PMA !!y5/EVb5KZI ID: f1b924 April 30, 2018, 11:58 p.m. No.5728   🗄️.is 🔗kun   >>5729 >>5733 >>5762 >>5814

>>5727

Excellent work, VA!

 

First pic attached is a revised layout of the f-2 portion of the c=6017 small square originally posted at >>5716.

 

The f-2 value of 132 has been divided into 8 symmetrical triangles with 4 remainder squares in light blue. The remainder of 4 coming from f-2 mod 8.

 

The inner blue square represents the smaller staircase number of 6 (or T(3)), while the outside dark blue is the larger staircase number of 10 (or T(4)). And again, the single orange square in the middle. Total of all squares is still f-1 = 133.

 

The second image is a quick calculation based on using these two staircase numbers as the triangular bases for the calculations originally explored in >>5525.

 

It appears that we can get very close to the solution in 5 steps.

 

Still need to determine a formula for splitting f into two staircase numbers and if this methodology applies to different test cases.

PMA !!y5/EVb5KZI ID: f1b924 May 3, 2018, 12:01 a.m. No.5746   🗄️.is 🔗kun   >>5747 >>5749 >>5757 >>5762

>>5690

Have been focusing on the following hints from VQC's recent post.

 

>For any idea, ensure that it scales.

 

>Using an arbitrary divisor for f, then each of the eight triangles will have one OR one of two … configurations in each triangle.

>The difference between those two configurations of a portion of f are that they are both staircase numbers where the base of one is a unit longer than the other.

 

The quickest iterative search so far has been using 2 different sized triangles. (see >>5522, >>5522, >>5525 for some background)

 

The problem here has been determining the different u values to use, and perhaps we now have a bit of a clue.

 

First pic attached is a spreadsheet calculation for c=14904371 showing the solution in 18 steps using the dual triangle approach. Previous best result using f-2 iterative search and valid n jumps was 1083 iterations. So a pretty good performance bump.

 

In this test case, the u1 and u2 values of 271 and 84 were chosen by trial and error. But there may be a way to get to those values - which is the reason for the post.

 

The second pic is the square layout for f-1. It is broken into 8 triangles of 368, 4 remainder squares, and 1 square in the middle. 1 + 8*368 + 4 = 2949 == f - 1.

 

One of the triangles has been subdivided into f-2 div 40 chunks of 73, with 2 separate remainder areas of 12 and 64, to try and figure out where these two configurations come from.

 

For the working solution:

 

u1 = 271 == (1+2) * 73 + 52 ???

u2 = 84 == (0+1) * 73 + 11 ???

 

The staircase numbers may be the factors that are applied to the arbitrary divisor of f (40 in this case). (0+1) and (1+2) in my guestimate.

 

But then where do the differences of 52 and 11 come from?

And why is the solution found with only 4 chunks of 73, when there are 5 available in the 368 triangle?

 

This may all be coincidental, of course, as these large u values would prevent this solution from working with small values of n.

PMA !!y5/EVb5KZI ID: f1b924 May 3, 2018, 6:47 p.m. No.5749   🗄️.is 🔗kun   >>5762

>>5746

A few visuals of what the full squares look like for the f-1 portion of c=14904371.

 

Color coding matches the single triangle breakdown posted in >>5747.

 

The darker center portions of the u2 and u1 images represent the "starting" values based on the staircase number * (f-2 div 40) chunks. The lighter area represents the "remainder" that I questioned earlier.

 

The "combined" image shows how the u1 and u2 squares mostly fit the starting f-1 original square.

 

It's possible that these 2 squares need to be searched for independently and then used to find the final answer.