CCi known issues list

[MichaelABurr]

• Empty intervals in RRi: (interval(2,-2))^2 = interval(4,4).
• Matrices (determinants, arithmetic operations)
• Polynomial rings (interval*variable)
• Missing ==
• determinant, this produces a strange last coordinate (all else is right): det matrix {{interval(1+2ii,3+4ii)}}
• larger determinants aren't working.
• Some new error in documentation (fixed -- Doug)

[MichaelABurr]

Pushed:

• Polynomial rings (interval*variable) is fixed.
• Maybe fixed inverses and determinants of interval matrices? At least fixed an error in multiplying CCi's in engine.
• Definitely fixed determinants (for 1-1 matrices). The problem was that several inputs were passed by reference, so when two were the same, changes to one also changed the other.
• Determinants not fixed (or M2 is using an unusual algorithm) for 2-2 matrices
• Larger determinants are working, but not necessarily optimally. They produce large regions (perhaps due to some sort of partial pivoting), but, as far as I can tell, they contain the determinant.
• I added == for CCi's , but there's an issue. toCCi(100,1/3,0) == toCCi(1/3,0) returns true, when it should return false. I think that what's happening is that the 1/3 is being rounded to 53 bits of precision too early, but I can't find where that happens.

[Rick3yHuang]

• ~~ interval(1*ii,-ii) returns [1,-1] (an empty interval)*ii ~~
• ~~interval(0,ii) returns [0,1]*ii, but interval(-ii,ii) returns an error ~~
• Should we have random(CCi) ?

[MichaelABurr]

• interval(1*ii,-ii) returns [1,-1] (an empty interval)*ii
  This is not an issue since interval should take the lower left corner of a box to the upper right corner of a box.

• interval(0,ii) returns [0,1]*ii, but interval(-ii,ii) returns an error
  This is an issue since it should work. Let's look at this today.

• Should we have random(CCi) ?
  We don't have a random(RRi), but we could. What behavior would you suggest for these functions?