The h Bar

Is your join too big/too little?

Then if I create a different query that just pulls those 11,000 records into a temp table, merge the temp table with all them joins and find the rows where the temp table's ID "is null" I get the right result again...

what do you mean by too big/too little?

confuuusing

with names redacted:

SELECT *
FROM TABLE1
WHERE YEAR(X) = YEAR(Y) AND MONTH(X) = MONTH(Y)

pulls 11,000 rows

As in using an outer join when it should be inner or inner when it should be outer

Simple

DROP TABLE TABLE1

then your problems are gone

SELECT DISTINCT *
FROM TABLE1
LEFT JOIN TABLE2 ON TABLE1.Z = TABLE2.Z
LEFT JOIN TABLE3 ON TABLE2.W = TABLE3.W
WHERE NOT (YEAR(X)=Year(Y) AND MONTH(X)=MONTH(Y))

But don't do that in the case you don't know what it means

12:16 AM

That second query has a lot more than 11,000 rows removed

from the version of that query where the "WHERE" statement is removed

the version of that query with the "WHERE" statement returns like 400k row, the version without returns 2.5 million

how does that work...-_-

meh, it's time for me to go home...

I'll figure it out later...

I dunno. I think the LEFT JOIN is going to duplicate some rows if you don't have a strict 1:1 mapping between the tables. I'm not terribly familiar with how joins work though. I can usually just access the db and poke on it until it's what I want

Sir Cumference

12:41 AM

Howdy all

I should poke it a little more...

Poking is the solution to all problems

I'm pretty sure rows from TABLE1 aren't being duplicated..

Look at those join columns and see if the results are what you expect or multiples of what you expect

Just SELECT Z, W or similar

1:01 AM

I've done a lot of checking counts and i haven't found any discrepancies

Unrelated, but have you ever heard of eli5? I just saw it mentioned and it looks neat eli5.readthedocs.io/en/latest/index.html

Haven't seen it before

Q: What's the difference between INNER JOIN, LEFT JOIN, RIGHT JOIN and FULL JOIN?

Maybe you should replace the left joins with inner joins?

Inner will have length $\text{len}(T_1 \cap T_2 \cap T_3)$ while second will have length $\text{len}((T_1 \cup T_2) - T_1 \cap T_2) + \text{len} ((T_2 \cup T_3) - T_2 \cap T_3)$

I think I have that right?

1253

What's the difference between INNER JOIN, LEFT JOIN, RIGHT JOIN and FULL JOIN in MySQL?

Actually the second one could be the length of the whole thing with a union in the middle instead of sum of two lengths. I don't really get to think of tables as sets usually

1:18 AM

Inner joins won't work

I need all the records in TABLE1 to be returned

The other tables are auxiliary information

Evidently those union minus intersection expressions are what the mathematicians use $T_1 \backslash T_2$ for

Hmmm. I suppose you could inner join on tables 2 and 3, but that should essentially be happening since the first join is a left join

Are the years and months always there?

I suppose the NOT may be picking up nulls too or something?

user 2646

4:14 AM

\o @Secret

5:40 AM

@DanielSank there is an option in your settings to turn off the left menu. That eliminates the wasted space. With the setting enabled the pages look like this:

The hamburger icon at the top left gives you the options that were on the left menu.

Elitzur–Vaidman bomb tester

pnas.org/content/114/19/4920

Ok I finally read and understood this article. Basically the central idea is very similar to the following:

The Elitzur–Vaidman bomb-tester is a quantum mechanics thought experiment that uses interaction-free measurements to verify that a bomb is functional without having to detonate it. It was conceived in 1993 by Avshalom Elitzur and Lev Vaidman. Since their publication, real-world experiments have confirmed that their theoretical method works as predicted.The bomb tester takes advantage of two characteristics of elementary particles, such as photons or electrons: nonlocality and wave-particle duality. By placing the particle in a quantum superposition, the experiment can verify that the bomb works...

So, one way to understand superposition is the quantum particle's wavefunctiom travels through all the paths. If one of the path is blocked, then it changes the interference in the other paths and thus result in a 0 or 1 be detected

Counterfactual communication go one step further by ensuring any photon that tries to travel through the channel to either be discarded and failed to reach its destination, or that the interference of the wavefunction makes it failed to even show up in there in the first place

But since it can still show up in other places, the information is still transferred, now encoded in the phase difference between the paths

Thetefore, if the wavefunction is treated as ontic, but contextual to the experiment setup, then what is being "transmitted" at the channel is literally the destructively interfered wavefunction, which ideally will result in no detection of anything

However I suspect that if I place a bomb in the transmission channel, then there is no way to prevent the bomb exploding within margin of error because the presence of the bomb may be enough to make that region no longer destructively interefered

So this method is still not recommended, but still superior when you are trapped in a room surrounded by bombs and you need to send an SOS out without blowing yourself up

because at least (unless the bombs are positioned in a way to take account even that) you push down certain death to a chance of death

5:57 AM

now i don't know whether it makes sense to talk about the travelling velocity of the wavefunction, because in practice any attempt to measure it you modify the interference as soon you slot something in

6:12 AM

@JohnRennie Where is my settings menu?

I've never seen such a thing.

@DanielSank In your profile:

pffft

That made the left nav bar go away, but the space is still completely wasted.

Thanks for the tip though :-)

@JohnRennie Which browser do you use?

@Abcd Chrome

@JohnRennie But now Google is stealing lots of data.

6:16 AM

@DanielSank what does the page look like now? Can you post an image?

@Abcd Google don't take any data from me that I'm not happy to give them. I don't have a problem with Google's data policies.

@JohnRennie you haven't read this

@JohnRennie it looks like that

@Abcd I've seen articles like that, though not that particular article, and I don't agree with the sentiments they express. Google provide me with a lot of important benefits for free and I'm happy to provide my online data in exchange.

rubbish

@JohnRennie Wow you are a fan of Google! You trust it so much.

6:21 AM

@DanielSank it's just centring a fixed width layout. Can't you unmaximise the window and shrink it down to the required size?

@Abcd I trust Google to some extent

The point is nothing is for free. If Google can't monetise the info they get from me they can't afford to provide me with all the cool stuff I use.

@JohnRennie I'm not sure what you mean, but if I have to work at all to prevent 1/4 of the front page from being wasted, I'd say we have some improvements to make.

Ah, perhaps I see what you mean: the page fits well on half my monitor.

Q: How fast does the collapse travel in a multiple slit experiment?

7:12 AM

1 hour ago, by Secret

now i don't know whether it makes sense to talk about the travelling velocity of the wavefunction, because in practice any attempt to measure it you modify the interference as soon you slot something in

Ok, one possible way to test this is take a simple inferometric setup:

The detectors $D_1,D_2$ are counted by the computer to plot a graph of which detector get clicked vs time

The half silver mirror $S$ is retractable with some frequency $f$

If the wave function is something that can be assigned a meaningful propagation velocity, then given an electron beam in the set up, when the mirror is raised, destructive interference occurs at $D_2$ while constructive interference occurred at $D_1$. If the mirror is lowered, then the reverse happens

The idea is to see how the plot changes as will increase $f$. Any delay between the detector signals switching between $1$ and $2$ after account for the velocity of the electron beam and the path length of the arms, will suggest "something" is travelling before the electron beam does

If the wave function velocity is instantaneous, or not applicable, then the changes will not be a function of $f$

Otherwise, it will be and suggested the ontic nature of the wave function

iopscience.iop.org/article/10.1088/1742-6596/410/1/012153/pdf

0

If we have a kilometer long million slit light source and make an observation on a slit at one of the ends, will the interference pattern fully collapse or partially collapse? is there any delay in the propagation of the collapse or is the collapse instant? Link to multiple slits. http://hyperp...

speed-of-light double-slit-experiment

physicsforums.com/threads/…

7:47 AM

arxiv.org/abs/1311.7127

And so referring back to this article, if the notion of "velocity of the wave function", or more precisely "How fast does the initial conditions of the quantum state updates throughout the experiment setup when we change the setting" make sense, then the wavefuncton will be the first concrete demonstration that something that by almost all notions of interactions, behaves like nothingness, is not only real, but can propagate

user1732

Euclid is said to have reported to Ptolemy at which time he was asked as to whether there was one key point or one set of finite ways of understanding all that was captured in the Elements. Euclid's reply is famous, "Sire, there is no Royal Road to the understanding of geometry."

physicsworld.com/a/…

However, it seems more likely that the wave function is really not a travelling entity, but inherently nonlocal, given how causality can be put into superposition, something that should not be possible if the update of the initial conditions took place at a finite time

phys.org/news/…

physicsworld.com/a/putting-quantum-noise-to-work

Will prepare a PSE on this later, hopefully it is not a duplicate

Dvij Mankad

8:03 AM

Hello, all! :) Would it be true to say that the spectrum of an operator that doesn't commute with the Hamiltonian of the system should be continuous? I have not put much of a rigorous thought into it but it seems that if we measure that operator and thus reduce the state to an eigenstate of the operator then if the operator is not commuting with the Hamiltonian then it would (except in special cases) be a superposition of several energy eigenstates--and thus [...]

[...] it would evolve to a state that is a superposition of more than one eigenstates of the operator (even in an infinitesimal time?). Thus, if the spectrum is discrete then the subsequent measurement would yield a drastically different result for the same operator. But if the spectrum is continuous then in an infinitesimal time, the state would only spread over infinitesimally close eigenstates of the operator and the value spitted by the next measurement would not be drastically different.

By "close" I mean "nearby" when I say "infinitesimally close eigenstates"

@DvijMankad no. The claim is false. The spectrum of a given operator is a function of the operator and nothing else. It does not depend on what the hamiltonian is.

It's always important to keep in mind that QM makes just as much sense on finite-dimensional systems.

If you're making a general claim about QM and one of its consequences is that the system must be infinite-dimensional, like your claim here, then the claim is automatically wrong.

Q: Updated theme with Left Navigation is live for Quantum Computing

8:33 AM

@CaptainBohemian the slide I posted is the first page of what is considered the first susy paper, the whole interaction argument is very interesting

user1732

9:16 AM

\o @Mithrandir24601

Mithrandir24601

9:46 AM

@user1732 o/

user1732

How's the quantum computing site doing?

4

You may have noticed that your site theme changed! As part of implementing the new unified themes across the network, we're gradually rolling out updated site themes for each site. As of today, we have enabled your updated site theme. Since your site design was done relatively recently and it's ...

discussion featured design

looks-very-cool

:-)

Dvij Mankad

10:08 AM

@EmilioPisanty Thanks! Yes, that makes sense. The check that any generic claim should necessarily not predict infinite dimensions is something I would keep in mind :) I wrote down a few steps and it is clear that the probability of obtaining a different operator eigenstate in a subsequent measurement made after an infinitesimal time is infinitesimally low--so my claim was based on a wrong intuition.

Even if the spectrum is discrete, the vector space is still continuous (in the sense that the field over which the vector space is defined is continuous) and thus, an infinitesimal time-evolution would cause the state to change its direction in the vector space only by an infinitesimal amount and this would reflect in the fact that the newly contributing eigenstates would only have an infinitesimally low contribution in this infinitesimal time duration.

"newly contributing eigenstates (of the operator)"*

By the way, it doesn't affect the falsity of my claim but I was thinking that it would be a restriction on the Hamiltonian that it be such that it commutes with all operators with discrete spectrums. I realize that all the operators would have their spectra dependent purely on themselves--in a straightforward mathematical manner. But I was thinking that the Hamiltonian could be restricted to behave as I described. Anyway, since the basic intuition behind all of this was wrong, it's moot.

But in addition, I think that it would be impossible to restrict the Hamiltonian in such a way because I guess there would arise an infinite number of independent restrictions that the Hamiltonian would have to follow. Right? @EmilioPisanty

Anonymous

11:33 AM

@Mithrandir24601 Autocompletion of user names in comment replies is an existing feature on both the Android and iOS Stack Exchange apps.

Why did the site change to an awful new format

1/3 of the page on the left filled with irrelevant stuff making it harder to read the post on the site, another 1/3 on the right with more irrelevant stuff (less irrelevant but still irrelevant), terrible

Anonymous

As for why they did not enable it on the mobile browsers, I do not know.

Anonymous

To be pedantic, however, it's not right to call it autocomplete. It's more like an equivalent of the "reply" button which is there in the chat.

Change is the only constant

Slereah

11:52 AM

Hello

Hi

@bolbteppa go into your profile, then the Edit Profile & Settings tab, then click Preferences on the left then tick the option Hide Left Navigation. That removes the wasted space on the left. You can still get the left menu by clicking on the hamburger icon.

@JohnRennie Ah that's great, thanks

Anonymous

12:09 PM

The left nav should be hidden by default. It's a terrible terrible design ~~idea~~ flaw anyway :P

@JohnRennie why do people say "hamburger icon" and expect that everybody will understand what that refers to?

It looks nothing like a hamburger

It's one of the perpetual mysteries about modern tech that just continues to befuddle me

@DvijMankad yes, that's roughly correct.

A paddy between two buns.

@DvijMankad "infinite number of independent restrictions" isn't necessarily a problem

"I'll gladly pay you Tuesday for a hamburger today."

Particularly because you haven't actually shown that they're independent.

The core problem is that you're introducing any restrictions at all.

12:13 PM

It is very useful to have those related questions on the right

^

Any operator can be chosen as the hamiltonian.

The only restrictions are that it needs to be hermitian, and that its spectrum needs to be bounded from below.

Anonymous

@EmilioPisanty It's more like a "double" hamburger :P

Slereah

@EmilioPisanty Well you don't have to

but it's more physical certainly

Anonymous

I prefer calling it a tribar menu

12:18 PM

—

_

—

another skullpatrol clone?

how many skullpatrols are there?

drone

Anonymous

@Secret Nah, just an username change

a skull nation

;-)

Sir Cumference

Morning

12:30 PM

Hi

5 hours ago, by Secret

O f888 there's a problem here

I don't want the electron beam to split when $S$ is lowered. Current setup will result in a 50:50 chance of D1 and D2 to click when S is lowered, and that probability will prevent us from measuring the response time of the wave function when the experiment setup is changed due to raising and lowering S

Dvij Mankad

12:45 PM

@EmilioPisanty I am probably missing the point but don't we put restrictions on the Hamiltonian all the time? Like it should be Hermitian (well, maybe that doesn't count) or that it should respect certain symmetries and so on? Or is it something completely different? In particular, can you explain a bit why it is a bad thing to impose restrictions on Hamiltonian? I thought that Hamiltonian is simply about laws of evolution and we can put all the restrictions in the world on laws of evolution.

Sir Cumference

@user2236 Hows life

Mithrandir24601

@user1732 it's not doing too badly, thanks :) We're not getting much in the way of questions, but that's normal for this stage. We have some very good questions though, so that does make me happy :)

@Blue ah, I don't use the app :P

@EmilioPisanty I potentially disagree with the bit about it having to be Hermitian (in principle, anyway)