A plane curve defined by an implicit equation
- ,
where F is a smooth function is said to be singular at a point if the Taylor series of F has order at least 2 at this point.
The reason for this is that, in differential calculus, the tangent at the point (x0, y0) of such a curve is defined by the equation
whose left-hand side is the term of degree one of the Taylor expansion. Thus, if this term is zero, the tangent may not be defined in the standard way, either because it does not exist or a special definition must be provided.
In general for a hypersurface
the singular points are those at which all the partial derivatives simultaneously vanish. A general algebraic variety V being defined as the common zeros of several polynomials, the condition on a point P of V to be a singular point is that the Jacobian matrix of the first-order partial derivatives of the polynomials has a rank at P that is lower than the rank at other points of the variety.
Points of V that are not singular are called non-singular or regular. It is always true that almost all points are non-singular, in the sense that the non-singular points form a set that is both open and dense in the variety (for the Zariski topology, as well as for the usual topology, in the case of varieties defined over the complex numbers).[1]
In case of a real variety (that is the set of the points with real coordinates of a variety defined by polynomials with real coefficients), the variety is a manifold near every regular point. But it is important to note that a real variety may be a manifold and have singular points. For example the equation y3 + 2x2y − x4 = 0 defines a real analytic manifold but has a singular point at the origin.[2] This may be explained by saying that the curve has two complex conjugate branches that cut the real branch at the origin.