Services & ResourcesWolfram ForumsMathGroup Archive

[Date Index] [Thread Index] [Author Index]

Re: FindInstance returning incorrect results

• To: mathgroup at smc.vnet.net
• Subject: [mg71390] Re: [mg71344] FindInstance returning incorrect results
• From: Adam Strzebonski <adams at wolfram.com>
• Date: Thu, 16 Nov 2006 00:53:54 -0500 (EST)
• References: <200611151144.GAA01342@smc.vnet.net> <C935A578-CC0A-4330-9AC5-9D9D5E6C1077@mimuw.edu.pl>
• Reply-to: adams at wolfram.com

This is a known bug that has been fixed in the development version.
To fix it in Mathematica 5.2 add the following code to your init.m file.

(*** begin fix ***)

Reduce`CPInstance;

System`ReduceDump`drltvars[f_, vars_] :=
    First/@Select[Transpose[
       {vars,
        Internal`DistributedTermsList[f, vars,
           CoefficientDomain -> RationalFunctions, Modulus -> 0,
           MonomialOrder -> DegreeReverseLexicographic,
           Sort -> False][[1, 1, 1]]}],
       #[[2]]=!=0&]

(*** end fix ***)

Now FindInstance finds a correct solution:

In[2]:= FindInstance[c*c== -1 && c*d == I, {c,d}]
Out[2]= {{c -> -I, d -> -1}}

In[3]:= FindInstance[c*c == -1 && c*d == I && c != 0, {c, d}]
Out[3]= {{c -> -I, d -> -1}}


Best Regards,

Adam Strzebonski
Wolfram Research

Andrzej Kozlowski wrote:
> *This message was transferred with a trial version of CommuniGate(tm) Pro*
> You are using FindInstance correctly and it seems you have found a bug. 
> Besides your another thing that suggest that something is wrong is this:
> 
> In[59]:=
> FindInstance[c*c == -1 && c*d == I && c != 0, {c, d}]
> 
>  From In[59]:=
> "TriangularSet in Reduce.m is incorrect"
> 
> Out[59]=
> {{c -> -I, d -> -1}}
> 
> 
> This is with
> 
> In[60]:=
> $Version
> 
> Out[60]=
> 5.1 for Mac OS X (October 25, 2004)
> 
> Andrzej Kozlowski
> 
> 
> 
> On 15 Nov 2006, at 20:44, Jason Grout wrote:
> 
>> I ran into the following problem with FindInstance today.  Is this a
>> known problem?  Am I using FindInstance correctly?  A simple example
>> where the problem comes up is in finding values c and d such that c^2=-1
>> and c*d=I.
>>
>> First, let's make sure a solution exists:
>>
>> Clear[c,d]; Solve[{c*c== -1, c*d== I},{c,d}]
>>
>> gives
>>
>> {{c -> -I, d -> -1}, {c -> I, d -> 1}}, which is correct.
>>
>> FindInstance[c*c== -1 && c*d == I, {c,d}] gives the _incorrect_ answer
>> {{c -> 0, d -> 0}}.
>>
>> FindInstance[c*c == -1 && c*d == I, {c,d}, Complexes]
>>   also gives {{ c -> 0, d -> 0}}, which again is an incorrect answer.
>>
>> However, if I ask for more than one solution, FindInstance gives correct
>> answers.  For example:
>>
>> FindInstance[c*c == -1 && c*d == I, {c,d}, 2] gives the answers that
>> Solve gave above.
>>
>> Am I using FindInstance incorrectly?  It seems odd that I get correct
>> answers only when I ask for more than one answer.  The documentation
>> mentions that FindInstance uses a faster method for generating an
>> example if only one example is needed--is that causing a problem here?
>> Can I be sure that if I ask for more than one answer, I'll get correct
>> answers?  On a related note, if FindInstance returns {}, does that mean
>> a solution does not exist, or merely that FindInstance couldn't find a
>> solution?
>>
>> Thanks,
>>
>> Jason Grout
>>
>> -- 
>> Jason Grout
>> jason-mathematica at creativetrax.com
>>
> 

• References:
  • FindInstance returning incorrect results
    • From: Jason Grout <jason-mathematica@creativetrax.com>