@CHM The best way to think of it: # (Slot[]) is, as the name implies, a slot (dummy variable, if you will) that is intended to be replaced when the function is applied to something. & (Function[]) just indicates that you are using a pure function.
blegh, speaking of bloatware. The latest upgrade for OS X was 1.3 GB but at least I now have Catalan, Croatian, Greek, Hebrew, Romanian, Slovak, Thai, and Ukrainian language support. If only I spoke any of them.
@JM i suppose 20 years of computer use in english are hard to forget. it's like physics, I can't discuss even basic concepts in physics in greek. when I talk to greek colleagues, half the discussion is in english
@JM on the other hand, in the System Preferences one can set the caps lock key to be instead the ctrl key. furthermore, ctrl-a and ctrl-e do the emacs thing in most GUI text windows. as far as I am concerned, this cancels out all the negatives
@JM (actually, in the terminal at least, you can kill and yank text a la emacs, and it goes to a kill ring separate from the clipboard--someone must have had a laugh at that)
debug is a command in DOS, MS-DOS, OS/2 and Microsoft Windows (only x86 versions, not x64) which runs the program debug.exe (or DEBUG.COM in older versions of DOS). Debug can act as an assembler, disassembler, or hex dump program allowing users to interactively examine memory contents (in assembly language, hexadecimal or ASCII), make changes, and selectively execute COM, EXE and other file types. It also has several subcommands which are used to access specific disk sectors, I/O ports and memory addresses.
Background
Traditionally, all computers and operating systems have included a main...
@Heike well, I only grew out of it when I discovered that I really wanted to do physics, and then I directed the monomania there. probably not so clever move in retrospect
I don't know what system is used to rank the sites (it must be some aggregate), but Mathematica.se is now ahead of both of the other two betas I'm "participating" in. Both, of which are older. If DSP isn't careful, we'll surpass them soon, too, despite them having twice the number of users. We have more at 200+!
My reading of the noncommercial licence terms for the Intel products is, you can use it if you're a student not being paid for your work, but not if you're an actual employee of the university. As such it might be feasible for you to get yourself a noncommercial licence even if it's not for your cluster administrators to buy a proper licence.
Well, it happens in big clusters... I would probably just install some decent checkpointing software and compile an MPI library set up to use it, and not worry about it too much.
Never used wien2k myself. It's pretty annoying with whatever software though if you're running a job that'll take a week or two over 256 CPUs on however many nodes, and one process stops talking for a while because some insensitive clod decided to parse through 500GB of scratch files using a bash script on that node, so the job crashes...
@OleksandrR the minimum price I'm seeing is $169 for the student development suite, but the I'm seeing is $400 for academic pricing of just the Fortran compiler. I can't tell if it comes with mkl.
Never had that happen, at least on the main clusters at my university. Once we liberated this one, the set up became home grown and accompanied by a gentleman's agreement not to use each others allotment. With at most 3 people using it, it works. I get a whole rack to myself as I'm the heaviest user.
@rcollyer well I'm not an administrator for our clusters nor am I in any way an employee of my university so I've tended to use the noncommercial licences: software.intel.com/en-us/articles/…
@rcollyer depends on your situation I suppose. I receive a stipend too, but in my case legally and for tax purposes a stipend isn't considered to be income or compensation so technically I'm still not an employee.
How could I get a sense of the progress made when running large computations ?
Specifically, how can I determine which index is ParallelTable[] currently evaluating?
I have look into the help but ProgressIndicator[] or EvaluationMonitor[] are either opaque or not appropriate.
I can't monitor ParallelTable:
Monitor[ParallelTable[Pause[3]; i, {i, 1, 10}], i]
just displays i until it is finished.
Do you guys know of alternatives?
@rcollyer What I can say for sure is that in the noncommercial download versions, whenever you download Intel C++ or Intel Fortran, MKL is included in that download. (No, I'm not in the US. No clue how it works in America.)
@rcollyer OK then. I guess that rules out using the noncommercial licence. Although I was under the impression that universities (in general) prefer not to consider their students as employees as it reduces the liabilities they have to them? For instance (as regards the US situation) OSHA only has jurisdiction where employees are involved, not students.
@rcollyer it works both ways, IMO. The situation of not being an actual employee with career development targets etc. gives you quite a bit of freedom, which I have used on occasion when I've felt like making comments critical of the competence displayed by higher-ups in the university administration.
@OleksandrR I had a friend at my undergrad institution who was partially an employee (the astronomy lab instructor), partially a student (working on 10 years plus when I knew him), and he used his student status to get a meeting with the President of the college regarding the building of an observatory. He had priced it all out, and now the college has a small observatory.
@rcollyer hm. Personally I don't mind makefiles. They're simple. Complicated build processes have a lot of opportunity to go wrong. (Ever tried building NumPy with Intel C? It's not fun, mainly due to the horrendous build system.)
@acl I remember two things from GR: the difference between an upper and a lower index (sort of), and that I was able to use an otherwise absurd approximation on a hw question. I was proud of the approximation.
@OleksandrR Personally, I hate tracking source files and dependencies between them. In make, you have to do that by hand, so, as you said, beyond a certain point it becomes nightmarish. boost-build will do dependency checking for you, and, to a large extent, you don't have to actually track the files yourself.
@rcollyer yes, I can see the advantages for package maintainers. But when the build system comes to embody a lot of extremely complicated logic and basically guesswork that might or might not have worked for any given person at any given time, then we (as users) have problems when we want to build in a slightly different way. That's the situation with NumPy, anyway.
@OleksandrR can't disagree with that. The primary reason I switched is I kept getting annoyed that I had to update my makefile whenever I added or removed a file, and since the project was in a lot of flux at that point, I hit my "I must automate this" point very quickly. My scripts to automate this process in make were buggy, at best, and completely useless at worst.
"As a biochemist and molecular biophysicist I studied a lot of this stuff and I must say, Schroedinger was my favorite and well, I had to do it." - Joe
@CHM that's indicative of quite some dedication to have all of that tattooed... and of a skilled tattooist not to drop a factor of 2 or change - to + ...
I saw one a while ago in which someone had tattooed the Y combinator on themselves. A bit more aesthetic than those three long and rather specific expressions, if you ask me.
@CHM I love the second comment: "Physicists dream of the day when a Theory of Everything can be reduced to tattoo that will fit on one bicep." Of course, the formula already exists: $\frac{\text{d}U}{\text{d}t}=0$. We just don't know the form of $U$.
I love working in condensed matter. we're one of the few fields in physics that can completely write down the equations of motion for our field, and it is completely useless.
@rcollyer There was the time it turned out that a bit too much catalyst was added to the reaction vat, resulting in a polymerized goop that I had to think deeply on how to dislodge...
I'm plotting this function: primePercent[x_Integer] := Module[{a}, a = Count[PrimeQ[Range[x]], True]; (a/x)*100 ] to know if the percentage of primes in a certain Range takes a steady value
The plot evaluated in under a minute for the first 10K naturals, but it's been 5 minutes and a lot of heat, but still no sign of the 100K plot.
Is there a way I could predict the time needed for evaluation?
Its dependence on set size obviously isn't linear.
@CHM It's really your prerogative, as OP. (Can't pressure you into accepting something you're not comfy with, see.) Just that an answer which is not fundamentally different from an earlier one getting something doesn't sit well with me. But hey, if you're fine with it, and Mr. Wizard is fine with it, I'm not pushing. I'll let you guys be the judge.
Say, 5 people prefer another answer, they "community accept" it, and whichever answer gets the most accepts, after receiving at least 5, for example, gets tagged as "community accepted"
@CHM I think "community accept" is redundant because what the community likes is indicated by the number of upvotes, but only you know for certain whether an answer addressed your question.
I'm trying to export the plot, but my computer is dead slow.
=)
@JM My interest in mathematics used to be very dim. When I was a kid, I found the math they taught us in primary/high school too... mechanical. Then came college, and I started to enjoy it more, probably because the teachers finally knew what the hell they were teaching, and liked it.
@CHM The math taught to kids is quite boring compared to what is actually done in mathematics. Sadly most people grow up thinking math is all about the crap they did as kids...
It's for use from the kernel that's not being controlled via the front end to begin with. I think you're thinking of FrontEndExecute which is for sending commands to the FE in the form of FrontEndTokens.
Without Developer`UseFrontEnd, Options[$FrontEndSession] gives an error from the kernel, though.
@David not off the top of my head. To be honest I rarely do anything of consequence with notebooks or the front end in general so I'm not the best person to ask. Maybe post it as a question?
I don't think asking it this way is good. Some people might "ask" the same question just because they are annoyed with Mathematica's behaviour. Perhaps "how to use it effectively" would be a better way to ask.
I'm not sufficiently annoyed to cast my super vote on this particular question, but if you guys feel strongly about it, you could consider telling the OP why the question does not sound too appealing to you. ;)
I voted for Rojo's answer. I believe it is the best way. If for some reason you cannot be that specific about your arguments you might use the converse:
nr = Except[_?OptionQ];
f[x_, y : nr : 2, z : nr : 3, OptionsPattern[]] := OptionValue["g"][x + y + z]
If for some further reason you need...
: signs are both for default arguments and named patterns
he gets a vote from me too
Funny to see all the different ways people misspell my name :) this time it's "Szabacks"
@yCalleecharan even though they're called "notebooks", the idea is to be able to produce publication-quality output using stylesheets for books, papers, reports, etc., so a more structured document is helpful. I've not used Mathcad, but I understand the intention there is to basically reproduce the totally unstructured, pencil and paper-type environment?
Thanks for the explanations (and for the useful link). I understand that this gives structure to a notebook and this is a very convenient feature. I am browsing through the book Mathematica by example by Abell/Braselton and it does not discuss notebooks as such.
@Oleksandr I have used Mathcad only for a short time but it is more like Word...so you can add sections and so on. The book that I was reading for Mathcad had a chapter dealing with creating templates for the worksheet (notebook for Mathematica) and I could not find a similar discussion (until now) for Mathematica.
@J.M The guide "Notebooks and Documents" is great. Thanks.
@yCalleecharan You can think of notebooks as an extension of the interactive command line. Except you can go back and change input lines or save the whole session. This way of working interactively is becoming increasingly popular. MATLAB now has "cell mode" (which in essence is the same thing) and the latest ipython has a htmlnotebook interface.
Well, for me it's not about mixing text and commands. It's about building up calculations incrementally instead of writing a whole program and running it.
I often see references to "paclets", for example the PacletManager` context, the PacletInfo.m file in some packages and next to downloaded computable data, $UserBadeDirectory/Paclets, etc.
What are "paclets", what is their relevance to end users, and how can one use them with packages?
@Szabolcs Oh, sorry, I completely misread your comment. I thought it said "Yes, I am arrogant" instead of ignorant.
I was puzzled why asking a question about paclets would make you arrogant (unless you had planned to write a 10 page answer to your own question explaining the ins and outs of the paclet manager).
The Elements of Style (1918), also known as Strunk & White, by William Strunk, Jr. and E. B. White, is a prescriptive American English writing style guide comprising eight "elementary rules of usage", ten "elementary principles of composition", "a few matters of form", a list of forty-nine "words and expressions commonly misused", and a list of fifty-seven "words often misspelled".
In 2011, Time magazine listed the writing style-guide as one of the 100 best and most influential books written in English since 1923.
History
Cornell University professor of English William Strunk, Jr., wrot...
@Heike They are probably averaging for a shorter time span. If I count posts/day since february 8 til the last post we had, I get 11.2 posts per day.
@Heike the grid lines on the graph are weekly, starting Sunday midnight. We probably have fewer posts because of the weekend now. I guess it would be reasonable to average for one complete week, so let me check the average since last Sunday ...
@Heike the average is 11.5876 for this week, 14.2574 for the week before and 14.8243 for the week before that