.

Yep, works now.

Overly tight error margins: expected 843253.1448715855 to be close to 843253.1448715873 +/- 1e-9

.

https://www.codewars.com/kata/5a77f725b17101edd5000020/ is the same problem with similar complexity requirements, although it only asks for the optimal time, not for an associated solution sequence.

I don't get the test case height: 1, width: 1, y: -4, x: -9, thought: 'Mirrors'

At time t = 0, the text 'M' starts at y = -4, x = -9 and does not appear in the view.

At time t = 1, the text 'Mi' starts at y = -3, x = -8 and does not appear in the view.

At time t = 2, the text 'Mir' starts at y = -2, x = -7 and does not appear in the view.

At time t = 3, the text 'Mirr' starts at y = -1, x = -6 and does not appear in the view.

At time t = 4, the text 'Mirro' starts at y = 0, x = -5 and does not appear in the view.

At time t = 5, y > 0 and the text can no longer enter the view.

Thus, I return 0n, but the test expects 5n. Why?

Both mine and Fbasham's solutions sometimes fail a random test. For example, both of our solutions fail the following test (with equal output):

Block:
[[0,0,3],[1,0,1],[0,1,0]]

Observations:
[["3B",2],["2C",2],["1B",2],["2C",2],["2A",2],["3C",0],["1C",1],["2A",3],["1C",3],["2A",0],["2B",2],["2A",3],["3C",0],["3C",2],["2A",3],["2B",2],["3A",3],["3B",3],["1A",2],["2C",2],["3A",2],["3B",1],["2A",3],["3B",2],["3A",1],["1C",3]]

: expected [ '1A', '2B' ] to deeply equal [ '1A', '2B', '3A' ]

Either our solutions or the author's solution is buggy.

Definitely an issue. Without sufficiently large values of n, a simple for loop will do. However, larger values of n will require values of r increasingly close to 1 to avoid overflow or underflow.

IMHO, this kata would work much better with integer-only inputs, with output taken modulo a large prime. This would eliminate all accuracy concerns.

I believe the reference solution is currently broken. It does not apply the modulus correctly, so it often produces wrong answers for large inputs.