[ADOL-C] Speed tapeless vs. tape-based

Andrea Walther andrea.walther at uni-paderborn.de
Tue Aug 16 05:22:17 EDT 2011


Hi,

> I am wondering in which situations a tape-based reverse mode can be
> expected to be faster than the tapeless forward mode for calculating
> first derivatives only.

the answer is really: It depends on the specific function evaluation
since especially for the tape-based version the tape access times
are hard to predict.

In the AD theory it is always assumed that these times are
negligible which may or may not hold on the computer.
This might explain the behaviour that you observed.

As far as I know, there are only tests for very special applications.
Hence I would hesitate to generalize them. I am aware of one 9 x 9
Jacobian were the tapeles mode was significant faster.

In theory, for the computation of a Jacobian of a function R^n -> R^m
the forward mode should be faster if m <= 3n/5

Best

Andrea Walther

-- 
Prof. Dr. Andrea Walther
Lehrstuhl fuer Mathematik und ihre Anwendungen
Institut fuer Mathematik
Universitaet Paderborn
Warburger Str. 100
33098 Paderborn

Email: andrea.walther at uni-paderborn.de
Phone: ++49 5251 602721
Fax:   ++49 5251 603728

**********



More information about the ADOL-C mailing list