- GiNaC-list - ginac.de

Possible bug in algebraic substitution
by Jan Rheinländer 30 Aug '26

30 Aug '26

Hi, maybe this is a bug, maybe combining "algebraic" and "no_pattern" is just not allowed. Here is the program: #include <iostream> #include <ginac.h> using namespace GiNaC; int main() { symbol x("x"); symbol y("y"); symbol A("A"); ex w(wild()); exmap repl; repl[w] = A; ex expr(x/y + w); ex result = expr.subs(repl, subs_options::algebraic | subs_options::no_pattern); std::cout << "Result: " << result << std::endl; return 0; } Expected output: x/y + A Actual output: A/y + A Apparently mul::algebraic_subs_mul() does not honour the no_pattern flag. ---------------- The practical reason why this problem arose is to achieve algebraic substitution in sums: #include <iostream> #include <ginac.h> using namespace GiNaC; int main() { symbol x("x"); symbol y("y"); symbol z("z"); symbol A("A"); symbol B("B"); ex w(wild()); exmap repl; repl[x + y + w] = A/B + w; ex expr = x + y + z; // Try to substitute x+y with A/B ex result = expr.subs(repl, subs_options::algebraic); std::cout << "Result: " << result << std::endl; return 0; } Expected output: A/B + z Actual output: z/B + z BTW, I noticed that expairseq::match() can do algebraic matching in sums. Why can it not be implemented for add::subs(), too? Regards, Jan

2 2

Wrongly generated C code
by Diddens, Christian (UT-TNW) 06 Aug '26

06 Aug '26

Dear GiNaC team, during debugging a complicated Jacobian generated with my FEM framework pyoomph via GiNaC, I was stuck, since the expressions were quite lenghty. I tracked down an issue with the C code generation of inverse powers ( x**^-1) when combining both an int (-1) and a float (-1.0) in the same expressions. It is now manually fixed in pyoomph y pre-parsing the expression, but of course, it should be patched in GiNaC itself. Below please find a code to reproduce it. Best regards, Christian PS: Thanks for GiNaC and CLN! It was the best decision to use it in my code! // GiNaC: print_csrc silently drops the reciprocal of a factor whose exponent is an inexact -1, i.e. // numeric(-1.0) rather than numeric(-1). // Seen on 1.8.7 (Ubuntu libginac-dev, shared, against libcln.so.6) and on 1.8.10 built from source. // // GiNaC compares numbers by value, so numeric(-1) and numeric(-1.0) are is_equal and hash alike; // the two expressions below are one and the same expression as far as every computation is // concerned. mul::do_print_csrc, however, asks two different questions about the exponent and gets // inconsistent answers (ginac/mul.cpp, do_print_csrc): // // // If the first argument is a negative integer power, it gets printed as "1.0/<expr>" // if (it == seq.begin() && it->coeff.info(info_flags::negint)) { ... c.s << "1.0/"; } // // // If the exponent is 1 or -1, it is left out // if (it->coeff.is_equal(_ex1) || it->coeff.is_equal(_ex_1)) // it->rest.print(c, precedence()); // ... // // Separator is "/" for negative integer powers, "*" otherwise // if (it->coeff.info(info_flags::negint)) c.s << "/"; else c.s << "*"; // // numeric::info(info_flags::negint) requires a CLN integer and is therefore false for -1.0, while // is_equal(_ex_1) compares by value and is therefore true. So neither the "1.0/" prefix nor the "/" // separator is emitted, and the exponent is left out all the same: x^(-1.0) inside a product is // printed as a plain multiplication by x. #include <ginac/ginac.h> #include <iostream> #include <sstream> #include <string> using namespace GiNaC; template <typename Context> static std::string csrc(const ex &e) { std::ostringstream os; e.print(Context(os)); return os.str(); } // `as_printed` is what the emitted C source literally says, written out as an expression again, so // that the two can be evaluated side by side. It is 0 where the output is correct. static void show(const std::string &what, const ex &e, const symbol &x, const ex &as_printed = 0) { std::cout << " " << what << "\n" << " expression : " << e << "\n" << " C source : " << csrc<print_csrc_double>(e) << "\n" << " expression @x=2 : " << e.subs(x == 2).evalf() << "\n"; if (!as_printed.is_zero()) std::cout << " that C @x=2 : " << as_printed.subs(x == 2).evalf() << " <-- WRONG\n"; std::cout << "\n"; } int main() { symbol x("x"); const ex exact = 3 * pow(1 + x, -1); // exponent is a CLN integer const ex inexact = 3 * pow(1 + x, numeric(-1.0)); // exponent is a CLN float const ex minus_one = numeric(-1), minus_one_point_zero = numeric(-1.0); std::cout << "The two exponents, and hence the two expressions, are equal to GiNaC:\n" << " -1 is_equal -1.0 = " << minus_one.is_equal(minus_one_point_zero) << "\n" << " equal hashes = " << (minus_one.gethash() == minus_one_point_zero.gethash()) << "\n" << " (exact - inexact).is_zero() = " << (exact - inexact).is_zero() << "\n" << " info(negint) of -1 and -1.0 = " << minus_one.info(info_flags::negint) << " and " << minus_one_point_zero.info(info_flags::negint) << " <-- the inconsistency\n\n"; std::cout << "They are not printed as the same C code:\n\n"; show("3*(1+x)^(-1) correct", exact, x); show("3*(1+x)^(-1.0) the reciprocal is gone", inexact, x, 3 * (1 + x)); std::cout << "Only an exponent of exactly -1 is affected, and only inside a product:\n\n"; show("(1+x)^(-1.0) correct - power::do_print_csrc handles it", pow(1 + x, numeric(-1.0)), x); show("3*(1+x)^(-2.0) correct - falls through to pow()", 3 * pow(1 + x, numeric(-2.0)), x); show("3*(1+x)^(-1/2) correct - not an integer either way", 3 * pow(1 + x, numeric(-1, 2)), x); show("x*(1+x)^(-1.0) the reciprocal is gone", x * pow(1 + x, numeric(-1.0)), x, x * (1 + x)); std::cout << "The other C source contexts print it the same way:\n" << " print_csrc_double : " << csrc<print_csrc_double>(inexact) << "\n" << " print_csrc_float : " << csrc<print_csrc_float>(inexact) << "\n" << " print_csrc_cl_N : " << csrc<print_csrc_cl_N>(inexact) << "\n" << "for comparison, the same expression with an exact exponent:\n" << " print_csrc_double : " << csrc<print_csrc_double>(exact) << "\n" << " print_csrc_float : " << csrc<print_csrc_float>(exact) << "\n" << " print_csrc_cl_N : " << csrc<print_csrc_cl_N>(exact) << "\n\n"; return 0; }

2 3

Slowdown in e.g. expansion with more variables defined
by Vitaly Magerya 10 Apr '26

10 Apr '26

Hi, all. I've noticed a worrying performance issue, and I was wondering if you could shed some light onto it. In ginsh, the time to do a simple expand() seems to get worse the more variables you've defined. Here's a test script: FIRST_RUN; time(e1=expand((2*x+3*y+5*z)^100)); time(e2=expand((2*x+3*y+5*z)^100)); time(e3=expand((2*x+3*y+5*z)^100)); time(e4=expand((2*x+3*y+5*z)^100)); time(e5=expand((2*x+3*y+5*z)^100)); time(e6=expand((2*x+3*y+5*z)^100)); time(e7=expand((2*x+3*y+5*z)^100)); time(e8=expand((2*x+3*y+5*z)^100)); time(e9=expand((2*x+3*y+5*z)^100)); UNASSIGN;; unassign('e1'): unassign('e2'): unassign('e3'): unassign('e4'): unassign('e5'): unassign('e6'): unassign('e7'): unassign('e8'): unassign('e9'): SECOND_RUN; time(f1=expand((2*x+3*y+5*z)^100)); time(f2=expand((2*x+3*y+5*z)^100)); time(f3=expand((2*x+3*y+5*z)^100)); time(f4=expand((2*x+3*y+5*z)^100)); time(f5=expand((2*x+3*y+5*z)^100)); time(f6=expand((2*x+3*y+5*z)^100)); time(f7=expand((2*x+3*y+5*z)^100)); time(f8=expand((2*x+3*y+5*z)^100)); time(f9=expand((2*x+3*y+5*z)^100)); The output is: FIRST_RUN 0.00923s 0.008251s 0.007379s 0.007837s 0.015366s 0.015251s 0.019319s 0.020222s 0.021082s UNASSIGN SECOND_RUN 0.00717s 0.006453s 0.006901s 0.007438s 0.015191s 0.012014s 0.015484s 0.018273s 0.019675s As you can see the expansion time for e9 is 2x that of e1, and if I'll continue, it will keep rising, to e.g. 20x at e100 and beyond. However, if I'll unassign the variables, the expansion time will be back to the original value. My guess would be that some kind of a caching system is at fault for this. Do you have any ideas as to which one exactly, and how to solve this? Thanks in advance, Vitaly

2 5

compiling date 22 / 03 / 2026
by Marko Riedel 23 Mar '26

23 Mar '26

Greetings. When I try to compile the latest version cloned from Git I get the following error: add.cpp: In member function ‘void GiNaC::add::print_add_sorted(const GiNaC::print_context&, const char*, const char*, const char*, unsigned int) const’: add.cpp:176:66: error: ‘std::ostream’ {aka ‘class std::basic_ostream<char>’} has no member named ‘str’ sorted_seq.emplace((std::ostringstream{} << sortex << it.rest).str(), it); I had been using the precompiled version that was included in OpenSuSE. Do I need to upgrade my installation? Best regards, Marko Riedel

2 3

pattern matching
by Marko Riedel 17 Mar '26

17 Mar '26

Greetings to all. I have two expressions that I would like to compare for tree structure. Here the tree terminal nodes are all from some set of symbols and I would like to know whether the two trees match where ideally I obtain a list of bindings that turn the first tree into the second. I tried building a wildcard pattern with different wildcards for each symbol and then invoking the match function. Unfortunately this did not work due to multiple wildcards appearing in sum terms, which the tutorial says will give ambiguous results like with a+$1+$2. For example, comparing L=Q(n,k)+Q(k,2) and Q(q,p)+Q(p,r) should give the consistent bindings (n => q, k => p, 2 => r). No such binding exists for L to Q(p,p)+Q(p,r) or Q(p,q)+Q(q,r)+Q(r,3) (different expression tree). The match Q(n+m+k,q) and Q(s+j+l, t) should go through and produce at least one binding such as (n=>s, m=>j, k=>l, q=>t). I understand that here we have permutations of the constituent terms of sums and products that need to be taken into account. Regards, Marko

1 1

Modification for ginac-1.8.10/check/exam_Gt.cpp (MacOS 26.3)
by 足立聡 17 Feb '26

17 Feb '26

Dear Developers of GiNaC, [0] Thank you very much for developing continuously GiNaC, which is a very nice C++ library to write programs for symbolic computation. I recently built ginac-1.8.10 from source code on my MacBook Pro. The compilation stage was successful but the check stage failed. To pass the check stage, I needed several modifications of the source file ginac-1.8.10/check/exam_Gt.cpp. I will explain these modifications below. My environment is the following: OS: macOS Tahoe 26.3 C++ compiler: g++ --version --> Apple clang version 17.0.0 (clang-1700.6.3.2) Target: arm64-apple-darwin25.3.0 Thread model: posix InstalledDir: /Applications/Xcode.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/bin [1] The build from the original source code of ginac-1.8.10 produced the following result: mkdir /tmp/AAA /tmp/AAA bzcat /usr/local/src/SymbolicComputation/GiNaC/ginac-1.8.10.tar.bz2 | tar xvf - cd ginac-1.8.10 ./configure make make check --> ... ... /bin/sh ../libtool --tag=CXX --mode=link g++ -g -O2 -o exam_inifcns_elliptic exam_inifcns_elliptic.o ../ginac/libginac.la -L/usr/local/lib -lcln libtool: link: g++ -g -O2 -o .libs/exam_inifcns_elliptic exam_inifcns_elliptic.o -Wl,-bind_at_load ../ginac/.libs/libginac.dylib -ldl -L/usr/local/lib /usr/local/lib/libcln.dylib /usr/local/lib/libgmp.dylib ld: warning: -bind_at_load is deprecated on macOS depbase=`echo exam_Gt.o | sed 's|[^/]*$|.deps/&|;s|\.o$||'`;\ g++ -DHAVE_CONFIG_H -I. -I../config -I./../ginac -I../ginac -DIN_GINAC -I/usr/local/include -g -O2 -MT exam_Gt.o -MD -MP -MF $depbase.Tpo -c -o exam_Gt.o exam_Gt.cpp &&\ mv -f $depbase.Tpo $depbase.Po exam_Gt.cpp:1891:17: error: call to 'pow' is ambiguous 1891 | const ex eps = pow(10,-30); | ^~~ /Applications/Xcode.app/Contents/Developer/Platforms/MacOSX.platform/Developer/SDKs/MacOSX.sdk/usr/include/c++/v1/__math/exponential_functions.h:161:65: note: candidate function [with _A1 = int, _A2 = int, $2 = 0] 161 | inline _LIBCPP_HIDE_FROM_ABI typename __promote<_A1, _A2>::type pow(_A1 __x, _A2 __y) _NOEXCEPT { | ^ ./../ginac/power.h:120:11: note: candidate function [with T1 = int, T2 = int] 120 | inline ex pow(const T1 & b, const T2 & e) | ^ 1 error generated. [2] The above error indicates that pow() in GiNaC and pow() in MacOS conflict as having the same name. Accordingly, I modified exam_Gt.cpp as follows: cd /tmp/AAA/ginac-1.8.10/check mv exam_Gt.cpp exam_Gt.cpp.org cat exam_Gt.cpp.org | sed -e 's/pow(10,-30)/GiNaC::pow(10,-30)/' > exam_Gt.cpp diff -u exam_Gt.cpp.org exam_Gt.cpp --> ------------------------------------------------------------------------------ --- exam_Gt.cpp.org 2026-02-10 01:49:00 +++ exam_Gt.cpp 2026-02-17 07:54:20 @@ -1888,7 +1888,7 @@ }; Digits = 40; - const ex eps = pow(10,-30); + const ex eps = GiNaC::pow(10,-30); auto evaluate = [&](const ex& points) { for (int deform = -1; deform <= +1; deform += 1) { ------------------------------------------------------------------------------ [3] Then the check stage proceeded as follows make check --> ... ... PASS: exam_paranoia PASS: exam_heur_gcd PASS: exam_match PASS: exam_parser PASS: exam_numeric PASS: exam_relational PASS: exam_powerlaws PASS: exam_collect PASS: exam_inifcns PASS: exam_inifcns_nstdsums PASS: exam_inifcns_elliptic ../config/test-driver: line 112: 73792 Abort trap: 6 "$@" >> "$log_file" 2>&1 FAIL: exam_Gt PASS: exam_differentiation PASS: exam_polygcd PASS: exam_collect_common_factors PASS: exam_normalization PASS: exam_sqrfree PASS: exam_factor PASS: exam_pseries PASS: exam_matrices PASS: exam_lsolve PASS: exam_indexed PASS: exam_color PASS: exam_clifford PASS: exam_archive PASS: exam_structure PASS: exam_misc PASS: exam_pgcd PASS: exam_mod_gcd PASS: exam_chinrem_gcd PASS: exam_function_exvector PASS: exam_real_imag PASS: check_numeric PASS: check_inifcns PASS: check_matrices PASS: check_lsolve PASS: check_cra PASS: time_dennyfliegner PASS: time_gammaseries PASS: time_vandermonde PASS: time_toeplitz PASS: time_lw_A PASS: time_lw_B PASS: time_lw_C PASS: time_lw_D PASS: time_lw_E PASS: time_lw_F PASS: time_lw_G PASS: time_lw_H PASS: time_lw_IJKL PASS: time_lw_M1 PASS: time_lw_M2 PASS: time_lw_N PASS: time_lw_O PASS: time_lw_P PASS: time_lw_Pprime PASS: time_lw_Q PASS: time_lw_Qprime PASS: time_antipode PASS: time_fateman_expand PASS: time_uvar_gcd PASS: time_Gt PASS: time_parser ============================================================================ Testsuite summary for GiNaC 1.8.10 ============================================================================ # TOTAL: 63 # PASS: 62 # SKIP: 0 # XFAIL: 0 # FAIL: 1 # XPASS: 0 # ERROR: 0 ============================================================================ See check/test-suite.log for debugging. Some test(s) failed. Please report this to ginac-list(a)ginac.de, together with the test-suite.log file (gzipped) and your system information. Thanks. ============================================================================ [3] I have investigated the above failure of exam_Gt and found that in exam_Gt.cpp the code fragment lst{lst{A,B}} is expected to produce { { A, B } } but actually produces { A, B } when g++ on my MacOS is used. Namely, the outside constructor lst in lst{lst{A,B}} is interpreted as the copy constructor. Accordingly, I modify lst{lst{A,B}} to lst().append(lst{A,B}) as follows: cat exam_Gt.cpp.org | sed -e 's/lst{$lst{[^}]*}$}/lst().append(\1)/' -e 's/pow(10,-30)/GiNaC::pow(10,-30)/' > exam_Gt.cpp [4] Now, the check stage becomes successful as follows: make check --> ... ... e_lw_M2 PASS: time_lw_N PASS: time_lw_O PASS: time_lw_P PASS: time_lw_Pprime PASS: time_lw_Q PASS: time_lw_Qprime PASS: time_antipode PASS: time_fateman_expand PASS: time_uvar_gcd PASS: time_Gt PASS: time_parser ============================================================================ Testsuite summary for GiNaC 1.8.10 ============================================================================ # TOTAL: 63 # PASS: 63 # SKIP: 0 # XFAIL: 0 # FAIL: 0 # XPASS: 0 # ERROR: 0 ============================================================================ Now, everything is fine. I do not know what is the most appropriate way to fix the code fragment lst{lst{A,B}} to produce { { A, B } } and please take the above modification as just a reference. Again, thank you very much for developing and maintaining GiNaC to help us write programs for symbolic commutation. Sincerely yours, Satoshi Adachi

1 0

Primary repo moved to Codeberg.org
by Richard B. Kreckel 15 Feb '26

15 Feb '26

Hi! TL;DR: If you have cloned the GiNaC git repository, please do: $ git remote set-url origin https://codeberg.org/ginac/ginac.git So it's been decided to modernize the environment a bit... Several people have expressed interest in doing things with GiNaC and kindly suggested that it would be easier to have a fashionable forge-based workflow where they can clone the repository, make changes, send pull requests, etc. Another consideration was the ongoing surge of distributed AI scrapers aggressively crawling our Gitweb instance in millisecond intervals, blatantly ignoring robots.txt, and repeatedly overwhelming the poor server with web traffic. After having looked at several forge platforms, we (i.e. Bruno Haible and myself) have decided to move the primary repo of both GiNaC and CLN to Codeberg.org: <https://codeberg.org/ginac/ginac> Codeberg e.V. is a Berlin-based non-profit organization dedicated to build and maintain supporting infrastructure for the creation, collection, dissemination, and archiving of Free and Open Source Software. Their Forgejo software is a free git-based hosting platform with most of the bells and whistles known from GitHub, GitLab, etc. Modern hip hackers will feel right at home there! ;-) I've set us up as an Organization <https://codeberg.org/ginac/> hosting both the CLN and GiNaC repos. It should be pretty transparent. If you need something special, don't hesitate to contact me! Along the way, this acquits us from having to come up with lame excuses to the pesky question where our issue tracker is. It is now here: <https://codeberg.org/ginac/ginac/issues> For now, the web site <https://www.ginac.de/> and the mailing lists will remain as they are. Happy hacking! -richy. -- Richard B. Kreckel <https://in.terlu.de/~kreckel/>

2 1

GiNaC 1.8.10 released
by Richard B. Kreckel 11 Feb '26

11 Feb '26

Hi, GiNaC 1.8.10 has been released. It includes a new class Gt for evaluating elliptic multiple polylogarithms with arbitrary arguments. (This is work by Claude Duhr, Florian Lorkowski, Robin Marzucca, Sofia Mauch, and Stefan Weinzierl.) The minimum C++ language standard level has been raised to C++14. This version is ABI-compatible to the previous version 1.8.9. Happy hacking! -richy. -- Richard B. Kreckel <https://in.terlu.de/~kreckel/>

2 1

No get_name() method of GiNaC::constant?
by Warren Weckesser 15 Jul '25

15 Jul '25

GiNaC::constant has a private string attribute called `name`. This is what is output when doing stream io. Is there a reason this class doesn't have a public get_name() method, so that code that generates code for another language can get the name instead of having to use, for example, stringstream? Warren

3 2

Can't compile example from the tutorial
by Warren Weckesser 03 Jul '25

03 Jul '25

I'm trying to compile the example from https://www.ginac.de/tutorial/#Input_002foutput. My complete version of the example, with 'include' and 'using' statements, and with an added declaration of x and y in main(), is: ----- #include <iostream> #include <ginac/ginac.h> using namespace std; using namespace GiNaC; static void my_print(const ex & e) { if (is_a<function>(e)) cout << ex_to<function>(e).get_name(); else cout << ex_to<basic>(e).class_name(); cout << "("; size_t n = e.nops(); if (n) for (size_t i=0; i<n; i++) { my_print(e.op(i)); if (i != n-1) cout << ","; } else cout << e; cout << ")"; } int main() { symbol x("x"), y("y"); my_print(pow(3, x) - 2 * sin(y / Pi)); cout << endl; return 0; } ----- I compile it with $ g++ --std=c++17 $(pkg-config --cflags ginac cln) custom_expression_output.cpp $(pkg-config --libs ginac cln) -o custom_expression_output which gives the following, with either g++ 11.4.0 or 12.3.0: ----- custom_expression_output.cpp: In function ‘void my_print(const GiNaC::ex&)’: custom_expression_output.cpp:16:9: error: parse error in template argument list 16 | if (is_a<function>(e)) | ^~~~~~~~~~~~~~ custom_expression_output.cpp:16:23: error: no matching function for call to ‘is_a<<expression error> >(const GiNaC::ex&)’ 16 | if (is_a<function>(e)) | ~~~~~~~~~~~~~~^~~ In file included from /home/warren/local/include/ginac/registrar.h:27, from /home/warren/local/include/ginac/basic.h:29, from /home/warren/local/include/ginac/ginac.h:28, from custom_expression_output.cpp:8: /home/warren/local/include/ginac/print.h:189:13: note: candidate: ‘template<class T> bool GiNaC::is_a(const GiNaC::print_context&)’ 189 | inline bool is_a(const print_context & obj) | ^~~~ /home/warren/local/include/ginac/print.h:189:13: note: template argument deduction/substitution failed: custom_expression_output.cpp:16:23: error: template argument 1 is invalid 16 | if (is_a<function>(e)) | ~~~~~~~~~~~~~~^~~ In file included from /home/warren/local/include/ginac/ginac.h:28, from custom_expression_output.cpp:8: /home/warren/local/include/ginac/basic.h:313:13: note: candidate: ‘template<class T> bool GiNaC::is_a(const GiNaC::basic&)’ 313 | inline bool is_a(const basic &obj) | ^~~~ /home/warren/local/include/ginac/basic.h:313:13: note: template argument deduction/substitution failed: custom_expression_output.cpp:16:23: error: template argument 1 is invalid 16 | if (is_a<function>(e)) | ~~~~~~~~~~~~~~^~~ In file included from /home/warren/local/include/ginac/ginac.h:30, from custom_expression_output.cpp:8: /home/warren/local/include/ginac/ex.h:955:13: note: candidate: ‘template<class T> bool GiNaC::is_a(const GiNaC::ex&)’ 955 | inline bool is_a(const ex &obj) | ^~~~ /home/warren/local/include/ginac/ex.h:955:13: note: template argument deduction/substitution failed: custom_expression_output.cpp:16:23: error: template argument 1 is invalid 16 | if (is_a<function>(e)) | ~~~~~~~~~~~~~~^~~ custom_expression_output.cpp:17:17: error: parse error in template argument list 17 | cout << ex_to<function>(e).get_name(); | ^~~~~~~~~~~~~~~ custom_expression_output.cpp:17:32: error: no matching function for call to ‘ex_to<<expression error> >(const GiNaC::ex&)’ 17 | cout << ex_to<function>(e).get_name(); | ~~~~~~~~~~~~~~~^~~ In file included from /home/warren/local/include/ginac/ginac.h:30, from custom_expression_output.cpp:8: /home/warren/local/include/ginac/ex.h:978:17: note: candidate: ‘template<class T> const T& GiNaC::ex_to(const GiNaC::ex&)’ 978 | inline const T &ex_to(const ex &e) | ^~~~~ /home/warren/local/include/ginac/ex.h:978:17: note: template argument deduction/substitution failed: custom_expression_output.cpp:17:32: error: template argument 1 is invalid 17 | cout << ex_to<function>(e).get_name(); | ~~~~~~~~~~~~~~~^~~ ----- If I change the first occurrence of "function" to "symbol" (just to see how it compiles), the expression is_a<symbol>(e) does not result in an error (but the compilation fails for other reasons). Any ideas of what is going on? Warren

2 3