Deep Mandelbrot Set Zoom 10^275

[Mrz00m]

<a href="http://www.youtube.com/v/0jGaio87u3A&rel=1&fs=1&hd=1" target="_blank">http://www.youtube.com/v/0jGaio87u3A&rel=1&fs=1&hd=1</a>

[cKleinhuis]

lol, another deepest zoom, sorry but 275 is by far not the deepest, it is a nice anim, but i prefer the more interesting locations,
try zooming into the elephant valley, like pauldelbrot is doing it in the mandelbrot safari, there is stuff to explore

[Dinkydau]

Oh, that one. It's way overrated. You can render a zoom deeper than this in a week with fractal extreme.

I think the deepest one was around 2^1700 or something, but low-res and extremely boring.

[Mrz00m]

deep yes, i figured it was fairly tight and well arranged w gd music.

[cKleinhuis]

i edited your post, it is enough to just paste the youtube url without the embedding code

[SeryZone]

Maybe, that's nobody read, but I got information, that this zoom has never been a record, because it uploaded at 26.01.2010, but phaumann's zoom 10^333 has uploaded at 24.05.2009.

Orson Wang wrote that 'he rendered it for 6 months', he started to render at ~July 2009, where 10^333 was uploaded...

<a href="http://www.youtube.com/v/x6DD1k4BAUg&rel=1&fs=1&hd=1" target="_blank">http://www.youtube.com/v/x6DD1k4BAUg&rel=1&fs=1&hd=1</a>

[Kalles Fraktaler]

Maybe, that's nobody read, but I got information, that this zoom has never been a record, because it uploaded at 26.01.2010, but phaumann's zoom 10^333 has uploaded at 24.05.2009.

Orson Wang wrote that 'he rendered it for 6 months', he started to render at ~July 2009, where 10^333 was uploaded...

I found an easy way to extend the maximum depth of Kalles Fraktaler to whatever. I read that increasing the stack size of a program would make it much slower, but from my tests so it far don't get noticeable slower with a stack that allows e101000. But I estimate that the easiest location, if it requires a million iterations, would take more than 25 hours just to calculate the reference. And when exploring to the minibrot the reference would need to be calculated MANY MANY times. And a movie with extreme speed would be 3 hours. Are you interested?

Here is a deep zoom in the elephants
<a href="http://www.youtube.com/v/_8lGiZ0yAzI&rel=1&fs=1&hd=1" target="_blank">http://www.youtube.com/v/_8lGiZ0yAzI&rel=1&fs=1&hd=1</a>

[Dinkydau]

I believe there is another way to find minibrots other than exploring, but I don't know how it works. It's more technical.

[youhn]

Polynomial root finding.

https://math.stackexchange.com/questions/404066/help-locating-mini-mandelbrots

Never done it though, so please don't ask the hard questions. 

[Kalles Fraktaler]

Polynomial root finding.

https://math.stackexchange.com/questions/404066/help-locating-mini-mandelbrots

Never done it though, so please don't ask the hard questions. 

Looks like having to solve huge high degree polynoms with the same precision as calculating the reference point.
Roots to high degree polynoms are usually found with the Newton Raphson method, which is also an iterative method.
And the polynom would have the same degree value as the number of iterations.
I doubt it would take less efforts...

[SeryZone]

I found an easy way to extend the maximum depth of Kalles Fraktaler to whatever. I read that increasing the stack size of a program would make it much slower, but from my tests so it far don't get noticeable slower with a stack that allows e101000. But I estimate that the easiest location, if it requires a million iterations, would take more than 25 hours just to calculate the reference. And when exploring to the minibrot the reference would need to be calculated MANY MANY times. And a movie with extreme speed would be 3 hours. Are you interested?

Here is a deep zoom in the elephants
<a href="http://www.youtube.com/v/_8lGiZ0yAzI&rel=1&fs=1&hd=1" target="_blank">http://www.youtube.com/v/_8lGiZ0yAzI&rel=1&fs=1&hd=1</a>

Very nice video, Kalle! Now, even 10^30000 is impossible to explore, but NOW, when limit is 10^101000, we should only optimize algorithms! I started to understand your 'CDecNumber' - that's very nice!!!

[Kalles Fraktaler]

Very nice video, Kalle! Now, even 10^30000 is impossible to explore, but NOW, when limit is 10^101000, we should only optimize algorithms! I started to understand your 'CDecNumber' - that's very nice!!!

Thanks!
The CDecNumber class is not used to calculate the references, it is only used to perform division to calculate the Magnification. It was created by IBM, I just added a wrapper class.
Instead CFixedFloat is the class that calculate the references, it is much faster than CDecNumber but it might be possible to optimize it further.
It should be based on 2 instead of 10, so that bit-wise operations can be used instead of division and modulus.
And it should be possible to speed it up significantly with SSE.

I have attached a deep zoom image. It should be the deepest image ever so far?
Position:
Re: -1.97
Im: 0
Magnification: 1e30000
Max iteration: 200'000
Computation time: 4hrs.

[SeryZone]

Thanks!
The CDecNumber class is not used to calculate the references, it is only used to perform division to calculate the Magnification. It was created by IBM, I just added a wrapper class.
Instead CFixedFloat is the class that calculate the references, it is much faster than CDecNumber but it might be possible to optimize it further.
It should be based on 2 instead of 10, so that bit-wise operations can be used instead of division and modulus.
And it should be possible to speed it up significantly with SSE.

I have attached a deep zoom image. It should be the deepest image ever so far?
Position:
Re: -1.97
Im: 0
Magnification: 1e30000
Max iteration: 200'000
Computation time: 4hrs.

e30000... You want to render another deepest zoom???

I happy to read this!!! your function calculate reference() do this or Calculate approx.()
Please, answer...

[Kalles Fraktaler]

e30000... You want to render another deepest zoom???

No. It took about a month to just find the e10000 minibrot...

I happy to read this!!! your function calculate reference() do this or Calculate approx.()
Please, answer...

CalculateReference calculates the reference.
CalculateApproximation calculates the series approximation.

[SeryZone]

No. It took about a month to just find the e10000 minibrot...CalculateReference calculates the reference.
CalculateApproximation calculates the series approximation.

xaxaxaxa))) Answer like a question xDDD Thanks, that's FixedFloat take less code than CDecNumber. Now I learn avx and sse (many data) and your code - classes.