-
VPLanet would probably be ~10x faster, and a more flexible if it used a production quality ODE solver. I think [LSODA (this is link to C version)](https://github.com/sdwfrost/liblsoda) is a good optio…
-
PyTorch Nightly (today)
Fedora 31
Anaconda Python 3.7
GPU: 970 (staying under 3.5 GB limit, last 0.5GB is lower quality memory)
Not completely confident it's a bug, if it isn't there should cert…
ghost updated
4 years ago
-
Updating the [`run_it_simple.jl`](https://github.com/pohlan/SimpleSheet.jl/blob/main/run_it_simple.jl) script to reproduce the diffeq script results. Works - as long as one set `e_v = 1`.
Added [`r…
-
The time-varying forcings are not being evaluated at the proper times at each of the RK4 stages. They are evaluated outside the RK4 stage loop and kept constant throughout. In practice, this is probab…
-
It is executing okay but validation is failing.
xylar updated
6 months ago
-
We want to remove the solution array and instead only store two "rows" at a time inside GeoAc_Propagate_RK4(). Steps:
1) We begin with the initial row, which GeoAc_SetInitialConditions() (GeoAc/GeoAc.…
-
Dear Juho,
My name is Anthony Hauser. I am working at the Institute of Social and Preventive Medicine (ISPM) in Bern, Switzerland, with Julien Riou. A few weeks ago, you applied some faster ODE-solve…
-
### Label
gempyor
### Priority Label
low priority
### Is your feature request related to a problem? Please describe.
Currently the function to plot compartments is attached to the `Compartments` …
-
Finish the RK4 numerical integrator
-
- numerical ode:
```
sage: y = var('y')
sage: desolve_rk4(x,y,ics=[0,0],end_points=1,step=0.5) # solution is x^2/2
[[0, 0], [0.5, 0.125], [1.0, 0.4999999999999999]]
sage: desolve_rk4(x*unit_step(1+…