The research of the group is now focused mainly on the structure and origin of the anomalous properties of liquid water. We have developed a new picture of ambient water based on fluctuations between two types of local structures connected to the anomalous properties of water. These become enhanced upon supercooling where thermodynamic response functions seem to diverge at a temperature of 228 K, i.e. below the temperature of homogeneous ice nucleation. The PI has been awarded an ERC Advanced Grant to develop theoretical techniques to improve simulations of the liquid and explore consequences of a two-state picture in terms of chemistry and consequences for aquatic life. These techniques include an improved force-field description of water, as well as techniques to include the spin-coupling of the protons and real-time TDDFT propagation of both electrons and protons. In a picture of water as locally fluctuating between more compact high-density and more open, tetrahedral conformations the question arises how gases dissolve in the liquid. Does O2 preferentially associate with the low-density regions and how then do fish extract it from the liquid? We work together with experimental colleagues in Japan on computing XES and RIXS of O2 and water and in fibrous structures modeling the interface between water and gills of fish. We perform EXAFS measurements with colleagues at University of Wisconsin at Madison of Argon in water as a proxy for O2. These data are then analyzed based on extensive MD simulations with high-quality force-fields from which we extract structures for EXAFS calculations. The hydrogens/protons in the water molecule are Fermions which leads to different rotational properties of water molecules dependent on the spin-coupling. To include this also the protons need to be included in the wave function description. Here we have extended the LOWDIN code to do non-orthogonal CI based on a sampling of rotational states on a grid to build the correct rovibrational wave functions. The aim is to investigate the effect of spin-coupling in different H-bonding situations. Furthermore, we are implementing real-time (RT) TDDFT simultaneous propagation of the electronic and nuclear density for molecular systems with the deMon2k code. The code is close to production and will be used in large-scale ab initio MD simulations of water without invoking the Born-Oppenheimer approximation. This will be a first-ever investigation of liquid water with proper wave function description of the protons including their overall spin-coupling. Finally, we determine how the local structure of water is reflected in x-ray absorption and emission spectra. Based on QM/MM RASSCF/RASSI for XES and the transition potential DFT approximation for XAS we find significantly more structure in the H-bonding network than given by the best force-fields that are available today. We need to verify this picture using other advanced techniques for spectrum calculations and plan to compute based on Bethe-Salpeter as well as TDDFT with an appropriate functional.