1 | | Good point, will do. |

| 1 | Good point, will do. It's clear for prime_ideals() and hnf_cmp() BUT in the latter case there is already a cmp function for ideals which is very similar but not good for my purposes: it does not first compare norms, and it uses the pari_hnf directly. So The simplest thing to do (replace the existing ideal cmp with mine) may cause a lot of annoying doctest changes. I'll try and see. |

| 2 | |

| 3 | curve_cmp() can be a method of the class for elliptic curves over number fields, or even (almost) for generic elliptic curves as it uses no special number field stuff, just compares the list of a-invariants. Not quite since I replace each ai with its list of coefficients and flatten to get a list of 5*d rational numbers where d is the degree of the field. |

| 4 | |

| 5 | curve_cmp_cm() is a bit of an experiment: for curves with (rational) CM only. |

| 6 | |

| 7 | I really need these comparison functions for ordering of the elliptic curves in a single isogeny class in a deterministic way. For CM isogeny classes it is nicer to group together the curves whose Endomorphism ring is the same (they are all orders in the same imaginary quadratic field, but can have different indices in the maximal order). So curve_cmp_cm() is only really relevant in the context of isogeny classes. Comments? |

| 8 | |

| 9 | Lastly there are a couple of utility functions concerning binary quadratic forms, which I will put elsewhere. |