In a previous thread entitled "Even More Simultaneous Equation Solutions" I reported the following results from analyzing a problem by Valentin Albillo with three methods on an HP-49 :
INV A * B B/A RREFI noted that the RREF results were the first ones which were better than those I had obtained with a solver on my Model 100. Rodger Rosenbaum responded that he received different results with his HP-49.-1.076 -0.6472 1.00148007239
0.999786 1.00110019 0.999999813411
0.407 -0.7415 1.00131966086
0.144 0.6554 1.00132720057
0.999124 0.9996398 1.00000091517
0.553 1.1592 0.998672819845
0.61 3.5385 0.997223232158
INV A * B B/A RREFAfter some playing around with the numbers I found that I can get a solution which looks very much like Rodger's RREF solution if I use a version of Valentin's problem in which all the elements have been multiplied by ten. The only differences are that my solution has the first element as 1.00124107685 and the seventh element as -4665.78990985 where Rodger's first element is a truncated version of my first element and his seventh element has a typographic error.1.00018 1.00018 1.00124
1.9043 1.9043 -33231081.9872
1.3779 1.3779 4704.81157655
1.00065 1.00065 4678.11919981
0.7689 0.7689 6770866.25288
0.553 0.553 -46565.78990985
-0.785 -0.785 -2226.81759925
When I do a B/A solution using all of the elements of the problem multiplied by a factor of ten I get
1.5286where I note that six of my seven elements are the same as Rodger's if I allow for some truncation, but my elements are in a different order. For the element which is not the same my solution yielded 1.5286 while Rodger's solution yielded 0.553 . To date I have not been able to psych out why the different order of the elements together with one definitely different element occur.
1.00018004
1.9043
1.3779
1.0006513
0.7689
-0.785