The h Bar

enumaris

12:02 AM

weird guy

Danu

@bolbteppa I think the comments by Sam Sinai are really great. Shame Ron doesn't want to hear it.

bolbteppa

Yeah

enumaris

the guy asking him to get cognitive therapy?

Physics Meta

1:01 AM

0

Q: Pedagogical questions about physics

Sparked by the discussion in the comments in the question How to teach modeling physical systems, especially if it is off-topic, I started looking for pedagogical questions about physics. I didn't find much and there is no site comparable to Mathematics Educators Stackexchange, where pedagogical ...

Petroglyph

2:35 AM

Is it okay to link the meta question I posted to my original question on Stack Exchange? I couldn't find a policy on that in the FAQ or while searching for it.

David Z

@Petroglyph It's fine to put a link to your original question in to the meta question. But I'd be wary of putting a link to the meta question in to your original question.

Not that we really have a policy against it, but I can't really think of any good reason that you would need to link to a meta question from a main-site question.

Petroglyph

All right, thank you :-)

John Rennie

3:40 AM

@ACuriousMind I don't. If it does have a meaning then I'd guess it stems from a Gaelic word.

Cows

4:26 AM

RITA ORA - I Will Never Let You Down (Video)

►

ok going to call me guardian force :P sip some tea and then watch HBO. Wrote some epic code today

I (we some superior minded coders) are going to disrupt this mofo!!

Rita Ora - Girls ft. Cardi B, Bebe Rexha & Charli XCX (Official Video)

►

.. . ... . . . .. . .. . . .

BETTA LEMME - BAMBOLA (Official Video)

►

Such code . . .

Anonymous

5:01 AM

@JohnRennie Morning! Do you have Visual Studio handy? I'm trying to reproduce a weird error a program (someone sent me is causing) and I'm not sure if it is a compiler issue or a faulty code (probably the latter, but I'm not getting any errors/warnings)

John Rennie

@Blue Yes, I have VS on this laptop.

What do you want me to test?

Anonymous

@JohnRennie Oh, great. I'm coming in 5 mins :)

Anonymous

I'm back....just uploading the code

Anonymous

Header file: paste.ofcode.org/7ddQ4qDrZzw8JVdcaSkVcK

Anonymous

Main file: paste.ofcode.org/Tjscq58uBkTqyekbq8sLfQ

Anonymous

5:14 AM

Sample Command Line Arguments:

Anonymous

L=100 p=0.3 d0=0.4 d1=0.6 n=1

Anonymous

L=100 p=0.3 d0=0.4 d1=0.6 n=100

Anonymous

L=1000 p=0.3 d0=0.4 d1=0.6 n=1

Anonymous

L=1000 p=0.3 d0=0.4 d1=0.6 n=100

Anonymous

These are the test cases

Anonymous

5:18 AM

The weird thing is that for the same test case, it is sometimes generating the output and sometimes it isn't.

John Rennie

D:\rhs\c>clusters L=100 p=0.3 d0=0.4 d1=0.6 n=1
# Seed: 10542474428453626471
# Iterations: 1
# Rows: 100
# Columns: 100
# P(1): 0.300000000
# P_diag(0): 0.400000000
# P_diag(1): 0.600000000

Anonymous

@JohnRennie Yup, that isn't the full output! However, sometimes for that test case it was generating the full output when I commented out the fflush statements

Anonymous

But when re-run, it didn't

Anonymous

Lemme see if I can generate the full output:

John Rennie

Ah yes, if I run it repeatedly I sometimes get the full output and sometimes don't

Anonymous

5:22 AM

@JohnRennie Exactly!

Anonymous

Lol, this is the weirdest code I've ever seen :P BTW is it showing any "warnings" on yours?

John Rennie

It compiles without warnings, if that's what you mean.

Anonymous

Yep, I meant that only

Anonymous

Same here

Anonymous

Umm...so any ideas what's going on? :/

Anonymous

5:24 AM

I'm trying to debug it myself....but couldn't spot much so far

Anonymous

It's using the POSIX thingy

Anonymous

(but I don't think that's a problem)

Anonymous

The issue probably is memory mismanagement of some kind

John Rennie

When it doesn't print anything it takes longer to run ...

Anonymous

Right

Anonymous

5:27 AM

Maybe it is generating the output but for some reason not displaying it? :/

Anonymous

Actually you can comment out the whole for (arg = 1; arg < argc; arg++) block

Anonymous

Will make it easier to debug...by just adjusting the default

John Rennie

The do_work function is crashing

Anonymous

Aaah...possible

Anonymous

`static inline void do_work(cluster_work *work)
{
size_t i = work->iterations;
while (i-->0) {
do_work_disjoint_set(work);
do_work_count(work);
do_work_update(work);
}
}`

Anonymous

5:32 AM

And this function is calling 3 other functions...hm

John Rennie

printf("burp 1 - %i\n", i);
		do_work_disjoint_set(work);
printf("burp 2 - %i\n", i);
		do_work_count(work);
printf("burp 3 - %i\n", i);
		do_work_update(work);
printf("burp 4 - %i\n", i);

just prints:
burp 1 - 0

So it's crashing in the first call to do_work_disjoint_set(work);

Anonymous

Interesting

Anonymous

Checking do_work_disjoint_set

Anonymous

`static void do_work_disjoint_set(cluster_work *work)
{
const size_t rows = work->rows;
const size_t cols = work->cols;
cluster_label *const matrix = work->matrix;
prng *const rng = &(work->rng);

/* Fill in the matrix */
{
const uint64_t limit = work->p_one;
size_t i = rows * cols;
while (i-->0)
matrix[i] = (2 * i) + prng_probability(rng, limit);
}

/* When we examine each cell, it is enough to merge it
left, up-left, up, and up-right. Essentially, we consider

John Rennie

@Blue Stick loads of printfs in and see where they stop printing anything

Anonymous

5:42 AM

Doing that!

Anonymous

`/* Generate a disjoint set in work->matrix, describing a matrix
filled with ones and zeroes.
*/
static void do_work_disjoint_set(cluster_work *work)
{
printf("Level 1\n");
const size_t rows = work->rows;
const size_t cols = work->cols;
cluster_label *const matrix = work->matrix;
prng *const rng = &(work->rng);

printf("Level 2\n");
/* Fill in the matrix */
{
const uint64_t limit = work->p_one;
size_t i = rows * cols;
while (i-->0)
matrix[i] = (2 * i) + prng_probability(rng, limit);

Anonymous

Output:

Anonymous

# Seed: 15726239593266041131
# Iterations: 1
# Rows: 100
# Columns: 100
# P(1): 0.500000000
# P_diag(0): 0.300000000
# P_diag(1): 0.700000000
Level 1
Level 2
Level 3
Level 4
Level x
Level y
Level x
Level y
Level x
Level y
Level x
Level y
Level x
Level y
Level x
Level y
Level x
Level y
Level x
Level y
Level x
Level y
Level x
Level y
Level x
Level y
Level x
Level y
Level x
Level y
Level x
Level y
Level x
Level y
Level x
Level y
Level x
Level y
Level x
Level y
Level x
Level y
Level x
Level y
Level x

Anonymous

It isn't reaching printf("Level 5\n");

Anonymous

5:59 AM

The last output is Level x

Anonymous

Not Level y

Anonymous

That gives some clue

Anonymous

` /* First element on row r. */
{
const size_t prev = r * cols;
const size_t curr = prev + cols;
const uint64_t limit = limits[LABEL_COLOR(matrix[curr])];
unsigned int merge = 0;

/* Merge up? */
merge |= (SAME_COLOR(matrix[curr], matrix[prev])) << 0;

/* Merge up-right? */
merge |= (SAME_COLOR(matrix[curr], matrix[prev + 1]) && prng_probability(rng, limit)) << 1;

switch (merge) {
case 1: /* Up */
djs_join2(matrix, curr, prev);
break;
case 2: /* Up-right */
djs_join2(matrix, curr, prev + 1);

Anonymous

Lemme put in some more print statements...lol

John Rennie

@Blue when C++ programs crash randomly that's usually a memory corruption problem. A heap or stack corruption. It generally involves rogue pointers or writing past the beginning or end of an array.

John Rennie

6:16 AM

@Blue when I run it in the VC IDE I get:

Anonymous

"read access violation"?

Anonymous

What's that

Anonymous

2

Q: C++ Read Access violation error

I am currently getting an "0xC0000005: Access violation reading location 0xcccccce0." error and I have tried diagnose the problem...I think the problem comes in when my rule of 3 that I have defined comes in play and points me to here. size_type size() const { // return length of sequence ...

c++

John Rennie

The value of the pointer djs is FD98D6B2 (see the error message). It's trying to read from an invalid memory address.

This looks like stack corruption to me

Anonymous

while (1)

Anonymous

6:24 AM

Infinite loop

Anonymous

Hmm

John Rennie

If you're using VS2017 create a new console app and paste in the code from clusters.cpp. Then you can use the VS debugger to try and track what's going on.

@Blue the loop contains a return that exits from it.

It might be worth putting in some trap for excessive looping though

Anonymous

@JohnRennie But that's in the else part, no?

Anonymous

What if that else condition is never satisfied

John Rennie

@Blue yes, so it could be getting stuck

Anyhow, I need to work now. Back ina while.

Anonymous

6:28 AM

No worries. Thanks

Anonymous

I'm seeing what I can do

Anonymous

Bit shift operations always seem error-prone to me

John Rennie

@Blue to be fair >>1 is just dividing by 2

Anonymous

Yeah, but I still didn't understand the code logic though :P Try to decipher what they're doing

Anonymous

6:43 AM

Basically, there's an invalid index from originating I guess

John Rennie

7:05 AM

@Blue Have you tried compiling it on Linux?

Just remove the stdafx include and it should compile.

Anonymous

Oh, I didn't. Trying

Mithrandir24601

7:29 AM

@JohnRennie Because Linux is the solution :)

John Rennie

I know you're only teasing, but for the record if you can run Visual Studio, i.e. if you have a PC it will run on, then you should run Visual Studio.

Anonymous

@JohnRennie It's working fine on Linux! What was the problem then ? Program was not compatible with Windows ? (Still couldn't figure out the reason)

Anonymous

Maybe Windows doesn't support some of the POSIX standards...still weird

John Rennie

7:49 AM

There are probably differences between gcc and the VS compiler.

Exactly what's causing the problem could be tricky to pin down

Anonymous

Well, it's working so all's good for now. I can postpone deciphering the code for a bit later XD I just need to check whether the output files match with the program which I wrote. :P

Anonymous

@JohnRennie Any way to loop from Ubuntu terminal with different command line arguments ? I don't to add another loop in the program. Something like 'for (p = 0.0 to 1.0) {....}'

John Rennie

I'm sure you can use shell command to do that. However I know next to nothing about shell scripting in Linux.

@Blue something like this

Anonymous

@JohnRennie That looks like it. Trying

Emilio Pisanty

8:10 AM

> I don't [want] to add another loop in the program.

^ sounds like you've got some re-factoring to do

Anonymous

@EmilioPisanty Umm...what ?

Emilio Pisanty

i.e. move the bulk of whatever you're calculating to main_calculation_function(parameters) and then bracket that in however many 'admin' for-loops your scan requires

adding a dimension to the scan should be an operation that works within a span of 10-20 lines of code

John Rennie

@EmilioPisanty it's someone else's code that is only needed for a short while.

Under the circumstances a quick hack is perfectly appropriate :-)

Emilio Pisanty

@JohnRennie "sounds like" $\neq$ "you absolutely need to"

;-)

in other news, wasn't there an extremely similar question to this one somewhere in the archives?

-1

Q: How to answer when the question is based on a wrong premise?

I ask this wondering about my answer to this recent question, though this applies in a more general setting. The situation is that the question asks about the physical interpretation of the typical curved-spacetime-as-rubber-sheet analogy, which as has been said multiple times on the parent site,...

discussion answers comments

I'm thinking some two years back or so

Slereah

8:46 AM

Emilio Pisanty

9:02 AM

@Slereah hmmmm. close, but not quite there.

we don't throw it out

we just, uh, "improve it"

in the same way that a kid with a box of crayons might "improve" the looks of that new gizmo you just bought

Slereah

9:20 AM

@EmilioPisanty I don't see what you mean...

2

Anonymous

10:10 AM

1

A: Looping with different command line arguments from Ubuntu terminal

Bash doesn't support floating point arithmetic - so you have to resort to something like bc - as in the example below. The easiest thing to do would be to place this in a Bash script. (You will need bc installed.) for ((i=0; i<=100; i++)) do x=`echo "scale=2; $i/100" | bc` y=`echo "scal...

Anonymous

@JohnRennie Finally done. Yay! :D (Also, learnt something useful today ^)

John Rennie

@Blue Cool :-)

Novalium Company

@Blue @JohnRennie wasuup.

John Rennie

Morning :-)

Novalium Company

I ordered a Kindle E-reader 5 yesterday and I'm so excited, it will arrive tommorow.

What have you been guys up to?

John Rennie

10:36 AM

@NovaliumCompany Kindle's are great for recreational reading. They aren't so good for textbooks because the screen is a bit small. For a textbook you really need to see at least a whole page at a time and preferably two pages - like a real book.

Novalium Company

12:09 PM

Ok @JohnRennie.

I haven't really tried to see if it's hard, so I can't give an opinion.

CooperCape

12:39 PM

ohai

John Rennie

2:47 PM

@NovaliumCompany I used a Kindle for years and I really like them.

Emilio Pisanty

3:41 PM

@JohnRennie courtesy ping - I reopened this one, I really don't think it's a dupe

Sir Cumference

2

Emilio Pisanty

this is correctly dupe-closed tho

enumaris

4:09 PM

gotta get my computer set upppppp

enumaris

4:22 PM

physics.stackexchange.com/questions/413149/… <--- I don't agree with this closure. That question is essentially just asking to compare/contrast Fusion and Fission which is certainly a valid physical question.

Novalium Company

@enumaris Does the water in a water bottle evaporate? (Come into an equilibrium)

enumaris

to some extent yes

Novalium Company

Maybe the air in the bottle is causing some effects?

enumaris

the air in the bottle effects how much water evaporates. If the bottle was rigid and you took out all the air more water would evaporate.

At a given temperature

Novalium Company

Yep makes sense.

enumaris

4:42 PM

mmmhm

Novalium Company

5:09 PM

My Kindle 5 arrives tommorow, I can't stop thinking about it... When it comes, you wouldn't see me for weeks :D

enumaris

I'm actually in a conundrum on what to read atm

Novalium Company

With billions of books existing, I'm sure you'll figure that out.

enumaris

the IT department is having problems setting up CUDA on my computer...

might be delayed further...

disappointing

Novalium Company

:: googleling what CUDA is ::

enumaris

developer.nvidia.com/…

this

Novalium Company

5:13 PM

Some programming extention?

enumaris

allows GPU acceleration for machine learning

Novalium Company

Is it hard to learn machine learning? I've been looking around the internet, but I'll probably read a book (on my new Kindle 5 xD)

enumaris

not hard

much simpler than physics imo :P

Novalium Company

Ok, I'll look around for a simple machine learning book.

enumaris

machine learning is a pretty broad topic tho

Novalium Company

5:18 PM

maybe neural networks then?

enumaris

all the way from linear/logistic regression up through advanced A.I.

neural networks are all the rage right now

Novalium Company

What do you mean?

enumaris

they are very popular

Novalium Company

Yep, I know.

enumaris

they generally fall under supervised learning though...except for the autoencoders

Novalium Company

5:20 PM

Supervised means that you have to teach it, right, with examples?

enumaris

yeah

labeled examples

Novalium Company

Yep, I actually have a little example of a supervised thing in Python, mind looking at it?

enumaris

uh sure

Novalium Company

Ah, I have deleted it. But it bascially uses a library which makes it super easy to enter some values, the system will find some sort of logic in them, and then, if you enter new values, you get the results in probability.

enumaris

kind of a vague description...

Novalium Company

5:25 PM

I know, I told you I'm new to this.

enumaris

but supervised learning is generally "given this input, predict a likely outcome"

Novalium Company

Yep.

enumaris

unsupervised is generally "how can I represent this input in a lower dimensional way?"

And Reinforcement learning is "GIMME ALL THE REWARDS"

rawr

Novalium Company

Unsupervised looks complicated, so I'm not digging there yet.

enumaris

it's not any more complicated than supervised learning imo

Novalium Company

5:26 PM

What do you mean in a lower dimensionl way?

Meh, I gues I'll read about it. (On my brand new Kindle 5 xDD)

enumaris

generally unsupervised learning is to cluster stuff

or reduce dimensionality

Novalium Company

Gimme an example?

enumaris

Say you have 5000 objects, and you describe each object in that set with a 200 dimensional vector: (x,y,z,...,200 variables here)

unsupervised learning generally tries to see if maybe you can split those 5000 objects into N (say 20) distinct groups OR if you can represent each of those objects with a N<200 dimensional vector (x,y,z,...,N variables here) without losing too much information.

Novalium Company

Got it, so basically simplifing without loosing too much info about the objects.

enumaris

yeah

so it can be applied as widely as from data compression to market segmentation

Novalium Company

5:34 PM

Seems cool.

enumaris

it's alright

deep learning is all the rage tho

There's some deep unsupervised learning algorithms I suppose...

Novalium Company

5:50 PM

Suppose I wanted to make a chat bot that I can chat with, you know, like it's a real person. What skills would I need to do that?

enumaris

A.I. god

skills

Novalium Company

Well, cleverbot have done it?

enumaris

No chat bot at the moment is "like a real person"

Even the google assistant is still a ways away

Novalium Company

Of course, but there are some pretty close results.

enumaris

most of that would be in the domain of natural language processing

Novalium Company

5:53 PM

For a real AI to exist, you basically have to create consciousness, which is God's job.

enumaris

well there's some debate on that lol

tho it's mostly philosophical

Novalium Company

It's imperfections at it's finest.

enumaris

i.e. if you create an A.I. which is indistinguishible from humans, at what level would you still say "it doesn't have consciousness"

sorry not i.e., I meant e.g.

Novalium Company

In my opinions, humans are pretty close to cracking the code for AI but we have to be careful.

I'll take a shower now, see you later.

Now imagine my Kindle arrives after 5 minutes :D

bolbteppa

Apparently the photo electric effect is a bound state problem in qft

What's going on with bound states in qft

Why do almost no standard qft books even mention them

enumaris

6:00 PM

scattering is so hot right now

scattering is like Hansel

bound states is like Zoolander

bolbteppa

ja

enumaris

QFT is obsessed with scattering prolly cus that's the only way to test those energy regimes at which QFT is necessary right...

ACuriousMind

@bolbteppa Because they're hard - since we effectively don't know the interacting Hilbert space, you can't investigate them as straightforwardly as you would in ordinary QM

enumaris

hmmm

bolbteppa

I thought all of qft was just scattering :(

enumaris

6:04 PM

pretty much

ACuriousMind

How could it be? E.g. confinement is clearly not a scattering phenomenon, but it is an important part of QCD

enumaris

u come up with some scattering experiment, you draw feynman diagrams, you try to calculate the S-matrix, you cry a little, you calculate the S-matrix, you cry a lot, and then you get your answer.

^important to not neglect steps 4 and 6

ACuriousMind

But rigorously showing confinement is hard - you can make heuristic arguments about Wilson loops and static quark pairs, but I'm not sure we know how confinement really works in full QCD. That's almost necessarily the realm of lattice computations since the coupling is large so perturbation theory doesn't work, and QFT lattice simulations are hard, too

enumaris

maybe the theorists haven't cried enough

neglected steps 4 and 6

bolbteppa

"With regard to phenomena, I recall wondering, as a student, why some of the fundamental things I studied in NRQM seemed to disappear in QFT. One of these was bound state phenomena, such as the hydrogen atom. None of the introductory QFT texts I looked at even mentioned, let alone treated, it. It turns out that QFT can, indeed, handle bound states, but elementary courses typically don’t go there.

Neither will we, as time is precious, and other areas of study will turn out to be more fruitful. Those other areas comprise scattering (including inelastic scattering where particles transmute types), deducing particular experimental results, and vacuum energy. "

Semiclassical

6:08 PM

Part of that is that the usual Monte Carlo method people use runs into a numerical sign problem when doing stuff like QCD

ACuriousMind

@Semiclassical Not the only problem, there's also e.g. fermion doubling

Semiclassical

Oh, sure

bolbteppa

So looking at chapter's 4 and 5 in the amazon toc view here:

amazon.com/Quantum-Electrodynamics-Second-Theoretical-Physics/…

It looks like chapter 4 is basically solving the hydrogen atom as a bound state problem treating the EM field as a classical field

Then chapter 5 is bound state problems where you use a quantum EM field

Semiclassical

That one just happens to be close to my heart because the issue is largely that the integrals involved are highly oscillatory

bolbteppa

They seem to solve the photoelectric effect, Zeeman, Stark

enumaris

6:11 PM

woah, landau and lifshitz did many more books than I thought lol

bolbteppa

I wonder what's wrong with what they do?

Why is it not good enough to be included in every qft book, I mean the photoelectric effect basically started quantum theory

enumaris

their EM book was quite good

bolbteppa

That EM book might be my bible, but I don't know it off by heart yet :(

Even half of it it's so insane

enumaris

did these guys just survey all of physics

lol

bolbteppa

Yeah

In a prison naturally

Course of Theoretical Physics

The Course of Theoretical Physics is a ten-volume series of books covering theoretical physics that was initiated by Lev Landau and written in collaboration with his student Evgeny Lifshitz starting in the late 1930s. It is said that Landau composed much of the series in his head while in an NKVD prison in 1938-1939. However, almost all of the actual writing of the early volumes was done by Lifshitz, giving rise to the witticism, "not a word of Landau and not a thought of Lifshitz". The first eight volumes were finished in the 1950s, written in Russian and translated into English in the late 1950s...

ACuriousMind

6:13 PM

@bolbteppa I don't think it's not "good" enough - but QFT doesn't make any particular prediction about the photoelectric effect we hadn't known without it, does it?

Semiclassical

In which case there’s at least a hope that saddle point approximation (aka the main workhorse behind semiclassical methods) has a hope of clarifying the matter

enumaris

maybe I'll get all their books...

ACuriousMind

The focus on scattering amplitudes is because it is a "true QFT" prediction - you can't explain what's going on any other way

bolbteppa

I would argue the other explanations are wrong :p

haha

enumaris

I might just have to try to read all of them before I die lol

bolbteppa

6:15 PM

That is the goal

enumaris

not sure I can reach that goal

bolbteppa

The first 3 chapters are phenomenal, some insane stuff about polarization density matrices leads to the Pauli-Lubanski pseudo-vector

So emission and absorption are basically bound state problems, that's why...

enumaris

is there QCD in that series

Semiclassical

The following is probably too simplistic but: nonperturbative QFT is a lot harder than pertubative QFT

bolbteppa

They don't make that clear that it's bound state problems as far as I can see browsing through it

enumaris

6:18 PM

if L+L can help me understand QFT that would be phenomenal

bolbteppa

No QCD, the earlier edition of this book has more on what they knew about the strong interactions, but it's from the 50's, Bjorken and Drell has a similar chapter full of guesses

No path integrals either, it's all just brute force (vicious) derivations based on thought

Semiclassical

With QED, that’s mostly fine because the coupling constant is small and therefore most of the asymptotic series you see can be well-approximated by a finite number of terms

With QCD, though, the coupling constant isn’t small

And therefore life becomes hard

bolbteppa

I think qcd is too modern to just learn the most up to date version, need to go through the history and see the delicacies, it's just so hard haha

Same with strings, which of course is about god damn Regge qcd stuff before qcd was invented :\

The known stuff, for which answers are there, is hard enough :(

enumaris

yeah if L+L was written in the 50's then no chance of much QCD as we know it there

Novalium Company

Ahh.. nothing feels better than a hot shower.

bolbteppa

6:26 PM

'Weinberg notes on p.560 that: the theory of relativistic effects and radiative corrections in bound states is not yet in entirely satisfactory shape. This quote from 1995 is still valid today, 20 years later.'

enumaris

pain

CooperCape

ohai

enumaris

6:41 PM

ohai

CooperCape

finished school today.

forever

unless I resit

who knows

a world of opportunity.

or not depending

ACuriousMind

@CooperCape Yay! So what are you doing here? Go party! :P

CooperCape

@ACuriousMind That's next week (not everyone's finished). Monday, wednesday

thursday and friday

Mithrandir24601

@CooperCape Nice! Just have 3 or 4 or 7 or 8 or who knows how many years of uni ahead of you :P Enjoy!

ACuriousMind

Ah, I see :)

CooperCape

6:45 PM

@Mithrandir24601 Here's hoping, never know might have to take a gap year.

That would be interesting

enumaris

graduated from college?

or high skoo?

CooperCape

Both in the Uk, but high schoo everywhere else.

because apparently we call what isn't college, college.

enumaris

hmmm

so you could have up to 12 more years of skoo left

assuming you do a 8 year phd

4 year undergrad

ACuriousMind

Solution: Don't do a PhD :P

CooperCape

I dunno I kinda wanna go into research.

But of course that's subject to change

bolbteppa

6:53 PM

3 year undergrad in uk, 1 year msc, 3-4 year phd

ACuriousMind

Yeah, a lot will happen between now and then. Don't sweat it ;)

CooperCape

Just let the wind take me.

bolbteppa

No broad education courses

Mithrandir24601

@enumaris That's twice as long as the one I'm doing

enumaris

7:32 PM

@Mithrandir24601 that's the maximum length for an experimental physics PhD at my alma mater

theorists are limited to 7 years

most people do 6 years

I did 7 years cus I'm lazy

5 years is generally the fastest...but if you can convince your committee and department and graduate department...maybe 3 years is doable

Loong

@enumaris Counting the years starting from what?

Mithrandir24601

@enumaris Blimey...

enumaris

It includes the Master's though

so maybe there's a slight difference there

Loong

ah

enumaris

Generally PhD programs here in the States include the Master's degree as an intermediate degree.

My program also gave me a C.Phil degree

dunno why

Mithrandir24601

7:38 PM

@enumaris That would be an extra year - most PhDs here would be 3-4 years but I did one with an extra year (sort of equivalent to doing a masters, to some approximation), so we get 4 years, or 4.5 years if we end up with an extension for whatever reason

enumaris

hmmm

yeah not sure why ours is longer :P

ACuriousMind

8:35 PM

@JohnRennie So, I did a bit of googling and apparently "Rennie" comes from the name "Reynold", which in turns seems to mean something like "counsel of the ruler".

enumaris

9:21 PM

I'm on a roll answering neural network questions in cross validated hehe

not many upvotes though.. :(

bolbteppa