit-width, convenience wrapper).
   */
  inline uint32_t packedWidth(void) const
  {
    return this->exponentWidth() + this->significandWidth();
  }
  /**
   * Get the exponent bit-width of the packed representation of this
   * floating-point format (= exponent bit-width).
   */
  inline uint32_t packedExponentWidth(void) const
  {
    return this->exponentWidth();
  }
  /**
   * Get the significand bit-width of the packed representation of this
   * floating-point format (= significand bit-width - 1).
   */
  inline uint32_t packedSignificandWidth(void) const
  {
    return this->significandWidth() - 1;
  }

 private:
  /** Exponent bit-width. */
  uint32_t d_exp_size;
  /** Significand bit-width. */
  uint32_t d_sig_size;

}; /* class FloatingPointSize */

/**
 * Hash function for floating point formats.
 */
struct CVC4_PUBLIC FloatingPointSizeHashFunction
{
  static inline size_t ROLL(size_t X, size_t N)
  {
    return (((X) << (N)) | ((X) >> (8 * sizeof((X)) - (N))));
  }

  inline size_t operator()(const FloatingPointSize& t) const
  {
    return size_t(ROLL(t.exponentWidth(), 4 * sizeof(uint32_t))
                  | t.significandWidth());
  }
}; /* struct FloatingPointSizeHashFunction */

/* -------------------------------------------------------------------------- */

/**
 * A concrete instance of the rounding mode sort
 */
enum CVC4_PUBLIC RoundingMode
{
  roundNearestTiesToEven = FE_TONEAREST,
  roundTowardPositive = FE_UPWARD,
  roundTowardNegative = FE_DOWNWARD,
  roundTowardZero = FE_TOWARDZERO,
  // Initializes this to the diagonalization of the 4 other values.
  roundNearestTiesToAway = (((~FE_TONEAREST) & 0x1) | ((~FE_UPWARD) & 0x2)
                            | ((~FE_DOWNWARD) & 0x4) | ((~FE_TOWARDZERO) & 0x8))
}; /* enum RoundingMode */

/**
 * Hash function for rounding mode values.
 */
struct CVC4_PUBLIC RoundingModeHashFunction
{
  inline size_t operator()(const RoundingMode& rm) const { return size_t(rm); }
}; /* struct RoundingModeHashFunction */

/* -------------------------------------------------------------------------- */

/**
 * This is a symfpu literal "back-end".  It allows the library to be used as
 * an arbitrary precision floating-point implementation.  This is effectively
 * the glue between symfpu's notion of "signed bit-vector" and CVC4's
 * BitVector.
 */
namespace symfpuLiteral {

/**
 * Forward declaration of wrapper so that traits can be defined early and so
 * used in the implementation of wrappedBitVector.
 */
template <bool T>
class wrappedBitVector;

using CVC4BitWidth = uint32_t;
using CVC4Prop = bool;
using CVC4RM = ::CVC4::RoundingMode;
using CVC4FPSize = ::CVC4::FloatingPointSize;
using CVC4UnsignedBitVector = wrappedBitVector<false>;
using CVC4SignedBitVector = wrappedBitVector<true>;

/**
 * This is the template parameter for symfpu's templates.
 */
class traits
{
 public:
  /** The six key types that symfpu uses. */
  using bwt = CVC4BitWidth;           // bit-width type
  using prop = CVC4Prop;              // boolean type
  using rm = CVC4RM;                  // rounding-mode type
  using fpt = CVC4FPSize;             // floating-point format type
  using ubv = CVC4UnsignedBitVector;  // unsigned bit-vector type
  using sbv = CVC4SignedBitVector;    // signed bit-vector type

  /** Give concrete instances of each rounding mode, mainly for comparisons. */
  static rm RNE(void);
  static rm RNA(void);
  static rm RTP(void);
  static rm RTN(void);
  static rm RTZ(void);

  /** The sympfu properties. */
  static void precondition(const prop& p);
  static void postcondition(const prop& p);
  static void invariant(const prop& p);
};

/**
 * Type function for mapping between types.
 */
template <bool T>
struct signedToLiteralType;

template <>
struct signedToLiteralType<true>
{
  using literalType = int32_t;
};
template <>
struct signedToLiteralType<false>
{
  using literalType = uint32_t;
};

/**
 * This extends the interface for CVC4::BitVector for compatibility with symFPU.
 * The template parameter distinguishes signed and unsigned bit-vectors, a
 * distinction symfpu uses.
 */
template <bool isSigned>
class wrappedBitVector : public BitVector
{
 protected:
  using literalType = typename signedToLiteralType<isSigned>::literalType;
  friend wrappedBitVector<!isSigned>;  // To allow conversion between types

// clang-format off
#if @CVC4_USE_SYMFPU@
  // clang-format on
  friend ::symfpu::ite<CVC4Prop, wrappedBitVector<isSigned> >;  // For ITE
#endif /* @CVC4_USE_SYMFPU@ */

 public:
  /** Constructors. */
  wrappedBitVector(const CVC4BitWidth w, const uint32_t v) : BitVector(w, v) {}
  wrappedBitVector(const CVC4Prop& p) : BitVector(1, p ? 1U : 0U) {}
  wrappedBitVector(const wrappedBitVector<isSigned>& old) : BitVector(old) {}
  wrappedBitVector(const BitVector& old) : BitVector(old) {}

  /** Get the bit-width of this wrapped bit-vector. */
  CVC4BitWidth getWidth(void) const { return getSize(); }

  /** Create a zero value of given bit-width 'w'. */
  static wrappedBitVector<isSigned> one(const CVC4BitWidth& w);
  /** Create a one value of given bit-width 'w'. */
  static wrappedBitVector<isSigned> zero(const CVC4BitWidth& w);
  /** Create a ones value (all bits 1) of given bit-width 'w'. */
  static wrappedBitVector<isSigned> allOnes(const CVC4BitWidth& w);
  /** Create a maximum signed/unsigned value of given bit-width 'w'. */
  static wrappedBitVector<isSigned> maxValue(const CVC4BitWidth& w);
  /** Create a minimum signed/unsigned value of given bit-width 'w'. */
  static wrappedBitVector<isSigned> minValue(const CVC4BitWidth& w);

  /** Return true if this a bit-vector representing a ones value. */
  CVC4Prop isAllOnes() const;
  /** Return true if this a bit-vector representing a zero value. */
  CVC4Prop isAllZeros() const;

  /** Left shift. */
  wrappedBitVector<isSigned> operator<<(
      const wrappedBitVector<isSigned>& op) const;
  /** Logical (unsigned) and arithmetic (signed) right shift. */
  wrappedBitVector<isSigned> operator>>(
      const wrappedBitVector<isSigned>& op) const;

  /**
   * Inherited but ...
   * *sigh* if we use the inherited version then it will return a
   * CVC4::BitVector which can be converted back to a
   * wrappedBitVector<isSigned> but isn't done automatically when working
   * out types for templates instantiation.  ITE is a particular
   * problem as expressions and constants no longer derive the
   * same type.  There aren't any good solutions in C++, we could
   * use CRTP but Liana wouldn't appreciate that, so the following
   * pointless wrapping functions are needed.
   */

  /** Bit-wise or. */
  wrappedBitVector<isSigned> operator|(
      const wrappedBitVector<isSigned>& op) const;
  /** Bit-wise and. */
  wrappedBitVector<isSigned> operator&(
      const wrappedBitVector<isSigned>& op) const;
  /** Bit-vector addition. */
  wrappedBitVector<isSigned> operator+(
      const wrappedBitVector<isSigned>& op) const;
  /** Bit-vector subtraction. */
  wrappedBitVector<isSigned> operator-(
      const wrappedBitVector<isSigned>& op) const;
  /** Bit-vector multiplication. */
  wrappedBitVector<isSigned> operator*(
      const wrappedBitVector<isSigned>& op) const;
  /** Bit-vector signed/unsigned division. */
  wrappedBitVector<isSigned> operator/(
      const wrappedBitVector<isSigned>& op) const;
  /** Bit-vector signed/unsigned remainder. */
  wrappedBitVector<isSigned> operator%(
      const wrappedBitVector<isSigned>& op) const;
  /** Bit-vector negation. */
  wrappedBitVector<isSigned> operator-(void) const;
  /** Bit-wise not. */
  wrappedBitVector<isSigned> operator~(void) const;

  /** Bit-vector increment. */
  wrappedBitVector<isSigned> increment() const;
  /** Bit-vector decrement. */
  wrappedBitVector<isSigned> decrement() const;
  /**
   * Bit-vector logical/arithmetic right shift where 'op' extended to the
   * bit-width of this wrapped bit-vector.
   */
  wrappedBitVector<isSigned> signExtendRightShift(
      const wrappedBitVector<isSigned>& op) const;

  /**
   * Modular operations.
   * Note: No overflow checking so these are the same as other operations.
   */
  wrappedBitVector<isSigned> modularLeftShift(
      const wrappedBitVector<isSigned>& op) const;
  wrappedBitVector<isSigned> modularRightShift(
      const wrappedBitVector<isSigned>& op) const;
  wrappedBitVector<isSigned> modularIncrement() const;
  wrappedBitVector<isSigned> modularDecrement() const;
  wrappedBitVector<isSigned> modularAdd(
      const wrappedBitVector<isSigned>& op) const;
  wrappedBitVector<isSigned> modularNegate() const;

  /** Bit-vector equality. */
  CVC4Prop operator==(const wrappedBitVector<isSigned>& op) const;
  /** Bit-vector signed/unsigned less or equal than. */
  CVC4Prop operator<=(const wrappedBitVector<isSigned>& op) const;
  /** Bit-vector sign/unsigned greater or equal than. */
  CVC4Prop operator>=(const wrappedBitVector<isSigned>& op) const;
  /** Bit-vector signed/unsigned less than. */
  CVC4Prop operator<(const wrappedBitVector<isSigned>& op) const;
  /** Bit-vector sign/unsigned greater or equal than. */
  CVC4Prop operator>(const wrappedBitVector<isSigned>& op) const;

  /** Convert this bit-vector to a signed bit-vector. */
  wrappedBitVector<true> toSigned(void) const;
  /** Convert this bit-vector to an unsigned bit-vector. */
  wrappedBitVector<false> toUnsigned(void) const;

  /** Bit-vector signed/unsigned (zero) extension. */
  wrappedBitVector<isSigned> extend(CVC4BitWidth extension) const;
  /**
   * Create a "contracted" bit-vector by cutting off the 'reduction' number of
   * most significant bits, i.e., by extracting the (bit-width - reduction)
   * least significant bits.
   */
  wrappedBitVector<isSigned> contract(CVC4BitWidth reduction) const;
  /**
   * Create a "resized" bit-vector of given size bei either extending (if new
   * size is larger) or contracting (if it is smaller) this wrapped bit-vector.
   */
  wrappedBitVector<isSigned> resize(CVC4BitWidth newSize) const;
  /**
   * Create an extension of this bit-vector that matches the bit-width of the
   * given bit-vector.
   * Note: The size of the given bit-vector may not be larger than the size of
   * this bit-vector.
   */
  wrappedBitVector<isSigned> matchWidth(
      const wrappedBitVector<isSigned>& op) const;
  /** Bit-vector concatenation. */
  wrappedBitVector<isSigned> append(const wrappedBitVector<isSigned>& op) const;

  /** Inclusive of end points, thus if the same, extracts just one bit. */
  wrappedBitVector<isSigned> extract(CVC4BitWidth upper,
                                     CVC4BitWidth lower) const;
};
}  // namespace symfpuLiteral

/**
 * A concrete floating point value.
 */
// clang-format off
#if @CVC4_USE_SYMFPU@
// clang-format on
using FloatingPointLiteral = ::symfpu::unpackedFloat<symfpuLiteral::traits>;
#else
class CVC4_PUBLIC FloatingPointLiteral
{
 public:
  // This intentional left unfinished as the choice of literal
  // representation is solver specific.
  void unfinished(void) const;

  FloatingPointLiteral(uint32_t, uint32_t, double) { unfinished(); }
  FloatingPointLiteral(uint32_t, uint32_t, const std::string&) { unfinished(); }
  FloatingPointLiteral(const FloatingPointLiteral&) { unfinished(); }
  bool operator==(const FloatingPointLiteral& op) const
  {
    unfinished();
    return false;
  }

  size_t hash(void) const
  {
    unfinished();
    return 23;
  }
};
#endif /* @CVC4_USE_SYMFPU@ */

class CVC4_PUBLIC FloatingPoint
{
 public:
  /**
   * Wrappers to represent results of non-total functions (e.g., min/max).
   * The Boolean flag is true if the result is defined, and false otherwise.
   */
  using PartialFloatingPoint = std::pair<FloatingPoint, bool>;
  using PartialBitVector = std::pair<BitVector, bool>;
  using PartialRational = std::pair<Rational, bool>;

  /** Constructors. */
  FloatingPoint(uint32_t e, uint32_t s, const BitVector& bv);
  FloatingPoint(const FloatingPointSize& size, const BitVector& bv);

  FloatingPoint(const FloatingPointSize& fp_size,
                const FloatingPointLiteral& fpl)
      : d_fp_size(fp_size), d_fpl(fpl)
  {
  }

  FloatingPoint(const FloatingPoint& fp)
      : d_fp_size(fp.d_fp_size), d_fpl(fp.d_fpl)
  {
  }
  FloatingPoint(const FloatingPointSize& size,
                const RoundingMode& rm,
                const BitVector& bv,
                bool signedBV);
  FloatingPoint(const FloatingPointSize& size,
                const RoundingMode& rm,
                const Rational& r);

  /**
   * Create a FP NaN value of given size.
   * size: The FP size (format).
   */
  static FloatingPoint makeNaN(const FloatingPointSize& size);
  /**
   * Create a FP infinity value of given size.
   * size: The FP size (format).
   * sign: True for -oo, false otherwise.
   */
  static FloatingPoint makeInf(const FloatingPointSize& size, bool sign);
  /**
   * Create a FP zero value of given size.
   * size: The FP size (format).
   * sign: True for -zero, false otherwise.
   */
  static FloatingPoint makeZero(const FloatingPointSize& size, bool sign);
  /**
   * Create the smallest subnormal FP value of given size.
   * size: The FP size (format).
   * sign: True for negative sign, false otherwise.
   */
  static FloatingPoint makeMinSubnormal(const FloatingPointSize& size,
                                        bool sign);
  /**
   * Create the largest subnormal FP value of given size.
   * size: The FP size (format).
   * sign: True for negative sign, false otherwise.
   */
  static FloatingPoint makeMaxSubnormal(const FloatingPointSize& size,
                                        bool sign);
  /**
   * Create the smallest normal FP value of given size.
   * size: The FP size (format).
   * sign: True for negative sign, false otherwise.
   */
  static FloatingPoint makeMinNormal(const FloatingPointSize& size, bool sign);
  /**
   * Create the largest normal FP value of given size.
   * size: The FP size (format).
   * sign: True for negative sign, false otherwise.
   */
  static FloatingPoint makeMaxNormal(const FloatingPointSize& size, bool sign);

  /** Get the wrapped floating-point value. */
  const FloatingPointLiteral& getLiteral(void) const { return this->d_fpl; }

  /**
   * Return a string representation of this floating-point.
   *
   * If printAsIndexed is true then it prints the bit-vector components of the
   * FP value as indexed symbols, otherwise in binary notation.
   */
  std::string toString(bool printAsIndexed = false) const;

  /** Return the packed (IEEE-754) representation of this floating-point. */
  BitVector pack(void) const;

  /** Floating-point absolute value. */
  FloatingPoint absolute(void) const;
  /** Floating-point negation. */
  FloatingPoint negate(void) const;
  /** Floating-point addition. */
  FloatingPoint plus(const RoundingMode& rm, const FloatingPoint& arg) const;
  /** Floating-point subtraction. */
  FloatingPoint sub(const RoundingMode& rm, const FloatingPoint& arg) const;
  /** Floating-point multiplication. */
  FloatingPoint mult(const RoundingMode& rm, const FloatingPoint& arg) const;
  /** Floating-point division. */
  FloatingPoint div(const RoundingMode& rm, const FloatingPoint& arg) const;
  /** Floating-point fused multiplication and addition. */
  FloatingPoint fma(const RoundingMode& rm,
                    const FloatingPoint& arg1,
                    const FloatingPoint& arg2) const;
  /** Floating-point square root. */
  FloatingPoint sqrt(const RoundingMode& rm) const;
  /** Floating-point round to integral. */
  FloatingPoint rti(const RoundingMode& rm) const;
  /** Floating-point remainder. */
  FloatingPoint rem(const FloatingPoint& arg) const;

  /**
   * Floating-point max (total version).
   * zeroCase: true to return the left (rather than the right operand) in case
   *           of max(-0,+0) or max(+0,-0).
   */
  FloatingPoint maxTotal(const FloatingPoint& arg, bool zeroCaseLeft) const;
  /**
   * Floating-point min (total version).
   * zeroCase: true to return the left (rather than the right operand) in case
   *           of min(-0,+0) or min(+0,-0).
   */
  FloatingPoint minTotal(const FloatingPoint& arg, bool zeroCaseLeft) const;

  /**
   * Floating-point max.
   *
   * Returns a partial floating-point, which is a pair of FloatingPoint and
   * bool. The boolean flag is true if the result is defined, and false if it
   * is undefined.
   */
  PartialFloatingPoint max(const FloatingPoint& arg) const;
  /** Floating-point min.
   *
   * Returns a partial floating-point, which is a pair of FloatingPoint and
   * bool. The boolean flag is true if the result is defined, and false if it
   * is undefined.
   */
  PartialFloatingPoint min(const FloatingPoint& arg) const;

  /** Equality (NOT: fp.eq but =) over floating-point values. */
  bool operator==(const FloatingPoint& fp) const;
  /** Floating-point less or equal than. */
  bool operator<=(const FloatingPoint& arg) const;
  /** Floating-point less than. */
  bool operator<(const FloatingPoint& arg) const;

  /** Return true if this floating-point represents a normal value. */
  bool isNormal(void) const;
  /** Return true if this floating-point represents a subnormal value. */
  bool isSubnormal(void) const;
  /** Return true if this floating-point represents a zero value. */
  bool isZero(void) const;
  /** Return true if this floating-point represents an infinite value. */
  bool isInfinite(void) const;
  /** Return true if this floating-point represents a NaN value. */
  bool isNaN(void) const;
  /** Return true if this floating-point represents a negative value. */
  bool isNegative(void) const;
  /** Return true if this floating-point represents a positive value. */
  bool isPositive(void) const;

  /**
   * Convert this floating-point to a floating-point of given size, with
   * respect to given rounding mode.
   */
  FloatingPoint convert(const FloatingPointSize& target,
                        const RoundingMode& rm) const;

  /**
   * Convert this floating-point to a bit-vector of given size, with
   * respect to given rounding mode (total version).
   * Returns given bit-vector 'undefinedCase' in the undefined case.
   */
  BitVector convertToBVTotal(BitVectorSize width,
                             const RoundingMode& rm,
                             bool signedBV,
                             BitVector undefinedCase) const;
  /**
   * Convert this floating-point to a rational, with respect to given rounding
   * mode (total version).
   * Returns given rational 'undefinedCase' in the undefined case.
   */
  Rational convertToRationalTotal(Rational undefinedCase) const;

  /**
   * Convert this floating-point to a bit-vector of given size.
   *
   * Returns a partial bit-vector, which is a pair of BitVector and bool.
   * The boolean flag is true if the result is defined, and false if it
   * is undefined.
   */
  PartialBitVector convertToBV(BitVectorSize width,
                               const RoundingMode& rm,
                               bool signedBV) const;
  /**
   * Convert this floating-point to a Rational.
   *
   * Returns a partial Rational, which is a pair of Rational and bool.
   * The boolean flag is true if the result is defined, and false if it
   * is undefined.
   */
  PartialRational convertToRational(void) const;

  /** The floating-point size of this floating-point value. */
  FloatingPointSize d_fp_size;

 protected:
  /** The floating-point literal of this floating-point value. */
  FloatingPointLiteral d_fpl;

}; /* class FloatingPoint */

/**
 * Hash function for floating-point values.
 */
struct CVC4_PUBLIC FloatingPointHashFunction
{
  size_t operator()(const FloatingPoint& fp) const
  {
    FloatingPointSizeHashFunction fpshf;
    BitVectorHashFunction bvhf;

    return fpshf(fp.d_fp_size) ^ bvhf(fp.pack());
  }
}; /* struct FloatingPointHashFunction */

/* -------------------------------------------------------------------------- */

/**
 * The parameter type for the conversions to floating point.
 */
class CVC4_PUBLIC FloatingPointConvertSort
{
 public:
  /** Constructors. */
  FloatingPointConvertSort(uint32_t _e, uint32_t _s) : d_fp_size(_e, _s) {}
  FloatingPointConvertSort(const FloatingPointSize& fps) : d_fp_size(fps) {}

  /** Operator overload for comparison of conversion sorts. */
  bool operator==(const FloatingPointConvertSort& fpcs) const
  {
    return d_fp_size == fpcs.d_fp_size;
  }

  /** The floating-point size of this sort. */
  FloatingPointSize d_fp_size;
};

/** Hash function for conversion sorts. */
template <uint32_t key>
struct CVC4_PUBLIC FloatingPointConvertSortHashFunction
{
  inline size_t operator()(const FloatingPointConvertSort& fpcs) const
  {
    FloatingPointSizeHashFunction f;
    return f(fpcs.d_fp_size) ^ (0x00005300 | (key << 24));
  }
}; /* struct FloatingPointConvertSortHashFunction */

/**
 * As different conversions are different parameterised kinds, there
 * is a need for different (C++) types for each one.
 */

/**
 * Conversion from floating-point to IEEE bit-vector (first bit represents the
 * sign, the following exponent width bits the exponent, and the remaining bits
 * the significand).
 */
class CVC4_PUBLIC FloatingPointToFPIEEEBitVector
    : public FloatingPointConvertSort
{
 public:
  /** Constructors. */
  FloatingPointToFPIEEEBitVector(uint32_t _e, uint32_t _s)
      : FloatingPointConvertSort(_e, _s)
  {
  }
  FloatingPointToFPIEEEBitVector(const FloatingPointConvertSort& old)
      : FloatingPointConvertSort(old)
  {
  }
};

/**
 * Conversion from floating-point to another floating-point (of possibly
 * different size).
 */
class CVC4_PUBLIC FloatingPointToFPFloatingPoint
    : public FloatingPointConvertSort
{
 public:
  /** Constructors. */
  FloatingPointToFPFloatingPoint(uint32_t _e, uint32_t _s)
      : FloatingPointConvertSort(_e, _s)
  {
  }
  FloatingPointToFPFloatingPoint(const FloatingPointConvertSort& old)
      : FloatingPointConvertSort(old)
  {
  }
};

/**
 * Conversion from floating-point to Real.
 */
class CVC4_PUBLIC FloatingPointToFPReal : public FloatingPointConvertSort
{
 public:
  /** Constructors. */
  FloatingPointToFPReal(uint32_t _e, uint32_t _s)
      : FloatingPointConvertSort(_e, _s)
  {
  }
  FloatingPointToFPReal(const FloatingPointConvertSort& old)
      : FloatingPointConvertSort(old)
  {
  }
};

/**
 * Conversion from floating-point to signed bit-vector value.
 */
class CVC4_PUBLIC FloatingPointToFPSignedBitVector
    : public FloatingPointConvertSort
{
 public:
  /** Constructors. */
  FloatingPointToFPSignedBitVector(uint32_t _e, uint32_t _s)
      : FloatingPointConvertSort(_e, _s)
  {
  }
  FloatingPointToFPSignedBitVector(const FloatingPointConvertSort& old)
      : FloatingPointConvertSort(old)
  {
  }
};

/**
 * Conversion from floating-point to unsigned bit-vector value.
 */
class CVC4_PUBLIC FloatingPointToFPUnsignedBitVector
    : public FloatingPointConvertSort
{
 public:
  /** Constructors. */
  FloatingPointToFPUnsignedBitVector(uint32_t _e, uint32_t _s)
      : FloatingPointConvertSort(_e, _s)
  {
  }
  FloatingPointToFPUnsignedBitVector(const FloatingPointConvertSort& old)
      : FloatingPointConvertSort(old)
  {
  }
};

class CVC4_PUBLIC FloatingPointToFPGeneric : public FloatingPointConvertSort
{
 public:
  /** Constructors. */
  FloatingPointToFPGeneric(uint32_t _e, uint32_t _s)
      : FloatingPointConvertSort(_e, _s)
  {
  }
  FloatingPointToFPGeneric(const FloatingPointConvertSort& old)
      : FloatingPointConvertSort(old)
  {
  }
};

/**
 * Base type for floating-point to bit-vector conversion.
 */
class CVC4_PUBLIC FloatingPointToBV
{
 public:
  /** Constructors. */
  FloatingPointToBV(uint32_t s) : d_bv_size(s) {}
  FloatingPointToBV(const FloatingPointToBV& old) : d_bv_size(old.d_bv_size) {}
  FloatingPointToBV(const BitVectorSize& old) : d_bv_size(old) {}

  /** Return the bit-width of the bit-vector to convert to. */
  operator uint32_t() const { return d_bv_size; }

  /** The bit-width of the bit-vector to converto to. */
  BitVectorSize d_bv_size;
};

/**
 * Conversion from floating-point to unsigned bit-vector value.
 */
class CVC4_PUBLIC FloatingPointToUBV : public FloatingPointToBV
{
 public:
  FloatingPointToUBV(uint32_t _s) : FloatingPointToBV(_s) {}
  FloatingPointToUBV(const FloatingPointToBV& old) : FloatingPointToBV(old) {}
};

/**
 * Conversion from floating-point to signed bit-vector value.
 */
class CVC4_PUBLIC FloatingPointToSBV : public FloatingPointToBV
{
 public:
  FloatingPointToSBV(uint32_t _s) : FloatingPointToBV(_s) {}
  FloatingPointToSBV(const FloatingPointToBV& old) : FloatingPointToBV(old) {}
};

/**
 * Conversion from floating-point to unsigned bit-vector value (total version).
 */
class CVC4_PUBLIC FloatingPointToUBVTotal : public FloatingPointToBV
{
 public:
  FloatingPointToUBVTotal(uint32_t _s) : FloatingPointToBV(_s) {}
  FloatingPointToUBVTotal(const FloatingPointToBV& old) : FloatingPointToBV(old)
  {
  }
};

/**
 * Conversion from floating-point to signed bit-vector value (total version).
 */
class CVC4_PUBLIC FloatingPointToSBVTotal : public FloatingPointToBV
{
 public:
  FloatingPointToSBVTotal(uint32_t _s) : FloatingPointToBV(_s) {}
  FloatingPointToSBVTotal(const FloatingPointToBV& old) : FloatingPointToBV(old)
  {
  }
};

/**
 * Hash function for floating-point to bit-vector conversions.
 */
template <uint32_t key>
struct CVC4_PUBLIC FloatingPointToBVHashFunction
{
  inline size_t operator()(const FloatingPointToBV& fptbv) const
  {
    UnsignedHashFunction< ::CVC4::BitVectorSize> f;
    return (key ^ 0x46504256) ^ f(fptbv.d_bv_size);
  }
}; /* struct FloatingPointToBVHashFunction */

/* Note: It is not possible to pack a number without a size to pack to,
 * thus it is difficult to implement output stream operator overloads for
 * FloatingPointLiteral in a sensible way. Use FloatingPoint instead. */

/** Output stream operator overloading for floating-point values. */
std::ostream& operator<<(std::ostream& os, const FloatingPoint& fp) CVC4_PUBLIC;

/** Output stream operator overloading for floating-point formats. */
std::ostream& operator<<(std::ostream& os,
                         const FloatingPointSize& fps) CVC4_PUBLIC;

/** Output stream operator overloading for floating-point conversion sorts. */
std::ostream& operator<<(std::ostream& os,
                         const FloatingPointConvertSort& fpcs) CVC4_PUBLIC;

}  // namespace CVC4

#endif /* CVC4__FLOATINGPOINT_H */
generated by cgit on debian on lair
contact matthew@masot.net with questions or feedback