i have discovered bug in CAS regarding solving limits of type a^(b*x)-c^(d*x) for x->+INF; mainly if b*Log(a) and d*Log© are almost equal - whole Limit changes sign between +/-INF (because if d*Log© > b*Log(a), 2nd exponentiel has a larger dynamis) - my experience is that a some step already after "revert", the solution remains +INF and keeping increasing b will result in crashes of CAS and whole calculator which can (about 50%) then (after reset) result in favorite message "MAIN MEMORIES CLEARED" (so make backup before). the crash is even graphical :-).
try: lim(3^(2x)-2^(3x), x, +INF) ->+INF=OK
try: lim(3^(2x)-2.1^(3x), x, +INF) ->bad response +INF, should be -INF
try: lim(3^(2x)-a^(3x), x, +INF) (2.2<=a<=2.6)->CRASH!
try: lim(3^(2x)-2.7^(3x), x, +INF) ->even syntax error message
try: lim(3^(2x)-2.8^(3x), x, +INF) ->bad response +INF again
you can work-out your own numbers in similar principle, the result will be probably similar (at least the treshold for +/-INF will bw shifted). These types of limits are "broken" on AFX
my rom is version 1.03; what's your experience?
