Multipolar effective one body waveform model for spin-aligned black hole binaries

We introduce TEOBiResumS_SM, an improved version of the effective-one-body (EOB) waveform model TEOBResumS for spin-aligned, coalescing black hole binaries, that includes subdominant gravitational waveform modes completed through merger and ringdown. Beyond the dominant (?,|m|)=(2,2) one, the more robust multipoles all over the parameter space are: (2, 1), (3, 3), (3, 2), (4, 4), and (5, 5). Modes as (3, 1), (4, 3), and (4, 2) can also be generated, but are less robust. The multipolar ringdown EOB waveform stems from suitably fitting many numerical relativity (NR) waveform data from the Simulating eXtreme Spacetimes (SXS) collaboration together with test-mass waveform data. Mode-mixing effects are not incorporated. The orbital (nonspinning) part of the multipolar waveform amplitudes includes test-mass results up to (relative) 6PN order and, for most modes, is Padé resummed. The m=odd waveform multipoles (up to ?=5) incorporate most of the currently available spin-dependent analytical information. Each multipolar amplitude is additionally orbital-factorized and resummed. Improving on previous work, we confirm that certain m=odd modes, e.g., the (2, 1), and even the (3, 1), may develop a zero (or a minimum) in the amplitude for nearly equal-mass binaries and for several combinations of the individual spins. A remarkable EOB/NR agreement around such zero is found for these modes. The new waveform, and radiation reaction, prompts a new NR-calibration of the spinning sector of the model, done with only 32 datasets. The maximum (2, 2) EOB/NR unfaithfulness F with Advanced LIGO noise against the SXS catalog (∼595 datasets) is always below 0.5% for binaries with total mass M as 10 M?≤M≤200 M?, except for a single outlier with max(F)∼0.85%. When (2, 1), (3, 3) and (4, 4) modes are included, one finds an excellent EOB/NR agreement up to M∼120 M?, above which the performance degrades slightly and moves above 3% We also point out that the EOB dynamics may develop unphysical features for large, antialigned, spins and this may impact the correct construction of the (2, 1) mode in some corners of the parameter space.