Computing desk
< January 9	<< Dec \| January \| Feb >>	January 11 >

Welcome to the Wikipedia Computing Reference Desk Archives
The page you are currently viewing is an archive page. While you can leave answers for any questions shown below, please ask new questions on one of the current reference desk pages.

January 10

In theory, how can you be sure that a result is right?

This is more of a theoretical question, not asking for tips to use this or that program. If you want to perform a calculation (one that cannot be calculated through other means, otherwise, why use a computer?), how, when, and after how many efforts can you be sure that a calculation is right?--Scicurious (talk) 01:33, 10 January 2016 (UTC)[reply]

Well, for a complex analysis, like finite element modeling, you can check if it matches real world results, like car crash tests. But, for something that can't be compared with real world results, like what happens when a huge meteor hits the Earth, then you're right, we are basically only guessing what will happen. StuRat (talk) 01:50, 10 January 2016 (UTC)[reply]

See Formal verification and Model checking for the theoretical concepts. Tevildo (talk) 12:25, 10 January 2016 (UTC)[reply]

Good scientists establish confidence, rather than claiming certainty. This is a very important philosophical point that is baked deeply into our scientific method. To place this in the context of the original question: we are never "sure" we are right; but we can achieve very high confidence in our correctness, asymptotically approaching certainty, by repeatedly testing hypotheses as many times as we like. In order to do this, we must formulate our work - in this case, some calculation - so that it has predictive power about some testable, observable claim. Nimur (talk) 15:22, 10 January 2016 (UTC)[reply]

Of course, in principle one can compute everything a computer can compute by hand - it just is entirely infeasible in practice. But in addition to the formal techniques suggested above, one way to gain more confidence is to compare the result of different, preferably independent, implementations addressing the same problem. --Stephan Schulz (talk) 15:55, 10 January 2016 (UTC)[reply]

As Stephan Schulz said, you can often calculate the same thing in two different ways and compare the results. For example when calculating more digits of π than ever before, I think it's standard to do it in two ways (summing two different series) and compare the final digits before announcing you've succeeded.

For problems in NP, the solver can produce a proof that the solution is correct, and it can be much simpler to check the proof than to solve the problem in the first place. Even if the solver is buggy and hit on the solution by accident, if the proof is valid then the solution is correct. -- BenRG (talk) 23:15, 10 January 2016 (UTC)[reply]

But one common problem with comparing two seemingly independent answers is that both may be based on the same hidden, but incorrect, underlying assumption or data. StuRat (talk) 23:44, 10 January 2016 (UTC)[reply]

In science theory the general principle is called Consilience. Vespine (talk) 04:28, 11 January 2016 (UTC)[reply]

For example, the age of the Earth can be found by multiple mechanisms, such as radioactive dating of the oldest rocks, geological means such as the rate of continental drift and the distances covered, and studies of life and the rate of genetic drift. However, there still may be underlying assumptions on which these are all based. For example, that time is constant. If everything used to happen slower, then the Earth and the universe could be older, potentially even infinitely old, and none of these seemingly independent methods would tell us that. StuRat (talk) 04:36, 11 January 2016 (UTC)[reply]

In this particular example, if "everything happens slower", then it's the same as if nothing happens slower. Everything is relative - if you have no yardstick, you cannot measure. --Stephan Schulz (talk) 13:37, 11 January 2016 (UTC)[reply]

That's why I included the case where the universe can be infinitely old (that is, the Big Bang took place infinitely long ago), as then there is a difference in that nothing could occur before then. StuRat (talk) 20:30, 11 January 2016 (UTC)[reply]

I'm pretty sure that some variation on the Halting problem prevents you from doing this in theory and in all cases. Some problems are trivial to check: If I ask you to calculate the square root of 14238970234.2342 using pencil and paper - it's going to take you a while - but finding out if the answer is correct (even with pencil and paper only) is quite a fast process because it takes only a single multiplication. It's hard to come up with problems where checking the answer takes significantly longer than calculating it - because the checker only has to repeat the calculation and add the (usually trivial) task of comparing the results. If I ask you to calculate the first million digits of PI, I'd have to check it by first calculating those digits myself - then comparing a million digits.

Sometimes there are shortcuts that only give you some idea of whether the answer is correct - but not 100% certainty.

Sometimes, knowing details of how the calculation is performed allows you to look specifically at the kinds of things that might happen if it went wrong and check just for those. If I was looking at a computer program to calculate prime numbers - testing it would be very, very costly because I'd have to compute all of those primes myself - but I can probably read the source code very carefully and check that it's OK and then trust it...but that's a difficult prospect in many cases.

SteveBaker (talk) 18:48, 11 January 2016 (UTC)[reply]

I want to download or make a special Text Tone for my iphone6

I searched my iTunes store for some special TEXT CODE NOISE. I don't like the xylophone tone or crazy noises that apple has to offer. I don't want suspects and Right now I'm using "RD2D Happy" and it okay, but I want something else. I am a police officer and I would like one of the Text Tones because the citizens would naturally think it just my handheld radio. We all remember the LAPD shows where two or there were three beeps before they get a call. This is the one I really want.

However, I would also like the transmission TONE CODE we heard from NASA as they transmitted to the astronauts on the Moon. I cannot find anything even close to either.

I’m at a lost and cannot find anything close. on their radio, I’ve searched and I’m about to give up. Does anybody have any ideas where to find these? or even make them? I just want the beeps, not the voice.

Somewhere, there may be the same type sounds that is in an unrelated Text Tone that sounds really similar, but I’ve yet to find it. Maybe someone with the equipment could isolate the sounds (I don’t want NASA and LAPD together, but separately)also, do you know of any developers I could contact to have this made? I/m sorry I don't not know how to sign my name to this post, so lets Creedkey (talk) 15:24, 10 January 2016 (UTC) go with §Jello[reply]

I'm not very sure which tones you mean: but the characteristic moon-mission beeps are probably what NASA calls "Quindar tones."

Quindar tones were used during the Apollo moon missions to trigger ground stations at the start and end of uplink. They were 2525 Hz to start uplink and 2475 Hz to terminate uplink, with one quarter second duration. The only reason you ever hear them is because of a technical error (usually, a slight frequency shift - analog radio circuits have weird problems!) Most of the time, the radio system should have been suppressing those tones.

Nimur (talk) 15:44, 10 January 2016 (UTC)[reply]

It is actually possible to record your own ringtones. First you need to get or make a mp3 of the sound: You can record an audio clip using free software like Audacity then save it as an MP3, then make the mp3 into a ringtone. This is precisely the process I followed to record some of my own ringtones using an analogue synthesizer I own. It's not as hard as it looks if you follow the tutorials. 1) Record the sound, 2) save it as an MP3, 3) import it into itunes, 4) make it into a ring tone. Vespine (talk) 04:24, 11 January 2016 (UTC)[reply]

browser language setting

On a French website, I see at the top of the frame: "It seems you don't speak French... [this site] is also available in English". I guess there's a browser setting. How do I change it? (Because I do speak French!) I use Firefox, but the question is applicable to other browsers. —Tamfang (talk) 22:31, 10 January 2016 (UTC)[reply]

In Firefox, goto Tools->Options in the main menu (or just Options in the toolbar menu), then select Content, and set the languages you are fluent in there. Other browsers will have similar options.-gadfium 23:00, 10 January 2016 (UTC)[reply]

Thanks. (On MacOS, rather than Tools→Options it's Firefox→Preferences.) —Tamfang (talk) 09:05, 12 January 2016 (UTC)[reply]