Solving Large Overdetermined Systems
- To: mathgroup at smc.vnet.net
- Subject: [mg88599] Solving Large Overdetermined Systems
- From: Hauke Reddmann <fc3a501 at uni-hamburg.de>
- Date: Fri, 9 May 2008 03:26:41 -0400 (EDT)
My equation system consists (at the moment - by taking all linear combinations, the cubics surely can be pruned too) of 18 quadratics and 360 cubics in 15 variables. Solve is out of question, for NSolve it's still too large (not mentioning that there are some 1-parametric solutions), so I turned to FindRoot - one found "random" solution would already be very revealing. But alas, FindRoot needs #eq=#var. Bummer. (Of course I can take 15 random equations and plug the solution back in, but the probability of 378 zeroes is zero...) Any other idea what I could try? -- Hauke Reddmann <:-EX8 fc3a501 at uni-hamburg.de Er-a svo gott sem gott kveða öl alda sonum, því að færra veit er fleira drekkur síns til geðs gumi.