CN108763777B - Method for establishing VLSI global layout model based on Poisson equation explicit solution - Google Patents
Method for establishing VLSI global layout model based on Poisson equation explicit solution Download PDFInfo
- Publication number
- CN108763777B CN108763777B CN201810543412.8A CN201810543412A CN108763777B CN 108763777 B CN108763777 B CN 108763777B CN 201810543412 A CN201810543412 A CN 201810543412A CN 108763777 B CN108763777 B CN 108763777B
- Authority
- CN
- China
- Prior art keywords
- electric field
- potential
- module
- equation
- layout
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/30—Circuit design
- G06F30/39—Circuit design at the physical level
- G06F30/392—Floor-planning or layout, e.g. partitioning or placement
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2111/00—Details relating to CAD techniques
- G06F2111/10—Numerical modelling
Abstract
The invention relates to a method for establishing a VLSI global layout model based on the explicit solution of Poisson equation, which expresses a circuit as a hypergraph model; simulating a VLSI circuit layout model into a two-dimensional electrostatic system, and converting density constraint into constraint of total potential energy N (v) =0 of the electrostatic system; establishing a partial differential equation set based on the Poisson equation, the boundary condition and the compatibility condition; establishing an analytic expression of a density function, and substituting the analytic expression into a partial differential equation set; determining expressions of electric potential and electric field according to the density function; determining the convergence of the electric potential and electric field expression; obtaining a solving expression of electric potential and electric field according to the partial sum; and obtaining the potential and electric field value of each grid by a rapid calculation method, weighting to obtain the potential and electric field of the module, and completing the VLSI circuit layout model establishment under the action of electric field force. The invention can provide a high-efficiency and practical global layout result, and can meet the requirements of the current VLSI global layout stage particularly for large-scale examples.
Description
Technical Field
The invention relates to the technical field of Very Large Scale Integration (VLSI) physical design automation, in particular to a VLSI global layout model building method based on Poisson's equation explicit solution.
Background
As technology advances into the deep nanometer era of billion transistor integration, the performance of layout tools dominates the overall quality of EDA tools. Therefore, in the near term, many placers have been developed. There are three main layout algorithms: simulated annealing based methods, partitioning based methods and analysis based methods. Recent studies have shown that analytic-type placers are generally able to achieve better placement quality and have good scalability.
In an analysis-based layout, one key technique is to reduce the overlap between modules to achieve a uniformly dispersed layout. For analysis-based layouts, many documents propose methods of reducing overlap, such as partitioning, cell movement, frequency control, partitioning, bell-shaped density control, helmholtz density control, and poisson density control. In these methods, poisson density control is employed by some mainstream placers, such as FDP, kraft werk, mFAR, and ePlace.
In the global layout, ePlace adopts the poisson equation, and in all theoretical placers based on ISPD2005 and ISPD2006 test examples, the optimal line length is achieved. The ePlace first converts each module to a positive charge, converting the module density to a potential energy constraint for disposal. And then modeling through electric potential and electric field generated by given charge density, establishing a Poisson equation, and setting a Norefman boundary condition and a compatibility condition according to the characteristics of the layout. Poisson's equation is a Partial Differential Equation (PDE) that is solved spectrally, ePlace can quickly calculate a potential and electric field distribution, so it can minimize line length while spreading the module quickly.
Poisson's equation is commonly used in many fields such as electrostatics, computer science, mechanical engineering, theoretical physics, astronomy, chemistry, and the like. For example, poisson's equation may use a biquadratic hermitian function to create a system with finite elements in a rectangular domain. In particular, poisson's equation is often used to describe the potential field distribution resulting from a given charge or mass density.
Solutions of the poisson equation fall into two categories: analytical solutions and numerical solutions. The analytical solution is an accurate solution and is also an explicit solution of partial differential equations. The numerical solution is obtained by numerical methods such as finite element method, numerical approximation method, interpolation method, etc. Since the numerical solution can only approximate the solution of the PDE, the numerical solution inevitably produces some numerical errors. Taking the half-space problem of characteristic strain as an example, people do a lot of work to minimize numerical errors. Generally, if an explicit solution for a PDE can be found, it is inherently superior to a numerical solution.
In the existing placers based on poisson density control, a numerical solution is used for solving poisson equations. In Kraft werk, a geometric multi-grid solver DiMePACK is used to solve the Poisson equation; in ePlace, a Fast Fourier Transform (FFT) is utilized to calculate the potential and electric field distribution. However, the poisson equation for different boundary conditions has different characteristics, and thus its solution is also different. It is generally considered that obtaining an explicit solution to a PDE is very challenging, if not impossible. Such as the lorentz force expression, is suitable and accurate in finite element simulations, but is not suitable for calculations.
Disclosure of Invention
The invention aims to provide a method for establishing a VLSI global layout model based on the Poisson equation explicit solution, so as to overcome the defects in the prior art.
In order to achieve the purpose, the technical scheme of the invention is as follows: a VLSI global layout model establishing method based on Poisson equation explicit solution comprises the following steps:
step S1: representing the circuit as a hypergraph model H = { V, E };
step S2: simulating a VLSI circuit layout model into a two-dimensional electrostatic system, and converting density constraint into constraint of total potential energy N (v) =0 of the electrostatic system;
and step S3: establishing a partial differential equation set based on the Poisson equation, the boundary condition and the compatibility condition;
and step S4: establishing an analytical formula of a density function, and substituting the analytical formula into a partial differential equation set;
step S5: determining expressions of electric potential and electric field according to the density function;
step S6: determining the convergence of the electric potential and electric field expression;
step S7: obtaining a solving expression of electric potential and electric field according to the partial sum;
step S8: and obtaining the potential and electric field value of each grid by a rapid calculation method, weighting to obtain the potential and electric field of the module, and completing the VLSI circuit layout model establishment under the action of electric field force.
Compared with the prior art, the invention has the following beneficial effects:
(1) The distribution of the modules is simulated by using an accurate density function, and numerical errors cannot be generated;
(2) Directly solving the Poisson equation on the basis of the accurate density function to obtain an explicit solution, and proving the convergence of the solution; because the analytical solution is adopted, no numerical error is generated when the electrostatic system is solved;
(3) The method not only realizes the optimization of the line length, but also ensures the solving speed. In comparison with two mainstream placers ePlace and NTUplace3, the results of using ISPD2005 and ISPD2006 test sets show that the algorithm of the invention can make the line length smaller. The solution of the fast Poisson equation of the software is used for replacing the density control, and the experimental result of embedding into NTUplace3 shows that the algorithm can reduce the wire length by 11 percent and greatly improve the result.
Drawings
FIG. 1 is a flow chart of a VLSI global layout model building method based on the explicit solution of Poisson's equation in the present invention.
Detailed Description
The technical scheme of the invention is specifically explained in the following by combining the attached drawings.
The invention relates to a method for establishing a VLSI global layout model based on Poisson equation explicit solution, which comprises the following steps as shown in figure 1:
(1) Representing the circuit as a hypergraph H = { V, E };
(2) Simulating the layout problem into a two-dimensional electrostatic system, and converting the density constraint into the constraint of the total potential energy N (v) =0 of the electrostatic system;
(3) Establishing a partial differential equation set by utilizing a Poisson equation, a boundary condition and a compatibility condition;
(4) Giving an analytic expression of the density function, and substituting the analytic expression into a partial differential equation set;
(5) Determining accurate expressions of the electric potential and the electric field by using a density function;
(6) The convergence of the potential and electric field expressions is proved;
(7) And obtaining a solving expression of the electric potential and the electric field by using the partial sum.
(8) And obtaining the electric potential and the electric field value of each grid by a quick calculation method, weighting to obtain the electric potential and the electric field of the module, and finishing the layout under the action of the electric field force.
Further, in step (1), the given layout area is [0, W ]]×[0,H]The VLSI circuit layout problem can be modeled as a hypergraph G (V, E), with the module represented as a set of vertices V = { V = } 1 ,v 2 ,...,v n Represents the net as a super-edge set E = { E = } 1 ,e 2 ,...,e r H, module v i Is w and is high i And h i The coordinate of the center point is (x) i ,y i ). Wherein i =1, 2, ·, n. The VLSI layout problem is to determine the optimal position of each module without overlap between modules, and the bus length is optimal:
there is no overlap between min W (v) s.t. modules (1)
Wherein W (v) is the total line length, calculated by the half perimeter line length (HPWL).
Further, in step S2, the layout problem is simulated as a two-dimensional electrostatic system. Depending on the position of the module in the layout area, the electric potential φ (x, y) and the electric field ξ (x, y) can be determined, where ξ (x, y) = (ξ) x ,ξ y ) = phi (x, y). Module i is regarded as a positive charge i, area A i Is expressed as a charge amount q i . By phi i =φ(x i ,y i ) And xi i =ξ(x i ,y i ) Representing the potential and electric field at charge i. The charge i is then dependent on the electric force F i =q i ξ i The movement is performed. Thus, the system potential energy may be defined asWherein N is i =q i φ i Representing the potential energy of charge i. And finally, converting the density constraint into a constraint of total potential energy N (v) =0 of the electrostatic system.
Further, in step (3), a partial differential equation set is established using poisson's equation, boundary conditions and compatibility conditions. Based on Gauss law, a Poisson equation of the electrostatic system is constructed, and a partial differential equation is obtained while satisfying layout boundary conditions and compatibility conditions:
where equation (2 a) gives the poisson equation, where ρ (x, y) is a density function; equation (2 b) is a Nooedeman boundary condition for avoiding a module running out of the boundary of the layout, where R andrespectively representing the boundary and the outer normal vector of the layout region R; equation (2 c) is a compatibility condition, giving the system of equations a unique solution.
Further, in step (4), the density function of the module i in the x direction is defined as:
to be compared with vector (x, y) = (x) 1 ,y 1 ,...x i ,y i ) Making a distinction byRepresenting a continuous variable. Likewise, the density function of module i in the y-direction can be defined asThe density function of module i isThe total density of all modules over the layout area is:
where n is the number of modules.
Further, in fig. 1, the "analytic solution" part is solved in the following specific manner:
Using the exact density function, poisson's equation, the relevant boundary conditions and compatibility conditions (2 a) - (2 c) become:
further, in step (5), by equations (6 a) and (6 b), their solutions can be obtained as follows:
wherein, a u,p Denotes the coefficient of each wave function, and u and p denote integer indices. To calculate the coefficient a u,p Substituting equation (7) into Poisson equation (6 a), and calculatingThe density function can be obtainedAnother expression form of (a):
multiplication on both the left and right sides of equation (8) simultaneouslyAnd integrated to obtain
The integration region of equation (9) is R = (0, w) × (0, h). Thus, to the right of equation (9), the first term takes on non-zero values only at μ =0, p = η, the second term only at η =0 and u = μ, and the third term only at p = η and u = μ, according to the orthogonality of the trigonometric functions. Thus at μ ≧ 1 and η =0, equation (9) reduces to:
it is thus possible to obtain:
also, the coefficient a can be obtained 0,η And a μ,η In order to satisfy equation (6 c), let a be in equation (7) 0,0 =0, then:
when μ, η ≧ 1, the formula (5) is taken into the formula (12 d) to obtain:
the following are also available:
and
note that in VLSI layout problem, a u,p Is calculated from the integral of the exact density function (5) and is more accurate than the discrete calculation in ePlace.
Further, in fig. 1, the "potential gradient" part is solved in the following manner:
known from Gauss's law, the electric fieldEqual to negative gradient of electric potentialOf formula (7)Is provided with
Further, in step (6), because of equation (9)Is an infinite series, and is thus to be provenIs convergent.
And (5) leading to 1.
Theorem 1.
And (3) proving that:
note that:
for a u,p U, p ≧ 1, by equation (13):
for the other two cases u =0, p ≧ 1 and u ≧ 1, p =0, there are:
and
thus, it is possible to provide
By theory 1, the above three infinite series are convergent, soThere is an upper bound of convergence, thereforeIs absolutely convergent.
Further, in step (7), according to theorem 1, in practice, the measurement is carried outOnly need to calculate in the calculationThe sum of (1) and (b). In addition to this, the present invention is,is thatThe negative gradient of (2) can also be usedIs approximated by a partial sum of (c). Because u is contained in equation (17) 3 Or p 3 Thus, therefore, it isThe convergence is fast and a more accurate solution can be obtained by iterating the computation partial sums only K times.
Therefore, the layout area can be divided into m × m grids of the same specification, and each grid is written as: b lj Where l =0, 1, · m-1, j =0, 1, · m-1, and m-1 denote the labels of the grids, then grid b lj The density of (a) is then:
module i in grid b lj Is determined by the size of the module, and the distance from the center point of module i and grid b lj Is inversely proportional, similar to the local smoothing and density scaling technique in ePlace, and the following density function can be obtained
WhereinRepresentation grid b lj Is used as the density function. Order (x) l ,y j ) Representation grid b lj Then the density function of all the grids approximated by equation (5) can be obtained as:
at the determined a u,p By substituting formula (18) into formulae (12 b), (12 c) and (12 d) at u =0 and p>In the case of = 1:
wherein x is l And y j Determined by the size and layout area of the grid, will
the remaining coefficients can also be obtained in the same manner. All coefficients a found u,p The following were used:
is determined byInfinite number of stages in theory 1Is absolutely convergent, so the following approximation can be made:
further, in step (8), the density of each mesh is calculated according to equation (18)In equations (21 a) - (21 b), a coefficient matrix a 'of m × m can be calculated by ignoring the summation of coefficients' u,p :
The coefficient matrix can be calculated by calling an FFT library once, and m is spent 2 Time to coefficient matrix a' u,p Is updated to a u,p Calculate all coefficients a u,p . After all the coefficients are calculated, phi (l, j) and xi (l, j) can be calculated by inverse fast Fourier transform, and the electric force F i =q i ξ i The module i is moved to complete the layout.
Further, in fig. 1, the "line length gradient" part is solved in the following specific manner:
in problem (1), W (v) is inconsequential, and direct optimization is difficult, so LSE line length is used to approximate HPWL. The LSE line length function in the x direction is:
wherein gamma is a smooth parameter, and the line length gradient can be obtained by solving the gradient of the function.
Further, the "module location optimization and parameter update" part in fig. 1 is implemented in the following specific manner:
in each iteration, a Nesterov method is adopted to solve the unconstrained minimization problem, and a new solution (x) is obtained through one iteration k+1 ,y k+1 ) The solution is the optimized module position, and then the penalty parameter λ is updated.
The above are preferred embodiments of the present invention, and all changes made according to the technical solutions of the present invention that produce functional effects do not exceed the scope of the technical solutions of the present invention belong to the protection scope of the present invention.
Claims (3)
1. A VLSI global layout model building method based on Poisson equation explicit solution is characterized by comprising the following steps:
step S1: representing the circuit as a hypergraph model G = { V, E }, wherein V is a vertex set, E is a hyperedge set, and a layout area is set to be [0, W ] × [0, H ];
step S2: simulating a VLSI circuit layout model into a two-dimensional electrostatic system, and converting density constraint into constraint of total potential energy N (v) =0 of the electrostatic system;
and step S3: establishing a partial differential equation set based on the Poisson equation, the boundary condition and the compatibility condition;
in step S3, based on gaussian law, a poisson equation of the electrostatic system is established, and partial differential equations are obtained while satisfying the layout boundary condition and the compatibility condition, where the partial differential equations are:
poisson equation, where ρ (x, y) is a density function:
where φ (x, y) represents the potential at charge;
noelman boundary condition for avoiding a module running out of a boundary of a layout, where R andthe boundary and the outer normal vector of the layout region R are respectively represented:
compatibility conditions, make the system of equations have a unique solution:
and step S4: establishing an analytic expression of a density function, and substituting the analytic expression into a partial differential equation set;
in step S4, the density function of the module i in the x direction is expressed as:
to be compared with vector (x, y) = (x) 1 ,y 1 ,...x i ,y i ) To distinguish byRepresents a continuous variable; define the density function of module i in the y-direction asThe density function of module i isThe total density of all modules over the layout area is:
wherein n is the number of modules;
redefining ρ (x, y):
wherein w i And h i Are respectively a module v i Width and height of (d);
from the density function, the poisson equation, boundary conditions and compatibility conditions become:
step S5: determining expressions of electric potential and electric field according to the density function;
step S6: determining the convergence of the electric potential and electric field expression;
step S7: obtaining a solving expression of electric potential and electric field according to the partial sum;
step S8: obtaining the potential and electric field value of each grid by a rapid calculation method, weighting to obtain the potential and electric field of the module, and completing the establishment of a VLSI circuit layout model under the action of electric field force;
in step S5, according to the poisson equation and the boundary condition obtained in step S4, the following solution is obtained:
wherein, a u,p Coefficients representing each wave function, u and p representing integer indices; to calculate the coefficient a u,p Substituting the solution into the Poisson equation obtained in step S4, and calculatingObtaining a density functionAnother expression of (a):
wherein the content of the first and second substances,
according to Gauss's law, electric fieldEqual to electric potentialA negative gradient of (d); from the above solutionIs provided with
In the step S7, the layout area is divided into m × m grids of the same specification, and each grid is denoted as b lj Wherein l =0, 1, · m-1, j =0, 1, · m-1, and m-1 represent the labels of the grids, and let
The potential is then:
the electric field is:
wherein:
in the step S8, a coefficient matrix a 'of m × m is calculated' u,p :
Calculating coefficient matrix by calling FFT library once and spending m 2 Time coefficient matrix a' u,p Is updated to a u,p Calculating all the coefficients a u,p (ii) a After all the coefficients are calculated, phi (l, j) and xi (l, j) are calculated by inverse fast Fourier transform, and the electric field force F i =q i ξ i And (5) moving the module i to complete the VLSI circuit layout model establishment.
2. The method for modeling a global layout of VLSI based on explicit poisson' S equation solution as claimed in claim 1, wherein in step S1, the layout area is [0,w ]]×[0,H]Given n modules and r nets, the VLSI circuit layout model is treated as a hypergraph G (V, E), and the modules are represented as a set of vertices V = { V = 1 ,v 2 ,...,v n Denoted as super-edge set E = { E 1 ,e 2 ,...,e r H, module v i Is w and is high i And h i The coordinate of the center point is (x) i ,y i ) I =1, 2, ·, n; the VLSI circuit layout model determines the optimal position of each module without overlap between modules and the bus length is optimal:
min W(v)
there is no overlap between s.t. modules
Wherein, W (v) is the total line length and is obtained by calculating the half perimeter line length.
3. A VLSI global layout model building method based on poisson' S equation explicit solution according to claim 2, characterized in that in step S2, according to the position of the module in the layout area, the electric potential Φ (x, y) and the electric field ξ (x, y) are determined, where ξ (x, y) = (ξ) x ,ξ y ) = - Δ φ (x, y); with module i as a positive charge i, area A i As quantity of electric charge q i By phi i =φ(x i ,y i ) And xi i =ξ(x i ,y i ) Are respectively shown inElectric potential and field at charge i, charge i being dependent on electric field force F i =q i ξ i Move with system potential energy ofWherein N is i =q i φ i Represents the potential energy of charge i; the density constraint is translated into a constraint of the total potential energy N (v) =0 of the electrostatic system.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810543412.8A CN108763777B (en) | 2018-05-30 | 2018-05-30 | Method for establishing VLSI global layout model based on Poisson equation explicit solution |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810543412.8A CN108763777B (en) | 2018-05-30 | 2018-05-30 | Method for establishing VLSI global layout model based on Poisson equation explicit solution |
Publications (2)
Publication Number | Publication Date |
---|---|
CN108763777A CN108763777A (en) | 2018-11-06 |
CN108763777B true CN108763777B (en) | 2023-02-28 |
Family
ID=64004604
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810543412.8A Active CN108763777B (en) | 2018-05-30 | 2018-05-30 | Method for establishing VLSI global layout model based on Poisson equation explicit solution |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108763777B (en) |
Families Citing this family (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113139361A (en) * | 2020-01-19 | 2021-07-20 | 上海复旦微电子集团股份有限公司 | Global layout method for 2.5D packaged FPGA |
CN111539167B (en) * | 2020-04-23 | 2023-02-21 | 福州立芯科技有限公司 | Layout method of ultra-large-scale integrated circuit considering atomization and proximity effect |
CN114722670B (en) * | 2022-04-02 | 2023-08-04 | 电子科技大学 | Potential distribution finite difference solving method for two-dimensional electrostatic particle model |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101872377A (en) * | 2010-06-12 | 2010-10-27 | 清华大学 | Method for restraining integrated circuit electricity supply network noise by using decoupling capacitance |
CN102323960A (en) * | 2011-04-19 | 2012-01-18 | 清华大学 | Layout module distribution density smoothing method considering degree of overlapping and wire length |
CN106980730A (en) * | 2017-03-31 | 2017-07-25 | 福州大学 | VLSI standard cell placement methods based on direct solution technology |
CN107526860A (en) * | 2017-03-31 | 2017-12-29 | 福州大学 | VLSI standard cell placement methods based on electric field energy modeling technique |
CN107563095A (en) * | 2017-09-22 | 2018-01-09 | 中国矿业大学(北京) | A kind of non-linear layout method of large scale integrated circuit |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2014133867A1 (en) * | 2013-02-27 | 2014-09-04 | The Regents Of The University Of California | Ic layout adjustment method and tool for improving dielectric reliability at interconnects |
-
2018
- 2018-05-30 CN CN201810543412.8A patent/CN108763777B/en active Active
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101872377A (en) * | 2010-06-12 | 2010-10-27 | 清华大学 | Method for restraining integrated circuit electricity supply network noise by using decoupling capacitance |
CN102323960A (en) * | 2011-04-19 | 2012-01-18 | 清华大学 | Layout module distribution density smoothing method considering degree of overlapping and wire length |
CN106980730A (en) * | 2017-03-31 | 2017-07-25 | 福州大学 | VLSI standard cell placement methods based on direct solution technology |
CN107526860A (en) * | 2017-03-31 | 2017-12-29 | 福州大学 | VLSI standard cell placement methods based on electric field energy modeling technique |
CN107563095A (en) * | 2017-09-22 | 2018-01-09 | 中国矿业大学(北京) | A kind of non-linear layout method of large scale integrated circuit |
Non-Patent Citations (4)
Title |
---|
An Analytical Placer for VLSI Standard Cell Placement;Jianli Chen 等;<IEEE TRANSACTIONS ON COMPUTER-AIDED DESIGN OF INTEGRATED CIRCUITS AND SYSTEMS>;20120712;第31卷(第8期);第1208-1211页 * |
ePlace: Electrostatics Based Placement Using Nesterov"s Method;Jingwei Lu 等;《2014 51st ACM/EDAC/IEEE Design Automation Conference》;20140821;第1-6页 * |
VEAP:基于全局优化的有效VLSI布局算法;孔天明 等;《半导体学报》;19970908;第18卷(第9期);第692-700页 * |
超大规模集成电路的布局算法研究;梁多勇;《中国优秀博硕士学位论文全文数据库(硕士) 信息科技辑》;20120315(第3期);第I135-162页 * |
Also Published As
Publication number | Publication date |
---|---|
CN108763777A (en) | 2018-11-06 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Zhang et al. | A hybrid boundary node method | |
Liu et al. | Model-reduced variational fluid simulation | |
CN108763777B (en) | Method for establishing VLSI global layout model based on Poisson equation explicit solution | |
Fabritius et al. | Improving the quality of finite volume meshes through genetic optimisation | |
CN113158527B (en) | Method for calculating frequency domain electromagnetic field based on implicit FVFD | |
Zhang et al. | Meshless regular hybrid boundary node method | |
Kelly et al. | Numerical solution of the two-phase incompressible Navier–Stokes equations using a GPU-accelerated meshless method | |
CN109726465A (en) | The three-dimensional method for numerical simulation streamed without viscous low speed based on non-structural curl grid | |
Yin et al. | Isogeometric bi-directional evolutionary structural optimization | |
Barral et al. | Anisotropic mesh adaptation in Firedrake with PETSc DMPlex | |
CN115577604A (en) | Simulation method and system for quickly constructing overall matrix and related equipment | |
Wang et al. | A projective transformation-based topology optimization using moving morphable components | |
Ghiu et al. | Discrete diagonal recurrences and discrete minimal submanifolds | |
Brański et al. | Galerkin Versions of Nonsingular Trefftz Method Derived from Variational Formulations for 2D Laplace Problem | |
Xu et al. | Parallelizing a high-order CFD software for 3D, multi-block, structural grids on the TianHe-1A supercomputer | |
Poursalehi et al. | An adaptive mesh refinement approach for average current nodal expansion method in 2-D rectangular geometry | |
Safinejad et al. | A local meshless RBF method for solving fractional integro-differential equations with optimal shape parameters | |
Galbraith et al. | A Discontinuous Galerkin Chimera scheme with implicit artificial boundaries | |
Spekreijse et al. | Semi-automatic domain decomposition based on potential theory | |
Wen et al. | Space–Time Conservation Element and Solution Element Method: Advances and Applications in Engineering Sciences | |
Habibi et al. | Extending a new two-grid waveform relaxation on a spatial finite element discretization | |
Gong | Immersed-interface finite-element methods for elliptic and elasticity interface problems | |
Yang et al. | A fast iterated orthogonal projection framework for smoke simulation | |
Bouchon et al. | An immersed method based on cut-cells for the simulation of 2d incompressible fluid flows past solid structures | |
Cui et al. | The asymptotic concentration approach combined with isogeometric analysis for topology optimization of two-dimensional linear elasticity structures |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |