on a pentium 4 using the SE batch mode I get these ratings for
Eioru's puzzles
- Code: Select all
536720080279083060008596372791632854624800793853947126000078030080360017307009048 # 4.5
012000406400602708608040209060504801020060507080907603246070985891256374300400162 # 4.6
172000308046000152305000607007031000000572000000640700708000269564000813209000574 # 4.7
370010046854960120000400000490256010500000004080049050000000000120604080700030001 # 7.1
007200403500000080009008006050600800003000000700102009100040607200001000074900010 # 4.7
000400870000003002000068504000000007025040908080000000700080003094016000030900040 # 5.0
200400030000000560006000100400060001000580200000003805048079000530000000000802000 # 5.0
060000040090204006004009780000000890400308002058000000087100400100906020020000010 # 4.7
000000000001234500030601070073000140040000050018000260020306010009542700000000000 # 4.699999999999999
003000700000020000609080104000308000034000510000205000805090406000060000002000900 # 4.8
000000000001020300043050610000708000079000840000203000064030190008010500000000000 # 4.8999999999999995
100000002030405060000000700007000200000306000008000400002000000090801030600000009 # 4.6
the 4th puzzle difference is suspect,
Eioru 4.6 vs. the batch SE 7.1
the 99999* could be an artifact of the java runtime math implementation
a way around this would be to code SE using an integer rating a multiple of 10 greater
than the listed ratings and use an explicit output format (c-style: printf("%d.%d", irating/10, irating%10);)