October 2020

[0scar]

Originally Posted by SmartMersenne

I really doubt that it will come anywhere close to July20.

I wish I could call you "pessimistic".
Looking at the past three years, Oct20 seems somehow similar to Nov17, another "find-constrained-algorithm" challenge. Nov17 was actually solved by 23 people only.
On the other hand, both Oct20 and July20 deal with linear-recurrenced sequences, candidate solutions can be analyzed with similar linear algebra methods.

About CHAOS, I found it useful to know that the following kind of instruction is permitted:
30 CHAOS 30, 40, 50, ...
so that the control flow graph has an edge from a one-line block to itself.

[Dieter]

Originally Posted by 0scar

I wish I could call you "pessimistic".
Looking at the past three years, Oct20 seems somehow similar to Nov17, another "find-constrained-algorithm" challenge. Nov17 was actually solved by 23 people only.
On the other hand, both Oct20 and July20 deal with linear-recurrenced sequences, candidate solutions can be analyzed with similar linear algebra methods.

About CHAOS, I found it useful to know that the following kind of instruction is permitted:
30 CHAOS 30, 40, 50, ...
so that the control flow graph has an edge from a one-line block to itself.

But that isn't necessary! There are solutions for both problems (0,2,2,... and 1970) without using this trick.

[0scar]

Originally Posted by Dieter

But that isn't necessary! There are solutions for both problems (0,2,2,... and 1970) without using this trick.

I agree at 100% with you.
Had it been necessary, writing it would have been too spoiling.
I only wrote that I found it useful (shorter code, faster reaching 1970)

[Dieter]

Originally Posted by 0scar

I agree at 100% with you.
Had it been necessary, writing it would have been too spoiling.
I only wrote that I found it useful (shorter code, faster reaching 1970)

How fast? My fastest is 1970 = 10th value of the sequence, but I search no more.

[0scar]

Originally Posted by Dieter

How fast? My fastest is 1970 = 10th value of the sequence, but I search no more.

The "trick" itself is just one more degree of freedom in writing code.
Of course I can remove all "tricky" edges from my solutions; the easiest way requires to keep the same number of incoming/outcoming edges by adding suitable nodes as new sources/destinations, which means some more CHAOS lines.
As an example, the length of my 7-step bonus solution grows from 17 to 19, not a problem, but I judge the "tricky" version more efficient.
So far, I found no "untricky" solutions which fit both the 6-step and the 20-line constraints.

[0scar]

I don't fully understand the second block of code within published base solution.
I suppose that new code lines must be inserted as follows:

10 A = a
20 B = b
24 Z = 42
25 JMP_ZERO A 75
30 JMP_ZERO B 80
40 X = A % B
50 A = B
60 B = X
70 JMP 30
75 CHAOS 76, 77, 80
76 CHAOS 75, 77, 80
77 CHAOS 75, 76, 80
80 RETURN A

If a and b are non-negative integers with a >= b, it makes sense.

What about line 24?
I argue that Gadi's program also computes the Answer to Life, the Universe, and Everything.
But it won't be returned before seven and a half million years...

[SmartMersenne]

Originally Posted by 0scar

I don't fully understand the second block of code within published base solution.
I suppose that new code lines must be inserted as follows:



If a and b are non-negative integers with a >= b, it makes sense.

What about line 24?
I argue that Gadi's program also computes the Answer to Life, the Universe, and Everything.
But it won't be returned before seven and a half million years...

The question was stupid to start with. Our goal is to find better and efficient solutions everywhere but in this question, we are intentionally adding stupidity (chaos) to the solution.

[SmartMersenne]

Originally Posted by 0scar

I wish I could call you "pessimistic".
Looking at the past three years, Oct20 seems somehow similar to Nov17, another "find-constrained-algorithm" challenge. Nov17 was actually solved by 23 people only.
On the other hand, both Oct20 and July20 deal with linear-recurrenced sequences, candidate solutions can be analyzed with similar linear algebra methods.

Yup, the number of solvers was barely able to reach 25, so clarification didn't have a special boost.

[0scar]

Originally Posted by SmartMersenne

The question was stupid to start with. Our goal is to find better and efficient solutions everywhere but in this question, we are intentionally adding stupidity (chaos) to the solution.

Originally Posted by SmartMersenne

Yup, the number of solvers was barely able to reach 25, so clarification didn't have a special boost.

As a joke, we could say that few people were stupid enough to solve the puzzle

[Dieter]

Originally Posted by 0scar

I don't fully understand the second block of code within published base solution.
I suppose that new code lines must be inserted as follows:



If a and b are non-negative integers with a >= b, it makes sense.

What about line 24?
I argue that Gadi's program also computes the Answer to Life, the Universe, and Everything.
But it won't be returned before seven and a half million years...

Perhaps line 24 contains a hint for the november challenge...

[0scar]

Originally Posted by Dieter

Perhaps line 24 contains a hint for the november challenge...

I still believe that variable "Z" stands for "Zaphod Beeblebrox" :-)