Roman Fedorov
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Roman Fedorov has a total of 15 co-authors. Here are the top co-authors by number of shared publications:
- Yan Soibelman
Department of Mathematics, Kansas State University, Manhattan, Ks., U.S.A.
4 shared publications - Alexander Soibelman
USC Dornsife
2 shared publications - Alexander Soibelman2 shared publications
- Anna S. Zhuk
Institute of Applied Computer Science, ITMO University , St. Petersburg 197101
1 shared publication - Artem G. Lada
Department of Microbiology and Molecular Genetics, University of California, Davis, CA 95616, USA
1 shared publication - Dmitrii E. Polev
Saint-Petersburg Pasteur Institute, Russia
1 shared publication
Latest Publications
- Open Access
Motivic invariants of moduli stacks of Higgs bundles and bundles with connections: results and speculations
We review some results and techniques from our papers devoted to the computation of motivic classes of stacks of parabolic Higgs budles and bundles with connections on a curve. In the last section we present some directions for future work, as well as some speculations. The latter include a generalization of the P=W conjecture inspired by the work of Maxim Kontsevich and the third author on the Riemann--Hilbert correspondence for complex symplectic manifolds as well as our running project on the motivic classes of the moduli stacks of nilpotent pairs on the formal disk and geometric Satake correspondence for double affine Grassmannians.
Open AccessMotivic invariants of moduli stacks of Higgs bundles and bundles with connections: results and speculations
We review some results and techniques from our papers devoted to the computation of motivic classes of stacks of parabolic Higgs budles and bundles with connections on a curve. In the last section we present some directions for future work, as well as some speculations. The latter include a generalization of the P=W conjecture inspired by the work of Maxim Kontsevich and the third author on the Riemann--Hilbert correspondence for complex symplectic manifolds as well as our running project on the motivic classes of the moduli stacks of nilpotent pairs on the formal disk and geometric Satake correspondence for double affine Grassmannians.
- 24 October 2025
- Published by European Mathematical Society - EMS - Publishing House GmbH in Journal of the European Mathematical Society
Unramified Grothendieck–Serre for isotropic groups
The Grothendieck–Serre conjecture predicts that every generically trivial torsor under a reductive group G over a regular semilocal ring R is trivial. We establish this for unramified R granted that G^{\mathrm{ad}} is totally isotropic, that is, has a “maximally transversal” parabolic R -subgroup. We also use purity for the Brauer group to reduce the conjecture for unramified R to simply connected G —a much less direct such reduction of Panin had been a step in solving the equal characteristic case of Grothendieck–Serre. We base the group-theoretic aspects of our arguments on the geometry of the stack \mathrm{Bun}_{G} , instead of the affine Grassmannian used previously, and we quickly reprove the crucial weak \mathbb{P}^{1} -invariance input: for any reductive group H over a semilocal ring A , every H -torsor \mathscr{E} on \mathbb{P}^{1}_{A} satisfies \mathscr{E}|_{\{t = 0\}}\simeq \mathscr{E}|_{\{t = \infty\}} . For the geometric aspects, we develop reembedding and excision techniques for relative curves with finiteness weakened to quasi-finiteness, thus overcoming a known obstacle in mixed characteristic, and show that every generically trivial torsor over R under a totally isotropic G trivializes over every affine open of \mathrm{Spec}(R) \setminus Z for some closed Z of codimension \ge 2 .
- 24 October 2025
- Published by European Mathematical Society - EMS - Publishing House GmbH in Journal of the European Mathematical Society
Unramified Grothendieck–Serre for isotropic groups
The Grothendieck–Serre conjecture predicts that every generically trivial torsor under a reductive group G over a regular semilocal ring R is trivial. We establish this for unramified R granted that G^{\mathrm{ad}} is totally isotropic, that is, has a “maximally transversal” parabolic R -subgroup. We also use purity for the Brauer group to reduce the conjecture for unramified R to simply connected G —a much less direct such reduction of Panin had been a step in solving the equal characteristic case of Grothendieck–Serre. We base the group-theoretic aspects of our arguments on the geometry of the stack \mathrm{Bun}_{G} , instead of the affine Grassmannian used previously, and we quickly reprove the crucial weak \mathbb{P}^{1} -invariance input: for any reductive group H over a semilocal ring A , every H -torsor \mathscr{E} on \mathbb{P}^{1}_{A} satisfies \mathscr{E}|_{\{t = 0\}}\simeq \mathscr{E}|_{\{t = \infty\}} . For the geometric aspects, we develop reembedding and excision techniques for relative curves with finiteness weakened to quasi-finiteness, thus overcoming a known obstacle in mixed characteristic, and show that every generically trivial torsor over R under a totally isotropic G trivializes over every affine open of \mathrm{Spec}(R) \setminus Z for some closed Z of codimension \ge 2 .
- Open Access
Motivic classes of irregular Higgs bundles and irregular connections on a curve
Let $X$ be a smooth projective curve over a field of characteristic zero and let $\mathcal D$ be an effective divisor on $X$. We calculate motivic classes of various moduli stacks of parabolic vector bundles with irregular connections on $X$ and of irregular parabolic Higgs bundles on $X$ with poles bounded by $\mathcal D$ and with fully or partially fixed formal normal forms. Along the way, we obtain several results about irregular connections and irregular parabolic Higgs bundles. In particular, we give a criterion for the existence of a connection on a higher level parabolic bundle and also develop homological algebra for irregular connections and irregular parabolic Higgs bundles. We also simplify our previous results in the regular case by re-writing the formulas for motivic classes in terms of the HLV generating function.Open AccessMotivic classes of irregular Higgs bundles and irregular connections on a curve
Let $X$ be a smooth projective curve over a field of characteristic zero and let $\mathcal D$ be an effective divisor on $X$. We calculate motivic classes of various moduli stacks of parabolic vector bundles with irregular connections on $X$ and of irregular parabolic Higgs bundles on $X$ with poles bounded by $\mathcal D$ and with fully or partially fixed formal normal forms. Along the way, we obtain several results about irregular connections and irregular parabolic Higgs bundles. In particular, we give a criterion for the existence of a connection on a higher level parabolic bundle and also develop homological algebra for irregular connections and irregular parabolic Higgs bundles. We also simplify our previous results in the regular case by re-writing the formulas for motivic classes in terms of the HLV generating function. - Open Access
Unramified case of Grothendieck-Serre for totally isotropic simply-connected group schemes
We prove a case of the Grothendieck-Serre conjecture: let $R$ be a Noetherian semilocal flat algebra over a Dedekind domain such that all fibers of $R$ are geometrically regular; let $G$ be a simply-connected reductive $R$-group scheme having a strictly proper parabolic subgroup scheme. Then a $G$-torsor is trivial, provided that it is trivial over the total ring of fractions of $R$. We also simplify the proof of the conjecture in the quasisplit unramified case. The argument is based on the notion of a unipotent chain of torsors that we introduce. We also prove that if $R$ is a Noetherian normal domain and $G$ is as above, then for any generically trivial torsor over an open subset $U$ of Spec $R$, there is a closed subset $Z$ of Spec $R$ of codimension at least two such the torsor trivializes over any affine open of $U-Z$.Open AccessUnramified case of Grothendieck-Serre for totally isotropic simply-connected group schemes
We prove a case of the Grothendieck-Serre conjecture: let $R$ be a Noetherian semilocal flat algebra over a Dedekind domain such that all fibers of $R$ are geometrically regular; let $G$ be a simply-connected reductive $R$-group scheme having a strictly proper parabolic subgroup scheme. Then a $G$-torsor is trivial, provided that it is trivial over the total ring of fractions of $R$. We also simplify the proof of the conjecture in the quasisplit unramified case. The argument is based on the notion of a unipotent chain of torsors that we introduce. We also prove that if $R$ is a Noetherian normal domain and $G$ is as above, then for any generically trivial torsor over an open subset $U$ of Spec $R$, there is a closed subset $Z$ of Spec $R$ of codimension at least two such the torsor trivializes over any affine open of $U-Z$. - Open Access
Generically isotropic reductive group schemes are locally isotropic
Let $R$ be a semilocal geometrically factorial Noetherian domain of characteristic zero. We show that a reductive $R$-group scheme is isotropic if it is generically isotropic. We derive various consequences, in particular for the Grothenieck-Serre conjecture and for homotopic invariance of torsors.Open AccessGenerically isotropic reductive group schemes are locally isotropic
Let $R$ be a semilocal geometrically factorial Noetherian domain of characteristic zero. We show that a reductive $R$-group scheme is isotropic if it is generically isotropic. We derive various consequences, in particular for the Grothenieck-Serre conjecture and for homotopic invariance of torsors. - Open Access
On the purity conjecture of Nisnevich for torsors under reductive group schemes
Let $R$ be a regular semilocal integral domain containing an infinite field $k$. Let $f\in R$ be an element such that for all maximal ideals $\mathfrak m$ of $R$ we have $f\notin\mathfrak m^2$. Let $\mathbf G$ be a reductive group scheme over $R$. Under an isotropy assumption on $\mathbf G$ we show that a $\mathbf G$-torsor over the localization $R_f$ is trivial, provided it is rationally trivial. We show that it is not true without the isotropy assumption. Finally, if $\mathbf G$ is a commutative group scheme of multiplicative type and the regular semilocal ring contains a field of characteristic zero, we prove an analogue of Nisnevich purity conjecture for higher \'etale cohomology groups. The first statement is derived from its abstract version concerning presheaves of pointed sets satisfying some properties. The counterexample is constructed by providing a torsor over a local family of affine lines that cannot be extended to the family of projective lines. The latter is accomplished using the technique of affine Grassmannians.Open AccessOn the purity conjecture of Nisnevich for torsors under reductive group schemes
Let $R$ be a regular semilocal integral domain containing an infinite field $k$. Let $f\in R$ be an element such that for all maximal ideals $\mathfrak m$ of $R$ we have $f\notin\mathfrak m^2$. Let $\mathbf G$ be a reductive group scheme over $R$. Under an isotropy assumption on $\mathbf G$ we show that a $\mathbf G$-torsor over the localization $R_f$ is trivial, provided it is rationally trivial. We show that it is not true without the isotropy assumption. Finally, if $\mathbf G$ is a commutative group scheme of multiplicative type and the regular semilocal ring contains a field of characteristic zero, we prove an analogue of Nisnevich purity conjecture for higher \'etale cohomology groups. The first statement is derived from its abstract version concerning presheaves of pointed sets satisfying some properties. The counterexample is constructed by providing a torsor over a local family of affine lines that cannot be extended to the family of projective lines. The latter is accomplished using the technique of affine Grassmannians. - Open Access
Satake equivalence for Hodge modules on affine Grassmannians
For a reductive group $G$ we equip the category of $G_\mathcal{O}$-equivariant polarizable pure Hodge modules on the affine Grassmannian $\mathrm{Gr}_G$ with a structure of neutral Tannakian category. We show that it is equivalent to a twisted tensor product of the category of representations of the Langlands dual group and the category of pure polarizable Hodge structures.Open AccessSatake equivalence for Hodge modules on affine Grassmannians
For a reductive group $G$ we equip the category of $G_\mathcal{O}$-equivariant polarizable pure Hodge modules on the affine Grassmannian $\mathrm{Gr}_G$ with a structure of neutral Tannakian category. We show that it is equivalent to a twisted tensor product of the category of representations of the Langlands dual group and the category of pure polarizable Hodge structures. - 5 November 2021
- Published by American Mathematical Society (AMS) in Transactions of the American Mathematical Society
- Vol. 375 (01) , 559-586
On the Grothendieck–Serre conjecture on principal bundles in mixed characteristic
Let $R$ be a regular local ring. Let $\mathbf {G}$ be a reductive $R$-group scheme. A conjecture of Grothendieck and Serre predicts that a principal ${\mathbf {G}}$-bundle over $R$ is trivial if it is trivial over the quotient field of $R$. The conjecture is known when $R$ contains a field. We prove the conjecture for a large class of regular local rings not containing fields in the case when $\mathbf {G}$ is split.- 5 November 2021
- Published by American Mathematical Society (AMS) in Transactions of the American Mathematical Society
- Vol. 375 (01) , 559-586
On the Grothendieck–Serre conjecture on principal bundles in mixed characteristic
Let $R$ be a regular local ring. Let $\mathbf {G}$ be a reductive $R$-group scheme. A conjecture of Grothendieck and Serre predicts that a principal ${\mathbf {G}}$-bundle over $R$ is trivial if it is trivial over the quotient field of $R$. The conjecture is known when $R$ contains a field. We prove the conjecture for a large class of regular local rings not containing fields in the case when $\mathbf {G}$ is split.- Open Access
On the Grothendieck-Serre Conjecture about principal bundles and its generalizations
Let $U$ be a regular connected affine semi-local scheme over a field $k$. Let $G$ be a reductive group scheme over $U$. Assuming that $G$ has an appropriate parabolic subgroup scheme, we prove the following statement. Given an affine $k$-scheme $W$, a principal $G$-bundle over $W\times_kU$ is trivial if it is trivial over the generic fiber of the projection $W\times_kU\to U$. We also simplify the proof of the Grothendieck-Serre conjecture: let $U$ be a regular connected affine semi-local scheme over a field $k$. Let $G$ be a reductive group scheme over $U$. A principal $G$-bundle over $U$ is trivial if it is trivial over the generic point of $U$. We generalize some other related results from the simple simply-connected case to the case of arbitrary reductive group schemes.Open AccessOn the Grothendieck-Serre Conjecture about principal bundles and its generalizations
Let $U$ be a regular connected affine semi-local scheme over a field $k$. Let $G$ be a reductive group scheme over $U$. Assuming that $G$ has an appropriate parabolic subgroup scheme, we prove the following statement. Given an affine $k$-scheme $W$, a principal $G$-bundle over $W\times_kU$ is trivial if it is trivial over the generic fiber of the projection $W\times_kU\to U$. We also simplify the proof of the Grothendieck-Serre conjecture: let $U$ be a regular connected affine semi-local scheme over a field $k$. Let $G$ be a reductive group scheme over $U$. A principal $G$-bundle over $U$ is trivial if it is trivial over the generic point of $U$. We generalize some other related results from the simple simply-connected case to the case of arbitrary reductive group schemes. - Open Access
Compensation for the absence of the catalytically active half of DNA polymerase ε in yeast by positively selected mutations in CDC28 gene
DNA polymerase ε (pol ε) participates in the leading DNA strand synthesis in eukaryotes. The catalytic subunit of this enzyme, Pol2, is a fusion of two ancestral B-family DNA polymerases. Paradoxically, the catalytically active N-terminal pol is dispensable, and an inactive C-terminal pol is essential for yeast cell viability. Despite extensive studies of strains without the active N-terminal half (mutation pol2-16 ), it is still unclear how they survive and what is the mechanism of rapid recovery of initially miserably growing cells. The reason for the slow progress is in the difficultly of obtaining strains with the defect. We designed a robust method for constructing mutants with only the C-terminal part of Pol2 using allele pol2rc-ΔN with optimized codon usage. Colonies bearing pol2rc-ΔN appear three times sooner than colonies of pol2-16 but exhibit similar growth defects: sensitivity to hydroxyurea, chromosomal instability, and an elevated level of spontaneous mutagenesis. UV-induced mutagenesis is partially affected; it is lower only at high doses in some reporters. The analysis of the genomes of pol2rc-ΔN isolates revealed the prevalence of nonsynonymous mutations suggesting that the growth recovery was a result of positive selection for better growth fueled by variants produced by the elevated mutation rate. Mutations in the CDC28 gene, the primary regulator of the cell cycle, were repeatedly found in independent clones. Genetic analysis established that cdc28 alleles single-handedly improve the growth of pol2rc-ΔN strains and suppress sensitivity hydroxyurea. The affected amino acids are located on the Cdc28 molecule’s two surfaces that mediate contacts with cyclins or kinase subunits. Our work establishes the significance of the CDC28 gene for the resilience of replication and predicts that changes in mammalian homologs of cyclin-dependent kinases may play a role in remastering replication to compensate for the defects in the leading strand synthesis by the dedicated polymerase. Author Summary: The catalytic subunit of the leading strand DNA polymerase ε, Pol2, consists of two halves made of two different ancestral B-family DNA polymerases. Counterintuitively, the catalytically active N-terminal half is dispensable while the inactive C-terminal part is required for viability. The corresponding strains show a severe growth defect, sensitivity to replication inhibitors, chromosomal instability, and elevated spontaneous mutagenesis. Intriguingly, the slow-growing mutant strains rapidly produced fast-growing clones. We discovered that the adaptation to the loss of the catalytic N-terminal part of Pol2 occurs during evolution by positive selection for a better growth fueled by variants produced by elevated mutation rates. Mutations in the cell cycle-dependent kinase gene, CDC28 , can single-handedly improve the growth of strains lacking the N-terminal part of Pol2. Our study predicts that changes in mammalian homologs of cyclin-dependent kinases may play a role in response to the defects of active leading strand polymerase.
Open AccessCompensation for the absence of the catalytically active half of DNA polymerase ε in yeast by positively selected mutations in CDC28 gene
DNA polymerase ε (pol ε) participates in the leading DNA strand synthesis in eukaryotes. The catalytic subunit of this enzyme, Pol2, is a fusion of two ancestral B-family DNA polymerases. Paradoxically, the catalytically active N-terminal pol is dispensable, and an inactive C-terminal pol is essential for yeast cell viability. Despite extensive studies of strains without the active N-terminal half (mutation pol2-16 ), it is still unclear how they survive and what is the mechanism of rapid recovery of initially miserably growing cells. The reason for the slow progress is in the difficultly of obtaining strains with the defect. We designed a robust method for constructing mutants with only the C-terminal part of Pol2 using allele pol2rc-ΔN with optimized codon usage. Colonies bearing pol2rc-ΔN appear three times sooner than colonies of pol2-16 but exhibit similar growth defects: sensitivity to hydroxyurea, chromosomal instability, and an elevated level of spontaneous mutagenesis. UV-induced mutagenesis is partially affected; it is lower only at high doses in some reporters. The analysis of the genomes of pol2rc-ΔN isolates revealed the prevalence of nonsynonymous mutations suggesting that the growth recovery was a result of positive selection for better growth fueled by variants produced by the elevated mutation rate. Mutations in the CDC28 gene, the primary regulator of the cell cycle, were repeatedly found in independent clones. Genetic analysis established that cdc28 alleles single-handedly improve the growth of pol2rc-ΔN strains and suppress sensitivity hydroxyurea. The affected amino acids are located on the Cdc28 molecule’s two surfaces that mediate contacts with cyclins or kinase subunits. Our work establishes the significance of the CDC28 gene for the resilience of replication and predicts that changes in mammalian homologs of cyclin-dependent kinases may play a role in remastering replication to compensate for the defects in the leading strand synthesis by the dedicated polymerase. Author Summary: The catalytic subunit of the leading strand DNA polymerase ε, Pol2, consists of two halves made of two different ancestral B-family DNA polymerases. Counterintuitively, the catalytically active N-terminal half is dispensable while the inactive C-terminal part is required for viability. The corresponding strains show a severe growth defect, sensitivity to replication inhibitors, chromosomal instability, and elevated spontaneous mutagenesis. Intriguingly, the slow-growing mutant strains rapidly produced fast-growing clones. We discovered that the adaptation to the loss of the catalytic N-terminal part of Pol2 occurs during evolution by positive selection for a better growth fueled by variants produced by elevated mutation rates. Mutations in the cell cycle-dependent kinase gene, CDC28 , can single-handedly improve the growth of strains lacking the N-terminal part of Pol2. Our study predicts that changes in mammalian homologs of cyclin-dependent kinases may play a role in response to the defects of active leading strand polymerase.