Yes, fair enough, I seem to have a small problem with Proper Nouns today. Call it a German moment, if you will.
It's not my field any more, but here's an example from when it was: I wrote some code to perform a particular Monte Carlo simulation. Within the code were a number of functions which performed specific numerical computations. They were deterministic, so I unit tested them as you might expect using known results computed by an alternate technique. I found bugs, I fixed them.
However, the core function of the program, the simulation itself, I did not and could not know the "correct" result of, as it was a numerical simulation using pseudo-random numbers (so repeatable, but possibly repeatably wrong).
I had no other way of generating a specimen set of results. So, how was I supposed to test it? If you can offer a practical solution, I'll be impressed. I'm expecting some more hand-waving and hubris, however.
The fact that you think it is possible to just knock up a "specified set of outputs" for this kind of code shows how unfamiliar you are with this kind of program.
Unfortunately, the purpose of simulation code is to determine those outputs! Yes, you could re-implement the entire algorithm in a different language. Yes, you can sometimes use analytic methods to find approximations which hold in special regions of parameter space. But traditional black or clear box testing, as if this were a business systems problem? Sorry, no dice.
Qualifier: I am a practicing Software Engineer (outside academia) with a background in scientific simulation.
Gratuitous car analogy: You don't need to take my car apart if you have access to the plans.
Unless, say, the thing you actually need from the innards of your car is the private key of the ignition system. Which isn't in the plans.
So no, access to the plans doesn't really help at all, in this specific situation.
Or you could, you know, just use perl 5 and have for my $x in ( 2..5 ) do exactly what you expect it to.
I quite like Python, but I don't really get why it has displaced perl as the scripting language of choice. It seems to make most simple tasks just slightly harder than they need to be.
Unless they're also going to disable booting an Ubuntu install DVD, I don't see how anyone could have a problem with this.
Windows is a commercial product. If you don't want to pay for it, use one of the entirely credible free alternatives.
And herein lies the rub - refuelling partway through the trip takes a few minutes with a petrol-engined vehicle. It takes hours with an electric vehicle.
Yes, overnight charging is clearly never going to work as a convenient mass-market solution.
So: standardise battery sizes and dimensions, and set up a battery exchange system. The batteries would be manufacturer-sealed and clearly display a tamper-proof lifetime indicator, so you'd know that the exchange model you were getting was good. Throw in a minimum wage attendant who actually did the manual work of changing the (relatively heavy battery units, and you're done.
A painfully simple and obvious solution to a non-problem -- why is it so difficult to imagine change?
Re:Second person shooter
on
Second Person
·
· Score: 1
i\partial\psi = H\psi, surely? Everybody uses units where \hbar = 1, and nobody bothers with the hat on the H. Oh, and only engineers put dots on the top of things.
What's amusing about this is that the article in question talks about Baron Cohen's experiences with anti-Semitism -- but the journalist took it at face value that he worked at Goldman Sachs because hey, after all, he is a Jew...
Seems to me like this is a classic example of why US lawsuits are a good thing (tm). They're preventing companies from rolling out products that could run over little kids without allowing the operator to override.
Who said the operator couldn't override it? I think the point the article was making was that in the US, people like to look for someone to blame other than themselves. So, if a three-year-old did get run over because of the driver's negligence when using the system, a US citizen would be likely to blame the car manufacturer for their own failings.
Note that I am not a US citizen, and therefore may display bias in this interpretation:-).
Anybody running RedHat and Fedora are strongly adviced to apply this patch!
But I am running SUSE! Am I adviced in similar fashion? Perhaps I too should applying patch lest SUSE found vulnerability also? Thankyou to www.fedora-redhat.com for adviced me in this helpful manner against remote attackers!
We've all made mistakes like this, I think. Somehow, you just get things backwards in your head once, and then fix it as a `definite truth' which you don't bother to look at again.
Usually, I find these kinds of mistake in my own work when someone else, who hasn't been tainted in the same way, points it out to me. I wonder why this kind of peer review didn't happen here?
The `anti-' name refers to the charge of the particle (electrons are assigned -1, positrons +1). So, when you say 1 + -1 = 0, you're right in the sense that e+ e- -> gamma, which is uncharged.
However, the energy of the particles is always positive (negative energy/mass is not observed), so in that case E_a + E_b = E_a + E_b.
Re:The choice between Beowulf and Big Iron...
on
Linux Clustering
·
· Score: 5, Insightful
The more complex a problem gets, the more likely you need one supercomputer as opposed to a cluster.
I'm not sure it is that simple. For some problems (e.g. Monte Carlo simulations), a more complex problem means more individual nodes are required, with very little inter-node communication. For other kinds of problem (finite element methods, maybe?), you're probably right.
In other words, the physical structure of the solution depends on the kinds of algorithms that you intend to run: there's not just one `correct' answer.
Slashdot Mirror
Yes, fair enough, I seem to have a small problem with Proper Nouns today. Call it a German moment, if you will.
It's not my field any more, but here's an example from when it was: I wrote some code to perform a particular Monte Carlo simulation. Within the code were a number of functions which performed specific numerical computations. They were deterministic, so I unit tested them as you might expect using known results computed by an alternate technique. I found bugs, I fixed them.
However, the core function of the program, the simulation itself, I did not and could not know the "correct" result of, as it was a numerical simulation using pseudo-random numbers (so repeatable, but possibly repeatably wrong).
I had no other way of generating a specimen set of results. So, how was I supposed to test it? If you can offer a practical solution, I'll be impressed. I'm expecting some more hand-waving and hubris, however.
The fact that you think it is possible to just knock up a "specified set of outputs" for this kind of code shows how unfamiliar you are with this kind of program.
Unfortunately, the purpose of simulation code is to determine those outputs! Yes, you could re-implement the entire algorithm in a different language. Yes, you can sometimes use analytic methods to find approximations which hold in special regions of parameter space. But traditional black or clear box testing, as if this were a business systems problem? Sorry, no dice.
Qualifier: I am a practicing Software Engineer (outside academia) with a background in scientific simulation.
Gratuitous car analogy: You don't need to take my car apart if you have access to the plans.
Unless, say, the thing you actually need from the innards of your car is the private key of the ignition system. Which isn't in the plans. So no, access to the plans doesn't really help at all, in this specific situation.
Or you could, you know, just use perl 5 and have for my $x in ( 2..5 ) do exactly what you expect it to. I quite like Python, but I don't really get why it has displaced perl as the scripting language of choice. It seems to make most simple tasks just slightly harder than they need to be.
Unless they're also going to disable booting an Ubuntu install DVD, I don't see how anyone could have a problem with this. Windows is a commercial product. If you don't want to pay for it, use one of the entirely credible free alternatives.
And herein lies the rub - refuelling partway through the trip takes a few minutes with a petrol-engined vehicle. It takes hours with an electric vehicle.
Yes, overnight charging is clearly never going to work as a convenient mass-market solution. So: standardise battery sizes and dimensions, and set up a battery exchange system. The batteries would be manufacturer-sealed and clearly display a tamper-proof lifetime indicator, so you'd know that the exchange model you were getting was good. Throw in a minimum wage attendant who actually did the manual work of changing the (relatively heavy battery units, and you're done. A painfully simple and obvious solution to a non-problem -- why is it so difficult to imagine change?
i\partial\psi = H\psi, surely? Everybody uses units where \hbar = 1, and nobody bothers with the hat on the H. Oh, and only engineers put dots on the top of things.
I take it all your encryption algorithms generate cipertext that begins "JFIF", then?
What's amusing about this is that the article in question talks about Baron Cohen's experiences with anti-Semitism -- but the journalist took it at face value that he worked at Goldman Sachs because hey, after all, he is a Jew...
So IBM don't care about Linux, for instance?
This would be what you're looking for, I think.
Who said the operator couldn't override it? I think the point the article was making was that in the US, people like to look for someone to blame other than themselves. So, if a three-year-old did get run over because of the driver's negligence when using the system, a US citizen would be likely to blame the car manufacturer for their own failings.
Note that I am not a US citizen, and therefore may display bias in this interpretation :-).
But I am running SUSE! Am I adviced in similar fashion? Perhaps I too should applying patch lest SUSE found vulnerability also? Thankyou to www.fedora-redhat.com for adviced me in this helpful manner against remote attackers!
You're complaining, on Slashdot, about the only UK train operator that provides laptop power points at every seat?
(And try the cookies -- yum!)
We've all made mistakes like this, I think. Somehow, you just get things backwards in your head once, and then fix it as a `definite truth' which you don't bother to look at again.
Usually, I find these kinds of mistake in my own work when someone else, who hasn't been tainted in the same way, points it out to me. I wonder why this kind of peer review didn't happen here?
The `anti-' name refers to the charge of the particle (electrons are assigned -1, positrons +1). So, when you say 1 + -1 = 0, you're right in the sense that e+ e- -> gamma, which is uncharged. However, the energy of the particles is always positive (negative energy/mass is not observed), so in that case E_a + E_b = E_a + E_b.
The more complex a problem gets, the more likely you need one supercomputer as opposed to a cluster.
I'm not sure it is that simple. For some problems (e.g. Monte Carlo simulations), a more complex problem means more individual nodes are required, with very little inter-node communication. For other kinds of problem (finite element methods, maybe?), you're probably right.
In other words, the physical structure of the solution depends on the kinds of algorithms that you intend to run: there's not just one `correct' answer.
the book's discussion of nmap starts on page 324
...and then later...
Chapter 3 [...] shows how to use [...] nmap for network enumeration.
Are chapters one and two gargantuan, or is it just too late at night for me?