Talk:Solving coin problems: Difference between revisions

 

"Maxima in Perl?" and "Solving the resulting translation": Solving the equations is auxiliary to the core machine translation task--this is the reason I left that to MAXIMA. Bobrow's original program and Norvig's implementation both code their own "solvers" based on constraint propagation. Norvig's is quite limited and can only solve for one variable. Most "coin problems" found in high school algebra texts require solving a simultaneous system of linear equations for two or three variables, so Norvig's implemenation just won't work here. If there is a module here on rosettacode for solving a system via guassian elimination, determinants, matrix inversion, or any other method, then that would be completely acceptable. Once the core program translates the english problem description to a system of equations, that system can be solved using any method whatsoever, e.g., a Java implementation could call the Java Algebra System (JAS) or it could call Mathematica's Java API. As long as it gets the correct answers to the original english input, anything goes. The only thing I'm not requiring in this task is translating those results back into english language sentences for output--something Bobrow's STUDENT program did. --[[User:2Powers|2Powers]] ([[User talk:2Powers|talk]]) 08:04, 18 May 2013 (UTC)