Re: Patterns_ to define linear operators?
- To: mathgroup at smc.vnet.net
- Subject: [mg71441] Re: Patterns_ to define linear operators?
- From: Peter Pein <petsie at dordos.net>
- Date: Sat, 18 Nov 2006 04:41:11 -0500 (EST)
- Organization: 1&1 Internet AG
- References: <ejgvpf$8tm$1@smc.vnet.net>
W. Craig Carter schrieb: > Hello, > I'd like to implement a rule that will factor out constants from > linear operators such as FourierTransform; i.e., > FourierTransform[a*f[x], x, k] to a*FourierTransform[f[x],x,k] > > Here is a method that seems sensible at first, but is potentially > dangerous: > > MyRule1 = > FourierTransform[a_ f_[y_], x_, k_] :-> a FourierTransform[f[y],x,k] > (which would fail if a was not free of x) > > An improvement is: > > MyRule2 = > FourierTransform[a_?(FreeQ[#,x]&), f_[y_], x_, k_] :-> a FourierTransform[f[y],x,k] > (which would fail if a were a product, say b x) > > Does anybody have advice on a robust way to implement this rule? > > Thanks, Craig Carter > Hi Craig, I recommend essentially the same as Dimitris does, but use ReplaceRepeated instead! myrule= lintrans_[a_. * b_,x_,w_] :> a*lintrans[b,x,w] /; FreeQ[a,x] && MemberQ[b,x,Infinity]; and compare ft[3*x*g[y]*y*h[x],x,w] /. myrule --> 3 ft[x y g[y] h[x],x,w] with ft[3*x*g[y]*y*h[x],x,w] //. myrule --> 3 y ft[x h[x],x,w] g[y] I guess, the second one is what you expect. Cheers, Peter