This paper describes how to compute algorithmically certain twisted signature invariants of a knot $K$ using twisted Blanchfield forms. An illustration of the algorithm is implemented on $(2,q)$-torus knots. Additionally, using satellite formulas for these invariants, we also show how to obstruct the sliceness of certain iterated torus knots.

We use the cobordism distance on the smooth knot concordance group $\mathcal {C}$ to measure how close two knots are to being linearly dependent. Our measure, $\Delta (\mathcal {K}, \mathcal {J})$, is built by minimizing the cobordism distance between all pairs of knots, $\mathcal {K}'$ and $\mathcal {J}'$, in cyclic subgroups containing $\mathcal {K}$ and $\mathcal {J}$. When made precise, this leads to the definition of the projective space of the concordance group, ${\mathbb P}(\mathcal {C})$, upon which $\Delta $ defines an integer-valued metric. We explore basic properties of ${\mathbb P}(\mathcal {C})$ by using torus knots $T_{2,2k+1}$. Twist knots are used to demonstrate that the natural simplicial complex $\overline {({\mathbb P}(\mathcal {C}), \Delta )}$ associated with the metric space ${\mathbb P}(\mathcal {C})$ is infinite-dimensional.

We give simple homological conditions for a rational homology 3-sphere $Y$ to have infinite order in the rational homology cobordism group $\unicode[STIX]{x1D6E9}_{\mathbb{Q}}^{3}$, and for a collection of rational homology spheres to be linearly independent. These translate immediately to statements about knot concordance when $Y$ is the branched double cover of a knot, recovering some results of Livingston and Naik. The statements depend only on the homology groups of the 3-manifolds, but are proven through an analysis of correction terms and their behavior under connected sums.