You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
$ CSDP_USE_JULIA_LAPACK=false julia -e 'import Pkg; Pkg.add("CSDP"); Pkg.build("CSDP")'
43
+
```
44
+
45
+
Note that if the environment variable `CSDP_USE_JULIA_LAPACK` is not set, it defaults
46
+
to using the system libraries if available and the Julia libraries otherwise.
35
47
36
-
To use CSDP with JuMP v0.18, do
48
+
To use CSDP with JuMP v0.19 and later, do
37
49
```julia
38
50
using JuMP
39
-
model =Model(solver=CSDPSolver())
51
+
model =Model(with_optimizer(CSDP.Optimizer))
40
52
```
41
-
and with JuMP development version, do
53
+
and with JuMP v0.18 and earlier, do
42
54
```julia
43
55
using JuMP
44
-
model =Model(with_optimizer(CSDP.Optimizer))
56
+
model =Model(solver=CSDPSolver())
45
57
```
46
58
47
59
## CSDP problem representation
@@ -146,18 +158,14 @@ For Windows, a pre-compiled DLL is downloaded (unless you configure the `build.j
146
158
147
159
148
160
## Next Steps (TODOs)
161
+
149
162
-[ ] Maybe port `libcsdp` to use 64bit Lapack, aka replace “some `int`s” by `long int` (the variables used in a Lapack call). Started in brach `julias_openblas64`
150
163
-[ ] Maybe think about an own array type to circumvent the 1-index problems in `libcsdp`.
151
164
-[ ] Map Julia's sparse arrays to `sparsematrixblock`.
152
165
-[ ] Upload `libcsdp.dll` for Windows via Appveyor deployment as described at
0 commit comments