What are solved by matrix clocks, but vector clocks cannot?

Question

I understand the need for vector clocks in terms of scalar logic clocks, without providing enough information to determine if there is an update conflict in updating the repository of key values, for example.

But I'm not sure which problem has not yet been solved by a vector clock, and then solved by a more bulky matrix clock?

Answer 1

, , , (== ). , , , , .

- , node . , , . node , node. , , node .

TSAE (timestamped anti-entropy). . 1992 (http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.88.7385&rep=rep1&type=pdf), 5.

Answer 2

( , ), , .

$vt_i [1\ldots n] $ $i $ $vt_i [k] $ $S_i $ $S_k $. "$ $ $k $, $\ ldots $".

$mt_i [1\ldots n, 1\ldots n] $ $S_i $ $mt_i [k, l] $ , $S_i $ $S_k $ $S_l $. "$ $, $k $ $l $, $\ ldots $".

.

[1]:

.

, , , , , , , , , ( ).

, , , .

, $ min_k (mt_i [k, i])\ge t $, $S_i $, $k $ $t $.

$\ le t $ .

[1] Ajay D. Kshemkalyani 2000