Hello,
I'm using the VNNLIB-Python library to parse the properties of the smart turn multimodal bench from the vnn comp (https://github.com/VNN-COMP/vnncomp2026_benchmarks/blob/main/benchmarks/smart_turn_multimodal_2026/2.0/vnnlib/instance_0.vnnlib.gz) but it fails on line 17259 because of a scientific notation.
(assert (and (>= X1[0,21,445] -8.915737271308899e-05) (<= X1[0,21,445] 0.09991084039211273)))
vnnlib._core.VNNLibException: Parse error: syntax error at line 17259, column 48: 'e'
The VNNLIB 2.0 standard only says for constants
Constants Numeric constant values may be referenced using basic standard integer or floating point syntax (e.g. 0, 0.0, -0.5).
So I'm not sure if scientific notation is actually allowed or not. If it is could the parser handle it ?
Best,
Augustin.
Hello,
I'm using the VNNLIB-Python library to parse the properties of the smart turn multimodal bench from the vnn comp (https://github.com/VNN-COMP/vnncomp2026_benchmarks/blob/main/benchmarks/smart_turn_multimodal_2026/2.0/vnnlib/instance_0.vnnlib.gz) but it fails on line 17259 because of a scientific notation.
The VNNLIB 2.0 standard only says for constants
So I'm not sure if scientific notation is actually allowed or not. If it is could the parser handle it ?
Best,
Augustin.