Cris Luengo

Gracias!

Pues estoy curioso por ver lo que dicen otros!

Pero "me acosté a las 2 de la noche".

Yo no lo diría así. Para mi serían las 2 de la madrugada.

I grew up in Alicante, yes.

1am would be "una de la noche" or maybe "madrugada". I can imagine it feeling differently if you're still up or have already gone to bed by then. :) "mañana" is usually after sunrise.

Honestly, this might be used differently in some region of Spain, I don't know. But "12 de la mañana" is standard for me. RAE uses the same phrase, so I'm guessing it's commonly used by a majority of speakers. We say "buenos días" until 2pm, then "buenas tardes". 2pm has always been the middle of the day for me. It's also when we eat, which might have to do with that.

@terdon You're right, in English noon means 12, you need to say "Solar noon" to refer to the moment when the sun is highest in the sky. But "meridiem", where "am" and "pm" come from, refers to the meridian transit, "the moment when the Sun contacts the observer's meridian" (i.e. solar noon). The etimology of the word "noon" is also quite interesting (I just learned this now, it used to mean 3 pm!): en.wikipedia.org/wiki/Noon#Etymology

@terdon Noon is when the sun is highest. Spain is in the wrong time zone, so 12 does not match with noon anywhere in its territory. It is por la mañana until 2 pm, then it's por la tarde. The RAE page you link says "las doce de la mañana".

@terdon In Spain, noon is at 2 pm, so 12 is definitely in the morning there.

By the way, I interviewed for postdoc positions in October/November, with a planned defence in early January (this being the Netherlands, I'd have my title paper work right after the defence). I would be hesitant to search for and agree to a job much further in advance than that, considering the uncertainties around the timing for finishing a PhD.

Yeah, I understand, I wouldn't be happy either. The "everything is fine" until all of sudden it isn't any more might be a cultural issue as well. "Asia" is a rather large region, but in some parts of Asia there's a cultural aversion to pointing out potential problems and conflicts. They might be thinking the room is on fire, but they'll just smile and nod at you.

Eletie, imagine you are a professor and you have a grant with money for a postdoc. The grant is time-limited (it is unusual for it not to be). You think you're getting a postdoc to start in September, then it becomes October or November, then it becomes February. Now you know you're not going to get the work you wanted to do with the grant completed. What do you do? You look for someone else that can start sooner. Because your #1 priority is to make best use of that grant that took you so much effort to get.

@Marianne013 In most countries, if there's a problem with your thesis it is communicated before the defence (viva). Problems can still come up during the defence (e.g. it becomes clear the candidate didn't do the work themselves), but this is highly unusual. Also, in the Netherlands, you get your official document signed and stamped right after the defence. All the administrative checks have been done beforehand. But that is quite unique I believe.

@RosemaryBlanchard Could be a fake address too. Did you look for the whois information for the web server? That can also be hidden behind 3rd parties though.

Why would they expose themselves to lawsuits for no financial gain? Yes, they can do anything, and they might not even act rationally (quite a lot of people seem to not act rationally). But these people are swindlers, they won't put in any effort if they don't get money for it.

Without a transfer of copyright, the publisher cannot publish the paper.

If this device allows for perpetual motion and free energy, you need to redesign how it works to make that impossible. I would stop reading right there. Implausible things are fun in science fiction, but breaking basic physics is a turn-off.

The problem with the two-pass algorithm is that it needs to read the data from memory twice. The computations are very cheap, likely what limits the speed is the memory access. If so, it's better to use an algorithm that goes over the data only once. Of course, timing it is the only way to know for sure.

I’m on my phone, I can’t type code right now. Just replace every instance of fft2 in your code with fft(…,axis=1).

I don’t know what you’re saying, I don’t understand where you’re coming from. Replacing fft2(…) with fft(…, axis=1) will always reduce the number of computations. If you find it slower, you’re doing something wrong.

Your answer has a fft2, which does M length N FFTs, and N length M FFTs. Using FFT instead will reduce the computational cost.

I don’t know, I didn’t see your code, just making an assumption. A 1D FFT of each row cannot be a slower than a 2D FFT of the whole array, because the latter starts with the former. The 2D FFT does 1D FFTs along rows, then along columns. We just need to skip that second part, which is wasted computation. If you found the 1D FFT slower, you’re doing something wrong.

np.fft.fft has an axis argument, no need to loop.

I would use a 1D FFT, not a 2D one.

@pacmaninbw done

I'm going to follow the question, see if it gets answered. :D

@pacmaninbw Discussion is always good. I learned stuff.

@Mast Indeed! I have experienced that first hand. I used to have the same problem with student's reports, the grammatical errors would sometimes obscure the real problems.

@pacmaninbw Since OP is only timing the call to np.linalg.solve, speeding up the code would mean using a different linear algebra solver. Sure, there might be an approach, one would have to figure out what type of problem it is, then figure out what the most efficient way to solve that is. MATLAB's `` does all of that for you.

"Can you please tell me why Python is an order of magnitude slower for only even linear solve?" MATLAB's linear solve function is apparently 10x faster than Python's in this case. Why? It just is. This has nothing to do with the code, as far as I can tell.

@Mast I certainly wouldn't volunteer for the diamond. You have my admiration and gratitude.

I don't think it's going to get a useful answer, but maybe I'm wrong... :)

Sorry for dragging this, wasn't my intention. I just voted to close and added a comment for the benefit of people googling to that question.

LOL

@Mast Thanks!

I follow the MATLAB tag here and on SO, and I get tired of seeing the same question. "Why is Python slower than MATLAB?". I see this at least once a week.

Yes, of course. And they likely have the last word. But I can voice my opinion, hopefully convert some to my cause. :)

@pacmaninbw Happy to chat. But I guess it all comes down to opinion in the end...

Does the UI list all the symbols in the language? Or is that just the most used symbols?

I'm sure there'll be a lot of questions. :)

I've got APL installed, I'm ready to start learning. :)

@Adám Thanks!

Indeed! :)

It makes the case for APL saying it's more compact than JS, with the example of the max operation: ⌈/. In MATLAB it's max(list). So I want to learn: how similar are MATLAB and APL?

I've followed a link starred in this chat, an intro book on APL.

In part because I can't make heads or tails from that code I read over at Code Review...

@Adám I am very curious about it.

@Ben: OP talks about people likning COVID vaccines to G5. Anyone doing that is an anti-vaxxer. You can be distrusting of the evidence all you want, but thinking that the vaccine would somehow purposefully damage you, contain a chip to track you, or some other way be related to 5G phone service is conspirational thinking, and requires a desire to discredit vaccines. That makes them an anti-vaxxer.

@Ben “generations of longitudinal follow-up data”? Which currently sold vaccine is that? Modern vaccines are very different from the ones that my parents got as children. None of them have that kind of age. And none of them needed decades of data for approval.