lambda_double.h
1 #ifndef SYMENGINE_LAMBDA_DOUBLE_H
2 #define SYMENGINE_LAMBDA_DOUBLE_H
3 
4 #include <cmath>
5 #include <limits>
7 #include <symengine/symengine_exception.h>
8 #include <symengine/visitor.h>
9 
10 namespace SymEngine
11 {
12 
13 template <typename T>
14 class LambdaDoubleVisitor : public BaseVisitor<LambdaDoubleVisitor<T>>
15 {
16 protected:
17  /*
18  The 'result_' variable is assigned into at the very end of each visit()
19  methods below. The only place where these methods are called from is the
20  line 'b.accept(*this)' in apply() and the 'result_' is immediately
21  returned. Thus no corruption can happen and apply() can be safely called
22  recursively.
23  */
24 
25  typedef std::function<T(const T *x)> fn;
26  std::vector<fn> results;
27  std::vector<T> cse_intermediate_results;
28 
30  cse_intermediate_fns_map;
31  std::vector<fn> cse_intermediate_fns;
32  fn result_;
33  vec_basic symbols;
34 
35 public:
36  LambdaDoubleVisitor() = default;
38  LambdaDoubleVisitor &operator=(LambdaDoubleVisitor &&) = default;
39  // delete copy constructor:
40  // https://github.com/symengine/symengine/issues/1674
41  LambdaDoubleVisitor(const LambdaDoubleVisitor &) = delete;
42  LambdaDoubleVisitor &operator=(const LambdaDoubleVisitor &) = delete;
43 
44  void init(const vec_basic &x, const Basic &b, bool cse = false)
45  {
46  vec_basic outputs = {b.rcp_from_this()};
47  init(x, outputs, cse);
48  }
49 
50  void init(const vec_basic &inputs, const vec_basic &outputs,
51  bool cse = false)
52  {
53  results.clear();
54  cse_intermediate_fns.clear();
55  symbols = inputs;
56  if (not cse) {
57  for (auto &p : outputs) {
58  results.push_back(apply(*p));
59  }
60  } else {
61  vec_basic reduced_exprs;
62  vec_pair replacements;
63  // cse the outputs
64  SymEngine::cse(replacements, reduced_exprs, outputs);
65  cse_intermediate_results.resize(replacements.size());
66  for (auto &rep : replacements) {
67  auto res = apply(*(rep.second));
68  // Store the replacement symbol values in a dictionary for
69  // faster
70  // lookup for initialization
71  cse_intermediate_fns_map[rep.first]
72  = cse_intermediate_fns.size();
73  // Store it in a vector for faster use in call
74  cse_intermediate_fns.push_back(res);
75  }
76  // Generate functions for all the reduced exprs and save it
77  for (unsigned i = 0; i < outputs.size(); i++) {
78  results.push_back(apply(*reduced_exprs[i]));
79  }
80  // We don't need the cse_intermediate_fns_map anymore
81  cse_intermediate_fns_map.clear();
82  symbols.clear();
83  }
84  }
85 
86  fn apply(const Basic &b)
87  {
88  b.accept(*this);
89  return result_;
90  }
91 
92  T call(const std::vector<T> &vec)
93  {
94  T res;
95  call(&res, vec.data());
96  return res;
97  }
98 
99  void call(T *outs, const T *inps)
100  {
101  if (cse_intermediate_fns.size() > 0) {
102  for (unsigned i = 0; i < cse_intermediate_fns.size(); ++i) {
103  cse_intermediate_results[i] = cse_intermediate_fns[i](inps);
104  }
105  }
106  for (unsigned i = 0; i < results.size(); ++i) {
107  outs[i] = results[i](inps);
108  }
109  return;
110  }
111 
112  void bvisit(const Symbol &x)
113  {
114  for (unsigned i = 0; i < symbols.size(); ++i) {
115  if (eq(x, *symbols[i])) {
116  result_ = [=](const T *x) { return x[i]; };
117  return;
118  }
119  }
120  auto it = cse_intermediate_fns_map.find(x.rcp_from_this());
121  if (it != cse_intermediate_fns_map.end()) {
122  auto index = it->second;
123  T *cse_intermediate_result = &(cse_intermediate_results[index]);
124  result_ = [=](const T *x) { return *cse_intermediate_result; };
125  return;
126  }
127  throw SymEngineException("Symbol not in the symbols vector.");
128  };
129 
130  void bvisit(const Integer &x)
131  {
132  T tmp = mp_get_d(x.as_integer_class());
133  result_ = [=](const T *x_) { return tmp; };
134  }
135 
136  void bvisit(const Rational &x)
137  {
138  T tmp = mp_get_d(x.as_rational_class());
139  result_ = [=](const T *x) { return tmp; };
140  }
141 
142  void bvisit(const RealDouble &x)
143  {
144  T tmp = x.i;
145  result_ = [=](const T *x) { return tmp; };
146  }
147 
148 #ifdef HAVE_SYMENGINE_MPFR
149  void bvisit(const RealMPFR &x)
150  {
151  T tmp = mpfr_get_d(x.i.get_mpfr_t(), MPFR_RNDN);
152  result_ = [=](const T *x) { return tmp; };
153  }
154 #endif
155 
156  void bvisit(const Add &x)
157  {
158  fn tmp = apply(*x.get_coef());
159  fn tmp1, tmp2;
160  for (const auto &p : x.get_dict()) {
161  tmp1 = apply(*(p.first));
162  tmp2 = apply(*(p.second));
163  tmp = [=](const T *x) { return tmp(x) + tmp1(x) * tmp2(x); };
164  }
165  result_ = tmp;
166  }
167 
168  void bvisit(const Mul &x)
169  {
170  fn tmp = apply(*x.get_coef());
171  fn tmp1, tmp2;
172  for (const auto &p : x.get_dict()) {
173  tmp1 = apply(*(p.first));
174  tmp2 = apply(*(p.second));
175  tmp = [=](const T *x) {
176  return tmp(x) * std::pow(tmp1(x), tmp2(x));
177  };
178  }
179  result_ = tmp;
180  }
181 
182  void bvisit(const Pow &x)
183  {
184  fn exp_ = apply(*(x.get_exp()));
185  if (eq(*(x.get_base()), *E)) {
186  result_ = [=](const T *x) { return std::exp(exp_(x)); };
187  } else {
188  fn base_ = apply(*(x.get_base()));
189  result_ = [=](const T *x) { return std::pow(base_(x), exp_(x)); };
190  }
191  }
192 
193  void bvisit(const Sin &x)
194  {
195  fn tmp = apply(*(x.get_arg()));
196  result_ = [=](const T *x) { return std::sin(tmp(x)); };
197  }
198 
199  void bvisit(const Cos &x)
200  {
201  fn tmp = apply(*(x.get_arg()));
202  result_ = [=](const T *x) { return std::cos(tmp(x)); };
203  }
204 
205  void bvisit(const Tan &x)
206  {
207  fn tmp = apply(*(x.get_arg()));
208  result_ = [=](const T *x) { return std::tan(tmp(x)); };
209  }
210 
211  void bvisit(const Log &x)
212  {
213  fn tmp = apply(*(x.get_arg()));
214  result_ = [=](const T *x) { return std::log(tmp(x)); };
215  };
216 
217  void bvisit(const Cot &x)
218  {
219  fn tmp = apply(*(x.get_arg()));
220  result_ = [=](const T *x) { return 1.0 / std::tan(tmp(x)); };
221  };
222 
223  void bvisit(const Csc &x)
224  {
225  fn tmp = apply(*(x.get_arg()));
226  result_ = [=](const T *x) { return 1.0 / std::sin(tmp(x)); };
227  };
228 
229  void bvisit(const Sec &x)
230  {
231  fn tmp = apply(*(x.get_arg()));
232  result_ = [=](const T *x) { return 1.0 / std::cos(tmp(x)); };
233  };
234 
235  void bvisit(const ASin &x)
236  {
237  fn tmp = apply(*(x.get_arg()));
238  result_ = [=](const T *x) { return std::asin(tmp(x)); };
239  };
240 
241  void bvisit(const ACos &x)
242  {
243  fn tmp = apply(*(x.get_arg()));
244  result_ = [=](const T *x) { return std::acos(tmp(x)); };
245  };
246 
247  void bvisit(const ASec &x)
248  {
249  fn tmp = apply(*(x.get_arg()));
250  result_ = [=](const T *x) { return std::acos(1.0 / tmp(x)); };
251  };
252 
253  void bvisit(const ACsc &x)
254  {
255  fn tmp = apply(*(x.get_arg()));
256  result_ = [=](const T *x) { return std::asin(1.0 / tmp(x)); };
257  };
258 
259  void bvisit(const ATan &x)
260  {
261  fn tmp = apply(*(x.get_arg()));
262  result_ = [=](const T *x) { return std::atan(tmp(x)); };
263  };
264 
265  void bvisit(const ACot &x)
266  {
267  fn tmp = apply(*(x.get_arg()));
268  result_ = [=](const T *x) { return std::atan(1.0 / tmp(x)); };
269  };
270 
271  void bvisit(const Sinh &x)
272  {
273  fn tmp = apply(*(x.get_arg()));
274  result_ = [=](const T *x) { return std::sinh(tmp(x)); };
275  };
276 
277  void bvisit(const Csch &x)
278  {
279  fn tmp = apply(*(x.get_arg()));
280  result_ = [=](const T *x) { return 1.0 / std::sinh(tmp(x)); };
281  };
282 
283  void bvisit(const Cosh &x)
284  {
285  fn tmp = apply(*(x.get_arg()));
286  result_ = [=](const T *x) { return std::cosh(tmp(x)); };
287  };
288 
289  void bvisit(const Sech &x)
290  {
291  fn tmp = apply(*(x.get_arg()));
292  result_ = [=](const T *x) { return 1.0 / std::cosh(tmp(x)); };
293  };
294 
295  void bvisit(const Tanh &x)
296  {
297  fn tmp = apply(*(x.get_arg()));
298  result_ = [=](const T *x) { return std::tanh(tmp(x)); };
299  };
300 
301  void bvisit(const Coth &x)
302  {
303  fn tmp = apply(*(x.get_arg()));
304  result_ = [=](const T *x) { return 1.0 / std::tanh(tmp(x)); };
305  };
306 
307  void bvisit(const ASinh &x)
308  {
309  fn tmp = apply(*(x.get_arg()));
310  result_ = [=](const T *x) { return std::asinh(tmp(x)); };
311  };
312 
313  void bvisit(const ACsch &x)
314  {
315  fn tmp = apply(*(x.get_arg()));
316  result_ = [=](const T *x) { return std::asinh(1.0 / tmp(x)); };
317  };
318 
319  void bvisit(const ACosh &x)
320  {
321  fn tmp = apply(*(x.get_arg()));
322  result_ = [=](const T *x) { return std::acosh(tmp(x)); };
323  };
324 
325  void bvisit(const ATanh &x)
326  {
327  fn tmp = apply(*(x.get_arg()));
328  result_ = [=](const T *x) { return std::atanh(tmp(x)); };
329  };
330 
331  void bvisit(const ACoth &x)
332  {
333  fn tmp = apply(*(x.get_arg()));
334  result_ = [=](const T *x) { return std::atanh(1.0 / tmp(x)); };
335  };
336 
337  void bvisit(const ASech &x)
338  {
339  fn tmp = apply(*(x.get_arg()));
340  result_ = [=](const T *x) { return std::acosh(1.0 / tmp(x)); };
341  };
342 
343  void bvisit(const Constant &x)
344  {
345  T tmp = eval_double(x);
346  result_ = [=](const T *x) { return tmp; };
347  };
348 
349  void bvisit(const Abs &x)
350  {
351  fn tmp = apply(*(x.get_arg()));
352  result_ = [=](const T *x) { return std::abs(tmp(x)); };
353  };
354 
355  void bvisit(const Basic &)
356  {
357  throw NotImplementedError("Not Implemented");
358  };
359 
360  void bvisit(const UnevaluatedExpr &x)
361  {
362  apply(*x.get_arg());
363  };
364 };
365 
367  : public BaseVisitor<LambdaRealDoubleVisitor, LambdaDoubleVisitor<double>>
368 {
369 public:
370  // Classes not implemented are
371  // Subs, UpperGamma, LowerGamma, Dirichlet_eta, Zeta
372  // LeviCivita, KroneckerDelta, FunctionSymbol, LambertW
373  // Derivative, Complex, ComplexDouble, ComplexMPC
374 
375  using LambdaDoubleVisitor::bvisit;
376 
377  void bvisit(const ATan2 &x)
378  {
379  fn num = apply(*(x.get_num()));
380  fn den = apply(*(x.get_den()));
381  result_ = [=](const double *x) { return std::atan2(num(x), den(x)); };
382  };
383 
384  void bvisit(const Gamma &x)
385  {
386  fn tmp = apply(*(x.get_args()[0]));
387  result_ = [=](const double *x) { return std::tgamma(tmp(x)); };
388  };
389 
390  void bvisit(const LogGamma &x)
391  {
392  fn tmp = apply(*(x.get_args()[0]));
393  result_ = [=](const double *x) { return std::lgamma(tmp(x)); };
394  };
395 
396  void bvisit(const Erf &x)
397  {
398  fn tmp = apply(*(x.get_args()[0]));
399  result_ = [=](const double *x) { return std::erf(tmp(x)); };
400  }
401 
402  void bvisit(const Erfc &x)
403  {
404  fn tmp = apply(*(x.get_args()[0]));
405  result_ = [=](const double *x) { return std::erfc(tmp(x)); };
406  }
407 
408  void bvisit(const Equality &x)
409  {
410  fn lhs_ = apply(*(x.get_arg1()));
411  fn rhs_ = apply(*(x.get_arg2()));
412  result_ = [=](const double *x) { return (lhs_(x) == rhs_(x)); };
413  }
414 
415  void bvisit(const Unequality &x)
416  {
417  fn lhs_ = apply(*(x.get_arg1()));
418  fn rhs_ = apply(*(x.get_arg2()));
419  result_ = [=](const double *x) { return (lhs_(x) != rhs_(x)); };
420  }
421 
422  void bvisit(const LessThan &x)
423  {
424  fn lhs_ = apply(*(x.get_arg1()));
425  fn rhs_ = apply(*(x.get_arg2()));
426  result_ = [=](const double *x) { return (lhs_(x) <= rhs_(x)); };
427  }
428 
429  void bvisit(const StrictLessThan &x)
430  {
431  fn lhs_ = apply(*(x.get_arg1()));
432  fn rhs_ = apply(*(x.get_arg2()));
433  result_ = [=](const double *x) { return (lhs_(x) < rhs_(x)); };
434  }
435 
436  void bvisit(const And &x)
437  {
438  std::vector<fn> applys;
439  for (const auto &p : x.get_args()) {
440  applys.push_back(apply(*p));
441  }
442 
443  result_ = [=](const double *x) {
444  bool result = bool(applys[0](x));
445  for (unsigned int i = 0; i < applys.size(); i++) {
446  result = result && bool(applys[i](x));
447  }
448  return double(result);
449  };
450  }
451 
452  void bvisit(const Or &x)
453  {
454  std::vector<fn> applys;
455  for (const auto &p : x.get_args()) {
456  applys.push_back(apply(*p));
457  }
458 
459  result_ = [=](const double *x) {
460  bool result = bool(applys[0](x));
461  for (unsigned int i = 0; i < applys.size(); i++) {
462  result = result || bool(applys[i](x));
463  }
464  return double(result);
465  };
466  }
467 
468  void bvisit(const Xor &x)
469  {
470  std::vector<fn> applys;
471  for (const auto &p : x.get_args()) {
472  applys.push_back(apply(*p));
473  }
474 
475  result_ = [=](const double *x) {
476  bool result = bool(applys[0](x));
477  for (unsigned int i = 0; i < applys.size(); i++) {
478  result = result != bool(applys[i](x));
479  }
480  return double(result);
481  };
482  }
483 
484  void bvisit(const Not &x)
485  {
486  fn tmp = apply(*(x.get_arg()));
487  result_ = [=](const double *x) { return double(not bool(tmp(x))); };
488  }
489 
490  void bvisit(const Max &x)
491  {
492  std::vector<fn> applys;
493  for (const auto &p : x.get_args()) {
494  applys.push_back(apply(*p));
495  }
496 
497  result_ = [=](const double *x) {
498  double result = applys[0](x);
499  for (unsigned int i = 0; i < applys.size(); i++) {
500  result = std::max(result, applys[i](x));
501  }
502  return result;
503  };
504  };
505 
506  void bvisit(const Min &x)
507  {
508  std::vector<fn> applys;
509  for (const auto &p : x.get_args()) {
510  applys.push_back(apply(*p));
511  }
512 
513  result_ = [=](const double *x) {
514  double result = applys[0](x);
515  for (unsigned int i = 0; i < applys.size(); i++) {
516  result = std::min(result, applys[i](x));
517  }
518  return result;
519  };
520  };
521 
522  void bvisit(const Sign &x)
523  {
524  fn tmp = apply(*(x.get_arg()));
525  result_ = [=](const double *x) {
526  return tmp(x) == 0.0 ? 0.0 : (tmp(x) < 0.0 ? -1.0 : 1.0);
527  };
528  };
529 
530  void bvisit(const Floor &x)
531  {
532  fn tmp = apply(*(x.get_arg()));
533  result_ = [=](const double *x) { return std::floor(tmp(x)); };
534  };
535 
536  void bvisit(const Ceiling &x)
537  {
538  fn tmp = apply(*(x.get_arg()));
539  result_ = [=](const double *x) { return std::ceil(tmp(x)); };
540  };
541 
542  void bvisit(const Truncate &x)
543  {
544  fn tmp = apply(*(x.get_arg()));
545  result_ = [=](const double *x) { return std::trunc(tmp(x)); };
546  };
547 
548  void bvisit(const Infty &x)
549  {
550  if (x.is_negative_infinity()) {
551  result_ = [=](const double * /* x */) {
553  };
554  } else if (x.is_positive_infinity()) {
555  result_ = [=](const double * /* x */) {
557  };
558  } else {
559  throw SymEngineException(
560  "LambdaDouble can only represent real valued infinity");
561  }
562  }
563  void bvisit(const NaN &nan)
564  {
565  assert(&nan == &(*Nan) /* singleton, or do we support NaN quiet/singaling nan with payload? */);
566  result_ = [](const double * /* x */) {
568  };
569  }
570  void bvisit(const Contains &cts)
571  {
572  const auto fn_expr = apply(*cts.get_expr());
573  const auto set = cts.get_set();
574  if (is_a<Interval>(*set)) {
575  const auto &interv = down_cast<const Interval &>(*set);
576  const auto fn_start = apply(*interv.get_start());
577  const auto fn_end = apply(*interv.get_end());
578  const bool left_open = interv.get_left_open();
579  const bool right_open = interv.get_right_open();
580  result_ = [=](const double *x) {
581  const auto val_expr = fn_expr(x);
582  const auto val_start = fn_start(x);
583  const auto val_end = fn_end(x);
584  bool left_ok, right_ok;
585  if (val_start == -std::numeric_limits<double>::infinity()) {
586  left_ok = !std::isnan(val_expr);
587  } else {
588  left_ok = (left_open) ? (val_start < val_expr)
589  : (val_start <= val_expr);
590  }
591  if (val_end == std::numeric_limits<double>::infinity()) {
592  right_ok = !std::isnan(val_expr);
593  } else {
594  right_ok = (right_open) ? (val_expr < val_end)
595  : (val_expr <= val_end);
596  }
597  return (left_ok && right_ok) ? 1.0 : 0.0;
598  };
599  } else {
600  throw SymEngineException("LambdaDoubleVisitor: only ``Interval`` "
601  "implemented for ``Contains``.");
602  }
603  }
604 
605  void bvisit(const BooleanAtom &ba)
606  {
607  const bool val = ba.get_val();
608  result_ = [=](const double * /* x */) { return (val) ? 1.0 : 0.0; };
609  }
610 
611  void bvisit(const Piecewise &pw)
612  {
613  SYMENGINE_ASSERT_MSG(
614  eq(*pw.get_vec().back().second, *boolTrue),
615  "LambdaDouble requires a (Expr, True) at the end of Piecewise");
616 
617  std::vector<fn> applys;
618  std::vector<fn> preds;
619  for (const auto &expr_pred : pw.get_vec()) {
620  applys.push_back(apply(*expr_pred.first));
621  preds.push_back(apply(*expr_pred.second));
622  }
623  result_ = [=](const double *x) {
624  for (size_t i = 0;; ++i) {
625  if (preds[i](x) == 1.0) {
626  return applys[i](x);
627  }
628  }
629  throw SymEngineException(
630  "Unexpectedly reached end of Piecewise function.");
631  };
632  }
633 };
634 
636  : public BaseVisitor<LambdaComplexDoubleVisitor,
637  LambdaDoubleVisitor<std::complex<double>>>
638 {
639 public:
640  // Classes not implemented are
641  // Subs, UpperGamma, LowerGamma, Dirichlet_eta, Zeta
642  // LeviCivita, KroneckerDelta, FunctionSymbol, LambertW
643  // Derivative, ATan2, Gamma
644 
645  using LambdaDoubleVisitor::bvisit;
646 
647  void bvisit(const Complex &x)
648  {
649  double t1 = mp_get_d(x.real_), t2 = mp_get_d(x.imaginary_);
650  result_ = [=](const std::complex<double> *x) {
651  return std::complex<double>(t1, t2);
652  };
653  };
654 
655  void bvisit(const ComplexDouble &x)
656  {
657  std::complex<double> tmp = x.i;
658  result_ = [=](const std::complex<double> *x) { return tmp; };
659  };
660 #ifdef HAVE_SYMENGINE_MPC
661  void bvisit(const ComplexMPC &x)
662  {
663  mpfr_class t(x.get_prec());
664  double real, imag;
665  mpc_real(t.get_mpfr_t(), x.as_mpc().get_mpc_t(), MPFR_RNDN);
666  real = mpfr_get_d(t.get_mpfr_t(), MPFR_RNDN);
667  mpc_imag(t.get_mpfr_t(), x.as_mpc().get_mpc_t(), MPFR_RNDN);
668  imag = mpfr_get_d(t.get_mpfr_t(), MPFR_RNDN);
669  std::complex<double> tmp(real, imag);
670  result_ = [=](const std::complex<double> *x) { return tmp; };
671  }
672 #endif
673 };
674 } // namespace SymEngine
675 #endif // SYMENGINE_LAMBDA_DOUBLE_H
T acos(T... args)
T acosh(T... args)
T asin(T... args)
T asinh(T... args)
T atan2(T... args)
T atan(T... args)
T atanh(T... args)
T back(T... args)
T ceil(T... args)
RCP< const Basic > get_den() const
Definition: functions.h:516
RCP< const Basic > get_num() const
Definition: functions.h:511
The base class for representing addition in symbolic expressions.
Definition: add.h:27
const RCP< const Number > & get_coef() const
Definition: add.h:142
vec_basic get_args() const override
Returns the list of arguments.
Definition: logic.cpp:234
The lowest unit of symbolic representation.
Definition: basic.h:97
Complex Double Class to hold std::complex<double> values.
Complex Class.
Definition: complex.h:33
rational_class real_
Definition: complex.h:38
RCP< T > rcp_from_this()
Get RCP<T> pointer to self (it will cast the pointer to T)
Integer Class.
Definition: integer.h:19
const integer_class & as_integer_class() const
Convert to integer_class.
Definition: integer.h:45
vec_basic get_args() const override
Returns the list of arguments.
Definition: functions.h:159
vec_basic get_args() const override
Returns the list of arguments.
Definition: functions.h:40
RCP< const Basic > get_arg() const
Definition: functions.h:36
vec_basic get_args() const override
Returns the list of arguments.
Definition: logic.cpp:302
RCP< const Basic > get_exp() const
Definition: pow.h:42
RCP< const Basic > get_base() const
Definition: pow.h:37
Rational Class.
Definition: rational.h:16
const rational_class & as_rational_class() const
Convert to rational_class.
Definition: rational.h:50
RealDouble Class to hold double values.
Definition: real_double.h:20
RCP< const Basic > get_arg1() const
Definition: functions.h:91
RCP< const Basic > get_arg2() const
Definition: functions.h:96
vec_basic get_args() const override
Returns the list of arguments.
Definition: logic.cpp:412
T clear(T... args)
T cos(T... args)
T cosh(T... args)
T data(T... args)
T end(T... args)
T erf(T... args)
T erfc(T... args)
T exp(T... args)
T find(T... args)
T floor(T... args)
T infinity(T... args)
T isnan(T... args)
T lgamma(T... args)
T log(T... args)
T max(T... args)
T min(T... args)
Main namespace for SymEngine package.
Definition: add.cpp:19
bool eq(const Basic &a, const Basic &b)
Checks equality for a and b
Definition: basic-inl.h:21
T pow(T... args)
T push_back(T... args)
T resize(T... args)
T signaling_NaN(T... args)
T sin(T... args)
T sinh(T... args)
T size(T... args)
Our less operator (<):
Definition: basic.h:228
T tan(T... args)
T tanh(T... args)
T tgamma(T... args)
T trunc(T... args)