HiLander | the derivative, for example, is a formally defined operator so the manipulations that lead to a closed form for the fibonacci numbers, for example, require no convergence |

euclidean | HiLander: I see.. but what about using the Maclaurin formula a_n = d^n f(0) / dx^n * 1/n! ? HiLander: can that be made formally? |

HiLander | if the resulting function is analytic, then yes, it'll have to work. the maclaurin formula for the coefficients is basically applying the formal derivative to the series and then noting that x=0 drops every term but the constant but, in the context of generating functions (or, as Cale prefers, generating series), whether or not the "function" exists other than at x=0 is basically never taken into account |

Cale | Which is why they're better referred to as series :) |

HiLander | haha, i knew i could summon him with only a word (granted, enumerative combinatorics isn't the area of combinatorics i study, so i may have misspoken, but i'm fairly sure i'm mostly right) |

Cale | You're really doing work in a ring of formal power series, and not usually concerned with convergence in the usual sense. There are some areas which attempt to make use of analytic convergence when it exists. I don't know much about them though. |

Parma_Quendion | wow, I've been spamming people with my away message overnight |

action | Parma_Quendion dehighlights "quen" |

Parma_Quendion | dehighlights "quen" |

Vivan | How solve the int[1/(x^2 + a)^(3/2)*dx] without a table...? |

euclidean | HiLander, Cale: Thanks for your help. I'm glad I can make all those computations without worrying about convergence :) |

evariste | but they're common thugs |

FatalError | Vivan, trig substitution |

futurist | "a_n / 1 + a_n _< ln(1+a_n) _< a_n by looking at the lower and upper sum for log(1+a_n) = INT(from 1 to 1+a_n)(1/x)dx" i don't understand why this is true.. |

BoundingBox | what gradient should be provided in an area where there is no such? |

nephlim | hey folks anyone awake? |

Mulder5 | chances rae, life exists |

nephlim | if I were doing an ANOVA to test differences between two groups, would n for each group need to be the same |

FatalError | I really need to stop IRCing in my sleep |

action | Mulder5 cant remember his stats |

Mulder5 | cant remember his stats |

action | nephlim neither |

nephlim | neither |

Kampen | nephlim n doesn't need to be the same |

nephlim | i don't want to change the numbers, because in this case, the males outnumber the females 2:1. but i'm not sure if the diff in n would skew the ANOVA results thanks Kampen |

Kampen | skew is what you're checking for, in a sense you're looking at variance, right? that's what anova is |

nephlim | i mean.... botch, screw with right all the examples in my SPSS book have equal n |

Kampen | what exactly are you trying to do? probably for simplicity |

nephlim | i'm doing an ANOVA to test for varaince between the sexes with the accruacy of their self diagnosis vs what i actually found in the lab |

Kampen | two-way anova? |

nephlim | so a list of male and female patients |