Parallel, 'large', dense matrix problems: Application to 3D sequential integrated inversion of seismological and gravity data

► Large dense matrices are easily handled with parallel algorithms. ► Optimization of gravity anomalies with LAPACK and ScaLAPACK routines. ► Parallel efficiency up to 75% for matrix size 170,255×170,016 (215.66 Gb). ► Density values obtained with a different number of processors have a maximum std of 10−9 kg/m3. ► The 3D high resolution density model beneath the European continent was recovered in 39 min, 14 s.