Known issues

§ mZeros((x+i)²,x) hangs (Priority 1)

§ SVD does not sort results correctly (1)

§ GMPFMul has a small memory leak (2)

§ MatFunc returns an incorrect result when the matrix is not diagonalizable (2)

§ The Position function returns incorrect results when searching for a symbolic list (2)

§ Cubic(Coef((x-i)³,x)) returns a list of undef’s (2)

§ Minimize is very slow (3)

§ PowerMod(3²,4⁷,6) incorrectly returns 0

§ MultMod(3²,4⁷,4) incorrectly returns 4

§ IncGamma(n,x) often returns incorrect results for complex x

§ NumRoots((x+1)ⁿ,x,-2,0) incorrectly returns 1 for n Î (3)

§ CoefList((a+b×i)×x³- (c+d×i)×x+(e-f×i),x) gives a result with no negative signs (3)

§ Cannot use the variables p, q, s and t in PDESolve, as they are reserved for use by PDESolve. Please use other variable names until this is fixed. (3)

§ Sylvestr cannot handle constant (degree 0) polynomials (3)

§ Laguerre({2,-3},x) returns undef (it should return x²/2+x+1) (3)

§ Gegenbau(2,-3,1) returns 0 (it should return 15) (3)

§ SubRes currently does not return the last subresultant (4)

§ Legendre gives incorrect results for large n, e.g. Legendre(1000,0) returns 0. This is because the calculation hits the maximum representable number (note the warning). (4)

§ Hermite(n,num) and Laguerre(n,num) give Domain Error for most values of num

§ FibNum(5000.) returns -¥

§ FibNum(32768) gives a Domain Error because of an issue with matrix powers in the AMS

§ PsiSum gives an incorrect result with a non-default lower limit for the sum

§ Groebner is currently in beta and is very slow. Groebner seems to spend most of its time in NormalF, so if you can optimize NormalF and/or Groebner, please let me know.

§ RTInt and Horowitz are also in beta, and may give Domain Errors. They currently work best for polynomials and rational functions.

§ Risch sometimes cannot solve transformed integrands when the resultant is constant in q.
Risch is a “framework” for future functionality and may not be particularly useful (in the sense of being able to handle integrals that the built-in symbolic integrator cannot do). Risch also sometimes incorrectly claims that the integral cannot be done in terms of elementary functions (e.g. for ln(x)²).

§ Note: For some bugs listed above, I’m planning to rewrite or significantly modify the function in question in the near future. The next release will be a bugfix-only release.