Numerical Algorithms for Nonlinear Dynamics
&
Existing Program Implementations

Software for bifurcation analysis
CANDYS/QA: [Un]necessary details

SGTND main    : :    Algorithms and Programms    : :    Software for bifurcation analysis    : :    CANDYS/QA    : :    Unnecessary details

russian      noframes
frames  

Analytical possibilities of СANDYS are summarized, as far as it was possible (in my restricted understanding) in tables (what-it-can и usability-and-requirements).

СANDYS/QA, in contrast with AUTO and CONTENT, uses Pioncare map for computing and continuation of the limit cycles. There are several options for choosing Poincare section in phase space.

Genuine "feature" of СANDYS/QA is, certainly, interface. The system of the nested menus is used, all this in a text (console) regime, with a set of commands being given from the keyboard, those commands being absolutely counter intuitive. Fortunately, the number of commands is not so big and it is possible to remember them, but firstly it is necessary to sit with a manual in hands.

One more peculiarity is a conceptually thought-out approach to representation of process of a solution in a graph form. User should get into the way the program functionality is organized - it seems to be unnecessarily complicated, especially after CONTENT, but this is kind of logic, maybe. Kinda German ordnung.

A case from my practice. I was studying my test problem (the system of six ODEs - all the same coupled Rossler systems) and have computed a cascade of five period doublings via variation of one of parameters. Then I wanted to perform path-following of few of these period doubling bifurcation points and issued a command to add one more active parameter. The program unexpectedly hang-up for a half an hour (it was Pentium IV 3000) and I've decided already, that somewhere there was a cycling, but, suddenly, it has produced something. It appeared, that CANDYS without any warnings was counting initial vectors for continuation on new parameter of all ( all !!! ) codimention-1 bifurcation points found in cut on one parameter. And, incidentally, it has began from the most deep level, and did not stop until has not finished its job (it takes an hour, or more). On the one hand, it is, certainly, a good thing - afterwards I have used some of these starting points. But the majority of them was not necessary for me - and I think this was a waste of time.