Editor, Experienced Forum User, Published Author, Skilled player
(1943)
Joined: 6/15/2005
Posts: 3250
I heard about this game early on (as being possibly the best Uzebox game) when Uzebox was first released in BizHawk.
Obviously it obviously looked like a nightmare to optimize. It's great that you got so many people interested in collaborating to make a TAS for this game. Also TASes on co-op modes are way more than twice as good as a TAS on a single-player mode.
I like this page.
Editor, Experienced Forum User, Published Author, Skilled player
(1943)
Joined: 6/15/2005
Posts: 3250
Last time I checked, Youtube's 3D mode is trash. I don't have 3D glasses, so the only way I can try to see it is with cross-eye view, but I couldn't figure out how to enable it in Youtube or if it was even possible.
At least Bizhawk has cross-eye view there (you can set it in VirtualBoy -> Preferences... -> ThreeDeeMode -> SideBySide) but unless you're trained in cross-eye view it's kind of difficult to do sometimes, and even then who knows about the actual art quality of the game itself.
Editor, Experienced Forum User, Published Author, Skilled player
(1943)
Joined: 6/15/2005
Posts: 3250
I think there should be a program these days that is capable of stitching them together (and keeping them at 60fps). Maybe Avidemux? I normally just use AviSynth because I tend to do a lot of post processing.
Editor, Experienced Forum User, Published Author, Skilled player
(1943)
Joined: 6/15/2005
Posts: 3250
Watched it just now. The TAS looks pretty good (I'm not familiar with ZX unfortunately) but the cutscenes are long. Models exist in this game essentially so you can cosplay as Mega Man characters? That's my guess. Looks cool anyway.
Also, nice "Movie finished playing." at 1:08:08. (I don't think DeSmuME AVI recording is supposed to put that there, but what do I know.)
Editor, Experienced Forum User, Published Author, Skilled player
(1943)
Joined: 6/15/2005
Posts: 3250
Encode (720p60 is available, always use 60fps for Mega Man):
Link to video
Looks interesting, seeing how the 2p controller exploits were used in the levels.
Editor, Experienced Forum User, Published Author, Skilled player
(1943)
Joined: 6/15/2005
Posts: 3250
It says in the description that it is a PC game.
Didn't expect otherwise; this is too good to be a NES/FC ROM.
Edit: Also ファミコン風 means Famicom-style which is a big indicator that the game is a PC game; if it were a NES/FC ROM, "Famicom-style" would not be the selling point.
Editor, Experienced Forum User, Published Author, Skilled player
(1943)
Joined: 6/15/2005
Posts: 3250
I haven't been following Pokemon save corruption recently (totally glitched runs just aren't my thing anymore), but I have a question. Did using glitch items (at least one which I believe allows you to execute RAM) ever factor into your TAS? I managed to find one on the JP version, but I can't remember if it was the same for the US version.
Editor, Experienced Forum User, Published Author, Skilled player
(1943)
Joined: 6/15/2005
Posts: 3250
Thanks! At least it's nice to get something submitted after a long time of not doing anything, even if it gets rejected.
Whatever, I was 6 months off. April Fools is overrated anyway. :)
Yeah, I actually hoped it would have ended in the negatives for the elusive one-death game over. Oh well, can't have everything I suppose.
Editor, Experienced Forum User, Published Author, Skilled player
(1943)
Joined: 6/15/2005
Posts: 3250
While -W(-1) is one solution (I'm not going to talk more about this; go see blackpenredpen if you're interested), this reasoning doesn't explain why there aren't any other solutions, since any z such that e^(e^z) = z (not necessarily e^z=z) could be a candidate solution for e^z = ln(z).
Technically, the problem isn't asking for a number. It's asking "Is the teacher correct?". Which is far easier to answer and justify.
This site answers the question, and also gives probabilities (up to n=17), for which the student coming forward would have heads; I see you also posted here going up to n=20.
As far as calculating the exact probability, I don't have a nice way to figure that out. But to answer the original question, "Is the teacher correct?":
No, the student is more likely to have flipped tails. Given that at least one student has neighbors that both flip heads, and a random candidate stepped forward, that student is more likely to have stepped forward the fewer candidates there are, which is more likely when that student has tails. It's related to similar Bayesian paradoxes like the Monty Hall problem and the Boy or Girl paradox.
Editor, Experienced Forum User, Published Author, Skilled player
(1943)
Joined: 6/15/2005
Posts: 3250
Forgot to reply despite watching it a week ago. The game still makes an entertaining TAS even up to now (that's probably why it has a star).
I can't remember what the previous run was like though; it would be nice if there was
a comparison encode.
Editor, Experienced Forum User, Published Author, Skilled player
(1943)
Joined: 6/15/2005
Posts: 3250
I assume by "closed-form expression" you are also excluding infinite sums.
Other than the circle and the degenerate ellipse (minor radius = 0), I don't think it is possible. However there are ways to approximate perimeter of an ellipse using infinite sums.
Editor, Experienced Forum User, Published Author, Skilled player
(1943)
Joined: 6/15/2005
Posts: 3250
I did a very quick (not very optimized) TAS of Crystal Clear 2.0 BABY, using the speedrun route (but not doing the 25 Rare Candies):
Link to video
Possible that Ursaring or some other Pokemon is faster.
Actually I did a TAS of 2.0 Beta a year ago. BABY is faster because you can beat 8 Gym Leaders + E4 to fight the Mt. Silver trainer, whereas in Beta you need 16 Gym Leaders + E4.
Editor, Experienced Forum User, Published Author, Skilled player
(1943)
Joined: 6/15/2005
Posts: 3250
Nice! I was trying to figure out how to get either of the two (there are two of them; one is a reflection of the other) similarity transformations from the affine transformation but couldn't quite figure it out.
By the way, the reasoning I gave for a parabola being a parabola under affine transformation is wrong; I assumed that it would hold for compositions of the transformations if it held for each transformation, which is wrong reasoning. After all, a rectangle remains a rectangle under the transformations I gave, but a rectangle is no longer a rectangle under a skew transformation.
Editor, Experienced Forum User, Published Author, Skilled player
(1943)
Joined: 6/15/2005
Posts: 3250
Looks like you need some room to air-dash onto the top of the door (an overhang or such), then slide down the door and switch to Zero at the last moment. There isn't enough room to do this on the final door leading to a Maverick boss battle. If you can find a glitch that allows this on a final door, then that would definitely save time.
Otherwise, it doesn't seem to save time, not even against Byte.
Editor, Experienced Forum User, Published Author, Skilled player
(1943)
Joined: 6/15/2005
Posts: 3250
y=x^2 and y=x^2/2 are similar; scaling the graph of y=x^2 uniformly by a factor of 2 gives the graph of y=x^2/2.
(It is also true that they are related by an uneven scaling, but that is beside the point.)
Yes, "parabolaness" is preserved under all (bijective) affine transformations; a parabola remains a parabola when you apply uneven x/y scaling or any isometry (rotation, reflection, translation), and since those operations generate the affine transformations, a parabola remains a parabola under any affine transformation.
Editor, Experienced Forum User, Published Author, Skilled player
(1943)
Joined: 6/15/2005
Posts: 3250
I assumed that Warp was referring to uniform scaling, based on what was said:
"scaling it equally in all directions" I assumed meant uniform scaling.
"uneven scaling" I assumed meant stretching x and y by unequal factors.
Edit: Also, I'm pretty sure that if you have uneven scaling together with all the isometries (rotation, reflection, translation), that is enough to give you all the affine transformations. So that probably explains why there isn't a special name for "similarity transformations plus uneven scaling", because that just gives you the affine transformations.
Editor, Experienced Forum User, Published Author, Skilled player
(1943)
Joined: 6/15/2005
Posts: 3250
We know that the sum of the first n terms of the harmonic series is given by:
Hn = ln(n)+γ+O(1/n)
where γ is the Euler-Mascheroni constant, and since the O(1/n) terms go to 0, we have:
lim[n→∞] sum[ i=n-2020 to 2n+2020 ] (1/i)
= lim[n→∞] ln(2n+2020)+γ) - (ln(n-2020)+γ)
= lim[n→∞] ln(2n+2020) - ln(n-2020)
= lim[n→∞] ln( (2n+2020)/(n-2020) )
= ln( lim[n→∞] (2n+2020)/(n-2020) )
= ln(2)
Do you like the harmonic series?
Editor, Experienced Forum User, Published Author, Skilled player
(1943)
Joined: 6/15/2005
Posts: 3250
Cantor's diagonal argument is not that.
Cantor's diagonal argument says only: Given any countable list of real numbers that I so choose, you can always find a real number not in this list, using the diagonal argument. This number you choose is dependent on the list I choose.
Cantor's diagonal argument is not about proving that a specific listing lacks some real number; it is not reducible to replacing every element with a natural number and then showing that there are real numbers that are not in the list. So replacing numbers does make a difference.
For the sake of argument, if it were possible to reduce an argument of uncountability to replacing elements with natural numbers and then showing that a number does not exist in the list, then I could "prove" that the set of rational numbers is uncountable as follows:
Suppose the rational numbers are countable. Then from a countable listing of rational numbers, I can replace every rational number with a natural number. Then in this new list, there is a rational number that does not exist in this list: namely 1/2. This is a "contradiction" and so I have "proved" that the set of rational numbers is uncountable.
That isn't what Cantor's diagonal argument is about.
Editor, Experienced Forum User, Published Author, Skilled player
(1943)
Joined: 6/15/2005
Posts: 3250
I'd say you're both right and wrong at the same time.
Right because, in spirit, you are using a special form of the chain rule:
dy/dx = (dy/du) / (dx/du)
Wrong because, pedagogically, use of differentials automatically makes any argument "wrong".
P.S. Don't take me seriously.
Editor, Experienced Forum User, Published Author, Skilled player
(1943)
Joined: 6/15/2005
Posts: 3250
If the axiom of countable choice (a weaker version of the axiom of choice) holds, then it is possible to index a countably infinite subset of any infinite set just by invoking the axiom, so there cannot be any smaller infinite cardinality than aleph-0.
If we are not allowed to assume any choice axioms, then I think it's up in the air.
The Calkin-Wilf sequence is one well-known way to order the positive rationals. It is defined by the breath-first traversal (level-order traversal) of an infinite binary tree whose root is 1/1 and the children of each node are defined as:
- left child of a/b is a/(a+b),
- right child of a/b is (a+b)/b.
The order of the positive rationals here is 1/1, 1/2, 2/1, 1/3, 3/2, 2/3, 3/1, ...
Editor, Experienced Forum User, Published Author, Skilled player
(1943)
Joined: 6/15/2005
Posts: 3250
I'll provide the answer I came up with. I find it interesting that the harmonic series (1 + 1/2 + 1/3 + ...) is involved in this question.
Notice that, because of the speeds, the hare should always wait until the tortoise is 20% into the race, regardless of how far the road has stretched. (The road stretches uniformly at all times, so the positions will always remain in proportion.) This means that we do not need to know how long it takes for the race to finish, only how long it takes for the tortoise to get to 20%.
To make things simpler, we can consider the road rescaled to 10 miles at all times, in order to rewrite the problem. So instead of the road stretching by 10 miles every minute, we consider the tortoise slowing down every minute, in a harmonic series-like fashion:
- The tortoise drives 1 mile in the first minute, then 1/2 mile in the next, then 1/3 mile in the next, ... How long then will it take the tortoise to get to 2 miles?
Which is a much simpler question:
- After 1 minute, tortoise is at 1 mi
- After 2 minutes, tortoise is at 1 + 1/2 = 3/2
- After 3 minutes, tortoise is at 3/2 + 1/3 = 11/6 = 22/12
- After 4 minutes, tortoise is at 22/12 + 1/4 = 25/12 > 2
So the tortoise reaches the 2-mile mark somewhere between 3 and 4 minutes. Linear interpolation shows that the tortoise reaches 24/12 at two thirds of a minute (40 seconds) after the 3-minute mark. So if the hare wants to finish the race at the exact same time as the tortoise, the hare should begin the race 3m 40s after the tortoise begins.
Note: How do we know the tortoise is able to finish the race, and it won't go on forever? Simple, 1+1/2+1/3+... diverges. So in fact the race could be a million miles long (stretching a million miles each time) and the tortoise will still eventually finish. Though, even for 10 miles it takes a long time (8.588 days, to be precise), and the amount of time required will increase exponentially based on the distance.
An image of the 8.588-day race is shown in the following solver's twitter post (sorry, the "GIF" (if it is even an actual .gif) is embedded so I can't extract it):
https://twitter.com/ali_geeee8/status/1285183324596535298