Doctest: segmentation fault with coefficients() on symbolic expressions

[videlec]

On sage 8.0

sage: (x^2/(1+x)).coefficients()
Segmentation fault (core dumped)

gdb shows that this is an infinite recursion in the Pynac library:

#0  GiNaC::add::integer_content (this=0x4095f00) at normal.cpp:95
#1  0x00007fff50051660 in GiNaC::ex::integer_content (this=this@entry=0x7fffff7ff410) at normal.cpp:81
#2  0x00007fff50076fdc in GiNaC::power::eval (this=0x7fffff800040, level=<optimized out>) at power.cpp:674
#3  0x00007fff4ff9b29e in GiNaC::ex::construct_from_basic (other=...) at ex.cpp:691
#4  0x00007fff50041ac3 in ex (other=..., this=0x7fffff800030) at ex.h:305
#5  GiNaC::mul::combine_ex_with_coeff_to_pair (this=0x40961a0, e=..., c=...) at mul.cpp:1228
#6  0x00007fff4ffa3084 in GiNaC::expairseq::evalchildren (this=this@entry=0x40961a0, level=1) at expairseq.cpp:1647
#7  0x00007fff50047390 in GiNaC::mul::eval (this=0x40961a0, level=<optimized out>) at mul.cpp:666
#8  0x00007fff4ff9b29e in GiNaC::ex::construct_from_basic (other=...) at ex.cpp:691
#9  0x00007fff50049ff6 in ex (other=..., this=0x7fffff800690) at ex.h:305
#10 GiNaC::mul::expand (this=0x4096030, options=0) at mul.cpp:1490
#11 0x00007fff4ff9ac0d in GiNaC::ex::expand (this=this@entry=0x7fffff800bf0, options=options@entry=0) at ex.cpp:84
#12 0x00007fff4ff9eec9 in expand (options=0, thisex=...) at ex.h:701
#13 GiNaC::match_monom (term=..., symb=..., vec=..., map=...) at ex.cpp:523
#14 0x00007fff4ff9f8c8 in GiNaC::ex::coefficients (this=this@entry=0x7fffff800e00, s=..., vec=...) at ex.cpp:576
#15 0x00007fff4ff9eed7 in GiNaC::match_monom (term=..., symb=..., vec=..., map=...) at ex.cpp:523
#16 0x00007fff4ff9f8c8 in GiNaC::ex::coefficients (this=this@entry=0x7fffff8011c0, s=..., vec=...) at ex.cpp:576
#17 0x00007fff4ff9eed7 in GiNaC::match_monom (term=..., symb=..., vec=..., map=...) at ex.cpp:523
#18 0x00007fff4ff9f8c8 in GiNaC::ex::coefficients (this=this@entry=0x7fffff801580, s=..., vec=...) at ex.cpp:576
#19 0x00007fff4ff9eed7 in GiNaC::match_monom (term=..., symb=..., vec=..., map=...) at ex.cpp:523
#20 0x00007fff4ff9f8c8 in GiNaC::ex::coefficients (this=this@entry=0x7fffff801940, s=..., vec=...) at ex.cpp:576
[...]

CC: @rwst

Component: symbolics

Keywords: bug, symbolics

Author: Ralf Stephan

Branch/Commit: 617a644

Reviewer: Travis Scrimshaw

Issue created by migration from https://trac.sagemath.org/ticket/23545

[rwst]

Description changed:

--- 
+++ 
@@ -1,4 +1,13 @@
 On sage 8.0
+
+```
+sage: (x/(1+x)).coefficients()
+[[x/(x + 1), 0]]
+sage: (x^2/(1+x)).coefficients()
+Segmentation fault (core dumped)
+```
+
+Originally reported as:
 
 ```
 sage: q,a,b = SR.var('q,a,b')

[rwst]

comment:1

Confirmed. I added a minimal case.

[jdemeyer]

Description changed:

--- 
+++ 
@@ -1,17 +1,33 @@
 On sage 8.0
 
 ```
-sage: (x/(1+x)).coefficients()
-[[x/(x + 1), 0]]
 sage: (x^2/(1+x)).coefficients()
 Segmentation fault (core dumped)
 ```
 
-Originally reported as:
+gdb shows that this is an infinite recursion in the Pynac library:
 
 ```
-sage: q,a,b = SR.var('q,a,b')
-sage: F = (a^2 + 6*a + 2*b)*q/(a^5 + 3*a^4*b + 3*a^3*b^2 + a^2*b^3) + (3*a + b)/(a^5 + 3*a^4*b + 3*a^3*b^2 + a^2*b^3)
-sage: F.coefficients()
-Erreur de segmentation (core dumped)
+#0  GiNaC::add::integer_content (this=0x4095f00) at normal.cpp:95
+#1  0x00007fff50051660 in GiNaC::ex::integer_content (this=this@entry=0x7fffff7ff410) at normal.cpp:81
+#2  0x00007fff50076fdc in GiNaC::power::eval (this=0x7fffff800040, level=<optimized out>) at power.cpp:674
+#3  0x00007fff4ff9b29e in GiNaC::ex::construct_from_basic (other=...) at ex.cpp:691
+#4  0x00007fff50041ac3 in ex (other=..., this=0x7fffff800030) at ex.h:305
+#5  GiNaC::mul::combine_ex_with_coeff_to_pair (this=0x40961a0, e=..., c=...) at mul.cpp:1228
+#6  0x00007fff4ffa3084 in GiNaC::expairseq::evalchildren (this=this@entry=0x40961a0, level=1) at expairseq.cpp:1647
+#7  0x00007fff50047390 in GiNaC::mul::eval (this=0x40961a0, level=<optimized out>) at mul.cpp:666
+#8  0x00007fff4ff9b29e in GiNaC::ex::construct_from_basic (other=...) at ex.cpp:691
+#9  0x00007fff50049ff6 in ex (other=..., this=0x7fffff800690) at ex.h:305
+#10 GiNaC::mul::expand (this=0x4096030, options=0) at mul.cpp:1490
+#11 0x00007fff4ff9ac0d in GiNaC::ex::expand (this=this@entry=0x7fffff800bf0, options=options@entry=0) at ex.cpp:84
+#12 0x00007fff4ff9eec9 in expand (options=0, thisex=...) at ex.h:701
+#13 GiNaC::match_monom (term=..., symb=..., vec=..., map=...) at ex.cpp:523
+#14 0x00007fff4ff9f8c8 in GiNaC::ex::coefficients (this=this@entry=0x7fffff800e00, s=..., vec=...) at ex.cpp:576
+#15 0x00007fff4ff9eed7 in GiNaC::match_monom (term=..., symb=..., vec=..., map=...) at ex.cpp:523
+#16 0x00007fff4ff9f8c8 in GiNaC::ex::coefficients (this=this@entry=0x7fffff8011c0, s=..., vec=...) at ex.cpp:576
+#17 0x00007fff4ff9eed7 in GiNaC::match_monom (term=..., symb=..., vec=..., map=...) at ex.cpp:523
+#18 0x00007fff4ff9f8c8 in GiNaC::ex::coefficients (this=this@entry=0x7fffff801580, s=..., vec=...) at ex.cpp:576
+#19 0x00007fff4ff9eed7 in GiNaC::match_monom (term=..., symb=..., vec=..., map=...) at ex.cpp:523
+#20 0x00007fff4ff9f8c8 in GiNaC::ex::coefficients (this=this@entry=0x7fffff801940, s=..., vec=...) at ex.cpp:576
+[...]
 ```

[rwst]

comment:3

The Pynac code does not check if the expression is a polynomial fraction in x. That case cannot be handled via coefficient() (the user should expand as a series). I will add a check to prevent this problem.

[rwst]

Upstream: Reported upstream. Developers acknowledge bug.

[rwst]

Branch: u/rws/segmentation_fault_with_coefficients___on_symbolic_expressions

[rwst]

Commit: 617a644

[rwst]

comment:5

Pynac master has the fix. This branch adds doctests.

---

New commits:

617a644

23545: segmentation fault with coefficients() on symbolic expressions

[rwst]

Dependencies: pynac-0.7.10

[rwst]

Changed upstream from Reported upstream. Developers acknowledge bug. to Fixed upstream, in a later stable release.

[rwst]

Author: Ralf Stephan