Home > Archive > APL > October 2007 > APL versu MatLab
You are viewing an archived Text-only version of the thread.
To view this thread in it's original format and/or if you want to reply to
this thread please [click here]
|
|
| Jeff Reid 2007-10-11, 9:58 pm |
| I've used APL in the past (back in the 1970's and 1980's).
In my current line of work (embedded firmware), I don't really
need this type of tool, but there are other co-workers that
use Matlab, for what appears to me to be the same reasons that
APL was used back in the old days.
Does anyone in this group know the pros and cons of Matlab versus APL?
| |
| Doug White 2007-10-11, 9:58 pm |
| Keywords:
In article <RUzPi.3581$W06.3410@newsfe17.phx>, "Jeff Reid" <jeffareid@hotmail.com> wrote:
>I've used APL in the past (back in the 1970's and 1980's).
>In my current line of work (embedded firmware), I don't really
>need this type of tool, but there are other co-workers that
>use Matlab, for what appears to me to be the same reasons that
>APL was used back in the old days.
>
>Does anyone in this group know the pros and cons of Matlab versus APL?
My understanding is that Matlab was originally created by some APL fans
because there was no APL available for the Mac at the time. I've taken a
Matlab course, and we have it at work. I never use it, because I have
APL. It is very polished, has lots of specialized add-on pakcages, and
is expensive.
Doug White
| |
| Jeff Reid 2007-10-12, 3:57 am |
| >>Does anyone in this group know the pros and cons of Matlab versus APL?
>
> My understanding is that Matlab was originally created by some APL fans
> because there was no APL available for the Mac at the time. I've taken a
> Matlab course, and we have it at work. I never use it, because I have
> APL. It is very polished, has lots of specialized add-on pakcages, and
> is expensive.
The guys I've known that use MatLab are usually involved with the analog
aspects of peripherals, such as servo and channel type stuff. One of the
"pros" is the ability to enter equations, and MatLab will solve them
for you, produce a graph, and/or produce a complex number graph.
I've never used it myself, just watch some of my co-workers using it.
Some of the equation, curve fitting, and similar stuff seems pretty
advanced.
Regarding price, I've browsed for APL environments and the ones I found
were around $1500 USA. Then again, MatLab for non-students is $1900, and
there seem to be thousands in add-on components for MatLab. I don't plan
on buying either anytime soon.
| |
| Jeff Reid 2007-10-12, 3:57 am |
| > Regarding price, I've browsed for APL environments and the ones I found
> were around $1500 USA. Then again, MatLab for non-students is $1900, and
> there seem to be thousands in add-on components for MatLab. I don't plan
> on buying either anytime soon.
Student version of MatLab is $99 (USA), and includes some add-ons, and
the add-on's are $59 (USA). Since I would never be a commercial user of
the software, it would be cheaper to sign up for classes at a local
college.
| |
| Paul Houle 2007-10-12, 3:57 am |
|
"Jeff Reid" <jeffareid@hotmail.com> wrote in message
news:J3DPi.6490$W06.266@newsfe17.phx...
>
> Student version of MatLab is $99 (USA), and includes some add-ons, and
> the add-on's are $59 (USA). Since I would never be a commercial user of
> the software, it would be cheaper to sign up for classes at a local
> college.
>
Try http://apl2c.com
It's free.
| |
|
|
| Dick Bowman 2007-10-12, 3:57 am |
| "Jeff Reid" <jeffareid@hotmail.com> wrote in
news:2UCPi.6328$W06.1640@newsfe17.phx:
>
> Regarding price, I've browsed for APL environments and the ones I
> found were around $1500 USA. [... deleted ...]
For personal non-commercial use Dyalog is around USD100.
| |
| Stephen Taylor 2007-10-12, 3:57 am |
| On Oct 12, 5:39 am, "Jeff Reid" <jeffar...@hotmail.com> wrote:
>
>
> The guys I've known that use MatLab are usually involved with the analog
> aspects of peripherals, such as servo and channel type stuff. One of the
> "pros" is the ability to enter equations, and MatLab will solve them
> for you, produce a graph, and/or produce a complex number graph.
> I've never used it myself, just watch some of my co-workers using it.
> Some of the equation, curve fitting, and similar stuff seems pretty
> advanced.
>
> Regarding price, I've browsed for APL environments and the ones I found
> were around $1500 USA. Then again, MatLab for non-students is $1900, and
> there seem to be thousands in add-on components for MatLab. I don't plan
> on buying either anytime soon.
See http://www.vector.org.uk/?area=dnld...nt/interpreters
for APL interpreters. Some are available free for personal or
educational use,
such as APLX and SAX on Linux.
Dyalog isn't free, but at about USD100 for personal use, offers its
latest
version complete with native class-instance support, scripting tools
and
advanced graphics.
Stephen Taylor
sjt@dyalog.com
| |
| Sam Sirlin 2007-10-12, 6:58 pm |
| Doug White wrote:
> Keywords:
> In article <RUzPi.3581$W06.3410@newsfe17.phx>, "Jeff Reid" <jeffareid@hotmail.com> wrote:
>
> My understanding is that Matlab was originally created by some APL fans
> because there was no APL available for the Mac at the time.
I don't think so. I have the original (free) fortran version, written by
cleve as an interactive interface to eispack,linpack. It's short for
matrix laboratory. Of course it has grown tremendously since then.
> I've taken a
> Matlab course, and we have it at work. I never use it, because I have
> APL. It is very polished, has lots of specialized add-on pakcages, and
> is expensive.
It is expensive, but not for students, hence is now essential for
engineering work, especially in research, controls. The newsgroup is
enormous.
Matlab has I think taken over the scientific/engineering interpreter
market. Key reasons are
- cheap for students
- trivial to code/understand (though many write loops where non-looping
solutions are more efficient, much like apl)
- has a nice built-in plot facility
- uses ascii (but allows latex commands on graphs)
- has solid numerical analysis (e.g. going back to linpack/eispack, but
now lapack)
- trivially extensible -- if you type
foo(x)
it looks for a file somewhere in your matlab path, either a compiled
function (foo.mex) or ascii text function/script (foo.m). This
means it is trivial to extend matlab with your own functions, and in
fact also to share with your friends/coworkers. There are extensive
libraries of all kinds of stuff on the mathworks web site.
The underlying language used to be rather hokey (only matrix objects).
But now has "boxed" objects (cell arrays), structures (analogous to c),
overloading, anonymous fns, and all kinds of data types. The syntax is
still a bit ugly, but very practical. It also has sets of toolboxes,
including simulink, a graphical tool for building system time-domain
simulations.
I think APL could have taken this spot (I used it for this in college),
but went towards finance rather than science, didn't have builtin
plotting, isn't as easily extensible (have to load libraries) and no
easy way to share functions (largely due to charset troubles).
Sam Sirlin
| |
| Ted Edwards 2007-10-12, 6:58 pm |
| Sam Sirlin wrote:
> I think APL could have taken this spot (I used it for this in college),
> but went towards finance rather than science, didn't have builtin
> plotting, isn't as easily extensible (have to load libraries) and no
> easy way to share functions (largely due to charset troubles).
I save my library functions as "groups" i.e. *.ATF files. I also have a
function called "need" which can be called recursively from within a
function to load one or more functions from one or more groups. e.g.
'Days' 'DR2J Sodano GPX_tracks' 'Pyth Rnd'need'date' 'gps' 'math' 'ed'
This has at least three important features:
Only one copy of a function is needed in the whole system.
Thus easier for maintenance.
Callable from within a function.
*.ATFs are smaller than the same stuff in WSs.
An ATF is a text file although it does contain APL characters. However
I've never had a problem e-mailing them .ZIP'd.
Ted
| |
| Curtis A. Jones 2007-10-13, 6:57 pm |
| On Oct 12, 12:02 pm, Sam Sirlin <swsir...@charter.net> wrote:
>
....
> Matlab has I think taken over the scientific/engineering interpreter
> market. Key reasons are
>
> - cheap for students
> - trivial to code/understand (though many write loops where non-looping
> solutions are more efficient, much like apl)
> - has a nice built-in plot facility
> - uses ascii (but allows latex commands on graphs)
> - has solid numerical analysis (e.g. going back to linpack/eispack, but
> now lapack)
....
Sam's description of MATLAB's success in the college/engineering
environment agrees well with my experience at San Jose State
University's College of Engineering. The language itself seems
inferior to APL but there's a huge collection of math functions and
tools for e.g. signal-processing systems simulations and for
statistics EASILY available. Mathworks has people out giving seminars
and providing strong support to users.
Bill Koko noted in 1998 at Minnowbrook and told me personally in
Scranton in 1999 that his colleagues at Bucknell (?) were using MATLAB
instead of APL because of the rich assortment of math functions
available with MATLAB. Curtis
| |
| Sam Sirlin 2007-10-13, 6:57 pm |
| Curtis A. Jones wrote:
> Bill Koko noted in 1998 at Minnowbrook and told me personally in
> Scranton in 1999 that his colleagues at Bucknell (?) were using MATLAB
> instead of APL because of the rich assortment of math functions
> available with MATLAB. Curtis
BTW there were originally (late 80's) some comercial competitors to
matlab: control-c and matrix-x (xmath) that were similar but not as
successful (not sure if they're still around). I think matlab "won" out
on price (for students) and extensibility.
There are some good free "competitors" that fill the same niche, octave
scilab, and R. Wikipedia has a pretty good list (but doesn't include APL...)
http://en.wikipedia.org/wiki/List_o...alysis_software
| |
| charles manoras 2007-10-14, 4:01 am |
|
"Stephen Taylor <editor@vector.org.uk>"
>
> Dyalog isn't free, but at about USD100 for personal use, offers its
> latest > version complete with native class-instance support, scripting
> tools and advanced graphics.
And still no complex numbers 25-30 yrs after APL2. :-(
| |
| ddooner 2007-10-15, 6:57 pm |
| I am losing my innocence with this post. I am interested in this
comparison between APL and Matlab. I am a mechanical engineer with a
university. I assign “small” computational assignments each semester and
allow students to use any computer language to solve the assignment. I
change the assignment each semester (to avoid copying projects from
previous semesters--a must). The majority of students use Matlab for the
assignments (common alternatives include MathCad, Excel, C, and Fortran).
C is the required language for the engineering students. I use exclusively
APL (specifically, APL2 and is free as part of the IBM academic initiative
program) to also code the assignment for results evaluation. Forget the
details and coding style with engineers--myself included; my dilemma is
promoting APL! It's not realistic to showcase examples using APL when no
faculty member as ever heard of APL. I do not know Matlab but I see its
parallel structure to APL. The added modules can be nice but we all have
a library of support routines with a nominal amount of programming. At a
glance, I still prefer APL and would like to see APL expand within the
engineering community. Specialty code for prototyping is engineering. To
the APL community, please help us in the university to showcase the beauty
of APL.
David Dooner
--
Message posted using http://www.talkaboutprogramming.com.../comp.lang.apl/
More information at http://www.talkaboutprogramming.com/faq.html
| |
| Richard 2007-10-18, 7:57 am |
| In a quick read I didnt see any mention of J.
J has a good interface to LAPACK, complex numbers and
much more of interest for engineering computing.
It is free.
J has a very active support group.
Since Iverson himself said J was "an APL"
then surely it sould be considered?
refer: www.jsoftware.com
| |
|
| On Oct 18, 11:32 am, Richard <hil...@melbpc.org.au> wrote:
> In a quick read I didnt see any mention of J.
> J has a good interface to LAPACK, complex numbers and
> much more of interest for engineering computing.
> It is free.
> J has a very active support group.
> Since Iverson himself said J was "an APL"
> then surely it sould be considered?
> refer:www.jsoftware.com
Welllll.....
Some of us really think it is the best APL.
Some think that it is not APL because it is missing the gr signs.
I started to use J especially because it was not giving me the
problems the gr signs were giving me with other APLs.
There are lot of reasons to use J and those of us who follow the J
forums see it has a growing number of followers too.
| |
| Bob Armstrong 2007-10-31, 9:57 pm |
| On Oct 11, 7:15 pm, "Jeff Reid" <jeffar...@hotmail.com> wrote:
> Does anyone in this group know the pros and cons of Matlab versus APL?
Interesting question . Interesting answers . But I sent the following
to my niece who is being exposed to Matlab ( and VB and C++ ) in her
applied math classes at Boulder :
> [ The ] comparisons but don't get into nitty-gritty of difference between the languages .
>
> My friend Devon McCormick said it does not have operators ( adverbs ) which take functions ( verbs ) as arguments .
>
> A sentence I typed this morning in K.CoSy to count the months between March 2006 and this month is :
>
> - -/ 12 _sv' ( 06 03 ; 07 10 )
> 19
> ...
> What would be the equivalent MatLab ?
Loren Shure from MatLab sat next to me on the "Thinking in Arrays"
panel at APL2007 . Afterwards she indicated that at least Matlab
understood the importance of boolean vectors and didn't have the hang
up about TRUE being different than 1 which Wolfram seems to .
What are the linguistic differences ?
|
|
|
|
|