List Archives >  Maple User Group List Archive >  Archive by date >  This Month By Date >  This Month By Topic

[MUG] yp := (t) -> %;

Search email archive for  

[MUG] yp := (t) -> %;
Author: Colin Campbell - IST    Posted: Mon, 05 Aug 2002 15:12:01 -0400
>> From: Colin Campbell - IST "campbell" Could people vote on the following proposal for Maple 9? 10? BACKGROUND: ^^^^^^^^^^^ You define a function: f := (t) -> sin(t^2); # ( ) on t not needed You differentiate it and get a result: diff( f(t), t ); You want to use the result to define a function. You'd like to type: fp := (t) -> %; # Method 1 Mmm ... pretty! but you have to type: fp := unapply(%, t); # Method 2 Mmm ... not pretty! QUESTION: ^^^^^^^^^ Would you support the idea of Maple 9? 10? permitting "Method 1"? NOTE: This would likely mean %, %%, ... could no longer be used in procedures. Some like Greg Fee would say that would be a good thing as using them tends to reduce readability! Please vote "yes" or "no" or suggest a better scheme for avoinding unapply. Colin ----------------------------------------------------------------------------- Colin Campbell U n i v e r s i t y o f W a t e r l o o Academic Computing Waterloo, Ontario, Canada N2L 3G1 Math & Computer 2048 www.ist.uwaterloo.ca/~campbell Electronic Workplace Group Voice: (519) 888-4567 x5327 Information Systems & Technology FAX: (519) 884-4398 "Across the country, computers are usually not used effectively in under- graduate ... courses. Often, they are not used at all. Problem-solving approaches and calculation methods are little influenced by ... computers." ---> http://www.asee.org/jee/papers/00220.PDF -----------------------------------------------------------------------------

[MUG] Re: yp := (t) -> %;
Author: Maple User Group    Posted: Wed, 7 Aug 2002 16:28:15 -0400 (
>> From: Maple User Group "maple_gr" -=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=- Date: Wed, 07 Aug 2002 02:07:47 +0200 From: Helmut Kahovec "helmut.kahovec" To: "maple-list" Subject: yp := (t) -> %; Well, I do not like your idea, sorry! BTW, I think that the following is pretty well legible: > restart; > f:=t->sin(t^2): > fp:=D(f); 2 fp := t -> 2 cos(t ) t Kind regards, Helmut -=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=- Date: Tue, 6 Aug 2002 17:29:55 -0700 (PDT) From: Robert Israel "israel" To: "maple-list" Subject: yp := (t) -> %; My vote would be an emphatic NO. The fundamental issue is this: when you define a function with proc or ->, the definition is not evaluated. Any variables occurring in that definition (including %) are evaluated only when the function is called, not when it is defined. So for example: > c:= 1; f:= x -> c*x; c:= 2; f(x); results in 2*x, not x. It's a clear, simple rule, and making an exception for % would mess it up. Another serious problem is that any existing code which used % inside procedures would be broken. Backwards compatibility is important. Whether % promotes or reduces readability is a side issue. Robert Israel "israel" Department of Mathematics http://www.math.ubc.ca/~israel University of British Columbia Vancouver, BC, Canada V6T 1Z2 On Mon, 5 Aug 2002, Colin Campbell - IST wrote: -=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=- Date: Wed, 07 Aug 2002 08:49:38 To: "maple-list" From: Bruce M Hartley "hartley" Subject: yp := (t) -> %; Colin, I have been using "unapply" to convert expressions to procedures and then to C subroutines. I can see little advantage in changing to a differnt method of "unapply"ing. Bruce Hartley Bruce Hartley Senior Research Associate Department of Exploration Geophysics Curtin University of Technology +61 8 9266 2297 Ph. +61 8 9266 3407 Fax. -=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=- Date: Tue, 6 Aug 2002 19:12:44 -0700 (MST) From: Matthias Kawski "kawski" To: "maple-list" Subject: yp := (t) -> %; I prefer 2 (use "unapply') I see this as part of a larger need to help clarify when two identical symbols (on the screen) represent the same quantity. I have had my fair share of trouble w/ escaped local variables etc. -- and would like to have a consistent and transparent scheme implemented throughout. Exceptions that "locally" facilitate / speed up the work make life harder as now one has to memorize when special tricks/shortcuts are no longer allowed. Method 1 reminds me of situations where I stumbled over that were similar in spirit, e.g. to whether the t in unapply(....,t) is the same as the t in unapply( diff(g(t),t)),...). Matthias ********************************************************** Matthias Kawski http://math.asu.edu/~kawski Dept. of Mathematics and Statistics "kawski" Arizona State University office: (480) 965 3376 Tempe, Arizona 85287-1804 home: (480) 893 0107 ********************************************************** -=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=- From: Stanley J Houghton "S.J.Houghton" Date: Wed, 7 Aug 2002 13:33:40 +0100 To: "maple-list" Subject: yp := (t) -> %; As a user of % in procedures (where it does not obscure readability), I would vote a definite "no" to this. However, my response deserves an explanation .. it is not just that I don't like change! The current %, %%, %%% nullary operators are very precisely defined in Maple (see the help pages). It is a shorthand notation and is there for a specific purpose. It is deliberately restricted to the current execution level (ignoring for loop variables and NULL) to avoid ambiguity. Computational results in calls to sub-procedures are ignored at the higher level as are computational results before entry to the current execution level. Changing the operator functions would introduce incompatibility and would need much more careful thought, design and definition than is given here. Forgive me but I consider the suggestion a bit of a "knee jerk" reaction to the notation. It is certainly not worth changing an existing useful definition simply for the ease of writing t->% instead of unapply(%,t), i.e. for a prettier notation. Nvertheless, I take your point about the readability of 'unapply' notation. Perhaps an acceptable alternative is to provide a new unapply operator (albeit different from '->') acting like unapply to construct a proc from evaluated arguments (at the same level) and provide the notation you want, e.g. (t) ->> %; (eqivalent to 'unapply(%,t);'). If you follow your approach, you have problems with > g(x);f(x);h(%%); i.e. does the %% now refer to the results of computing g(x) or does it refer to the penultimate computed expression in f(x)? and note also that using your approach > t^2; f:=(t)->%; the first t would be global and different from the formal argument t. Hence my preference and I hope the explanation helps. I await others' comments. Regards Stan Houghton | >> From: Colin Campbell - IST "campbell" | | Could people vote on the following proposal for Maple 9? 10? | You define a function: | | f := (t) -> sin(t^2); # ( ) on t not needed | | You differentiate it and get a result: | | diff( f(t), t ); | | You want to use the result to define a function. You'd like to type: | | fp := (t) -> %; # Method 1 Mmm ... pretty! | | but you have to type: | | fp := unapply(%, t); # Method 2 Mmm ... not pretty! | | Would you support the idea of Maple 9? 10? permitting "Method 1"?

[MUG] Re: yp := (t) -> %;
Author: Maple User Group    Posted: Wed, 7 Aug 2002 16:27:47 -0400 (
>> From: Maple User Group "maple_gr" | >> From: Colin Campbell - IST "campbell" | | Could people vote on the following proposal for Maple 9? 10? | You define a function: | | f := (t) -> sin(t^2); # ( ) on t not needed | | You differentiate it and get a result: | | diff( f(t), t ); | | You want to use the result to define a function. You'd like to type: | | fp := (t) -> %; # Method 1 Mmm ... pretty! | | but you have to type: | | fp := unapply(%, t); # Method 2 Mmm ... not pretty! | | Would you support the idea of Maple 9? 10? permitting "Method 1"? -=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=- Date: Tue, 06 Aug 2002 11:13:22 -0700 From: Allan Wittkopf "wittkopf" Subject: yp := (t) -> %; To: "maple-list" Even prettier: > f := (t) -> sin(t^2); 2 f := t -> sin(t ) > fp := D(f); 2 fp := t -> 2 cos(t ) t - Allan Wittkopf -=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=- Date: Tue, 6 Aug 2002 14:30:43 -0400 (EDT) From: Fred Chapman "fwchapma" To: "maple-list" Subject: yp := (t) -> %; Yes, I vote in favor of Method 1. In this particular example, you could just do fp := D(f), which is much simpler. In general, however, I think it would be very convenient and desirable for t -> % to mean the same thing as unapply(%, t). ---------------- http://www.scg.uwaterloo.ca/~fwchapma/ ---------------- Frederick W. Chapman, Ph.D. Student UW Office: Math & Computer 5162 Department of Applied Mathematics UW Phone: (519) 888-4567 x6672 University of Waterloo E-Mail: "fwchapman" Waterloo, Ontario, N2L 3G1, Canada Curriculum Vitae: see home page -=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=- From: "Richard Quint" "rquint" To: "maple-list" Subject: yp := (t) -> %; Date: Tue, 6 Aug 2002 11:44:08 -0700 You can already do this > f:=x->sin(x^2); 2 f := x -> sin(x ) > fp:=D(f); 2 fp := x -> 2 cos(x ) x > fp(t); 2 2 cos(t ) t -=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=- From: Joe Riel "joer" To: "maple-list" Date: Tue, 06 Aug 2002 11:40:17 -0700 Subject: yp := (t) -> %; I'm inclined to vote against it, though I haven't given it much thought. Here is an operator that does what you want. > `&->` := proc() unapply(args[-1],args[1..-2]) end: > f := t &-> sin(t^2): > diff(f(t),t); 2 2 cos(t ) t > fp := t &-> %; 2 fp := t -> 2 cos(t ) t Joe Riel -=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=- Date: Tue, 6 Aug 2002 14:41:35 -0500 From: Phil Mendelsohn "mend0070" To: "maple-list" "campbell" Subject: yp := (t) -> %; Yes, as long as it doesn't break any multivariable combinations. I.e., you can still specify the mapped symbols arbitrarily, i.e., fp:= (z, t) -> % should work, even if t doesn't show up in the expression. Why? 'cause if you make a mistake, you don't want the value of % to have changed underneath you and do all the calculations over again. (Why % is slippery in the first place.) Cheers, Phil Mendelsohn -- www.rephil.org / University of Minnesota "To misattribute a quote is unforgivable" -- Anonymous -=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=- Date: Tue, 6 Aug 2002 13:50:36 -0700 (PDT) From: John Kurtzke "kurtzke" To: "maple-list" Subject: yp := (t) -> %; Colin, Maybe I am missing something, but what about D? f := t -> sin(t^2); g := D(f); Maple responds with; g := t -> 2 cos(t^2) t Ask Maple what g(t is and you get: g(t); 2*cos(t^2)*t (except Maple doesn't put in the asterisks). g(sqrt(Pi)); -2 sqrt(Pi) Is this what you want? (I'm using Maple 7, but this works on Maple 6, ...) john -- John F. Kurtzke, C.S.C. Department of Mathematics 278 Buckley Center University of Portland Portland, OR 97203 503-943-7377 "kurtzke"

Previous by date: [MUG] Re: A bug?, Maple User Group
Next by date: [MUG] AW: A solution,  Thomas Richard
Previous thread: [MUG] gradient of a vector field, Rob Scott
Next thread: [MUG] A solution, C W