Antony Lewis
Antony Lewis
The model is certainly physically wrong in various ways, though not obviously wrong order of magnitude? [detailed_heating_analysis.py](https://github.com/user-attachments/files/22049769/detailed_heating_analysis.py) This is the agent's plotting code, it looked OK at quick glance but...
This enhancement in the non-linear Pk is presumably not physical.
I have submitted a possible gfortran patch (https://gcc.gnu.org/pipermail/gcc-patches/2025-June/686998.html). There's a docker image with the fix build of gfortran at cmbant/docker-gcc-build:gcc15-fix
Now fixed on gcc 16 master branch https://gcc.gnu.org/cgit/gcc/commit/?id=d1f05661fa6c8a6ea6f59ad365a84469100e425e Updated docker cmbant/docker-gcc-build:devel based on latest master.
Should now also be in released gfortran 15.2+, updated conditionals accordingly (https://github.com/cmbant/CAMB/commit/c14f814cf2b8782456b62fb4edc64d93f7504aa5) and closing.
The current behavior seems reasonable. Not clear dict_x_alt has to be defined to be equivalent in all cases; e.g. dict_x could be the correct way to override a dict b...
That would be great. But to avoid bloating the main code, it might be better to make it an external sampler package inheriting from one of the standard classes (i.e....
I now switched to not installing classy on py12/numpy2 tests
The best thing is likely to change the likelihood to e.g. use al0_m1 rather than al0_-1.