Computing desk
< July 15	<< Jun \| July \| Aug >>	July 17 >

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.

July 16[edit]

Run VST Effects on Raspberry Pi 2?[edit]

Hello everyone. I just got my self a Raspberry Pi 2 and I would like to know how to run VST 2(the .DLL VST files) effects on the Pi. I am running Raspbian. Would anyone know how to do this? I plan on integrating the Pi into my electric guitar to do advanced DSP via VST effects straight in my guitar. Yes I know that I will have to use an Analog to digital converter to convert the analog signal to digital PCM data. Thanks for your help in advance. —SGA314 I am not available on weekends (talk) 15:35, 16 July 2015 (UTC)[reply]

If anyone has done this with the first gen pi, please do explain how you did it. I would think it would work for the second gen pi as well. —SGA314 I am not available on weekends (talk) 19:30, 16 July 2015 (UTC)[reply]

In search of software for reading letter by letter[edit]

Is there a software that can read letter by letter of a word/pdf document rather than simply reading whole words? Mak2202 (talk) 17:26, 16 July 2015 (UTC)[reply]

JAWS allows you to read letter by letter. It automatically goes into letter-by-letter mode when it hits a word that it doesn't know (usually an acronym), but you can force it. Mine is configured so 5 on the number pad goes into letter-by-letter mode and then pressing insert-5 goes back to word mode. Pressing insert-5-5 makes it spell one word, but then keep reading words after that. 209.149.113.45 (talk) 18:02, 16 July 2015 (UTC)[reply]

Instead of searching for software that reads letter by letter, you should consider adapting the text to trigger the letter by letter mode. If you process the text you want to be read from, for example, "Encyclopedic content must be ..." to "e.n.c.y.c.l. ..." and feed it into any software, it will be read letter by letter.--YX-1000A (talk) 15:06, 17 July 2015 (UTC)[reply]

Calculating the Time Complexity of Quantum Gates[edit]

How do we calculate the complexity of a step in a quantum algorithm that apply unitary operator on the qubits? For instance, we say that Grover's algorithm runs is sqrt(N) time, but it only applies one unitary operator, since all the loop is only a composition of unitary matrices, and thus - could be replaced with one unitary operator. So, why don't we consider Grover's algorithm to run in constant time, as for an algorithm that doesn't use a loop? In the general case, how do we measure the time takes to algorithm to apply unitary operator on its qubits? 80.246.136.199 (talk) 19:15, 16 July 2015 (UTC)[reply]

Gates can be treated as taking constant time because their behavior doesn't scale with the problem size. It's fine to include a constant-time Grover-solving gate with k inputs in your gate set, but it will only work for input sizes up to k. After that you'll need more of them, or more other gates, and in the N→∞ limit you'll need O(√N) of them. -- BenRG (talk) 21:39, 16 July 2015 (UTC)[reply]

(It may be worth adding that people are not always good at following this rule. For example in classical complexity analyses it's common to treat integer addition as a constant-time operation even when the integers being added grow indefinitely with the problem size. Really they should build a circuit out of fixed-size adders, which might add a factor of log N to the total time if the additions are a bottleneck.) -- BenRG (talk) 21:45, 16 July 2015 (UTC)[reply]

First, just to make sure I understood you correctly: any gate on constant number of qubits takes constant time, and every gate on arbitrary N qubits takes like the number of gates on constant number of qubits that their composition is this gate.

If I understood correctly, I have two more questions:

First, composing operator on constant number of qubits always results in an operator on contant number of qubits, so we can't define the time complexity of a N-qubits gate using the number of gates that their composition is this gate...? For instance, in Grover's algorithm in each iteration we use (or compose) N-qubits Hadamard gate, rather than a constant-qubits Hadamard gate, and thus, the result is a N-qubits gate.
Second, according to this, applying Hadamard gate (on arbitrary N) takes more than constant time. So since in each iteration in Grover's algorithm we apply Hadamard gate (for sqrt(N) iterations), so its total time compltexity should be more than sqrt(N)?

Thank you very much! :) 80.246.137.208 (talk) 07:30, 17 July 2015 (UTC)[reply]

??? 185.32.178.38 (talk) 06:00, 18 July 2015 (UTC)[reply]

Well, a "gate on arbitrary N qubits" isn't really a gate. It's a family of gates parametrized by N. Any particular gate or circuit takes O(1) time. By analogy, counting up to 100 takes O(1) (=O(100)) time, counting to a googol takes O(1) time, counting to any n less than a googolplex takes O(1) time, but counting to any arbitrarily large n takes O(n) time. (Or maybe O(n log n) time, depending on how you do it.)

Grover's algorithm is a bad/weird example because instead of providing the input in the usual way (encoded in the input qubits), it's provided as a gate, and the gate is treated as taking O(1) time to evaluate regardless of the function. As you say, the article also treats the operator U_s as taking constant time independent of N, which makes even less sense. I think the article is just wrong. The interesting thing about Grover's algorithm is that it takes O(√N) evaluations of the black-box function. This is interesting because (a) you obviously can't do better than N−1 ∊ O(N) evaluations classically, and (b) the BBBV paper (the article's first reference) proved that you can't do better than O(√N) evaluations quantum mechanically, so Grover's algorithm is optimal. The article is wrong to suggest that it takes O(√N) time (i.e., fundamental steps, i.e., gates). -- BenRG (talk) 06:05, 18 July 2015 (UTC)[reply]

I looked now in [Grover's article], and he himself said that it takes O(√N) steps... 185.32.178.38 (talk) 06:59, 18 July 2015 (UTC)[reply]

He seems to mean iterations when he says steps. It's a pretty poorly written article, honestly. -- BenRG (talk) 07:07, 18 July 2015 (UTC)[reply]

To conclude, what's the time complexity of Grover's algorithm? 80.246.136.20 (talk) 07:32, 18 July 2015 (UTC)[reply]

Each iteration requires one evaluation of the function plus, according to this answer, log N additional gates. Since (as that answer points out) it takes log N steps just to read the input, the additional gates can be neglected in any interesting case, so the time complexity is √N times the complexity of evaluating the function. -- BenRG (talk) 08:44, 18 July 2015 (UTC)[reply]

So, in the general case, any gate on log(N) qubits takes log(N) time? if not - why (and when) not? 18:18, 18 July 2015 (UTC) — Preceding unsigned comment added by 185.32.178.108 (talk)

Any circuit that affects Ω(k) qubits has to consist of at least Ω(k) gates (k = log N in this case), but it could consist of more. My only source for there being just Ω(log N) extra gates in this case was the stackexchange.com post. It could be wrong.

The other complication that I've been avoiding is that gate count and execution time are different, since you might (depending on the quantum computing hardware) be able to operate on different qubits in parallel. In that case you could affect all qubits in constant time. But U_s, since it actually mixes all qubits, would probably still take Ω(log N) time, and U_ω would probably take at least as long for similar reasons. But I'm not really sure. There are weird quantum hardware setups like measurement-based quantum computing where the time complexity might be different for all I know. -- BenRG (talk) 05:16, 19 July 2015 (UTC)[reply]

After reading again what you've written I now fully understand it. Thank you very much for your help! 5.29.9.245 (talk) 08:28, 21 July 2015 (UTC)[reply]

Respectsale malware[edit]

I have malware called respectsale that displays ads on Google Search and ad links on some buzzwords in a web page. The problem is, I've tried all steps to remove it with no avail. Panda Antivirus, Malwarebytes, SuperAntiSpyware, Adwcleaner, all will not find it. It will not show as an installed program if I go to control panel and click "uninstall a program". The only thing I have found is in Chrome by bringing up the task manager (shift+esc) it shows as an extension that is running, but it will not show if I actually go to the extensions page. How do I remove it? KonveyorBelt 21:49, 16 July 2015 (UTC)[reply]

Google, when asked [1], finds

and many other similar pages. Have you tried any of them? --CiaPan (talk) 09:11, 17 July 2015 (UTC)[reply]

Those all tell me to remove it from the Windows Control Panel list of installed programs but it doesn't show up there, nor in Chrome if I go to the list of extensions. KonveyorBelt 22:53, 18 July 2015 (UTC)[reply]

Update: Malwarebytes found a different malware in the same location /Appdata/Roaming/Google/Chrome/Default and I cleared the entire folder and it all went away. KonveyorBelt 00:44, 19 July 2015 (UTC)[reply]