todo.org

schedule 10/15~11/21

temp

change system message from stdout to stderr

“do heuristic by line twice”

6. After TAAI

交大論文1: An efficient algorithm for solving nonograms

交大論文2: An Efficient Approach to Solving Nonograms

use bitset instead int…

use better procedure

1. implementation

make independent scanner

make scanner interface 10/21

test by original data 10/22

make independent printer

make printer interface 10/28

test by original data 10/29

2. test

test correctness

ONGOING if there’s wrong problem, record and analyze 11/5

test speed

ONGOING calculate worse case and record it 11/5

3. optimization

ONGOING refact 10/15~10/30(simple), 11/5~11/21(combine 2., 3.)

O3 warnings

seperate heuristic and board

ONGOING optimize heuristic 11/5~11/21

ONGOING optimize dfs 11/5~11/21

use row and col to fill

4. analyze

ONGOING 1. nonogram papaers 10/15~10/30

5. misc

check how to use makefile implicit and some techiniques

record

==23312== Conditional jump or move depends on uninitialised value(s) ==23312== at 0x804FA8F: Board::init(int, int, std::vector<std::vector<RowLimit, std::allocator<RowLimit> >, std::allocator<std::vector<RowLimit, std::allocator<RowLimit> > > >, std::vector<std::vector<ColLimit, std::allocator<ColLimit> >, std::allocator<std::vector<ColLimit, std::allocator<ColLimit> > > >, char*, int) (board.h:100) ==23312== by 0x80492F2: main (nonogramReader.h:48) ==23312==

fix updatelimitbylimit_col bug

	15x15	25x25
1		90.08(row) stuck(row+col)
2		0(row) 2.3(row+col)
1~100	0.62
1~200	2.54
1~290	17.6
1000	13.72

need to find why 25x25 would stuck(it’s fine if only use row dfs)

add col dfs

	15x15	25x25
1		stuck
2		stuck
1~100	1.66
1~200	6.96
1~290	41.16
1000	38.32

maybe doheuin… is too slow and backup cost is high

v1.0 Thu Oct 22 11:51:26 2015 +0800

	15x15	25x25
1		771
2		stuck
1~100	4.85
1~200	18.26
1~290	73.04
1000	21.17

commit 9c9165c48 Wed Oct 21 16:58:14 2015 +0800

	15x15	25x25
1		stuck
2		stuck
1~100	4.85
1~200	18.26
1~300	stuck in 275
1000	stuck