The experience from LEP and HERA experiments suggests that the software will be as important as hardware, and some 1000 person years are needed for CMS software development. There are a number of different aspects of software activity, when the impact of RDMS participants may be essential.

10.1. Detector performance studies and CMS physics program development.

During last 3 years we made essential contribution to the different physics channels study from the point of view of their observability at the CMS detector. The range of our interests is wide: light Standard Model (SM) Higgs, very heavy SM Higgs, strongly interacting symmetry breaking models, SUSY Higgses, SUSY particles, B-physics, heavy ions, etc. We plan to continue this activity, involving theoretical groups for search of new interesting physical channels and further Monte Carlo generators development. Next years, following the general collaboration trends, the physics performance studies will include the detailed instrumental simulations of the CMS detector response.

10.2. Detector optimization study.

Here our interests are mainly connected with the corresponding hardware interests: electromagnetic (ECAL), hadronic (HCAL) and very forward (VF) calorimeters; preshower detector; forward muon detectors. These studies imply a number of software activities levels: GEANT simulations, software support of different detector subsystems R&D, physics simulations.

10.3. Physics tools.

The development of optimal algorithms for the different objects recognition by means of their detector response is extremely important for the entire experiment to be successful. This task may be formalized as a set of very complicated pattern recognition problems demanding a use of new modern mathematical approaches for the solution. At the moment, only a few physicists take part in track finding, B - tagging, - tagging, jet reconstruction algorithms development. We plan to involve much more manpower here.

10.4. General software development.

At present the major activity concentrates on the development of GEANTbased simulation package CMSIM / CMANA which will provide a self consistent simulation model of the experimental data flow. This package will be the base for future physics performance and detector optimization studies and development and verification of event reconstruction software and algorithms. We participate in the development of the GEANT description of HCAL and VF, fast Monte Carlo simulations of hadronic detector response, jet finding and track finding software.

10.5 Special software development.

Simulation package MARS'94 is designed for the study of the wide scope of problems related to the simulation of particles transport through matter, nuclearelectromagnetic cascades energy deposition and absorbed dose accumulation, radiation background generation, muon convoy simulation etc. MARS'94 is an inclusive program. It is fast and supports a user-friendly powerful geometrical package. MARS'94 will be used together with FLUKA code for the CMS simulation problems, like the accelerator generated background, radiation shielding design, environmental problems related to the collider detector operation, induced radioactivity calculations, radiation hardness studies.