1 This file records noteworthy changes.
3 1.8.0 (14 October 2020)
4 * New routines for the numerical evaluation of iterated integrals like
5 elliptic multiple polylogarithms or iterated integrals of modular forms.
6 * Stronger normalization methods for expressions with powers. Where this
7 is safe, GiNaC now tries replacing various kinds of powers in order to
8 find and cancel common subexpressions.
9 * Improved CMake build.
11 1.7.11 (3 August 2020)
12 * Fix elusive bugs in factor() and in expand().
15 * Fix collect_common_factors() for hidden zero arguments.
16 * Fix build on modern systems (C++20 and Python3).
19 * Fix unarchiving of overloaded functions.
22 1.7.8 (7 October 2019)
23 * Fix pseries::evalf(), broken since 1.7.0.
24 * Fix a corner-case bug in H_evalf().
26 1.7.7 (22 September 2019)
27 * Fix unarchiving empty containers.
28 * Remove obsolete exhashmap implementation.
29 * Fix issues with zeta computation.
33 * Fix a bug in mul::series() about asymptotic order term.
34 * Fix cl_N code generation for small negative integers.
35 * Fix multiple polylog computation related to trailing zeros and user-specified
36 signs for small imaginary parts.
37 * Fix CMake build with Python3.
39 1.7.5 (17 February 2019)
40 * Add Markowitz-ordered Gaussian eliminatin.
41 * Avoid un-normal zeros in matrix::solve().
42 * Accept exprseq arguments in lsolve(eqns, vars).
43 * Add overload of matrix::rank(solve_algo).
44 * Fix case in G_do_hoelder with real x values which are not cl_R.
45 * Fix an elusive bug in expairseq/add ctor, introduced in 1.7.0.
47 1.7.4 (19 February 2018)
48 * Fix ABI compatibility with 1.7.[0-2] (broken unintentionally in 1.7.3).
50 1.7.3 (1 February 2018)
51 * Speed up some corner cases of polynomial factorization.
52 * Speed up some corner cases of gcd computation.
53 * Speed up factorization of pre-factored polynomials.
54 * Make matrix::solve() work with hidden (non-normal) zeros.
55 * Fix bug in evaluation of real and imaginary parts of powers.
56 * Fix some LaTeX output, add new function symbol::get_TeX_name().
58 1.7.2 (10 January 2017)
59 * Fix memory leak in operators returning relationals, introduced in 1.7.0.
60 * Make macro GINAC_DECLARE_REGISTERED_CLASS() work outside namespace GiNaC.
61 * Fix some build issues.
63 1.7.1 (2 October 2016)
64 * Fix crash in basic::subs(), visible in classes which don't override it.
65 * Fix convergence failure in Li_projection.
66 * Fix cliffor_bar() and clifford_star().
67 * Improve performance of clifford_to_lst().
68 * Fix compilation issues, reduce compiler warnings.
71 * Make use of C++11 features:
72 - Use new language features 'auto', 'nullptr', 'constexpr', range-based for
73 loops, the contextual keyword 'override', static 'noexcept' operator, etc.
74 - Add initializer list ctors for container<>, lst and nested initializer
75 lists ctor for class matrix.
76 - Add many rvalue reference overloads for ctors from std containers. (Note
77 that whether they really speed up things depends on whether evaluation
78 leaves the container unchanged or not.)
79 - Rvalue reference overloads for ctors from std containers made it possible
80 to remove the manual 'discardable' hack and std::auto_ptr<T> for good.
81 - New dynallocate<T>(args...) helper functions obsolete manual allocation
82 on the heap and setting the dynallocated status flag.
83 - Got rid of C++98ish algorithm features (bind2nd, etc.) and replaced them
85 * Make add::eval(), mul::eval() more careful so they return an evaluated,
86 canonical expression without compromise.
87 * Remove 'level' argument of all .eval() methods. This has become possible by
88 ensuring that add::eval() and mul::eval() evaluate their own structure
89 without compromise (this was not guaranteed before). If all ex are evaluated,
90 any object which is an aggregate of such ex will only ever have to
91 evaluate itself, obsoleting the need for recursing.
92 * Remove 'level' arguments of all .evalf() and .normal() methods. They were
93 modeled after the .eval() methods.
94 * Make relational::rhs() and lhs() nonvirtual.
95 * Add support for power::info(info_flags::real).
96 * Add fderivative::derivatives() method.
97 * Remove broken info_flags::algebraic.
98 * Remove unmaintained support for Cint.
99 * Remove unfinished andcode for hash table-based expairseq.
100 * Remove deprecated ex::to_rational(lst) and ex::to_polynomial(lst).
101 * ginac-excompiler installs in $LIBEXECDIR now, not in $BINDIR.
102 * Many minor performance enhancements.
104 1.6.7 (29 February 2016)
105 * Fix elusive bug in sqrfree(), normal(), numer(), and denom().
107 1.6.6 (20 December 2015)
108 * Fix elusive bug in add::eval().
109 * Several minor performance enhancements.
112 * Fix pow(+(...),2).expand() where the sum contained roots of expressions.
113 * Speed-up pow(+(...),n).expand() where n > 3.
114 * Fix a crash in PolyLogs of rational numbers.
117 * Fix pow(+(...),2).expand() where the sum contained roots of rationals.
118 * Various improvements for info(info_flags::nonnegative).
119 * Make power::expand() (x*p)^c -> x^c * p^c, if p>0.
120 * Fix normal() where nested functions occur multiple times.
121 * Add optional explicit derivatives to functions and use it in abs(),
122 imag_part(), real_part().
123 * Evaluate abs(z)^(2n) -> z^n * conjugate(z)^n.
125 1.6.3 (27 November 2014)
126 * New expand_options::expand_transcendental. GiNaC will apply additional
127 rules when expanding expressions if this option is specified, i.e.
128 log(a*b) -> log(a) + log(b)
129 exp(a + b) -> exp(a)*exp(b)
130 * Functions can define a custom expand() and info() now.
131 * Automatically rewrite (x^{-1})^c -> x^{-c} for positive c.
132 * Transform abs(x)^n -> x^n if x is real and n is even.
133 * Improved power::info(): power with a real base and even exponent
134 reports that it's non-negative.
135 * Fixed spurious match failure due to expairseq::match() bug:
136 > match(sin(y)*exp(b)+sin(x)*exp(a), sin($0)*exp(a)+exp(b)*sin($1))
138 * Fixed power::is_polynomial() and mul::is_polynomial() computation.
139 * factor_univariate correctly handles polynomials over rationals (no
141 * Fixed an infinite loop in factor_multivariate due to incorrect handling
142 of polynomials having a leading coefficient being a sum with a non-trivial
144 * Fixed numeric evaluation of generalized multiple polylogarithms for real
145 and negative argument.
146 * Avoid calling log(1-x) in numeric evaluation of classical polylogarithm.
147 * Fixed numeric evaluation of multiple polylogarithms (correctly set
148 the signs of the imaginary parts in the intermediate expression).
149 * Fixed numeric evaluation of Nielsen polylogs for arguments close to
150 the 6th root of unity or its conjugate.
151 * Avoid infinite recursion in series expansion of logarithm.
152 * numeric::info(info_flags::nonnegative) correctly handles complex numbers
153 with non-zero imaginary part.
154 * Fixed several build issues:
155 - libreadline and dlopen() misdetection.
156 - multiple definition of symbol on MinGW.
157 - auto* tools and texinfo being unhappy in various ways.
158 * GiNaC can be built with CMake now.
160 1.6.2 (6 November 2011)
161 * Fixed the parser to read GiNaC::lst again.
162 * Fixed a compile warning (relevant to openSUSE build).
163 * Fixed a bug concerning the subsitutions in indexed expressions
164 (subs_options::no_index_renaming was ignored).
167 * Fixed several internal assertions for the debug version of GiNaC.
168 * More verbose report of (auto-)configuration problems.
169 * Fixed a compile problem with ginsh using gcc 4.6.
172 * Fixed compilation issues with clang.
173 * Fixed compilation issues with VC++.
174 * Fixed a segfault in fsolve() and improved its numerical algorithm.
175 * Fixed several memory leaks in power::eval().
176 * Fixed the substitution algorithm that led to wrong results (exp(log(x))->x).
177 * Fixed a bug in add:eval() (real_part() had returned wrong results).
178 * Fixed infinite loop bug when unarchiving realsymbol and possymbol.
179 * Fixed bugs in the GCD code.
180 * Fixed empty string bug in symbol::get_name().
181 * Fixed bugs concerning STL vector sizes that can potentially cause segfaults.
182 * Fixed a bug in is_polynomial().
183 * Removed autogen dependencies.
184 * Improved simplify_indexed() algorithm.
185 * Changed the connection between library version and package version.
186 * power::series() can handle now more expressions with singularities in the exponent.
187 * Added conjugate() methods to functions cosh, sinh, tanh.
190 * Changed default behaviour of conjugate() for user-defined functions to avoid
191 wrong evaluations. A user-defined conjugate_func must now be explicitly
193 * Fixed: Parser now handles abbreviations as advertized in the manual.
194 * Fixed two bugs in the GCD code (one infinite loop, one miscalculation).
195 * Fixed dangerous iterator use.
196 * Fixed a memory leak in excompiler.
197 * Fixed several syntax issues wrt to gcc 4.5 and wrt to compile option -std=cxx0x.
198 * Fixed a wrong URL and a texinfo problem in the manual.
199 * Added degree_vector() utility function.
201 1.5.7 (29 March 2010)
202 * Fixed a bug in the GCD code (infinite loop in pgcd()).
203 * Fixed a bug in atan2() (division by zero error for atan2(x,0) if x is a
204 negative and composite term).
205 * Products now correctly set info_flags::negative and info_flags::negint.
207 1.5.6 (28 January 2010)
208 * Fixed compile problem with gcc 3.4.
209 * Improved excompiler: The ginac-excompiler script will only be installed if
210 excompiler is supported and enabled.
212 1.5.5 (04 November 2009)
213 * Correct libtool versioning.
215 1.5.4 (02 November 2009)
216 * Default parser now can again read in user defined classes.
217 * Add mul::info() and add::info() support for numeric info_flags (GiNaC now can
218 detect that the sum or product of a number of positive expressions is again a
219 positive expression).
220 * Fixed cast that caused compile error on 64bit machines.
221 * Fixed incorrect uses of STL iterators.
222 * Removed subdirectory debian.
223 * Output LaTeX-symbols using \mathrm, instead of \mbox.
226 * The parser now understands user-defined functions by default. The identifier
227 names can now contain underscores.
230 * Fixed documentation of return_type_info()
231 * Fixed cstdio include issue.
232 * GiNaC now requires at least version 1.2.2 of cln.
234 1.5.1 (25 February 2009)
235 * Fixed compiling problems with GNU g++ 3.4.
236 * Work around weird C++ RTTI behaviour on woe32 so GiNaC works on MinGW again.
238 1.5.0 (17 February 2009)
239 * Added polynomial factorization.
240 * New, faster (recursive descent) expression parser.
241 * Faster GCD computation.
242 * gcd() now allows the user to override (some of the) heuristics.
243 * Writing extensions classes is simpler now:
244 - Replaced custom RTTI by standard C++ RTTI.
245 - Definition of archiving methods now optional when declaring a new GiNaC class.
246 * Fixed recursion in polynomial divide that caused a significant slowdown in
248 * Improved lsolve() of systems containing non-numeric coefficients.
249 * Removed implicit conversions from cl_N to numeric.
250 * Improved configuration and compatibility.
253 * Fixed a bug in the multiple polylogarithms Li/G that caused imaginary parts
254 of the arguments to be ignored.
255 * Fixed archiving of complex numbers.
256 * Made the behaviour of class function more consistent with respect to
258 * Fixed bug in power::is_polynomial.
259 * Improved heur_gcd() so it can handle rational polynomials.
260 * Improved build scripts.
262 1.4.3 (04 April 2008)
263 * Fixed bug in numerical evaluation of multiple polylogarithms and
264 alternating multiple zeta values introduced in version 1.4.2.
265 * Nielsen polylog now invalidates its lookup tables in case the precision
266 (Digits) has been changed.
267 * Added new checks for recent bugs in the numerical evaluation of Li and zeta.
269 1.4.2 (03 April 2008)
270 * Fixed VPATH building of documentation.
271 * Fixed bug in numerical evaluation of multiple polylogarithms for arguments
272 equal to certain roots of unity (thanks to Jianqiang Zhao).
273 * Fixed check for memory leakage in parser.
274 * Made internal function coerce() standard compliant.
275 * Improved lsolve() of systems containing non-numeric coefficients.
276 * Improved automatic test programs. Now they work on MinGW and Cygwin as well.
278 1.4.1 (21 November 2007)
279 * Fixed memory leak in ginac_yylex().
280 * Fixed memory leaks in mul::eval() and power::eval().
281 * Fixed macro checking for version of libreadline (Mac OS X).
282 * Fixed broken preprocessor instruction in excompiler.cpp.
283 * Fixed broken module loading in excompiler.cpp.
284 * info(info_flags::has_indices) now works for sums and products.
285 * Improved mul::expand().
286 * Improved CLN output.
288 1.4.0 (31 August 2007)
289 * New tinfo mechanism.
290 * Removed rtt from class symbol.
291 * Faster archiving by adding a map from strings to indices in the atoms vector.
292 * Improved unarchiving: algorithms order N instead of order N^2.
293 * Added compile function, excompiler class.
295 * Added step function to GiNaCs built-in functions.
296 * Added is_polynomial() method.
297 * Added real_part() and imag_part() methods.
298 * Added matrix::is_zero_matrix() method.
299 * Added evalm() method to pseries class.
300 * Improved dummy index renaming.
301 * Many improvements for class clifford.
302 * New flag for positive symbols/numbers, added possymbol class.
303 * Added programming examples in doc/examples
305 1.3.8 (28 August 2007)
306 * Drop support of ginac-config and ginac.m4. Please use pkg-config instead.
307 * atan2(y,x) branch cut correction.
308 * Fixed bug in series expansion.
309 * Additional transformations for mul and power.
310 * Clifford units now honor representation labels.
311 * Correct csrc output for idx and fderivative.
312 * Improved handling of convergence transformationis for Li/G.
313 * Fix compilation issues with prereleases of GCC 4.3.
314 * info_flags::expanded added.
315 * input_parser also accepts pow().
316 * texinfo.tex removed from package.
318 1.3.7 (05 February 2007)
319 * Fixed bug in expansion of power.
320 * Fixed bugs in functions S() (infinite loop), H() and zeta() (wrong results).
321 * Rebuilt bison related files with bison version 2.3 (fixes parse error bugs).
322 * Improved power::eval().
323 * Improved libreadline detection.
325 1.3.6 (13 December 2006)
326 * Better worst case behavior in heuristic gcd.
327 * Improved collecting for sparse multivariate polynomials.
328 * Improved collect_common_factors.
329 * Fixed bug in numerical evaluation of multiple polylogs.
330 * Fixed numerical integration of complex functions.
331 * Fixed bug in tensor::replace_contr_index().
333 1.3.5 (17 August 2006)
334 * Re-built bison related files that caused bugs with gcc 4.
335 * Fixed bugs in fderivative::match_same_type(), expairseq::match(),
336 expand_dummy_sum(), and expairseq::map().
337 * Fixed template specialization for class container to be gcc 4.2 compliant.
338 * Output in csrc mode is now -x instead of -1.0*x.
339 * Fixed tutorial and reference generation in html/pdf/ps format.
340 * Modified autotool configuration to allow for compilation on MinGW.
341 * Debian packaging files updated.
343 1.3.4 (12 April 2006)
344 * More general index dimensions are now possible.
345 * Improved algebraic substitutions.
346 * Fixed wrong matching in .has().
347 * Fixed bug in differentiation of a power.
348 * collect_common_factors also works with negative powers of common factors now.
349 * Code clean-up and preparations for upcoming cln 1.2.
351 1.3.3 (24 October 2005)
352 * Fixed bug occuring in algebraic substitutions with expressions involving
354 * Fixed bug that selected the wrong algorithm for determinants of purely
355 numeric matrices with at least one floating point element.
356 * Fixed bug in function H that caused an infinite recursion for arguments
358 * Fixed const-correctness in printing handler for GCC 4.0.2.
359 * Made lookup in adaptivesimpson precision-aware.
360 * Added series expansion for functions (classical) Li and S around x==0.
361 * Added fsolve() numerical univariate real-valued function solver.
362 * Added functions sub_matrix() and reduced_matrix().
363 * Small cleanups. Less warnings with latest GCC.
366 * GCD avoids to produce expanded expressions.
367 * Fixed bug in expanding expressions containing dummy indices.
368 * Fixed static initialization order bug.
369 * collect_common_factors now works with powers.
370 * Modernized configuration scripts.
373 * integral() and eval_integ() can be used from ginsh.
374 * Integrals can be series-expanded.
375 * Fixed a library initialization problem.
376 * GiNaC compiles from tarball even if lex/flex is missing.
377 * Fixed bugs in canonicalize_clifford(), clifford_prime() and clifford_to_lst().
378 * clifford_moebius_map(), remove_dirac_ONE() and LaTeX output of Clifford
379 objects now care about representation labels.
381 * Better output for slashed expressions
383 1.3.0 (19 October 2004)
384 * The Clifford classes have been generalized to allow working with Clifford
385 algebras generated by arbitrary symmetric tensors or matrices. Also, a lot
386 of new functions for Clifford algebras have been added, including
387 automorphisms and Moebius transformations. [V. Kisil]
388 * Added some basic symbolic and numeric integration facilities. [C. Dams]
389 * The multiple polylogarithm Li() now evaluates numerically for arbitrary
391 * New functions G(a,y) and G(a,s,y) added (another notation for multiple
394 1.2.4 (12 October 2004)
395 * Added ex::unitcontprim() to compute the unit, content, and primitive parts
396 of a polynomial in one go.
397 * binomial(n, k) evaluates for non-integer arguments n.
398 * Li(2,x) now evaluates for +-I.
400 * Fixed bug in Li(n,x) (if Li(2,x) was calculated with high precision the
401 enlargement of the look-up table caused a segmentation fault).
402 * Fixed another bug in the series expansion of powers, and a bug in
405 1.2.3 (13 August 2004)
406 * Added variants of dirac_trace() and color_trace() that take the trace over
407 more than one representation label by specifying a set or list of labels.
408 * diracgamma::contract_with() uses Chisholm identities in 4 dimensions to
409 produce more compact results.
410 * Fixed a bug in the series expansion of powers.
411 * The --enable-html-doc and --enable-ps-doc configure options are gone.
412 Documentation in HTML, DVI, PostScript, and PDF formats is now built with
413 "make html", "make dvi", "make ps", and "make pdf", respectively. The only
414 documentation that gets built by default are the tutorial in .info format
417 1.2.2 (3 August 2004)
418 * Added const_preorder_iterator and const_postorder_iterator classes (and
419 associated methods ex::pre/postorder_begin/end()) providing tree traversal
421 * Fixed the LaTeX output of the varidx class.
422 * Fixed bugs in series expansion and complex conjugation.
423 * (p.i*p.i).get_free_indices() returns the correct result () instead of (.i).
424 * Symbolic functions without any eval(), evalf() etc. functions now work
426 * integer_content(), content(), and primpart() now also work for polynomials
427 with rational coefficients (calculating the LCM of coefficients'
429 * Added method matrix::rank().
430 * Added function resultant().
431 * Added integer_content() function to ginsh.
433 1.2.1 (23 April 2004)
434 * Fixed infinite recursion in atan2_evalf() and improved atan2_eval().
435 * Added automatic evaluations for trigonometric functions with negative
436 arguments (e.g. sin(-2) -> -sin(2)).
437 * Fixed a static initialization order goof-up.
438 * Fixed various bugs in series expansion.
440 1.2.0 (19 March 2004)
441 * Added a structure<T> template class for the easy creation of user-defined
443 * Added support for (acyclic) visitors, to allow cleaner implementations of
444 algebraic algorithms.
445 * Added a const_iterator class that can be used instead of op()/nops().
446 * Completely revamped the implementation of expression output. It is now
447 possible to add new output formats, to change the behavior of predefined
448 formats at run-time, and to have different output styles for algebraic
450 * Symbols can be made non-commutative.
451 * Added a method ex::conjugate() and a function conjugate() for complex
452 conjugation. Symbols can be declared as real or complex-valued.
453 * Improved the speed of subs(), normal(), to_rational() and to_polynomial()
454 by the use of maps instead of lists. The old forms
455 subs(const lst & ls, const lst & lr, unsigned options)
456 to_rational/to_polynomial(lst & repl)
457 are still available for compatibility, but using the new forms
458 subs(const exmap & m, unsigned options)
459 to_rational/to_polynomial(exmap & repl)
460 is more efficient, especially when the number of replacements is large.
461 * quo(), rem(), prem(), sprem(), decomp_rational(), unit(), content(),
462 primpart() and matrix::charpoly() now take a "const ex &" instead of a
464 * Redundant expressions (two ex pointing to different objects are found to be
465 equal in compare()) are now actively deleted/fused to conserve memory and
466 speed up subsequent comparisons. This behavior can be suppressed on a
467 per-object level with status_flags::not_shareable. Lists and matrices are
468 not shareable by default.
469 * Lots of internal streamlining and optimizations.
470 * Caveats for class implementors:
471 - basic::copy() and basic::destroy() are gone; classes derived from
472 basic can use the defaults for the assignment operator and copy
474 - basic::subs(), basic::normal(), basic::to_rational() and
475 basic::to_polynomial() take 'exmap' objects instead of lists.
476 - basic::subs() now descends into subexpressions (if accessible via
477 nops()/op()/let_op()). If you have a custom implementation of subs()
478 that calls basic::subs() after substituting subexpressions, this needs
479 to be changed to a call to subs_one_level().
480 - lst::thislst() and exprseq::thisexprseq() renamed to thiscontainer().
481 - thiscontainer() and associated constructors now take a std::auto_ptr.
482 - Overloading basic::print() is now deprecated. You should use
483 print_func<>() class options instead.
485 1.1.7 (11 March 2004)
486 * Fixed a bug in canonicalize_clifford().
487 * Series expansion now works predictably. All terms with the exponent of the
488 expansion variable smaller than the given order are calculated exactly. If
489 the series is not terminating, the Order function is (at least) of the given
492 1.1.6 (22 January 2004)
493 * Added a function option "dummy()" which means "no options". This simplifies
494 the implementation of symbolic functions which are not to be further
496 * Removed a bug in the numerical evaluation of Li() that caused the system
497 to hang for certain parameter combinations.
498 * Fixed a bug in the calculation of hash values for indices that could lead
499 to wrong results or bogus error messages from simplify_indexed().
500 * Fixed a bug in the evaluation of harmonic polylogarithms for complex
501 arguments with positive imaginary part.
503 1.1.5 (5 November 2003)
504 * Harmonic polylogarithms now numerically evaluate for arbitrary arguments
505 (parameter must still be positive integers).
506 * The zeta function now can also be given a lst as a parameter in which case
507 it becomes a multiple zeta value. The use of mZeta is deprecated.
508 * The order of parameters for the multiple polylogarithm has been corrected.
509 * Documentation for the nested sums functions zeta, harmonic polylog, multiple
510 polylog, etc. has been added.
512 1.1.4 (17 October 2003)
513 * Lists and matrices can now be initialized from comma-separated lists of
514 expressions, like this:
524 This is both faster and produces much smaller code than the old constructors
525 lst(ex, ex, ...) and matrix(unsigned, unsigned, lst), especially in the case
526 of matrices, and is now the recommended way to create these objects.
527 * The function mZeta now evaluates much faster for arbitrary parameters. The
528 harmonic and multiple polylogarithms evaluate considerably faster and check
529 for convergence. The order of parameters for the harmonic polylogarithm
532 1.1.3 (22 August 2003)
533 * Added new symbolic functions for better integration with nestedsums:
534 (multiple) polylogarithm Li(), Nielsen's generalized polylogarithm S(),
535 harmonic polylogarithm H(), and multiple zeta value mZeta().
536 * New exhashmap<T> template intended as a drop-in replacement for
537 std::map<ex, T, ex_is_less> using GiNaC's hashing algorithms.
539 1.1.2 (11 August 2003)
540 * Fixed a bug in the unarchiving of sums and products: terms were not
541 reordered in a canonical way.
542 * Fixed a bug in normal()/numer_denom(): denominator was not made unit
543 normal if it was a simple number.
544 * Improved the speed of subs() in some cases.
547 * lst (and exprseq) provide iterators for read-only element access. For
548 sequential access this is one order faster than using op().
549 * Implemented relational::subs() (this was done in 1.0.9 but inadvertently
550 omitted from the 1.1 branch).
551 * pole_error and do_taylor are available to library users.
552 * Added on-line help and Tab-completion for print(), iprint(), print_latex()
553 and print_csrc() in ginsh.
556 * Removed deprecated macros is_ex_a, is_ex_exactly_a and friends for good.
557 * The scalar_products mechanism allows the specification of an index dimension.
558 * Removed dirac_gamma6/7().
559 * Added ex::to_polynomial().
560 * subs() accepts an optional "options" argument. The option
561 subs_option::subs_algebraic enables "smart" substitutions in products and
563 * Added stream manipulators "dflt", "latex", "python", "python_repr", "tree",
564 "csrc", "csrc_float", "csrc_double", "csrc_cl_N", "index_dimensions" and
565 "no_index_dimensions" to control the output format. Calling basic::print()
566 directly is now deprecated.
567 * Made the hashing more simple and efficient.
568 * Caveats for class implementors:
569 - basic::subs(): third argument changed from "bool" to "unsigned"
570 - unarchiving constructor and basic::unarchive(): "const" removed from
572 - basic::let_op() should only be implemented if write access to
573 subexpressions is desired
574 - simplify_ncmul() renamed to eval_ncmul()
575 - simplified_ncmul() renamed to hold_ncmul()
576 - nonsimplified_ncmul() renamed to reeval_ncmul()
578 1.0.14 (1 March 2003)
579 * Improved the C-source output: complex numbers are printed correctly (using
580 the STL complex<> template or cln::complex()), rational numbers use cl_RA()
581 in the CLN output, and small integers are printed in a more compact format
582 (e.g. "2.0" instead of "2.0000000e+00").
583 * function_options::set_return_type() and function_options::do_not_evalf_params()
586 1.0.13 (27 January 2003)
587 * Contracting epsilon tensors with Euclidean indices now works.
588 * Improved dummy index symmetrization in sums.
589 * Added dirac_gammaL/R(), which can be used instead of dirac_gamma6/7()
590 but are single objects, to allow for a more compact notation of Dirac
592 * Powers with negative numeric exponents are printed as fractions in the
594 * Added symbolic_matrix() for the convenient creation of matrices filled
596 * Added collect_common_factors() which collects common factors from the
598 * simplify_indexed() converts "gamma~mu*p.mu" to "p\".
600 1.0.12 (30 October 2002)
601 * Fixed a bug in power::expand() that could produce invalid expressions.
602 * The input parser no longer ignores extra data following accepted input.
603 * Improved the CLN C-source output (integers are printed as integers, and
604 floating point numbers include the precision).
605 * Fixed a problem in the LaTeX-output of negative fractions.
606 * Added print_latex() and print_csrc() to ginsh.
607 * The sprem() function is now public.
609 1.0.11 (18 September 2002)
610 * Fixed a possible memory corruption in contractions of indexed objects with
611 delta or metric tensors.
612 * Computing the derivative of a power series object with respect to a symbol
613 that is not the expansion variable now works correctly.
614 * Several bugfixes in code generation.
616 1.0.10 (24 July 2002)
617 * Powers of indexed objects are now parenthesized correctly in LaTeX output.
618 * Input parser handles indices (they have to be specified in the same list
620 * Added some limited support for subspaces in the idx and tensor classes.
621 * Fixed a bug in canonicalize() (antisymmetric canonicalization of an already
622 sorted list containing two or more equal objects failed to return 0).
625 * simplify_indexed() now raises/lowers dummy indices to canonicalize the index
626 variance. This allows some simplifications that weren't possible before,
627 like eps~a.b~c~d*X.a*X~b -> 0 and X.a~a-X~a.a -> 0.
628 * Implemented relational::subs().
629 * Fixed bug in simplify_ncmul() for clifford objects.
631 1.0.8 (31 March 2002)
632 * Improvements in memory usage of the expand() methods.
634 1.0.7 (18 March 2002)
635 * Fixed LaTeX output of indexed and matrix objects.
636 * Fixed matrix::pow(n) for n==0 and added helper functions to create unit
637 matrices "ex unit_matrix(unsigned, unsigned)".
640 * "(x+1).subs(x==x-1)" now returns the correct result "x" instead of "x-1".
642 1.0.5 (27 January 2002)
643 * (l)degree(s), coeff(s, n) and collect(s) were extended to accept expressions
644 of any class (except add/mul/ncmul/numeric) for "s". They should even work
645 if "s" is a "power" object, as long as the exponent is non-integer, but with
646 some limitations. For example, you can "collect(a*2^x+b*2^x, 2^x)" to get
647 "(a+b)*2^x", but "degree(2^(3*x), 2^x)" yields 0 instead of 3).
648 * Fixed a small output bug.
650 1.0.4 (24 January 2002)
651 * Speedup in expand().
652 * Faster Bernoulli numbers (Markus Nullmeier).
653 * Some minor bugfixes and documentation updates.
655 1.0.3 (21 December 2001)
656 * Fixed a bug where quo() would call vector::reserve() with a negative
658 * Fix several bugs in code generation.
660 1.0.2 (19 December 2001)
661 * Input parser recognizes "sqrt()", which is also used in the output.
662 * divide(a,b,q) only modifies q if the division succeeds; also, divide(a,b,a)
664 * Fixed small bug in dummy index renaming which could cause it to not
665 recognize renamable indices in some cases.
666 * power::degree() and power::ldegree() throw an exception when encountering
667 a non-integer exponent.
668 * Add output-support for Python bindings.
670 1.0.1 (22 November 2001)
671 * Function sqrfree() handles a few more cases now.
672 * Class relational has real canonical ordering now.
673 * Handle obscene libreadline version numbers when building ginsh.
675 1.0.0 (6 November 2001)
676 * Some internal reorganization resulting in a general speed-up.
677 * The last 3 evaluated expressions in ginsh are now referred to with the
678 tokens '%', '%%' and '%%%'. The old '"', '""' and '"""' remain for
679 compatibility but may be removed in a future version of GiNaC.
681 0.9.4 (20 September 2001)
682 * Functions have better support for external scripting languages.
683 * Interface cleanups and bugfixes.
684 * Fix silly bug in evalf() that prevented things like 2^Pi being computed.
686 0.9.3 (16 August 2001)
687 * series expansion now much more consistent for small order expansion.
688 * lsolve() accepts algorithmic hint as parameter.
691 * Epsilon tensor is more functional.
692 * simplify_indexed() is better at detecting expressions that vanish for
694 * Several little bugfixes and consistency enhancements.
697 * Ctors of class numeric are not explicit any more. All built-in callers for
698 pseudofunctions are now templated and default to ex arguments which relaxes
699 the need for explicit ctors.
700 * New functions/methods:
702 - remove_first(), remove_last(), sort() and unique() for lists
703 - symmetrize_cyclic()
705 * Instead of just totally symmetric or antisymmetric, complex symmetries
706 can now be defined for indexed objects. Symmetries are described by a
707 tree of "symmetry" objects that is constructed with the sy_none(),
708 sy_symm(), sy_anti() and sy_cycl() functions. The symmetry of a function
709 with respect to its arguments can also be defined (this is currently
710 only used for the Beta function).
711 * Generalized map() to take a function object instead of a function pointer.
712 This allows passing an arbitrary number of additional state to the
713 function being called.
714 * color_trace(), dirac_trace(), diff(), expand(), evalf() and normal() work
715 better with container classes, e.g. using color_trace() on a relation will
716 take the trace on both sides, using diff() on a matrix differentiates every
718 * diff() works properly with non-commutative products and indexed objects.
719 * New option flag "expand_function_args" for expand().
720 * Supplement some (now deprecated) macros by inlined template functions:
721 - is_of_type(foo, type) -> is_a<type>(foo)
722 - is_ex_of_type(foo, type) -> is_a<type>(foo)
723 - is_exactly_of_type(foo, type) -> is_exactly_a<type>(foo)
724 - is_ex_exactly_of_type(foo, type) -> is_exactly_a<type>(foo)
725 - ex_to_foobar(baz) -> ex_to<foobar>(baz)
726 * rem(c, p[x], x) (c: numeric, p[x]: polynomial) erroneously returned p[x]
728 * Small bugfixes in pattern matching.
729 * Updated libtool to version 1.4.
732 * In the output and in ginsh, lists are now delimited by { } braces, and
733 matrices are delimited by single [ ] brackets.
734 * simplify_indexed() renames dummy indices so, e.g., "a.i*a.i+a.j*a.j" gets
735 simplified to "2*a.i*a.i" (or "2*a.j*a.j").
736 * New functions/methods:
737 - canonicalize_clifford() (helpful when comparing expressions containing
739 - symmetrize() and antisymmetrize()
740 - numer_denom() (return numerator and denominator in one call)
741 - map() (apply function to subexpressions)
742 - evalm() (evaluate sums, products and integer powers of matrices)
743 * Added a new function match() for performing pattern matching. subs() and
744 has() also accept patterns as arguments. A pattern can be any expression,
745 optionally containing wildcard objects. These are constructed with the
746 call "wild(<unsigned>)" and are denoted as "$0", "$1" etc. in the output
748 * Positive integer powers of non-commutative expressions (except matrices)
749 are automatically expanded.
750 * Removed cint subdirectory, ginaccint is a separate package now due to
751 packaging considerations.
752 * Several little bugfixes.
755 * color and clifford classes are functional and documented.
756 * New "spinidx" class for dotted/undotted indices.
757 * Predefined spinor metric tensor (created by spinor_metric()).
758 * Symbols can have a LaTeX name, e.g. symbol s("s", "\\sigma");
759 * LaTeX output of indexed objects is much nicer.
760 * Fixed some build problems (with recent libreadline).
761 * Semantics of arithmetic operators now follows the C++ rules more strictly.
763 0.8.2 (24 April 2001)
764 * degree(), ldegree(), coeff(), lcoeff(), tcoeff() and collect() work with
765 non-symbols as the second argument in ginsh.
766 * the argument to collect() can be a list of objects in which case the
767 result is either a recursively collected polynomial, or a polynomial in
768 a distributed form with terms like coeff*x1^e1*...*xn^en, as specified by
769 the second argument to collect().
770 * Several bugfixes (including a nasty memory leak in .normal()).
771 * class matrix: solve() doesn't call algorithms redundantly any more and
772 inverse() falls back to solve() which works in more general cases.
774 0.8.1 (16 April 2001)
775 * degree(), ldegree(), coeff(), lcoeff(), tcoeff() and collect() can now
776 be used with constants, functions and indexed expressions as well, so you
777 can use it to collect by powers of Pi or sin(x), or to find the coefficient
780 - it only works with symbols, constants, functions and indexed expressions,
781 trying to find the coefficient of, e.g., "x^2" or "x+y" won't work;
782 - it does not know about dummy index summations; the coefficient of
783 gamma~0 in p.mu*gamma~mu should be p.0 but is returned as 0;
784 - using coeff(), tcoeff(), lcoeff() or collect() on elements of
785 noncommutative products might return wrong or surprising results.
786 * subs() no longer only substitutes symbols and indices but performs a more
787 general "syntactic substitution", i.e. it substitutes whole objects in sub-
788 expressions. You can subs((a+b)^2,a+b==3) and get 9, but subs(a+b+c,a+b==3)
791 - substituting numerics (subs(expr, 2==4)) will not replace then in all
792 occurences; in general, you shouldn't substitute numerics, though.
793 * Added preliminary (re)implementations of color and clifford classes.
794 * simplify_indexed(): contraction of symmetric and antisymmetric tensors
796 * Replaced the various print*() member functions by a single print() that
797 takes a print_context object that determines the output formatting. This
798 should make it easier to add more output types, such as LaTeX output,
799 which is based on work by Stefan Weinzierl.
800 * Added functions to retrieve the properties stored in archive objects
801 outside of unarchive() (for printing or debugging purposes).
802 * Some bugfixes (indexed objects, archive writing).
803 * .collect() on non-polynomials is now algebraically correct.
805 0.8.0 (24 March 2001)
806 * Complete revamp of indexed objects. Instead of multiple classes for
807 indexed things and their indices there is now only one "indexed" class
808 and two types of indices: "idx" for simple indices and "varidx" for
809 indices with variance. There are predefined delta, epsilon and metric
810 tensors, and a function simplify_indexed() that performs canonicalization
811 and dummy index summations. Matrix objects can be indexed for doing simple
813 * Added an option "expand_indexed" to expand() to perform expansion of
814 indexed objects like (a+b).i -> a.i + b.i
815 * Renamed get_indices() to get_free_indices(), which no longer returns
816 dummy indices and checks the consistency of indices in sums.
817 * sqrfree() factorization fixed and improved syntactically.
818 * subs() works on matrices.
819 * Matrices can be constructed from flat list of elements; diagonal matrices
820 can be constructed from list of diagonal elements with diag_matrix().
821 * Fixed memory leak in expand().
822 * Operator% for objects of class ncmul has gone. Use operator* now for that
823 case too, which is much more natural.
825 0.7.3 (28 February 2001)
826 * Several bugfixes and minor performance tunings.
827 * Added a section to the tutorial about adding new algebraic classes to GiNaC.
828 * Closed many in-source documentation gaps.
830 0.7.2 (17 February 2001)
831 * Several bugfixes in power series expansion, one of them critical.
833 0.7.1 (7 February 2001)
834 * Fix problems with Cint that were caused by CLN's overloaded operator new.
835 * Fix compilation errors with GCC3.
836 * normal() handles large sums of fractions better and normalizes the exponent
837 of power expressions.
838 * expand() always expands the exponent and transforms x^(a+b) -> x^a*x^b.
839 * Some bugfixes of series expansion around branch cuts of special functions.
841 0.7.0 (15 December 2000)
842 * Requires CLN 1.1 now. Class numeric doesn't use an indirect pointer to the
843 actual representation any more. This is a speedup.
844 * mul::expand() was reengineered to not allocate excess temporary memory.
845 * Non-integer powers of a symbol are treated as constants by (l)degree() and
846 coeff(). Using these functions on an expression containing such powers used
847 to fail with an internal error message. The side-effect is that collect()
848 can be used on expressions which are not polynomials.
849 * Added a man page for the ginac-config script.
850 * Ctor of numeric from char* honors Digits.
852 0.6.4 (10 August 2000)
853 * Complete revamp of methods in class matrix. Some redundant (and poor)
854 implementations of elimination schemes were thrown out. The code is now
855 highly orthogonal, more flexible and much more efficient.
856 * Some long standing and quite nasty bugs were discovered and fixed in the
857 following functions: add::normal(), heur_gcd(), sr_gcd() and Order_eval().
860 * Derivatives are now assembled in a slightly different manner (i.e. they
861 might 'look' different on first sight). Under certain circumstances this
862 can result in a dramatic speedup because it gives hashing a better chance,
863 especially when computing higher derivatives.
864 * Some series expansions of built-in functions have been reengineered.
865 * The algorithm for computing determinants can be chosen by the user. See
866 ginac/flags.h and ginac/matrix.h.
867 * The Dilogarithm (Li2) now has floating point evaluation, derivative and a
868 proper series expansion.
869 * Namespace 'std' cleanly disentangled, as demanded by ISO/EIC 14882-1998(E).
870 * Some minor bugfixes, one major lsolve()-bugfix and documentation updates.
873 * ginaccint.bin is now launched by a binary program instead of by a scripts.
874 This allows us to write #!-scripts. A small test suite for GiNaC-cint was
876 * Several minor bugfixes.
879 * Cleanup in the interface to Cint. The required version is now Cint 5.14.38.
880 * Several bugfixes in target install.
883 * IMPORTANT: Several interface changes make programs written with GiNaC
884 much clearer but break compatibility with older versions:
885 - f(x).series(x,p[,o]) -> f(x).series(x==p,o)
886 - series(f(x),x,p[,o]) -> series(f(x),x==p,o)
887 - gamma() -> tgamma() (The true Gamma function, there is now also
888 log(tgamma()), called lgamma(), in accord with ISO/IEC 9899:1999.)
889 - EulerGamma -> Euler
890 * #include'ing ginac.h defines the preprocessor symbols GINACLIB_MAJOR_VERSION,
891 GINACLIB_MINOR_VERSION, and GINACLIB_MICRO_VERSION with the respective GiNaC
892 library version numbers.
893 * Expressions can be constructed from strings like this:
894 ex e("2*x+y", lst(x, y));
895 * ex::to_rational() provides a way to extend the domain of functions like
896 gcd() and divide() that only work on polynomials or rational functions (the
897 good old ex::subs() method reverses this process)
898 * Calling diff() on a function that has no derivative defined returns the
899 inert derivative function "Derivative".
900 * Several new timings in the check target. Some of them may be rather rude
901 at your machine, feel free to interrupt them.
903 0.5.4 (15 March 2000)
904 * Some algorithms in class matrix (notably determinant) were replaced by
905 less brain-dead ones and should now have much better performance.
906 * Checks were completely reorganized and split up into three parts:
907 a) exams (small regression tests with predefined input)
908 b) checks (lenghty coherence checks with random input)
909 c) timings (for coherence and crude benchmarking)
910 * Behaviour of .evalf() was changed: it doesn't .evalf() any exponents.
911 * Expanded expressions now remember they are expanded to prevent
912 superfluous expansions.
913 * Small bugfixes and improvements in the series expansion.
915 0.5.3 (23 February 2000)
916 * A more flexible scheme for registering functions was implemented,
917 allowing for remembering, too.
920 0.5.2 (16 February 2000)
921 * Mainly fixes a bug in the packaging of release 0.5.1.
923 0.5.1 (14 February 2000)
924 * Fixes a small number of bugs.
926 0.5.0 (7 February 2000)
927 * Expressions can be written ("archived") to files and read therefrom.
928 * Addition of GiNaC-cint, which lets you write complete programs in
929 an interactive shell-like manner in your favoured programming
932 0.4.1 (13 December 1999)
933 * Series Expansion of Gamma function and some other trigonometric
934 functions at their poles works now.
935 * Many more evaluations of special functions at points where
937 * info_flags::rational doesn't return true for complex extensions
938 any more---use info_flags::crational for the old behaviour.
939 info_flags::integer and -::cinteger work similarly, the same
940 holds for types like info_flags::rational_polynomial.
942 0.4.0 (26 November 1999)
943 * First public release.