Research Statement

Hiroyuki Chihara

21 November 2026

My main research field is mathematics, in particular, mathematical analysis. I will state my past research, current research, and future research.

Past Research (1993-2019)

I split this paragraph into four parts.When I was a graduate student, I joined the research of the mathematical analysis of the initial value problem for dispersive partial differential equations on the Euclidean space. We say that the evolution equation is dispersive if the highest order of differentiation in time variable is strictly smaller than that in space variables, and its initial value problem is solved in both directions in time. The Schrödinger evolution equations in quantum mechanics and the KdV equation in classical mechanics are typical examples of dispersive PDEs.

Initial value problem for semilinear Schrödinger equations (1993-2009): I first studied the IVPs for semilinear Schrödinger evolution equations whose solutions do not satisfy the classical energy estimates due to the appearance of the bad first order terms ([1,2,3,4,5,6,7,12]). Based on pseudodifferential calculus, I established time-local existence theorem using the theory of linear Schödinger evolution equations and my own idea of diagonalization ([2,7]). This idea was highly rated since this paper solved difficult open problem at that time. Furthermore I also studied time-global existence theorem for small data ([3,4,5]), gain of regularity according to the decay of the data ([7,12]), and other model equations arising in classical mechanics ([6]).

Linear Dispersive PDEs (2000-2005): After that I mainly studied the local smoothing effect of linear dispersive equations and related topics ([8,9,10,11]). In case of the Schrödinger evolution equation of a free particle on the Euclidean space, the solution gains extra smoothness of order 1/2 locally in the space variables. If all the Hamilton flow generated by the principal symbol of the equation go to infinity, then the local smoothing effect occurs. In [8] I studied the local smoothing effect of dispersive pseudodifferential equations of order \(m>1\) on the Euclidean space. On one hand I studied the variable coefficient case under so-called the non-trapping condition of the Hamilton flow generated by the principal symbol making full use of pseudodifferential calculus. On the other hand I studied the constant coefficient case without lower order terms, and obtained the final results for this case using my own idea of decomposition of the frequency space. [8] is the most quoted paper in my papers. In [9] I studied the initial value problem for linear Schrödinger evolution equations on the torus. In this case smoothing effect never occurs and the restriction on the bad first order terms becomes stronger. So I proved that the initial value problem is \(L^2\)-well-posed, which is the initial value problem possesses a unique solution in the framework of \(L^2\), if and only if the real vector field of the bad first order terms has a scalar potential.

Geometric Analysis of Dispersive Flows (2009-2015): The motion of vortex filaments and the Heisenberg spin chain systems is the motion of curves on the 2-sphere, and formulated as some nonlinear dispersive systems from the real line to the 2-sphere. In mathematics these models are generalized to some dispersive equations describing the motion of curves on compact Kähler manifolds. In particular the second order equation is called the Schrödinger map equation and their solutions are called Schrödinger flows. I studied the dispersive flows on compact almost Hermitian manifolds to understand the effect of the Kähler property in terms of linear partial differential equations ([14,15,16,17]). The most significant results is given in [15], where I established the time-local existence theorem of Shrödinger map equations from closed Riemannian manifolds to compact almost Hermitian manifolds. In this case the classical energy estimates do not hold due to the lack of the Kähler property. I succeeded in removing this obstruction by firstly introducing pseudodifferential operators acting on the sections of pullback bundles defined by a mapping between manifolds. I overcame several difficulties for this by using partitions of unity on both manifolds and some other ideas.

Some Topics in Functional Analysis (2009-2019): In 2007 I learned Sjöstrand’s analytic microlocal analysis on the Euclidean space. The basic idea of this is to convert functions on the Euclidean space to entire functions of the Segal-Bargmann space using some kind of Fourier integral operators with complex-valued phase functions. I studied some problems in functional analysis related to this framework ([13,18,19,20,21]). I omit the detail.

Current Research (2020-)

Since 2020 I have been working on geometric tomography and microlocal analysis in integral geometry ([22,23,24,25,26,27,28]).

Streaking Artifacts: [25] and [27] are concerned with streaking artifacts in CT images. This is caused by beam hardening effect of the energy levels of X-rays when a human body contains metal regions such as dental implants and stents in blood vessels. Park, Choi and Seo first studied this phenomenon for X-ray transform on the plane containing finite number of convex metal regions, and proved that this is the propagation of microlocal singularities along the common tangent lines of the metal regions (Comm. Pure Appl. Math., 2017). Paracios, Uhlmann and Wang proved that the metal streaking artifacts are conormal distributions whose singular supports are common tangent lines. In [25] I studied this for the d-plane transform on the \(n\)-dimensional Euclidean space with \(n \geqq 2\) and \(1 \leqq d \leqq n-1\), and obtained the generalization, that is, the metal streaking artifacts are conormal distributions whose singular supports are conic hypersurfaces consisting of common tangent lines. Unfortunately, however, I could not understand some meanings of lemmas and theorems in [25]. So in [27] I studied geodesic X-ray transform on compact, simple (no conjugate point) and nontrapping Riemannian manifolds with strictly convex boundary. I proved the same results as in [25] and could understand all the questions about the geometric meanings arising in [25]. For example, I found that all the Jacobi fields along any common tangent geodesic implicitly play a crucial role is describing microlocal singularities in the space of all the normal geodesics. For this purpose I supposed that the manifold is two-dimensional or a space of constant curvature. This condition guarantees that all the Jacobi fields can be written as the product of a scalar function and a parallel transport of a tangent vector. I think that we cannot derive similar results without this strong condition, and that the result of [27] is the final one for the streaking artifacts for geodesic X-ray transform.

Double Fibration Transforms: It is very easy to learn the inversion formula of the X-ray transform on \(\mathbb{R}^2\), but it is very hard to understand the reason why the inversion formula holds. Probably it is impossible to do it in the Euclidean space setting. The reason is that the Euclidean space has no conjugate points as a Riemannian manifold. More precisely, if a Riemannian manifold has no conjugate points, then the canonical relation of the X-ray transform \(\mathcal{X}\) satisfies so-called the Bolker condition, and the normal operator \(\mathcal{X}^T\circ\mathcal{X}\) becomes an elliptic pseudodifferential operator of order \(-1\), and has a parametrix (approximated inverse operator). If a Riemannian manifold admits conjugate points, what happens the invertibility of \(\mathcal{X}\)? Holman and Uhlmann (J. Diff. Geom., 2018) studied this problem. Under some condition on the distribution of conjugate points with the same orders, they proved that the normal operator \(\mathcal{X}^T\circ\mathcal{X}\) splits into an elliptic pseudodifferential operator and some Fourier integral operators of different orders determined by the order of conjugate pairs. In [26] I studied the same problem for an integral transform defined by a double fibration Z, which is an embedded submanifold of a product manifold \(\mathcal{G}\times{X}\) satisfying some conditions. There are many examples of this type integral transforms: \(d\)-plane transform on the Euclidean space, geodesic X-ray transform, geodesic Radon transform, light-ray transform, many kinds of ray transforms determined by null bicharacteristic flows of real principal-type pseudodifferential operators, and etc. In case of the geodesic X-ray transform, \(\mathcal{G}\) is the set of all normal geodesics, \(X\) is a Riemannian manifold, and \(Z\) is the set of all the pairs consisting of each geodesic and its all points. In this setting \(X\) does not have a metric and there are many differences between this setting and the case of geodesic X-ray transform. I proved exactly the same results of Holman-Uhlmann in this general setting in [26].

Future Research

I state some concrete research plans in my sight at this time.

That is all.