| 1 |
|
| 2 |
,-------------------------------------------------------------------------------, |
| 3 |
| This is a very messy and svd-centric TODO. The items predominantly relate to | |
| 4 |
| other graph-related applications and helper applications, mostly *NOT* to mcl.| |
| 5 |
'-------------------------------------------------------------------------------' |
| 6 |
|
| 7 |
- Thread locality, pthread_attr_getaffinity_np, sched_setaffinity |
| 8 |
lstopo numactl --local-alloc |
| 9 |
MCLX_THREAD_POLICY "spread", "compact" |
| 10 |
|
| 11 |
- mclxCopy_loc(mx matrix, mx ordering) |
| 12 |
thread locality-aware copying |
| 13 |
|
| 14 |
$$$$ unify tf(), mcxi, mcxsubs, mcxarray languages |
| 15 |
|
| 16 |
$$? clm info best for regime where clssize ~ node degree? |
| 17 |
|
| 18 |
- mcxarray clean-up .. table output of job division. |
| 19 |
|
| 20 |
$ mcx{load,dump} restrict tab: report intersect and difference (useful sanity-check feedback for user). |
| 21 |
|
| 22 |
$ mcxload etc and sif modes; |
| 23 |
allow value on the source node. |
| 24 |
allow spaces if tabs are present. |
| 25 |
|
| 26 |
- make mcxarray available as library call with callback for pairwise vector comparison |
| 27 |
|
| 28 |
- hidden options in clm close should be part of other clm mode. |
| 29 |
|
| 30 |
- why are mclxSub and mclxChangeDomains not more alike? |
| 31 |
(physical moving of column vectors in the latter). |
| 32 |
|
| 33 |
? matrix read on clustering (intra-edges only -- help --force-connected) |
| 34 |
|
| 35 |
$$$ enable -j 0-2 [one job takes multiple groups] |
| 36 |
|
| 37 |
$$$$ mcxi modality for strict domain identity / subsumption. |
| 38 |
|
| 39 |
$$$ mclExpandVector / mclvKBar should use pval rather than float? |
| 40 |
|
| 41 |
$ -tf add loop transformations. #diag(max, sum, center, laplace) |
| 42 |
|
| 43 |
$ mcxi uses 'int' for integers. Why not long? |
| 44 |
|
| 45 |
$ mcxdump JSON output. |
| 46 |
|
| 47 |
- domain selection in mcxi (skel?) |
| 48 |
|
| 49 |
- skeleton creation in mcx query /mcxi ? |
| 50 |
|
| 51 |
- fast colsizes in mcx query with binary format. |
| 52 |
|
| 53 |
- domain masking in mcxi mask maskc maskv mask |
| 54 |
sel selinv mask maskinv maskr maskc maskrinv? masacinv? too many primitives. cr mask |
| 55 |
|
| 56 |
- mcx query -vary-threshold with 20x21 matrix L == 0 - better NA. |
| 57 |
|
| 58 |
- clm protocols -> generic treatment of thresholding and transformation? |
| 59 |
check mcx/mcl/clm family / mcxio / mclfaq / clmprotocols manual page structure. |
| 60 |
(proposed new) mcledge.azm and (existing) mcxio.azm, mclfamily.azm, sec_networktools.azm |
| 61 |
|
| 62 |
- (clxdo) easy way to generate top and bottom clusterings. |
| 63 |
paste <(seq 0 1 $((10-1))) <(yes 0 | head -n 10) | recol 21 | mcxload -123 - |
| 64 |
|
| 65 |
- re-read functionality (now also with base clustering ....) stick it all in a separate file. |
| 66 |
|
| 67 |
|