phantom and Amuse unit scaling factors don't match

[Yrisch]

Describe the bug

During the start of Phantom, The scaling factor initialising does not take the converter value.. It always chooses 1 Myr in second 1 Msun and the the corresponding length to match G=1 instead. It does not interfere with the conversion between Amuse and Phantom but if I try to use physical units in Phantom, Particles pos/vel/internal energy will be in the wrong scaling.

Ps : Strangely, the set_unit is correctly made before the call of set_unit_length/mass/time. I can't find these three callsl in the source..

To Reproduce Steps to reproduce the behavior: The converter is fixed in the Phantom interface with values (0.1 pc ,1Msun, and the corresponding time to match G=1) Setup a Phantom object. during the gas_particles.add_particles the code run setters for ulength,umass,utime. you'll see the result in the Phantom log file.

Ps : add "print*,utime,udist,umass" in amuse_set_unit_length before any change.

Expected behavior the setter should set the right value.

Logs

Using NATIVE inverse sqrt

Running on AMD Ryzen 7 4800HS with Radeon Graphics 16 cpus @ 3.83 Ghz, cache size 512 KB Running in openMP on 16 threads dtmax: 5.3012566233989994E-004 Ps : print of udist/time/mass in set_unit_length : 3.0860000000000000E+017 14881140623620.354 1.9891000000000000E+033 set_unit_length called: utime/mass/dist: 14881140623620.354 1.9891000000000000E+033 5.0936369248881318E+017 set_unit_mass called: utime/mass/dist: 14881140623620.354 1.9889199999999999E+033 3.0859069098660026E+017 set_unit_time called: utime/mass/dist: 31557600000000.000 1.9889199999999999E+033 5.0936369248881293E+017

Environment (please complete the following information):

• OS and version: [e.g. popOS 20.4]
• Compiler: [e.g. gcc9.4]

[Yrisch]

Erratum : the Post-scriptum set_units is made manually with the good value in amuse_initialize_code, so not related with the converter value.

[Yrisch]

The fixed values correspond to the default values given in the Interface.py. So in short, in the phantomcode.gas_particles.add_particles method there are 3 calls of set_unit but without the converter values in arguments. The default ones are chosen instead. Resulting in a mismatch between the Amuse converter values and the Phantom values.

[rieder]

Yes you're right. Phantom (for now) uses fixed units and should not be called with a converter. This is a bug - it either should accept and use a converter or it should not accept one.

The reason for this is that Phantom internally uses fixed units for the cooling module - using it with a converter and different units resulted in wrong behaviour. I haven't gotten around to fixing this.

[rieder]

Before this bug is resolved, at least the following needs to happen:

• [ ] Write a test that runs Phantom without and with (h2) cooling
• [ ] Test with two different converters passed to Phantom
• [ ] Verify the answers are the same (within N decimals) regardless of the converter