problems with maxima inequalities

[sagetrac-dsm]

'Sage Version 4.4.1, Release Date: 2010-05-02'

mac 10.4 32 bit running on 10.5.8.

sage: solve([2*x==3, x < 10], x)
[[x == (3/2)]]
sage: solve([2*x==3, x > 10], x)
[]
sage: solve([2*x==3, x == 10], x)
[]
sage: solve([2*x==3, x == 3/2], x)
[[x == (3/2)]]

sage: solve([2*x==3, x < 4, x > 4], x)
[]

all work as expected, but:

sage: solve([2*x==3, x != 5], x)
-----------------------------------------------------------
!TypeError                                 Traceback (most recent call last)

/Applications/sage/devel/sage-main/build/sage/<ipython console> in <module>()
/Applications/sage/local/lib/python2.6/site-packages/sage/symbolic/relation.pyc in solve(f, *args, **kwds)
    670                 s = []
    671 
--> 672         sol_list = string_to_list_of_solutions(repr(s))
    673         if 'solution_dict' in kwds and kwds!['solution_dict']==True:
    674             if isinstance(sol_list![0], list):
]/Applications/sage/local/lib/python2.6/site-packages/sage/symbolic/relation.pyc in string_to_list_of_solutions(s)
    455     from sage.structure.sequence import Sequence
    456     from sage.calculus.calculus import symbolic_expression_from_maxima_string
--> 457     v = symbolic_expression_from_maxima_string(s, equals_sub=True)
    458     return Sequence(v, universe=Objects(), cr_str=True)
    459 
/Applications/sage/local/lib/python2.6/site-packages/sage/calculus/calculus.py in symbolic_expression_from_maxima_string(x, equals_sub, maxima)
   1527         return symbolic_expression_from_string(s, syms, accept_sequence=True)
   1528     except !!!!SyntaxError:
-> 1529         raise !TypeError, "unable to make sense of Maxima expression '%s' in Sage"%s
   1530     finally:
   1531         is_simplified = False
!TypeError: unable to make sense of Maxima expression '[[x==3/2,-7/2!==0]]' in Sage

--

The !== which is causing trouble is due to the

if equals_sub:
            s = s.replace('=','==')

lines in symbolic_expression_from_maxima_string. This could be fixed by changing the replace to a regexp, or adding a hack s = s.replace('!==', '!=') afterwards.

Component: symbolics

Keywords: sd40.5

Author: Douglas McNeil

Reviewer: Karl-Dieter Crisman

Merged: sage-5.1.beta5

Issue created by migration from https://trac.sagemath.org/ticket/8969

[sagetrac-dsm]

Description changed:

--- 
+++ 
@@ -38,7 +38,7 @@
 
 /Applications/sage/local/lib/python2.6/site-packages/sage/calculus/calculus.py in symbolic_expression_from_maxima_string(x, equals_sub, maxima)
    1527         return symbolic_expression_from_string(s, syms, accept_sequence=True)
-   1528     except !!SyntaxError:
+   1528     except !!!SyntaxError:
 -> 1529         raise TypeError, "unable to make sense of Maxima expression '%s' in Sage"%s
    1530     finally:
    1531         is_simplified = False
@@ -50,11 +50,11 @@
 The "!==" which is causing trouble is due to the
 
 if equals_sub:
-            s = s.replace('=','==')
+            s = s.replace('=','==')
 
 lines in  symbolic_expression_from_maxima_string.  This could be fixed by changing the replace to a regexp, or adding a hack s = s.replace('!==', '!=') afterwards.
 
-This deals with the obvious problem but not the underlying one, which is that the result is still IMHO underprocessed: 
+This deals with the obvious problem but not the underlying one, which is that the result is still IMHO underprocessed:
 
 MODIFIED_sage: solve([2*x==3, x != 4], x)
 [[x == (3/2), (-5/2) != 0]]
@@ -66,9 +66,9 @@
 
 when I wanted [].
 
-In fact, even in cases not involving "!=", it's possible for maxima output -- %union([x = 3/2,  -5/2 # 0]) --  to be insufficiently processed,IMHO:
+In fact, even in cases not involving "!=", it's possible for maxima output -- %union([x = 3/2,  -5/2 # 0]) --  to be insufficiently processed,IMHO:
 
-sage: solve([2*x==3, (x-4)!^2 > 0], x)
+MODIFIED_sage: solve([2*x==3, (x-4)!^2 > 0], x)
 [[x == (3/2), (-5/2) != 0]]
 
 ISTM the extra information about what condition maxima used isn't worth the inconvenience of having to postprocess the solutions to see if one exists.

[kcrisman]

Description changed:

--- 
+++ 
@@ -1,6 +1,8 @@
 'Sage Version 4.4.1, Release Date: 2010-05-02'
+
 mac 10.4 32 bit running on 10.5.8.
 
+```
 sage: solve([2*x==3, x < 10], x)
 [[x == (3/2)]]
 sage: solve([2*x==3, x > 10], x)
@@ -12,63 +14,60 @@
 
 sage: solve([2*x==3, x < 4, x > 4], x)
 []
-
-
+```
 all work as expected, but:
 
+```
 sage: solve([2*x==3, x != 5], x)
----------------------------------------------------------------------------
-TypeError                                 Traceback (most recent call last)
+-----------------------------------------------------------
+!TypeError                                 Traceback (most recent call last)
 
 /Applications/sage/devel/sage-main/build/sage/<ipython console> in <module>()
-
 /Applications/sage/local/lib/python2.6/site-packages/sage/symbolic/relation.pyc in solve(f, *args, **kwds)
     670                 s = []
     671 
 --> 672         sol_list = string_to_list_of_solutions(repr(s))
     673         if 'solution_dict' in kwds and kwds!['solution_dict']==True:
     674             if isinstance(sol_list![0], list):
-
-/Applications/sage/local/lib/python2.6/site-packages/sage/symbolic/relation.pyc in string_to_list_of_solutions(s)
+]/Applications/sage/local/lib/python2.6/site-packages/sage/symbolic/relation.pyc in string_to_list_of_solutions(s)
     455     from sage.structure.sequence import Sequence
     456     from sage.calculus.calculus import symbolic_expression_from_maxima_string
 --> 457     v = symbolic_expression_from_maxima_string(s, equals_sub=True)
     458     return Sequence(v, universe=Objects(), cr_str=True)
     459 
-
 /Applications/sage/local/lib/python2.6/site-packages/sage/calculus/calculus.py in symbolic_expression_from_maxima_string(x, equals_sub, maxima)
    1527         return symbolic_expression_from_string(s, syms, accept_sequence=True)
-   1528     except !!!SyntaxError:
--> 1529         raise TypeError, "unable to make sense of Maxima expression '%s' in Sage"%s
+   1528     except !!!!SyntaxError:
+-> 1529         raise !TypeError, "unable to make sense of Maxima expression '%s' in Sage"%s
    1530     finally:
    1531         is_simplified = False
-
-TypeError: unable to make sense of Maxima expression '[[x==3/2,-7/2!==0]]' in Sage
-
+!TypeError: unable to make sense of Maxima expression '[[x==3/2,-7/2!==0]]' in Sage
+```
 --
 
-The "!==" which is causing trouble is due to the
+The `!==` which is causing trouble is due to the
 
+```
 if equals_sub:
             s = s.replace('=','==')
-
-lines in  symbolic_expression_from_maxima_string.  This could be fixed by changing the replace to a regexp, or adding a hack s = s.replace('!==', '!=') afterwards.
+```
+lines in  `symbolic_expression_from_maxima_string`.  This could be fixed by changing the replace to a regexp, or adding a hack `s = s.replace('!==', '!=')` afterwards.
 
 This deals with the obvious problem but not the underlying one, which is that the result is still IMHO underprocessed:
 
-MODIFIED_sage: solve([2*x==3, x != 4], x)
-[[x == (3/2), (-5/2) != 0]]
+```
+MODIFIED_sage: solve([2*x==3, x != 4], x)[[BR]][[x == (3/2), (-5/2) != 0]]
 
 when I wanted [[x == (3/2)]], or
 
-MODIFIED_sage: solve([2*x==3, x != 3/2], x)
-[[x == (3/2), 0 != 0]]
+MODIFIED_sage: solve([2*x==3, x != 3/2], x)[[BR]][[x == (3/2), 0 != 0]]
 
 when I wanted [].
-
+```
 In fact, even in cases not involving "!=", it's possible for maxima output -- %union([x = 3/2,  -5/2 # 0]) --  to be insufficiently processed,IMHO:
 
+```
 MODIFIED_sage: solve([2*x==3, (x-4)!^2 > 0], x)
 [[x == (3/2), (-5/2) != 0]]
-
+```
 ISTM the extra information about what condition maxima used isn't worth the inconvenience of having to postprocess the solutions to see if one exists.

[kcrisman]

comment:2

DSM, is this really all one ticket? It's kind of confusing.

[sagetrac-dsm]

Changed keywords from none to sd40.5

[sagetrac-dsm]

Description changed:

--- 
+++ 
@@ -53,21 +53,3 @@
 ```
 lines in  `symbolic_expression_from_maxima_string`.  This could be fixed by changing the replace to a regexp, or adding a hack `s = s.replace('!==', '!=')` afterwards.
 
-This deals with the obvious problem but not the underlying one, which is that the result is still IMHO underprocessed:
-
-```
-MODIFIED_sage: solve([2*x==3, x != 4], x)[[BR]][[x == (3/2), (-5/2) != 0]]
-
-when I wanted [[x == (3/2)]], or
-
-MODIFIED_sage: solve([2*x==3, x != 3/2], x)[[BR]][[x == (3/2), 0 != 0]]
-
-when I wanted [].
-```
-In fact, even in cases not involving "!=", it's possible for maxima output -- %union([x = 3/2,  -5/2 # 0]) --  to be insufficiently processed,IMHO:
-
-```
-MODIFIED_sage: solve([2*x==3, (x-4)!^2 > 0], x)
-[[x == (3/2), (-5/2) != 0]]
-```
-ISTM the extra information about what condition maxima used isn't worth the inconvenience of having to postprocess the solutions to see if one exists.

[sagetrac-dsm]

Author: Douglas McNeil

[sagetrac-dsm]

comment:4

It was all one ticket in my head, but maybe it makes more sense to separate them. Might as well address the low-hanging fruit.

[kcrisman]

comment:5

The problem wasn't really our translation of Maxima's inequality (#, which we finally fixed a while ago) but rather that we then had this little hack already.

But all of your tests already work in Sage 5.0, because of the # replacement. It really has to test the original bug report example, otherwise this is trivial. I suggest

sage: from sage.calculus.calculus import symbolic_expression_from_maxima_string as sefms 
sage: sefms("x != 3") == SR(x != 3) 
True
sage: sefms("x # 3") == SR(x != 3) 
True
sage:  solve([2*x==3, x != 5], x)
[[x == (3/2), (-7/2) != 0]]

[sagetrac-dsm]

comment:6

Oy, you're right that I truncated the original bug test (!). But we do need to fix !== to round-trip neq, so I think it's worth doing.

[sagetrac-dsm]

revised version