Matlab versus VBA
Mike Soloman
all their calculations. When I came on board I brought introduced
them to Matlab and started producing Matlab scripts to do work.
Someone at work asked me "What can Matlab do that VBA cannot?" I
don't know VBA so I couldn't answer. If anyone here can explain when
Matlab would be a better choice from Excel/VBA please let me know.
Thanks.
Mike Soloman
Warpstar Technologies
Duane Hanselman
Some possible answers:
1) Cleve Moler and Jack Little are not on the Forbes top 100 list of
the richest people, but Bill Gates is. Therefore Excel and VBA are a
better choice.
2) When people post requests for "facial identification,"
"fingerprint identification," "Neural network solution to anything,"
"jpeg 2000 code," or other homework and thesis problems, etc., to
newgroups for Excel/VBA people always promptly supply the needed
code, even if they have to take time out of their busy schedules to
write the code from scratch. This doesn't happen here on cssm because
MATLAB users are more snobbish that Excel users. Therefore Excel and
VBA are a better choice.
3) Excel and VBA work on Windoze machines only. Therefore Microsoft
didn't waste time on resources to support alternative operating
systems like Linux. As a result, Excel and VBA are less buggy and
more productive because all resources were devoted to one operating
system as opposed to being diluted by supporting multiple operating
systems. MATLAB did not follow this superior approach. Therefore
Excel and VBA are a better choice.
4) Excel and VBA are cheaper than MATLAB. Therefore Excel and VBA are
a better choice.
5) Even office administrative assistant(s) have Excel on their
computers. As a result, they can make meaningful contributions to the
design and analysis code written by engineers. The folks in accounts
payable and accounts receivable can too for the same reason. This
increases company productivity. Administrative assistants and the
folks in the business office will never learn MATLAB. So MATLAB is
less productive. Therefore Excel and VBA are a better choice.
6) Charts and spreadsheet data from Excel are easier to incorporate
into Powerpoint slides. Therefore those who use Excel and VBA are
more likely to impress the boss and therefore are more likely to get
promoted. You won't get a star on your forehead for using MATLAB.
Therefore Excel and VBA are a better choice.
Can anyone else help out here?
Scott Seidman
@webcrossing.raydaftYaTP:
That would really depend upon the sort of calculations you're doing,
wouldn't it? If you need to multiply a column by a constant, fine, go with
VBA/Excel. If you need to reliably numerically solve a stiff differential
equation, you might think otherwise.
FWIW, VBA is a full programming language, as is Matlab. There are thus
real theorems that apply that essentially state that there is nothing
Matlab can do that VBA cannot, and vice versa. There are things that are
easier or faster to implement in Matlab, and things that are easier/faster
to do in VBA/Excel. (Note that when I say "faster" I'm refering to
development time, not execution time, which is arguably less important for
the type of decision you're talking about).
--
Scott
Reverse name to reply
billc
>My dog Matt doesn't excel at anything. He's a Lab.
I could probability compile all the fetches he's made to the
SIMUlated GUI sausage LINKs over the years. Every datefun he loves
the trapz games we play. Just ASCII him (he's NOT STRJUST ANY ALPHA
MAX, WHILE he'll TRY to make you QUIVER with the rest of the PACK.
He's especially good at WAVPLAY at the beach WHILE SQRTing a CASE of
STREAMLINEs. No big DEAL WHICH you can understand.
His microsoft version is too dainty to get this job done.
How much have I LEFT out ?
Joseph
Seriously though, what do you plan on using MatLab or Excel for? Any
matrix manipulation will be easier and faster in Matlab. Not just
because it has a really cool name that includes part of matrix in it
but rather because it includes code optimized for handling matrices
(i.e. Lapac)
Plus, there are a heck of a lot of built in functions...so you don't
always have to reinvent the wheel (more than likely a less-than
efficient wheel).
engr.student
Mike,
There is no perfect tool. There are better tools for use on a given
job, and there are worse tools. Excel is a step up from the abacus,
and is great for showing 100x100 pieces of data per sheet, and doing
simple math. It can also show simple graphs, and it runs well on
windows. Its math is built to work with at the level of any high-
school student, so arithmetic, algebra, statistics, and some finance.
If engineering was only those, then there would be no engineers
because they would already be replaced by pocket calculators. Serious
engineering is all about, ahem ALL ABOUT, differential equations,
especially partial differential equations. The tools that work with
those are not what your average high school graduate has ever seen
before. They just stare at the squiglies, and panick.
Excel has NONE of the plotting, number crunching, and symbolic
manipulation capability of matlab. You could attempt it raw in vba,
but one line in matlab is a book in vba.
Plotting: The vector field plotting capabilities are really amazing.
This is something that real people in computational fluids use to
display their results. The best excel can do is trivial compared to
the best matlab can do. Excel has maybe a 5 basic plots, and then
variations on the theme. Matlab has dozens of plots, including
subplots, TeX annotation. Matlab can make AVI movies from its plots.
Matlab plots can make fully functional games of tetris.
You can take the symbolic jacobian of a tensor function, then convert
the output to an evaluatable function, and interate it until
convergence (think optimal newton-gauss-seidel nonlinear optimization)
in matlab. How do you do that in excel?
You can read in data from a file (csv), wire it into text strings that
comprise part of a system of symbolic ordinary differential equations,
and solve those equations. You can then take the symbolic solution,
evaluate it over a field, and plot it. You can wire it into
equations, then do symbolic laplace transforms, deconvolutions, and
state-space representations of it.
You can wire together a dozen equations with a dozen unknowns
including all constructable functions into a deterministic system, and
then algebraically solve for all the variables in matlab. You can do
symbolic integration. Show me an excel command that will give you a
symbolic answer for the integral of exp -x^2 dx.
Matlab has beautifully optimized functions like ode45 (think runga-
kutta) that would be a nightmare to make or even begin to interpret in
excel, and even worse in vba. If you feed ode45 a stiff problem and
it crashes, all you have to do is change 3 characters (ode45 ->
ode23s) and you are using a TOTALLY different optimized solver.
Its a powerful tool. Right now maple is talking about having labview
and documentation, but Im not sure about all that. For me matlab is a
tool that I can use as an engineer to do my job.. better.
Oh, and if your secretary is doing the engineering, why is the
engineer getting paid? Fire the engineer and let the secretary do the
work.
-mm
engr.student
> 6) Charts and spreadsheet data from Excel are easier to incorporate
> into Powerpoint slides. Therefore those who use Excel and VBA are
> more likely to impress the boss and therefore are more likely to get
> promoted. You won't get a star on your forehead for using MATLAB.
> Therefore Excel and VBA are a better choice.
>
> Can anyone else help out here?
Duane,
I've been using powerpoint for about a decade, and have had it hang in
some particularly painful professional situations. When Im seriously
presenting, I make a pdf presentation and use adobe. It works on
every system, and has very consistent representations on every
system. The colors are good, the fonts are good. Not so across
powerpoint. You can have the wrong version of powerpoint on the
presentation computer, and be hosed. If it has acrobat, it shows the
show.
Thats been my experience with ppoint.
-mm
K. A. Bachman
Excel is great, but I have found that its solver cannot handle large
problems, e.g., a large optimization problem. Actually, neither can
MATLAB.
Steven Lord
"engr.student" <engr.s...@gmail.com> wrote in message
news:1172272158.2...@h3g2000cwc.googlegroups.com...
engr. student, Joseph, I think your sarcasm detectors may be miscalibrated.
The level of sarcasm they detected in Duane's post was significantly lower
than the level my sarcasm detector did. [Mine's a heavy-duty model -- if it
hasn't broken in the 10 or so years I've been reading newsgroups, it's not
going to break.] You might want to take them into the repair shop and have
them tested :)
--
Steve Lord
sl...@mathworks.com
Duane Hanselman
I didn't consider my posting to be sarcastic. The adjective silly or
stupid is more appropriate. I guess I assumed that folks would
recognize who I am and immediately chuckle out loud. Oh well, such is
life on a newsgroup...
Smile folks, my posting was entirely in jest. It was completely
facetious. I do not like MS products. I use their operating system
only and may move over to Linux soon. I bought my first copy of
PC-MATLAB from (now CEO) Jack Little in 1986 when he was selling it
out of his garage in California after using "Classic MATLAB" starting
in 1983. Back then when I called for tech support, I would get Jack
on the phone!
Duane
Steve Amphlett
I have followed this thread with interest and have one additional
comment to add. The scenario (one common to me):
You have a data file containing loads of channels of measured data
taken from a piece of machinery (e.g. an engine) that isn't quite
behaving right and you want to know why. Can a spreadsheet help?
Can thousands of lines of VBA help? Can a fantastically versatile
interpreted environment help?
I'm obviously biased here, but to me the main strength of ML is that
it gives you the ability to investigate without having to write
(invest in) thousands of lines of code - the entry barrier is really
low.
One particular bit of analysis from (10+) years ago springs to mind
(bear with me, this is long, but representative of real life). We
had a big noisy diesel engine on a test bed and were investigating
the sources of various torsional excitation orders. For some unknown
reason, 4th stood out, which isn't what you'd expect for an I4.
After thinking about it for a while, I wondered if it was really 4th
or some kind of superposition of other sources to give what looked
like a pure 4th order source. The thought took several minutes; the
investigation and subsequent proof took mere seconds in ML. Try that
with VBA and/or excel.
Hum-drum processing and presentation of boring data is ideal for
spreadsheets. Real interatction with your data requires a real
interactive tool.