Re: N[] does not work inside Replace[] ?

• To: mathgroup at smc.vnet.net
• Subject: [mg53196] Re: [mg53185] N[] does not work inside Replace[] ?
• From: Bob Hanlon <hanlonr at cox.net>
• Date: Tue, 28 Dec 2004 23:12:14 -0500 (EST)
• Sender: owner-wri-mathgroup at wolfram.com

```Use RuleDelayed

s=Rule[a,10/3];

s//N

a -> 3.3333333333333335

s /. Rule[x_,y_] :> Rule[x,N[y]]

a -> 3.3333333333333335

b = Range[5]/3

{1/3, 2/3, 1, 4/3, 5/3}

b // N

{0.333333,0.666667,1.,1.33333,1.66667}

b /. x_Rational :> N[x]

{0.3333333333333333, 0.6666666666666666, 1,
1.3333333333333333, 1.6666666666666667}

Bob Hanlon

>
> From: yehuda ben-shimol <benshimo at bgu.ac.il>
To: mathgroup at smc.vnet.net
> Date: 2004/12/28 Tue AM 06:30:32 EST
> To: mathgroup at smc.vnet.net
> Subject: [mg53196] [mg53185] N[] does not work inside Replace[] ?
>
> Hi,
> I just found a small but annoying problem with operating N[] inside a
> replace rule
> s was actually part of an answer I got from Solve[] and I wanted to
> leave the Rule, but change the right hand side to a machine real number.
> A reduced form of the problem is given below. I'm using version 5.0.0
> (ver 5.1 is on the way ... )
>
> s = Rule[a, 10/3]
> and the output is
> a->10/3
>
> case 1:
> now use a replace rule with N[]
> s /. Rule[x_, y_] -> Rule[x, N[y]]
> and the result is still a->10/3
>
> case 2:
> using some other header in place of N[]
> s /. Rule[x_, y_] -> Rule[x, g[y]]
> and the results is as expected
> a->g[10/3]
>
> Trying to convert the second argument to a machine real number in
> different way
>
> case 3:
> s /. Rule[x_, y_] -> Rule[x, 1. y]
> results with
> a -> 3.33333
> as expected.
>
>
> Using Trace[] for the three cases reveals that N[y] is replaced by y
> very early in the evaluation process, so it explains this unexpected
> behavior.
>
> I tried then a simpler example
> b=Range[5]/3
> and then tried the same replacement idea
> b/.x_Rational->N[x]
> and still the results leave the rational numbers unchanged
>
> but with the b/.x_Rational->g[x] or b/.x_Rational->1.0 * x Mathematica
> works as expected.
>
> So, again, is this a bug? If it is so in version 5.0.0, does it still
> exists in version 5.1?
>