jerviedog

`sc.tl.rank_genes_groups(adata, 'leiden', method='logreg',max_iter=1000)` worked.

> Same issue here. I solved the problem by replacing readsdmx::read_sdmx with rsdmx::readSDMX in the function get_data_structure. Solution confirmed:)

I am learning optimization, so I don't know what the result mean. But the csdp result said it solved your problem. Here is the output: csdp input.txt ``` CSDP 6.2.0...

But your primal-dual gap was quite small though.

FYI: My compilation process was: 1. modified the Makefile: export CFLAGS=-m64 -march=native -mtune=native -Ofast -fopenmp -ansi -Wall -DBIT64 -DUSEOPENMP -DSETNUMTHREADS -DUSESIGTERM -DUSEGETTIME -I../include export LIBS=-L../lib -lsdp -Wl,--no-as-needed -lmkl_intel_lp64 -lmkl_intel_thread -lmkl_core...