Parallel calculation method

A technology of parallel computing and computing methods, applied in computing, organic chemistry methods, computational theoretical chemistry, etc., can solve problems such as shortening the integration calculation time

CN1714349AInactive Publication Date: 2005-12-28JAPAN SCI & TECH CORP +1
0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Publication Date
2005-12-28
Estimated Expiration
Not applicable · inactive patent

Smart Images

  • Figure 1
    Figure 1
  • Figure 2
    Figure 2
  • Figure 3
    Figure 3
Patent Text Reader

Abstract

A parallel calculation method is a calculation method using a parallel computer with multiple processors. When converting from the atomic orbital base layer (rs|tu) of the two-electron integration to the molecular orbital base layer (ab|cd), for the above multiple processors, designate the labels r and s of the atomic orbitals to be calculated, and on each of the above-mentioned multiple processors, complete the specified labels R and S, and the labels t and u of the atomic orbitals to be calculated combined to process. As a result, the granularity is uniform, multiple conventional processors can be connected while maintaining high performance, the cost of high-speed computing computers can be reduced, and cheap local memory can be utilized, so a large main storage area can be secured for the computer as a whole.
Need to check novelty before this filing date? Find Prior Art

Description

technical field

[0001] The present invention relates to a kind of comprehensive self-consistent field theory (Multi configuration Self-Consistent Field: MSCF) method and configuration interaction (Configuration Interaction: CI) method in the molecular orbital method that has no experience. Parallel Computational Methods for Energy and Energy Gradients. Background technique

[0002] Among non-empirical molecular orbital calculation methods, the above-mentioned MCSCF method and CI method are known as representative calculation methods that can use electron correlation. In the MSCCF method, the differential of the total energy E and the coordinates of the nucleus, that is, the energy gradient (force acting on the nucleus) is obtained as in Equations (1) and (2).

[0003] E = Σ ab MO γ ab h ab + 1 2 ...

Examples

Embodiment Construction

[0056] Embodiments of the present invention will be described below.

[0057] The inventor of the present application believes that: because the two-electron integral of the atomic orbital base layer and the differential of the atomic nucleus coordinates are independent of each other, in the transformation from the atomic orbital base layer to the molecular orbital base layer, the labels r and s of the atomic orbital are parallelized, that is , for the above-mentioned plurality of processors, designate the labels r and s of the atomic orbitals to be calculated, and process the data separately to make the granularity uniform by processing all combinations of t and u by each processor. Although the marks r and s are parallelized, since all combinations of t and u are calculated by each processor, an average of 10,000 floating decimal point operations are required for each two-electron integral, so each processor needs to perform 10000N 2 Times of operations can ensure a suffici...