I have never understood why the rules/standards for counting turns follow the "range of motion metric" and not the "efficiency metric". (c) Anthony Snyder.
The efficiency metric implies physically minimal time expenditures for executing a solution.
Here we can agree with K. Marx that "any savings are savings in time".
Both a human and a robot can simultaneously rotate any number of faces of a 3x3x3 cube located on one axis.
I would venture to suggest that at the moment the most optimal metric is the one with an unsettled name - the metric of simultaneous axial quarter turn [taking into account WCA Rotations (entire puzzle)]:
Time Robot Metric - TRM (t), defined as follows: quarter turn of any number of faces located on one axis is counted as 1 turn.
Denoted by the lowercase letter t.
The time expenditures for solving a cube in the Time Robot Metric(t) metric are minimal.
For example, turns (RL), (R’L’), (R’L), (RL’), and rotations x, y, z, (y z), (x2 z) be considered one move.
Here is an example of the world record of the robot RustyCuber (0.203s):
y U’D’ L2 U’ B2 U’D’ R2L2 B2 R’ UD F’ D’ B’ L FB RL’ B’ D F (24f, 29q, 22t)
The inspection (y) only takes into account the rotation of the robot's recording camera.
“Optimal solver”: R D B2 U B R2 F’ R D2 F L’ U’ R D R L2 U2 L’ (18f,23q,22t)
In the efficiency metric: D’ L2 B’ U’ F’ U D’ F’ L' D2 F B L2 F L D’ B’ U’ F2 (19f,23q,21t)
The solution of the RustyCuber robot, executed in 10 ms, in the TRM metric is only 1t more than the optimal one, which currently requires significantly more computation time.
Conclusion: The efficiency metric allows us to evaluate the approach of the solution to the optimal one for both the robot and the human. To do this, it is sufficient to subtract the value of the TRM(t) from the value of the QTM(q).
The shortest solutions of the superflip require 24 moves in the QTM:
R' U2 B L' F U' B D F U D' L D2 F' R B' D F' U' B' U D' ( 22f, 24q, 22t)
U R2 F' R D' L B' R U' R U' D F' U' D' B' L' F' B' D' L' (23f, 24q, 22t)
U D' R F' D L' B L' U' R' D' B' U' D L' F D' R B' R U L D B (24f, 24q, 22t)
U D' R F U' D' L D' F R U' R U' D' F U' F L B' U F' B' L B' (24f,24q, 20t)
And in the efficiency metric the solution has fewer moves:
R U D' R F L U' F B R U' F' U' L R2 B F U2 L2 R2 B F (22f, 26q, 19t)
There is another nuance to using this metric when a solution may have different degrees of effectiveness due to the permutation of moves:
D2 R D' R F' L' U R B' U B' D B' U' F R' F2 L2 U2 L2 U' (21f, 26q*, 26t)
vs:
R D' R F' L' U R B' U B' D B' U' F R' F2 L2 U2 L2 U' D2 (21f, 26q*, 25t)
And the icing on the cake - solution of the superflip composed with four spot (26q*):
F B U’ R L’ U’ D R’ F2 L’ R’ U D B’ R2 U D’ F2 L’ U2 D2 (21f, 26q, 18t)
I have not come across the meaning of the G-d number for the axial simultaneous turn metric. Is it time to clarify it?
The efficiency metric implies physically minimal time expenditures for executing a solution.
Here we can agree with K. Marx that "any savings are savings in time".
Both a human and a robot can simultaneously rotate any number of faces of a 3x3x3 cube located on one axis.
I would venture to suggest that at the moment the most optimal metric is the one with an unsettled name - the metric of simultaneous axial quarter turn [taking into account WCA Rotations (entire puzzle)]:
Time Robot Metric - TRM (t), defined as follows: quarter turn of any number of faces located on one axis is counted as 1 turn.
Denoted by the lowercase letter t.
The time expenditures for solving a cube in the Time Robot Metric(t) metric are minimal.
For example, turns (RL), (R’L’), (R’L), (RL’), and rotations x, y, z, (y z), (x2 z) be considered one move.
Here is an example of the world record of the robot RustyCuber (0.203s):
y U’D’ L2 U’ B2 U’D’ R2L2 B2 R’ UD F’ D’ B’ L FB RL’ B’ D F (24f, 29q, 22t)
The inspection (y) only takes into account the rotation of the robot's recording camera.
“Optimal solver”: R D B2 U B R2 F’ R D2 F L’ U’ R D R L2 U2 L’ (18f,23q,22t)
In the efficiency metric: D’ L2 B’ U’ F’ U D’ F’ L' D2 F B L2 F L D’ B’ U’ F2 (19f,23q,21t)
The solution of the RustyCuber robot, executed in 10 ms, in the TRM metric is only 1t more than the optimal one, which currently requires significantly more computation time.
Conclusion: The efficiency metric allows us to evaluate the approach of the solution to the optimal one for both the robot and the human. To do this, it is sufficient to subtract the value of the TRM(t) from the value of the QTM(q).
The shortest solutions of the superflip require 24 moves in the QTM:
R' U2 B L' F U' B D F U D' L D2 F' R B' D F' U' B' U D' ( 22f, 24q, 22t)
U R2 F' R D' L B' R U' R U' D F' U' D' B' L' F' B' D' L' (23f, 24q, 22t)
U D' R F' D L' B L' U' R' D' B' U' D L' F D' R B' R U L D B (24f, 24q, 22t)
U D' R F U' D' L D' F R U' R U' D' F U' F L B' U F' B' L B' (24f,24q, 20t)
And in the efficiency metric the solution has fewer moves:
R U D' R F L U' F B R U' F' U' L R2 B F U2 L2 R2 B F (22f, 26q, 19t)
There is another nuance to using this metric when a solution may have different degrees of effectiveness due to the permutation of moves:
D2 R D' R F' L' U R B' U B' D B' U' F R' F2 L2 U2 L2 U' (21f, 26q*, 26t)
vs:
R D' R F' L' U R B' U B' D B' U' F R' F2 L2 U2 L2 U' D2 (21f, 26q*, 25t)
And the icing on the cake - solution of the superflip composed with four spot (26q*):
F B U’ R L’ U’ D R’ F2 L’ R’ U D B’ R2 U D’ F2 L’ U2 D2 (21f, 26q, 18t)
I have not come across the meaning of the G-d number for the axial simultaneous turn metric. Is it time to clarify it?
Last edited:
