I've put the 'behind the scenes' on how the number line was made on Patreon. Spoiler: it involves a spreadsheet. www.patreon.com/posts/49699535 PS This video was previous sponsored by a VPN but that has since expired. Please now enjoy it sponsor free!
Mat. Check out my math vid's i made. Of a program that no one has made. Please. I am sick and i may die. I don't know yet as i have not gotten tested yet. But i will. Talk to me i want to give them to you as tools for teaching. Freely.
PIA has been bought by Kape Technologies(formerly crossrider), since that time in court proving they didn't log. OVPN is currently the only proven non-logger from a court-case that is still the same company. Other VPN's are unproven (PIA is among them now, read into Kape Technologies and their crossrider days making malware and adware) and NordVPN had a data breach and didn't inform their customers that they might have been leaking their data untill a year after. With everything online, a small provider might have sub-par security, but they are also a smaller target but it's always a risk. VPN's are not a risk free privacy guarantee. Not only have huge companies suffered data breaches, but the "hiding from your ISP" argument is *ONLY* valid if you trust your VPN provider more than your ISP.
You thought the Parker square was named after Matt Parker. Actually, Matt himself is merely the human example of the Parker. (Love you Matt, nobody makes math stuff educational and hilariously like you do.)
@@ChrisHarringtonMinneapolis Yes, of the sequence of "largest prime gap up to N," that one is my current favorite. 3 consecutive decades that are empty of primes: 1330's, 1340's, 1350's. I call it "The Grand Canyon." The "South Rim" is 1327, and the "North Rim" is 1361. The 8 numbers in that span that aren't divisible by 2, 3, or 5, factor as follows: 1331 = 11³ 1333 = 31·43 1337 = 7·191 1339 = 13·103 1343 = 17·79 1349 = 19·71 1351 = 7·193 1357 = 23·59 [Incidentally, today's (2021 Apr 6) Julian Day Number, 2,459,311, is prime.] Fred
I'm so happy someone still remembers Future Mark. Those benchmarks ROCKED! Wait, wrong FutureMark... (just search for it here on YT, there are videos of all of them!)
I thought the "here's log base #" bit was a bit ha ha for people who already know what's happening but I have a feeling that people who don't already know a lot about logs would probably be scratching their heads. It needed a bit more explanation.
There actually is little-o notation which is like a stronger version of the big-O notation. (en.wikipedia.org/wiki/Big_O_notation#Little-o_notation ) I think the equation on screen is correct, so he should've called it "little-o".
When Matt accidentally implies that his function cannot be growing faster in any way than the function he is talking about at that point even after multiply the function by a constant
By the way, changing the base of a log only scales it by a constant amount. That is, log_a (x) = c * log_b (x) where c = 1 / log_b (a). So for _any_ log plot, changing the base of the log would not affect the shape of the plot. It just changes the scale of the plot.
@@standupmaths Hmm.. I think wheel sieve(of primordials) is more intuitive for showing prime gaps. Each successive primordial wheel sieve is made up of its predecessor?
There aren't enough theoretical multiverses, each containing our universe's quantity of atoms, in order to write each digit of what you just said on the surface of each atom.
I was really expecting a Matt Parker complicated script writing and timing special where when we were talking about looking for a gap of 8 he would at some point look down and just point at one scrolling across the bottom of the screen "Oh! there's one!"
Funnily enough, there's a gap of 8 at 7:40 (just before he starts talking about the factorial proof) and at 9:30 (just as he finishes talking about it), but none in between
Looking at the graph, I have my own conjecture about the primorials/jumping-champions connection but I don't know if it's been considered already. As Matt points out at 19:45, the top of the line is all the multiples of 6. The ones he highlights as suspicious contenders, who are raised slightly above the others, are all multiples of 30 until 210 which is raised even more from the other lines. My suspicion here is that the 'thickness' of this line is actually the result of multiple lines being overlaid, with each line sharing the same common factors. So one line for powers of two, one for multiples of only 2 and 3, for 2,3 and 5 and so on. In the Silva paper in the description, they highlight the multiples of 6 in another colour and I think it would be interesting to see the same for the rest of the primorials which, by their definition, would be the lowest value for each of their respective lines. Each line then, is more popular than the last as numbers grow higher but lower numbers are more frequent for any given line, which is why it takes time for each champion to jump to the top. As an afterthought, this might explain the bumpiness of the lines, too. There are sets of unique prime factors that are non-primorial (ignoring the odds) - 2*5, 2*7, 2*3*7 and so on. From that we would expect bumps at 10, 14, 20, 28, 40, 42... At least up to that far, the graph looks to me like it meets expectations.
So usually there's an enormous wait between new papers released about prime gaps, but suddenly there were two papers released right next to each other? ... Let's call it: "the twin paper conjecture."
On other fora I've heard "steam engine time" used to mean the moment when conditions are ripe for some innovation to occur, so suddenly a whole bunch of people make the leap at once.
I've dipped the tiniest tip of a toe into the deep lake that is prime number theory, and what most gets me is just how simple and breezy this video can come off as, all the concepts being so easy to explain, yet underlying them is no doubt some extraordinarily complex mathematics.
I came looking for this, sortof. @ 9:49 he says the gap is 89-97. Then mentions 87 at ur stamp. I was confused, and now im More confused cuz apparently i missed a joke too... :(
There is actually a seminar by Terence Tao on prime gaps uploaded to TH-cam by UCLA from just after they published their papers. It provides some cool insight into what happened at the time.
@@blindleader42 it is easy for small numbers (1 to 5) cause you can brute force it with google. Mathematicians are still unsure on values as small as 6 though.
16m42s: "A day later, on the 21st of August, 2014, someone else proved the same thing a different way." [Shows title & Abstract of a paper by James Maynard.] Hey, he's not just "someone else;" he's that famous prime-o-phile from the Numberphile channel! Fred
It's not big O notation, obviously its a small o. Small o is much stricter than big O. If f in O(g) it means that f(n) will be smaller than a constant times g(n) after some n great enough If f in o(g) it means that f(n)/g(n) tends to zero as n tends to infinity. So while both are Landau notation, big O acts as a ≤ while little o acts as
My gosh, the primorials fact blew my mind, it's crazy how clearly there must be some underlying structure to the primes, and how much it brings about such neat patterns, yet it completley illudes us.
"... because they are all odd numbers the gaps are always even..." 1 not being a prime i could accept but now 2 is also left on the side that i can not allow!
It's interesting you should point this out: because the only reason 2 was declassified as a prime was convention - to avoid having to say "any prime except 2" or "take any odd prime". In this case it avoids having to add the qualification "all prime gaps, except the gap between 2 and 3, are even.
and the gap between 2 and 3! they're consecutive primes too! yet there's no point eternally in the bottom left corner of all of his graphs for the single gap of 1 that appears
As the "top point on the line" increases from 6 to 30 to 210, etc the shape of the line doesn't change. The resolution of the plot gets very much smaller and the earlier, smaller, numbers are just smushed into the band under the top point. As 2 is when the top point is 6.
I was going to say, since the bottom right is roughly (ln (no. of primes))^2, it will continue on WAY faster than each next primorial taking over. However, when all 150 million were animated starting from small numbers, the slope of the line definitely looks like it drops with more and more primes. Also, I think Matt should try skipping a large amount of the first primes to make these calculations, such as going from the 140 millionth to 170 millionth primes.
As Matt exemplifies in his presentation, time for pure mathematicians is merely the succession of numbers. He constantly refers to the gaps getting bigger "quickly" as the number X in the lower boundary equation gets bigger. What an educator! I've been enthralled from beginning to end. Thank you!
Suggestion for your 1M subscriber special: complain about all the times past Matt wasn't excited enough about graphs or maths in general. That was fun!
Hey Matt, I greatly appreciate that your VPN ad spot was honest and not misleading! Too many TH-camrs read off BS/misleading/incorrect scare tactics in their ad spots in order to get more sales. I'm glad you were honest about what a VPN does; and didn't go off and say that without a VPN, hackers can steal all your data. It's sad that I have to actually praise people for *not* spreading misinformation, but well... that's where we are at the moment.
Thanks! I did slip-up and say without a VPN your ISP can see your search terms, which is not true for Google using https. So it’s not perfect! I’ll correct that next time.
Since the log base doesn't matter, the graph should be animated such that the log base is always the frequency of gaps of size 2. That way the animation will always grow from 0, and you have an absolute reference point.
I like logs too! A log house is long-lasting and cool, you can make wood statues out of logs, logs have an industry of their own! Logs are just so amazing and useful.
Just a reminder: For expressions like log log log ... log x, one can always use the recomposition notation: $\log \overset n \circ x$, where n is the number of logs. Another reminder: awesome video!
I kinda love that these big numbers you're talking about (like, 10^|my overdraft|) are infinitesimal fractions of huge numbers like Graham's Number and Tree (3), which are themselves, by definition, infinitesimal fractions of the entire number line. It blows my mind that mathematicians can construct and manipulate such big numbers, while simultaneously recognising that these numbers are trivially small. For the first three or four minutes, I was wondering if you were heading towards the Riemann Hypothesis, but then you went somewhere I wasn't expecting.
The video I watched before this was a video about Rick and Morty and I'm not sure if the algorithm is just that good or if an amazing coincidence just happened
You might want to check out the video I just made (look for the one and only video on my channel). It's very bland - no sound, and only comprehensible if you saw Matt's video. But it's an extended version of that animation at 6:30 :-)
In the part where you do 8 factorial and then reduce it to the greatest common multiple, you could just use primorials. The reason this still works is that the product +2, +4, or +8 all are composite because 2 divides into them. So you actually wouldn't have to multiply 2 three times, but just once.
6:20 The animation has the horizontal axis labeled with half the gap, but you can tell by the multiples of 30 and where they stay higher on the line that it's actually scaled by the gap instead of half the gap. At the very end of the animation, yes, the scale suddenly changes to half the gap.
Thank you so much for explaining the functions! I've seen other functions before but couldn't understand their meanings. You made it so much easier! Great job!
20:18 I was very proud of myself when I'd predicted "Oooh, the next peak will be at 2310 because that's 210*11, and 210 is 7*30 !" a few seconds before he mentioned this.
I love these videos! They always make me confused since i didnt have an oppitunity to study maths past my GCSEs, but it all so facinating from what i can get
fantastic video! I applaud the video editing. When you pinpointed the individual points on the graph with your finger (the ones that take the lead eventually for common gap size), I have no idea how you were able to do that . And the running timeline at the bottom was great, something extra to look at
"As a number theorist i have a favorite numerical sequence. Did you know that if you take the number 41 and add first two, then four then six etcetera. To get the sequence 41,43,47,53 etc. That the first forty numbers are all primes. And that no similar numerical sequence of that lenght exists." - General Michael O'Toole. RAMA the video game. Based on the works of Arthur C Clarke and Gentry Lee. :)
What rolls down stairs Alone or in pairs, And over your neighbor's dog? What's great for a snack, And fits on your back? It's log, log, log It's log, it's log, It's big, it's heavy, it's wood. It's log, it's log, it's better than bad, it's good. " Everyone wants a log You're gonna love it, log Come on and get your log Everyone needs a log Log log log
(this sent me down a rabbit hole of the evolution of the Slinky ad--the jingle originated in the 60s, but "without a care" became the better-rhyming "alone or in pairs" in the 70s)
hi matt! apologies for this probably long comment! firstly, i absolutely love all of your videos you have such a way of telling mathematical stories without losing any of the maths itself which i love so much! i, and i think some other people online, have noticed that you often will use singular they/them pronouns for people and according to reddit this is also true for much of Humble Pi. I thought this was cool! after also hearing a professor of mine (physics, so i was asking about how the uni may try to better express that people who use gender neutral pronouns are welcome in this area) discuss the use of gender neutral pronouns by academics (something i still haven’t fully been able to understand, maybe just for ease? or confidentiality?) this is what i had just assumed was what you were doing. And then this video! at 14:11 you referred to past matt (which in some way is you but i don’t do philosophy) with they/them pronouns! which i, again, thought was very cool. i can’t find anything online about you discussing your gender and obviously if this is something you’d rather not explicitly discuss because that is your personal life then that is very cool and understandable. i don’t really? have a question um i apologise if this has been a waffle i just wanted to see if you had anything to add onto this, i am nonbinary and really appreciate this sort of stuff of moving to normalise the use of gender neutral pronouns. especially in stem fields!! ❤️
To be honest: I respond to any of he/him/they/them and don’t mind anything else as long as it’s not malicious. I actually refer to myself as sometimes they/them for the same reason I do other people much of the time (and 100% of the time if they are hypothetical people like the examples in my book) which is to normalise non-gender-specific language. I hope that makes sense!
@@standupmaths absolutely! thank you for clarifying and responding! i think what you’re doing as a maths educator and curiosity-inspirerer(?) is so wonderful
The way I think about primes from a non-mathematical perspective is that they are recursively self-destructive. What I mean by that is that for every prime we find, every future multiple of that prime can no longer be a prime (self destructive and therefore recursively defined), and every prime is therefore defined by the *lack* of a prime divisor of itself earlier in the sequence. The more primes we find, the larger the future gaps will be because every new prime removes infinitely many future potential primes. These gaps *would* increase linearly with the primes if it weren’t for the fact that the multiples of primes *overlap* with each other at an increasing rate as more primes are found (not sure how to word this better), thus giving us a *logarithmic* increase in the gaps instead of a linear one. Another way to look at the logarithmic nature of prime gaps is from the fact that when verifying that a number is prime, you never need to check any prime higher than the square root of the number you’re checking. In other words, as the number you’re checking increases linearly, the amount of numbers that could prevent it from being a prime only increase logarithmically, and thus the number of gaps can only increase logarithmically. This recursively self-destructive definition of primes is part of why primes are so difficult to get a solid grasp on. Every prime’s very existence is defined by *not* being a multiple of a previous prime.
Unless I've misunderstood what is being said here, I think I have a small correction for around 15:10 . . . log(x)*log(log(x)) actually does get bigger than log(x) for large enough x. For example, using base 10, log(log(10^100)) = 2; since log is an increasing function, the difference only grows as x gets larger. Since we're considering x arbitrarily large, we shouldn't think of log(log(log(log(x)))) as "small" either, except compared to the denominator. We definitely shouldn't be thinking of log(log(log(x)))^2 as smaller than log(log(log(x))) in the large x limit because eventually it exceeds 1. In fact, log(log(log(log(x))))/(log(log(log(x))))^2 goes to zero as x gets large. BUT log(log(x)) blows up even faster. A rigorous way to see that the expression outlined in green is bigger than log(x) for large x is to take the limit of that expression divided by log(x) as x goes to infinity, e.g., using L'Hopital's rule.
Hi Matt and all. I am sure you all know this already, but I still thought I'd comment. Is there a pattern to Prime Numbers? The answer is a resounding YES. Unfortunately for those searching, it is an inverse pattern. In other words, rather than looking for where the primes are, we need to look for where they are not and why. Obviously, the numbers that are not prime are those which are multiples of them. One thing in our favour is that these have repeatable patterns. As such, if we have a list of numbers and eliminate all the mutiples we will be left with only primes. [See below] So, for any value of X we have to check for divisibility for all primes less than the root of X. (The reason we only need to check up to the root of X is because any number after that would be a corresponding factor to one of the numbers before the root.) For 10, whose factor pairs are 1*10 and 2*5 and whose root is 3.16... we would only have to check for the values 2 and 3. Dividing by 2 gives us the 'whole number' 5, so it's not prime. For 11, has only the factor pair 1*11 and root 3.31... we would still only check for 2 and 3. 11/2= 5.5 and 11/3= 3.67, as neither result is an 'whole number', 11 must be prime. Numbers eliminated (1 to 31): For 2: - - - 4 - 6 - 8 - 10 - 12 - 14 - 16 - 18 - 20 - 22 - 24 - 26 - 28 - 30 - For 3: - - - - - 6 - - 9 -- -- 12 -- -- 15 -- --18 -- -- 21 -- --24 -- --27 -- -- 30 - For 5: - - - - - - - - - 10 -- -- -- -- 15 -- -- -- -- 20 -- -- -- -- 25 -- -- -- -- 30 - Left: 1 2 3 5 7 11 13 17 19 23 29 and 31 are all Prime. The reason for varying gaps is due to numbers being in or out of synchronization. When 'in sync' less numbers are eliminated thus an higher chance of primes. Regarding 'Twin Primes'. As you can see when we combine and remove the 2 and 3 rows, all mutliples of 6 (2*3) have a number either side that can not eliminated by 2 nor 3. 1 2 3 - 5 - 7 - - -- 11 -- 13 -- -- -- 17 -- 19 -- -- -- 23 -- 25* -- -- -- 29 -- 31 However, for the first time 5 starts removing numbers ie 25 (5*5). At 30 (2*3*5) as all numbers are 'in sync' 29 and 31 are both spared. As numbers become higher the more numbers are removed as a result of more prime multiples being introduced. These would start from their squares, as prior multiples would be accounted for by previous numbers. (7 from 49, 11 from 121, 13 from 169 etc.). Happy Priming :)
Excellent video!! I never thought about the shape of that 'line' with such a huge numbers taken into consideration, but that is actually a great question!
3:55 No, it‘s not! The probability that a number is prime is 100% if it‘s not a multiple of any number below it except 1. If it is, then the probability is 0%.
nitpicks! at 12:46, you use a little o for big O notation - thats kinda confusing because there is a little o notation, which one are you talking about?
Little O means that your function can’t grow faster than any function even after multiplying the function you are comparing. In practice, it means that to be little-o of a function means you really grow slower than a family of functions (as opposed to big-o meaning to grow slower than or at the same rate as a family of functions).
I've put the 'behind the scenes' on how the number line was made on Patreon. Spoiler: it involves a spreadsheet. www.patreon.com/posts/49699535
PS This video was previous sponsored by a VPN but that has since expired. Please now enjoy it sponsor free!
Love PIA! Learned about it from LinusTechTips and been using it since for years!
Mat. Check out my math vid's i made. Of a program that no one has made. Please. I am sick and i may die. I don't know yet as i have not gotten tested yet. But i will. Talk to me i want to give them to you as tools for teaching. Freely.
You do realize you said "because they're all odd numbers" when referring to the first one million primes, right?
Hey Matt, I spotted a small mistake for the corrections list. At 5:11, the GAP axis shows the numbers 0-16 which should be 0-160.
PIA has been bought by Kape Technologies(formerly crossrider), since that time in court proving they didn't log. OVPN is currently the only proven non-logger from a court-case that is still the same company. Other VPN's are unproven (PIA is among them now, read into Kape Technologies and their crossrider days making malware and adware) and NordVPN had a data breach and didn't inform their customers that they might have been leaking their data untill a year after. With everything online, a small provider might have sub-par security, but they are also a smaller target but it's always a risk.
VPN's are not a risk free privacy guarantee. Not only have huge companies suffered data breaches, but the "hiding from your ISP" argument is *ONLY* valid if you trust your VPN provider more than your ISP.
Poor Past Matt, always getting interrupted by that know-it-all from the slightly less distant past.
Story of my life.
You thought the Parker square was named after Matt Parker. Actually, Matt himself is merely the human example of the Parker.
(Love you Matt, nobody makes math stuff educational and hilariously like you do.)
@@standupmaths Stop lying! We all know you're future Matt. You're not fooling anyone. Stop bullying past Matt!
@@zerid0 Well he’s definitely lying, he’s the even less distant Matt who can occasionally provide even more corrections.
I love the interruptions its so funny
Nice to see you and past Matt finally doing a colab, long overdue
I just love the prime gaps sliding over the screen as the video progresses. It's such a nice detail.
"34 OMG!!"
Thanks! I was really proud of that. Fun fact: it was generated in a spreadsheet!
@@standupmaths I'd be surprised and disappointed if it was done any other way.
@@ChrisHarringtonMinneapolis Yes, of the sequence of "largest prime gap up to N," that one is my current favorite.
3 consecutive decades that are empty of primes: 1330's, 1340's, 1350's.
I call it "The Grand Canyon." The "South Rim" is 1327, and the "North Rim" is 1361.
The 8 numbers in that span that aren't divisible by 2, 3, or 5, factor as follows:
1331 = 11³
1333 = 31·43
1337 = 7·191
1339 = 13·103
1343 = 17·79
1349 = 19·71
1351 = 7·193
1357 = 23·59
[Incidentally, today's (2021 Apr 6) Julian Day Number, 2,459,311, is prime.]
Fred
I was unreasonably happy at 9:23 when it became longer than the width of the screen
I'll let future Mark finish this comment...
Edit: Future Mark here. Past Mark put me in a bit of a spot since i've nothing to add. Thanks past Mark!
Present Mutt here. Nothing to add from this time period either
Hang on, that's not Future Mark; you're in the past now!
@@vigilantcosmicpenguin8721 were you talking to me? Because to past you i am from the future, so not a lie! 🤣
Time is a social construct
I'm so happy someone still remembers Future Mark. Those benchmarks ROCKED!
Wait, wrong FutureMark... (just search for it here on YT, there are videos of all of them!)
"log base I don't care" was often the answer I gave in exams
Log, base-eleventeen.
It's imaginary...
I thought the "here's log base #" bit was a bit ha ha for people who already know what's happening but I have a feeling that people who don't already know a lot about logs would probably be scratching their heads. It needed a bit more explanation.
Logging camp is a log base.
Ah, computer science major I take it.
I took 3 years of calculus way back when I was young I don't remember ever covering "log" or "e".
>Matt: this is big O notation
>also Matt: *uses a small o to represent it *
And calls it ‘big zero’ at 15:31
There actually is little-o notation which is like a stronger version of the big-O notation. (en.wikipedia.org/wiki/Big_O_notation#Little-o_notation )
I think the equation on screen is correct, so he should've called it "little-o".
When Matt accidentally implies that his function cannot be growing faster in any way than the function he is talking about at that point even after multiply the function by a constant
Yes, that should be little o. Totally my fault. On several levels.
thats matt's schtick now
I absolutely love that Matts WiFi is called “one small step for LAN”
The best on i have seen was
Too fly for a wifi
I wonder if the password is "one giant leap for LANkind". too easy to hack, maybe.
@@vincentpelletier57 It's gonna be a parker password. It'll be as you say except it's arbitrarily misspelled
@@Kram1032 Makes sense
One of my favorites "Rebellious Amish".
By the way, changing the base of a log only scales it by a constant amount. That is, log_a (x) = c * log_b (x) where c = 1 / log_b (a).
So for _any_ log plot, changing the base of the log would not affect the shape of the plot. It just changes the scale of the plot.
Use base 1 or 0
@@happygimp0 oof, infinite and zero scale
👏👏
Even easier to see it using the change of base rule
log_b(a) = log_x(a) / log_x(b).
The divider is constant for all different values of a
Nice ME system you got there.
@@cubixthree3495 Thanks :)
I read the title as (gasp!) and was wondering what was so exiting
same
Primes. Primes are so exciting.
@@standupmaths Exactly! What is more exciting than primes? Nothing. Not even getting a new guitar.
@@standupmaths Hmm.. I think wheel sieve(of primordials) is more intuitive for showing prime gaps. Each successive primordial wheel sieve is made up of its predecessor?
@@standupmaths have you tried taking an unnatural log (log to the base π) of something?
"There's a gap between two primes the size of Graham's number. We can prove this exists, first take the factorial."
I spot a problem.
I can help! The factorial ends with more than 7.6 trillion 0's.
Btw Graham's number ends with 7.
There aren't enough theoretical multiverses, each containing our universe's quantity of atoms, in order to write each digit of what you just said on the surface of each atom.
Its odd! But what about using twice the size.
@@anawesomepet how do you know it ends in 7.6 trillion 0's?
@@ERROR-ei5yv for n! there are (Summation from k=1 to infinity of the integer part of n/5^k) trailling zeros, thats how
"As big as it need to be gosh darn it"
Mathematics is a really objective and precise in nature, yes.
It’s precisely as vague as it needs to be
@@gamersgonnagam3 perhaps "exactly as vague as it can get away with"?
Astronomers see nothing unusual with that statement.
@@SgtKOnyx I think that might be engineering, actually
I was really expecting a Matt Parker complicated script writing and timing special where when we were talking about looking for a gap of 8 he would at some point look down and just point at one scrolling across the bottom of the screen "Oh! there's one!"
Funnily enough, there's a gap of 8 at 7:40 (just before he starts talking about the factorial proof) and at 9:30 (just as he finishes talking about it), but none in between
@@LARAUJO_0 is that a gap in the gaps !?
"Zeroth things first..." That is the best thing this guy does. 0-indexing is important.
Shouldn’t it be “Zeroth things zeroth”
Looking at the graph, I have my own conjecture about the primorials/jumping-champions connection but I don't know if it's been considered already.
As Matt points out at 19:45, the top of the line is all the multiples of 6. The ones he highlights as suspicious contenders, who are raised slightly above the others, are all multiples of 30 until 210 which is raised even more from the other lines.
My suspicion here is that the 'thickness' of this line is actually the result of multiple lines being overlaid, with each line sharing the same common factors.
So one line for powers of two, one for multiples of only 2 and 3, for 2,3 and 5 and so on. In the Silva paper in the description, they highlight the multiples of 6 in another colour and I think it would be interesting to see the same for the rest of the primorials which, by their definition, would be the lowest value for each of their respective lines.
Each line then, is more popular than the last as numbers grow higher but lower numbers are more frequent for any given line, which is why it takes time for each champion to jump to the top.
As an afterthought, this might explain the bumpiness of the lines, too. There are sets of unique prime factors that are non-primorial (ignoring the odds) - 2*5, 2*7, 2*3*7 and so on. From that we would expect bumps at 10, 14, 20, 28, 40, 42... At least up to that far, the graph looks to me like it meets expectations.
I like that there are GAPS in the video with future Matt interrupting!
The probability that Future Matt interrupts Past Matt is log log n
@@diamondsmasher But how about the odds that Future-Future-Matt interrupts Future-Matt interrupting Past-Matt?
Now we need to calculate the time gaps between these interruptions. Are they behaving primorial?
They are known as Parker gaps
Stealth pun!
I just love the idea that Matt spends his free time reading "giant chalkboards covered in math"
No comment.
he's trapped in the Chalk Dimension, trying to calculate a route out.
I bet he also uses his vpn for tracking down dark-web sources of Hagoromo chalk.
@@gcewing You need to go through some really sketchy back-alleys for the _really_ good stuff.
@@standupmaths In fact, this is a comment.
>Big Zero
chungus*
@@philkaw say chungus but replace the "chu" with "amo"
Amumugos
Amonges
Amongus
So usually there's an enormous wait between new papers released about prime gaps, but suddenly there were two papers released right next to each other?
... Let's call it: "the twin paper conjecture."
Nice.
Hillarious
Do the gaps between papers get larger?
On other fora I've heard "steam engine time" used to mean the moment when conditions are ripe for some innovation to occur, so suddenly a whole bunch of people make the leap at once.
@@vigilantcosmicpenguin8721 its because each paper gets thicker.
"Big zero" spotted! Glad you, the author of Humble Pi, left it in.
I've dipped the tiniest tip of a toe into the deep lake that is prime number theory, and what most gets me is just how simple and breezy this video can come off as, all the concepts being so easy to explain, yet underlying them is no doubt some extraordinarily complex mathematics.
Very true, best example is the paper containing the proof of the ternary golbach conjecture lmao.
Holy crap the editing these videos must take. Aside from the enthusiasm, I have a lot of respect for the time and effort you put in. Thanks Matt!
"In this case is 840. I mean, it is nt 87, but it is a lot smaller"
- Matt Parker
I love out of context quotes.
People will think I'm strange now when I'm working my exams and I whisper "Future Matt? Any help on this one?"
Oh man, Yeah! Future Matt - hear our prayers! Answer our math/s questions and elevate the quality of our calculations!
Now my suggested videos include: “Making a log carving robot”
Following that channel keeps me happy
should have been used private internet access (tm)
Lol, someone's (ro)bot isn't intelligent
10:54 "840! I mean it's not 87 but it's a lot smaller." Lovely Parker sentence.
I came looking for this, sortof. @ 9:49 he says the gap is 89-97. Then mentions 87 at ur stamp. I was confused, and now im More confused cuz apparently i missed a joke too... :(
840! is way bigger than 87
Wasn't 840!, but rather plain 840, which is much smaller than 8!
There is actually a seminar by Terence Tao on prime gaps uploaded to TH-cam by UCLA from just after they published their papers. It provides some cool insight into what happened at the time.
I somehow missed that. Will check it out. Tao is amazing.
To be fair, you need to have a really high IQ to predict the date of the next Rick and Morty season
I must have a really high IQ then, because I know the date of the season 5 premier.
@@blindleader42 it is easy for small numbers (1 to 5) cause you can brute force it with google. Mathematicians are still unsure on values as small as 6 though.
@@yyeeeyyyey8802 OK. I predict season 6 sometime in 2022... or never.
Would you believe they announced the date between me filming this and release it. You’re welcome.
I can find a lower bound on the date. But it's not very impressive.
16m42s: "A day later, on the 21st of August, 2014, someone else proved the same thing a different way."
[Shows title & Abstract of a paper by James Maynard.]
Hey, he's not just "someone else;" he's that famous prime-o-phile from the Numberphile channel!
Fred
Drinking game: take a shot everytime a gap of 2 appears at the bottom
younger matt starting at 487 saved lives
If you and your mates (who are betting on another numbers) are cursed with immortality, you'll be the most sober guy in the room.
You have no right being this funny and simultaneously educational. I love it.
It's not big O notation, obviously its a small o. Small o is much stricter than big O.
If f in O(g) it means that f(n) will be smaller than a constant times g(n) after some n great enough
If f in o(g) it means that f(n)/g(n) tends to zero as n tends to infinity.
So while both are Landau notation, big O acts as a ≤ while little o acts as
Thank you.
Was gonna comment this ty
My gosh, the primorials fact blew my mind, it's crazy how clearly there must be some underlying structure to the primes, and how much it brings about such neat patterns, yet it completley illudes us.
If this youtube thing doesn't work out at least we know you have the pointing skills to be a weatherman
Matt: "Anything I say from now on assume it's a sensible case"
Us: No, I don't think I will
"... because they are all odd numbers the gaps are always even..." 1 not being a prime i could accept but now 2 is also left on the side that i can not allow!
I don’t know it’s the only even prime, hardly fits in with the others ;P
@@Lanthardol as math teachers like to jest, 2 is the oddest prime of all.
It's interesting you should point this out: because the only reason 2 was declassified as a prime was convention - to avoid having to say "any prime except 2" or "take any odd prime". In this case it avoids having to add the qualification "all prime gaps, except the gap between 2 and 3, are even.
@@ig2d there isn’t a gap between 2 and 3
3 minus 1 is 2. If we take 2 out of the primes club, can we bring 1 back in?
Ooh, time for my favourite maths joke!
"What sound does a drowning number theorist make?"
logloglogloglog...
i almost ordered a custom t-shirt with that printed on it, its my favourite joke too
Matt at 0:33: "Because they're all odd numbers…"
The number 2: 🥺
yeah, definitely an odd prime for sure.
and the gap between 2 and 3! they're consecutive primes too!
yet there's no point eternally in the bottom left corner of all of his graphs for the single gap of 1 that appears
@@DagothXil Speaking of gaps, is it actually relevant to say there's a gap between consecutive numbers?
It's around 11:18 where i stopped watching a math video but started watching a magician's performance.
One small comment: The papers seem to use little O notation, not big O. The difference is that the bound is strict.
As the "top point on the line" increases from 6 to 30 to 210, etc the shape of the line doesn't change. The resolution of the plot gets very much smaller and the earlier, smaller, numbers are just smushed into the band under the top point. As 2 is when the top point is 6.
I was going to say, since the bottom right is roughly (ln (no. of primes))^2, it will continue on WAY faster than each next primorial taking over. However, when all 150 million were animated starting from small numbers, the slope of the line definitely looks like it drops with more and more primes.
Also, I think Matt should try skipping a large amount of the first primes to make these calculations, such as going from the 140 millionth to 170 millionth primes.
7:31 "Arbor Terry" Love that guy. Always planting trees.
As Matt exemplifies in his presentation, time for pure mathematicians is merely the succession of numbers. He constantly refers to the gaps getting bigger "quickly" as the number X in the lower boundary equation gets bigger. What an educator! I've been enthralled from beginning to end. Thank you!
Suggestion for your 1M subscriber special: complain about all the times past Matt wasn't excited enough about graphs or maths in general. That was fun!
Hey Matt, I greatly appreciate that your VPN ad spot was honest and not misleading! Too many TH-camrs read off BS/misleading/incorrect scare tactics in their ad spots in order to get more sales. I'm glad you were honest about what a VPN does; and didn't go off and say that without a VPN, hackers can steal all your data. It's sad that I have to actually praise people for *not* spreading misinformation, but well... that's where we are at the moment.
Thanks! I did slip-up and say without a VPN your ISP can see your search terms, which is not true for Google using https. So it’s not perfect! I’ll correct that next time.
Since the log base doesn't matter, the graph should be animated such that the log base is always the frequency of gaps of size 2. That way the animation will always grow from 0, and you have an absolute reference point.
I like logs too! A log house is long-lasting and cool, you can make wood statues out of logs, logs have an industry of their own! Logs are just so amazing and useful.
“It’s called Big G, because it looks for big gaps” 😂
I imagine Big G is a gangster boss
@@3Ppaatt my thought exactly 😂
"I'd like you to meet Big G from Chicago."
Honestly just some of the best STEAM communication I'm subscribed to; I just love the enthusiasm and passion and humor.
For clarification, whenever someone refers to log without a base, it is ALMOST ALWAYS log base e (or ln).
Just a reminder:
For expressions like log log log ... log x, one can always use the recomposition notation: $\log \overset n \circ x$, where n is the number of logs.
Another reminder: awesome video!
its worth noting that when the base of the logs change, the scale of the plot changes as well. its not the same number, but its just scales the axis
He also pulled a sneaky Y-axis flip for 0.001, it started rising in the negative direction
@@ilurv2eetpie yup, though you could argue that this is just scaling as well
I wish i could watch thus channel while learning in a middle school. I envy nowadays students have this opportunity.
I was hoping future Matt would keep interrupting after the second one. I was not disappointed.
I kinda love that these big numbers you're talking about (like, 10^|my overdraft|) are infinitesimal fractions of huge numbers like Graham's Number and Tree (3), which are themselves, by definition, infinitesimal fractions of the entire number line. It blows my mind that mathematicians can construct and manipulate such big numbers, while simultaneously recognising that these numbers are trivially small.
For the first three or four minutes, I was wondering if you were heading towards the Riemann Hypothesis, but then you went somewhere I wasn't expecting.
The video I watched before this was a video about Rick and Morty and I'm not sure if the algorithm is just that good or if an amazing coincidence just happened
Blame future Matt.
YT does not make coincidences, I mean what a mistake... Shoot, this is going nowhere...
Mate... That animation at 6:30 is brilliant.... Seriously well played!
You might want to check out the video I just made (look for the one and only video on my channel).
It's very bland - no sound, and only comprehensible if you saw Matt's video. But it's an extended version of that animation at 6:30 :-)
0:34. "Because they're all odd numbers". The Parker Two
In the part where you do 8 factorial and then reduce it to the greatest common multiple, you could just use primorials. The reason this still works is that the product +2, +4, or +8 all are composite because 2 divides into them. So you actually wouldn't have to multiply 2 three times, but just once.
Matt: So I've written some Python code...
Matt's Laptop: pleeez haalp
6:20 The animation has the horizontal axis labeled with half the gap, but you can tell by the multiples of 30 and where they stay higher on the line that it's actually scaled by the gap instead of half the gap. At the very end of the animation, yes, the scale suddenly changes to half the gap.
20:46...we managed to “prove” that it “implies”... 😆 I love these.
Thank you so much for explaining the functions! I've seen other functions before but couldn't understand their meanings. You made it so much easier! Great job!
Not all prime numbers are even. 2 became prime against all odds.
ALL. BETS. ARE. OFF.!!!
Thanks Matt for finally making a video on this topic! I have been waiting patiently for this video :-). Absolutely love your channel!
The Rick and Morty comparison is something I didn't know I needed today.
Beautiful animated scatter plot of how the prime-gap changes. Thanks for making my day.
Future Matt appearing and scribbling everywhere gave me Emperor’s New Groove vibes
All this talk about prime gaps reminds me about runs of sequential Collatz sequences with the exact same length. Blows my mind!
20:18
I was very proud of myself when I'd predicted "Oooh, the next peak will be at 2310 because that's 210*11, and 210 is 7*30 !" a few seconds before he mentioned this.
210 != 7*30! 😉
@@RedGorillaa Yes it is. Use a calculator.
@@smergthedargon8974 You've been foiled by the unintentional factorial.
@@smergthedargon8974 7*30! = 7*30*29*28*...3*2*1 != 210 😉
@@Euler13 Oh, so you're just being a smartass.
I love these videos! They always make me confused since i didnt have an oppitunity to study maths past my GCSEs, but it all so facinating from what i can get
I couldn't avoid getting distracted every time twin primes appeared
fantastic video! I applaud the video editing. When you pinpointed the individual points on the graph with your finger (the ones that take the lead eventually for common gap size), I have no idea how you were able to do that . And the running timeline at the bottom was great, something extra to look at
So you couldn't wait another few seconds so the bottom bar could reach 2000? :) My OCD feels a bit of anxiety for being left at only 1973...
.. and who was born in 1973?....
"As a number theorist i have a favorite numerical sequence. Did you know that if you take the number 41 and add first two, then four then six etcetera. To get the sequence 41,43,47,53 etc. That the first forty numbers are all primes. And that no similar numerical sequence of that lenght exists." - General Michael O'Toole. RAMA the video game. Based on the works of Arthur C Clarke and Gentry Lee. :)
What rolls down stairs
Alone or in pairs,
And over your neighbor's dog?
What's great for a snack,
And fits on your back?
It's log, log, log
It's log, it's log,
It's big, it's heavy, it's wood.
It's log, it's log, it's better than bad, it's good. "
Everyone wants a log
You're gonna love it, log
Come on and get your log
Everyone needs a log
Log log log
To what tune do I sing this?
(this sent me down a rabbit hole of the evolution of the Slinky ad--the jingle originated in the 60s, but "without a care" became the better-rhyming "alone or in pairs" in the 70s)
Can I just say that I appreciate the "bonus" of the continuous prime-line that keeps going on the bottom the whole time? :)
0:08 That's why he's so smart!!
I kinda like the edits, to clarify. It has a nice pace to it, and you addressing your past self is quite funny.
hi matt! apologies for this probably long comment! firstly, i absolutely love all of your videos you have such a way of telling mathematical stories without losing any of the maths itself which i love so much! i, and i think some other people online, have noticed that you often will use singular they/them pronouns for people and according to reddit this is also true for much of Humble Pi. I thought this was cool! after also hearing a professor of mine (physics, so i was asking about how the uni may try to better express that people who use gender neutral pronouns are welcome in this area) discuss the use of gender neutral pronouns by academics (something i still haven’t fully been able to understand, maybe just for ease? or confidentiality?) this is what i had just assumed was what you were doing. And then this video! at 14:11 you referred to past matt (which in some way is you but i don’t do philosophy) with they/them pronouns! which i, again, thought was very cool. i can’t find anything online about you discussing your gender and obviously if this is something you’d rather not explicitly discuss because that is your personal life then that is very cool and understandable. i don’t really? have a question um i apologise if this has been a waffle i just wanted to see if you had anything to add onto this, i am nonbinary and really appreciate this sort of stuff of moving to normalise the use of gender neutral pronouns. especially in stem fields!! ❤️
To be honest: I respond to any of he/him/they/them and don’t mind anything else as long as it’s not malicious. I actually refer to myself as sometimes they/them for the same reason I do other people much of the time (and 100% of the time if they are hypothetical people like the examples in my book) which is to normalise non-gender-specific language. I hope that makes sense!
@@standupmaths absolutely! thank you for clarifying and responding! i think what you’re doing as a maths educator and curiosity-inspirerer(?) is so wonderful
@@standupmathsVery cool.
9:39 the timing on that was absolutely impeccable.
The way I think about primes from a non-mathematical perspective is that they are recursively self-destructive.
What I mean by that is that for every prime we find, every future multiple of that prime can no longer be a prime (self destructive and therefore recursively defined), and every prime is therefore defined by the *lack* of a prime divisor of itself earlier in the sequence. The more primes we find, the larger the future gaps will be because every new prime removes infinitely many future potential primes.
These gaps *would* increase linearly with the primes if it weren’t for the fact that the multiples of primes *overlap* with each other at an increasing rate as more primes are found (not sure how to word this better), thus giving us a *logarithmic* increase in the gaps instead of a linear one.
Another way to look at the logarithmic nature of prime gaps is from the fact that when verifying that a number is prime, you never need to check any prime higher than the square root of the number you’re checking. In other words, as the number you’re checking increases linearly, the amount of numbers that could prevent it from being a prime only increase logarithmically, and thus the number of gaps can only increase logarithmically.
This recursively self-destructive definition of primes is part of why primes are so difficult to get a solid grasp on. Every prime’s very existence is defined by *not* being a multiple of a previous prime.
This is a very good explanation. Thanks!
“Biding its time” “lying in wait”
Sonic underground reference is not one i’d expect to see!
I was just gonna search this up...
Unless I've misunderstood what is being said here, I think I have a small correction for around 15:10 . . .
log(x)*log(log(x)) actually does get bigger than log(x) for large enough x. For example, using base 10, log(log(10^100)) = 2; since log is an increasing function, the difference only grows as x gets larger.
Since we're considering x arbitrarily large, we shouldn't think of log(log(log(log(x)))) as "small" either, except compared to the denominator. We definitely shouldn't be thinking of log(log(log(x)))^2 as smaller than log(log(log(x))) in the large x limit because eventually it exceeds 1. In fact, log(log(log(log(x))))/(log(log(log(x))))^2 goes to zero as x gets large. BUT log(log(x)) blows up even faster.
A rigorous way to see that the expression outlined in green is bigger than log(x) for large x is to take the limit of that expression divided by log(x) as x goes to infinity, e.g., using L'Hopital's rule.
The biggest prime gap you will see scrolling by the bottom of the screen is 34. To see it, just go to 16:35 :)
Hi Matt and all. I am sure you all know this already, but I still thought I'd comment.
Is there a pattern to Prime Numbers? The answer is a resounding YES.
Unfortunately for those searching, it is an inverse pattern. In other words, rather than looking for where the primes are, we need to look for where they are not and why.
Obviously, the numbers that are not prime are those which are multiples of them. One thing in our favour is that these have repeatable patterns. As such, if we have a list of numbers and eliminate all the mutiples we will be left with only primes. [See below]
So, for any value of X we have to check for divisibility for all primes less than the root of X. (The reason we only need to check up to the root of X is because any number after that would be a corresponding factor to one of the numbers before the root.)
For 10, whose factor pairs are 1*10 and 2*5 and whose root is 3.16... we would only have to check for the values 2 and 3. Dividing by 2 gives us the 'whole number' 5, so it's not prime.
For 11, has only the factor pair 1*11 and root 3.31... we would still only check for 2 and 3. 11/2= 5.5 and 11/3= 3.67, as neither result is an 'whole number', 11 must be prime.
Numbers eliminated (1 to 31):
For 2: - - - 4 - 6 - 8 - 10 - 12 - 14 - 16 - 18 - 20 - 22 - 24 - 26 - 28 - 30 -
For 3: - - - - - 6 - - 9 -- -- 12 -- -- 15 -- --18 -- -- 21 -- --24 -- --27 -- -- 30 -
For 5: - - - - - - - - - 10 -- -- -- -- 15 -- -- -- -- 20 -- -- -- -- 25 -- -- -- -- 30 -
Left: 1 2 3 5 7 11 13 17 19 23 29 and 31 are all Prime.
The reason for varying gaps is due to numbers being in or out of synchronization. When 'in sync' less numbers are eliminated thus an higher chance of primes.
Regarding 'Twin Primes'. As you can see when we combine and remove the 2 and 3 rows, all mutliples of 6 (2*3) have a number either side that can not eliminated by 2 nor 3.
1 2 3 - 5 - 7 - - -- 11 -- 13 -- -- -- 17 -- 19 -- -- -- 23 -- 25* -- -- -- 29 -- 31
However, for the first time 5 starts removing numbers ie 25 (5*5). At 30 (2*3*5) as all numbers are 'in sync' 29 and 31 are both spared.
As numbers become higher the more numbers are removed as a result of more prime multiples being introduced. These would start from their squares, as prior multiples would be accounted for by previous numbers. (7 from 49, 11 from 121, 13 from 169 etc.).
Happy Priming :)
I only liked this comment for all the complicated things in it
I wonder whether there's a point at which it just makes sense to re-make an entire video? :) But I enjoyed it
🔥 fire.
Have you ever heard that saying, burning down the house............ For the insurance money? That would be one case.
@@Robert_McGarry_Poems Whoa, his videos are insured?
Excellent video!! I never thought about the shape of that 'line' with such a huge numbers taken into consideration, but that is actually a great question!
3:55 No, it‘s not! The probability that a number is prime is 100% if it‘s not a multiple of any number below it except 1. If it is, then the probability is 0%.
I’ve seen a lot of people get worked up about this. Interesting!
Innuendo with the ad in the end. Nice
nitpicks! at 12:46, you use a little o for big O notation - thats kinda confusing because there is a little o notation, which one are you talking about?
all the linked papers in the description use little o, so i'd assume he actually means little o.
@@randomdude9996 Yeah, I think it should be little o. Otherwise there would be no point having 1 in "1 + o(1)", as 1 + O(1) is just the same as O(1).
I'm so disappointed at how good that Rick and Morty joke is, because I don't have any maths friends that would appreciate how spot on it truly is.
@sixequalszero except you already heard the joke
You must be the one to convert them to mathematics.
I really thought Freeze Frame Matt was going to subtly move or even talk back, ala Under Dunn
That "Ooh matrices" at 24:10 was so in-character
You said big-Oh notation at 12:44, but just to clear that is a little-Oh (which is also a type of big-Oh notation), right?
Oops, I just left a comment asking exactly the same thing before seeing this
He later calls it Big-Zero, but then says it gets smaller as x gets bigger, so I think it is supposed to be a Little-Oh
Little O means that your function can’t grow faster than any function even after multiplying the function you are comparing.
In practice, it means that to be little-o of a function means you really grow slower than a family of functions (as opposed to big-o meaning to grow slower than or at the same rate as a family of functions).
I've never been so disappointed in Past Matt. Many thanks to Future Matt for being so awesome.
Past Matt: please take future Matt to one side and have a quiet word with him - things were fine as they were before he kept interrupting!
The moving orange prime gap line plot math thing at the bottom reaches 1951 at the end of the video
Me whenever I remember that 2 is prime:
*You may be on this council, but we do not grant you the rank of master*
Who does number 2 work for?