[Date Index] [Thread Index] [Author Index]
Re: Why does Reduce work this way ...?
On 13 Jun 2006, at 14:07, jackgoldberg at comcast.net wrote: > Hi folks, > > This post is related to a previous series of posts about ReplaceAll > in a set of inequalities. > I have "reduced" the problem to this unexpected situation: > > In FullForm[-3<=x<=1] > Out LessEqual[-3,x,1] > > In Reduce[-3<=x<=1] > Out -3<=x<=1 > > In FullForm[%2] > Out Inequality[-3, LessEqual, y, LessEqual, 1] > > In Reduce[ LessEqual[-3,x,1] > Out -3 ≤ x ≤ 1 > > In FullForm[%4] > Out Inequality[-3, LessEqual, y, LessEqual, 1] > > This defies my understanding. What is qoing on with Reduce? > > Jack > Mathematica has a (not very well documented) function Inequality which can be used to express complicated inequalities in a compact form; for example: In:= Inequality[a, Less, b, Less, c] Out= a < b < c In:= Inequality[a, Greater, b, Less, c] Out= a > b && b < c In:= Inequality[a, Greater, b, Less, c, Equal, d] Out= a > b && Inequality[b, Less, c, Equal, d] In:= Inequality[a, Less, b, LessEqual, c] Out= Inequality[a, Less, b, LessEqual, c] As you can see, sometimes these functions reduce to a simpler form, like a < b < c, which has FullForm: Less[a,b,c] and sometimes not. The forms LessEqual[a,b,c] and Inequality[a,LessEqual,b,LessEqual,c] are semantically equivalent (and have the same but syntactically distinct but normally they evaluate to themselves. The relation between them is a little unusual; note than using the Cell:Convert To menu (in other words the FrontEnd) and converting Inequality [a,LessEqual,b,LessEqual,c] to StandardForm or to TraditionalForm will result in Less[a,b,c] but doing this: In:= Inequality[a,Less,b,Less,c]//StandardForm Out//StandardForm= a<b<c In:= FullForm[%] Out//FullForm= Inequality[a,Less,b,Less,c] does not. Note than the Inequality form is much more general than forms with Heads Less, Greater, LessEqual etc, which I think is the reason why Reduce often returns expressions with head Inequality . If you find this a problem you can always use LogicalExpand@Reduce instead of Reduce, which will take longer and give you a more complex answer but which, I think, will not contain any expressions with Head Inequality. Andrzej Kozlowski